Jouniafof Range 9&Nagment

Journal of Range Management, Volume 55, Number 4 (July 2002)

Item Type Journal; text

DOI 10.2458/azu_jrm_v55i4_board

Publisher Society for Range Management

Journal Journal of Range Management

Download date 08/10/2021 06:50:44

Item License http://rightsstatements.org/vocab/InC/1.0/

Version Final published version

Link to Item http://hdl.handle.net/10150/650593 Jouniafof Range 9&nagment

TABLE OF CONTENTS: VOL. 55, NO. 4, July 2002 FEATURE ARTICLE 318 Shrub control and streamflow on rangelands: A process based viewpoint by Bradford P. Wilcox Animal Ecology 327 Early weaning and length of supplementation effects on beef calves by A.J. Pordomingo 336 Density and reproductive success of Florida grasshopper sparrows following fire by Michael F. Delany, Stephen B. Linda, Bill Pranty, and Dustin W. Perkins

341 Large ungulate habitat preference in Chobe National Park, Botswana by Uyapo J. Omphile and Jeff Powell 350 Plains larkspur (Delphinium geyeri) grazing by cattle in Wyoming by James A. Pfister, Dale R. Gardner, Bryan L. Stegelmeier, Anthony P. Knight, James W. Waggoner, Jr., and Jeffery 0. Hall Grazing Management 360 Elk and cattle forage use under a specialized grazing system by Lacey E. Halstead, Larry D. Howery, George B. Ruyle, Paul R. Krausman, and Robert J. Steidl Hydrology 367 Sediment movement and filtration in a riparian meadow following cattle use by R.R. McEldowney, M. Flenniken, G.W. Frasier, M.J. Trlica, and W.C. Leininger Measurement/Sampling 374 Calibrating fecal NIRS equations for predicting botanical composition of diets by John W. Walker, Scott D. McCoy, Karen L. Launchbaugh, Merrita J. Fraker, and Jeff Powell 383 Short-term monitoring of rangeland forage conditions with AVHRR imagery by David P. Thoma, Derek W. Bailey, Daniel S. Long, Gerald A. Nielsen, Marl P. Henry, Meagan C. Breneman, and Clifford Montagne Plant-Animal 390 Prior feeding practices do not influence locoweed consumption by M.H. Ralphs, G. Greathouse, A.P. Knight, D. Doherty, J.D. Graham, B.L. Stegelmeier, and L.F. James Published bimonthly-January, March, May, July, 394 Clipping and precipitation influences on locoweed vigor, mortality and toxicity September, November by M.H. Ralphs, D.R. Gardner, J.D. Graham, G. Greathouse, and A.P. Knight Copyright 2002 by the Society for Range Plant Ecology Management 400 Long-term impacts of livestock grazing on Chihuahuan Desert rangelands by INDIVIDUAL SUBSCRIPTION is by membership in Joseph M. Navarro, Dee Gait, Jerry Holechek, Jim McCormick, and Francisco the Society for Range Management. Molinar LIBRARY or other INSTITUTIONAL SUBSCRIP- cycles broom snakeweed in the Colorado Plateau and Snake TIONS on a calendar year basis are $95.00 for the 406 Population of United States postpaid and $112.00 for other coun- River Plains by M.H. Ralphs and K.D. Sanders tries, postpaid. Payment from outside the United States should be remitted in US dollars by interna- 412 Monitoring a half-century of change in a hardwood rangeland by Kerry L. Heise tional money order or draft on a New York bank. and Adina M. Merenlender BUSINESS CORRESPONDENCE, concerning 420 Fingerprint composition of seedling root exudates of selected grasses by Johan back issues, and subscriptions, advertising, reprints, F. Dormaar, Bonnie C. Tovell, and Walter D. Willms related matters, should be addressed to the Managing Editor, 445 Union Blvd., Suite 230, Book Reviews Lakewood, Colorado 80228. 424 Restoration Ecology and Sustainable Development. Edited by Krystyna M. Urbanska, EDITORIAL CORRESPONDENCE, concerning J. Edward. and The Political Economy of Agricultural, manuscripts or other editorial matters, should be Nigel R. Webb, and Peter addressed to the Editor, Gary Frasier, 7820 Stag Natural Resources, and Environmental Policy Analysis. By Wesley F. Peterson. Hollow Road, Loveland, Colorado 80538. Page proofs should be returned to the Production Editor, 3059A Hwy 92, Hotchkiss, CO 81419-9548..

INSTRUCTIONS FOR AUTHORS appear on the inside back cover of most issues. THE JOURNAL OF RANGE MANAGEMENT (ISSN 0022-409X) is published bimonthly for $56.00 per year by the Society for Range Management, 445 Union Blvd., Ste 230, Lakewood, Colorado 80228. PERIODICALS POSTAGE paid at Denver, Colorado and additional offices. POSTMASTER; Return entire journal with address change-Address Service Requested to Society for Range Management, 445 Union Blvd., Suite 230, Lakewood, Colorado 80228.

PRINTED IN USA Editor-In-Chief Associate Editors ELAINE E. GRINGS MICHAEL H. RALPHS SAM ALBRECHT YUGUANG BAI USDA-ARS USDA-ARS 445 Union Blvd., Ste 230 Dept. Plant Sciences Fort Keogh-LARRL Poisonous Plant Lab Lakewood, Colorado 80228 University of Saskatchewan 243 Fort Keogh Rd. 1150 E 1400 N (303) 986-3309 51 Campus Drive Miles City, Montana 59301 Logan, Utah 84341-2881 Fax: (303) 986-3892 Saskatoon, Saskatchewan e-mail address: S7N 5A8 MARSHALL HAFERKAMP LARRY REDMON [email protected] CANADA USDA-ARS TAREC Editor/Copy Editor Fort Keogh-LARRL P.O. Drawer E Texas 74684 GARY FRASIER/JO FRASIER KLAAS BROERSMA 243 Fort Keogh Rd. Overton, 7820 Stag Hollow Road Agriculture and Agri-Food Canada Miles City, Montana 59301 NEIL RIMBEY Loveland, Colorado 80538 Rd. 3015 Ord University of Idaho e-mail address: British Columbia ROBERT A. MASTERS Kamloops, 16952 S. 10th Ave. [email protected] Dow AgroSciences LLC V2B 8A9 CANADA Caldwell, Idaho 83607 Production Editor 3618 South 75th Street Lincoln, Nebraska 68506 PATTY RICH JOE E. BRUMMER CAROLYN HULL SIEG Society for Range Management Mt. Meadows Res. Ctr. S.W. Science Complex 3059A Hwy 92 P.O. Box 598 MITCHEL McCLARAN 2500 South Pine Knoll Hotchkiss, Colorado 81419-9548 Gunnison, Colorado 81230 University of Arizona Flagstaff, Arizona 86001 e-mail address: 301 Biological Science East [email protected] DAVID ELDRIDGE Tucson, Arizona 85721-0001 STEVE WARREN Book Review Editor Dept. Land and Water Conserv. Colorado State University M. NAETH DAVID L. SCARNECCHIA School of Geography ANNE Center for Environmental Dept of Natural Res. Sci. University of New South Wales University of Alberta Management of Military Land Washington State University Sydney, New South Wales Dept. Renewable Resources Dept. of Forestry Sciences Bldg. Pullman, Washington 99164-6410 2052 AUSTRALIA 751 General Services Fort Collins, Colorado 80523 e-mail address: Edmonton, Alberta scarneda@ mail.wsu.edu ROBERT GILLEN T6G 2H1 CANADA Electronic JRM Editor USDA-ARS ROBERT PEARCE M. KEITH OWENS Southern Plains Range Res. Sta. 5028 Highway 6 Texas A&M University 200018th Street Research Center Woodward, Oklahoma 73801 Bishop, California 93514 1619 Garner Field Road Ulvade, Texas e-mail address: [email protected] THE SOCIETY FOR RANGE MANAGEMENT, founded in 1948 as the American Society of Range Management, is a nonprofit association incorporated under the laws of the State of Wyoming. It is recognized exempt from Federal income tax, as a scientific and educa- tional organization, under the provisions of Section 501(c)(3) of the Internal Revenue Code, and also is classed as a public foundation as described in Section 509(a)(2) of the Code. The name of the Society was changed in 1971 by amendment of the Articles of Incorporation. The objectives for which the corporation is established are: President RODNEY K. HEITSCHMIDT -to properly take care of the basic rangeland resources of soil, plants, and water; USDA-ARS Ft. Keogh LARRL -to develop an understanding of range ecosystems and of the principles applicable to the Rt 1, Box 2021 management of range resources; Miles City, Montana 59301-9801 1st Vice-President -to assist all who work with range resources to keep abreast of newfindings and BOB BUDD techniques in the science and art of range management; Red Canyon Ranch 350 Red Canyon Rd -to improve the effectiveness of range management to obtain from range resources the Lander, Wyoming 82520-9417 products and values necessary for man's welfare; 2nd Vice-President -to create a public appreciation of the economic and social benefits to be obtainedfrom MORT KOTHMANN the range environment; Texas A&M University Dept. Rangeland Ecology & Mgt. -to promote professional development of its members. College Station, Texas 77843-0001 Executive Vice-President Membership in the Society for Range Management is open to anyone engaged in or interested in SAM ALBRECHT any aspect of the study, management, or use of rangelands. Please contact the Executive Vice- 445 Union Blvd. Suite 230 President for details. Lakewood, Colorado 80228-1259 (303) 986-3309 Fax: (303) 986-3892 e-mail address: [email protected] Directors The Journal of Range Management is a publication of the Society for Range 2000-2002 Management. It serves as a forum for the presentation and discussion of facts, ideas, and philosophies RICHARD H. HART pertaining to the study, management, and use of rangelands and their several resources. Accordingly, all USDA-ARS material published herein is signed and reflects the individual views the authors and is High Plains Grasslands Station of not necessari- 8408 Hildreth Rd. ly an official position of the Society. Manuscripts from anyone-nonmembers as well as members-are Cheyenne, Wyoming 82009-8809 welcome and will be given every consideration by the editors. Editorial comments by an individual are DON KIRBY also welcome and, subject to acceptance by the editor, will be published as a "Viewpoint." North Dakota State University Animal & Range Science In Cooperation With: Some of the articles appearing in The Journal of Range Management (JRM) Fargo, North Dakota 58105 are presented in cooperation with The American Forage and Grassland Council (AFGC). This cooperation consists of JRM acceptance of professional papers in forage grazing man- A 2001-2003 FG JOHN TANAKA agement and related subject areas from AFGC members and the appointment of 2 AFGC C7 " Eastern Oregon Agr. Res. Center-Union affiliated associate editors to JRM's Editorial Staff. The American Forage and Grassland P.O. Box E Council Offices: P.O. Box 94, Georgetown, Texas 78627; Larry Jeffries, President; Dana Tucker, Union, Oregon 97883 Executive Secretary. GREG TEGART BCMAFF 1690 Powick Rd, Suite 2000 Kelowna, BC V1X 7G5 CANADA Contribution Policy: The Society for Range Management may accept donations of real and/or personal property subject to limitations set forth by State and Federal law. All donations shall be 2002-2004 subject to management by the Executive Vice President as directed by the Board of Directors and their JOHN MALECHEK discretion in establishing and maintaining trusts, memorials, scholarships, or other Utah State University types of funds. Dept. of Rangeland Resources Individual endowments for designated purposes can be established according to Society policies. Gifts, UMC 5230 bequests, legacies, devises, or donations not intended for establishing designated endowments will be Logan, Utah 84322-0001 deposited into the SRM Endowment Fund. Donations or requests for further information on Society MARTIN VAVRA policies can be directed to the Society for Range Management, Executive Vice-President, 445 Union EOARC Blvd., Suite 230, Lakewood, Colo. 80228-1259. We recommend that donors consult Tax Advisors in HC 71 Box 451, Hwy 205 regard to any tax consideration that may result from any donation. Burns, Oregon 97720-9807 The term of office of all elected officers and directors begins in February of each year during the Society's Annual Meeting.

JOURNAL OF RANGE MANAGEMENT 55(4) July 317 J. Range Manage. 55: 318-326 July 2002 Shrub control and streamflow on rangelands: A process based viewpoint

BRADFORD P. WILCOX

The author is an Associate Professor in the Range/and Ecology and Management Department, Texas A&M University, College Station, Tex. 77843.

Abstract Resumen

In this paper, the linkage between streamflow and shrub cover En este articulo se examina la conexion entre las corrientes y on rangelands is examined, with a focus on the extensive Texas la cobertura de los pastizales con enfoque a los pastizales exten- rangelands dominated by mesquite and juniper. The conclusions sivos del Texas dominados por mezquite y junipero. Las conclu- drawn are consistent with results from field studies and with our siones a las que se llego son consistentes con los resultados de los understanding of runoff processes from rangelands. Whether estudios de campo y con nuestro entendimiento de los procesos and how shrub control will affect streamflow depends on shrub de escurrimiento de los pastizales. Si el control de arbustos afec- characteristics, precipitation, soils, and geology. Precipitation is ta, y la manera en que afectara las corrientes, depende de las perhaps the most fundamental of these factors: there is little if caracteristicas de los arbustos, la precipitacion, los suelos y la any real potential for increasing streamflow where annual pre- geologia. La precipitacion es quiza el mas fundamental de estos cipitation is below about 500 mm. For areas in which precipita- factores: Hay muy poco potencial para incrementar las corri- tion is sufficient, a crucial indicator that there is potential for entes en lugares donde la precipitacion anual es menor de increasing streamflow through shrub control is the presence of aproximadamente 500 mm. En areas donde la precipitacion es springs or groundwater flow to streams. These conditions often suficiente, un indicador crucial de que hay potencial para incre- occur at locations where soils are shallow and underlain by frac- mentar las corrientes mediante el control de arbustos es la pres- tured parent material. Under such conditions, reducing shrub encia de de manantiales o flujo de agua subterranea a las corri- cover may increase streamflows because water that would other- entes. Estas condiciones a menudo ocurren en localidades donde wise be lost through interception by the canopy instead moves los suelos son someros y estan sobre material parental fractura- into the soil and quickly travels beyond the root zone. If, on the do. Bajo tales condiciones, reduciendo la cobertura de arbustos other hand, there is no obvious subsurface connection between se pueden incrementar las corrientes, porque el agua que seria the hillslope and the stream channel and when runoff occurs it perdida a traves de intercepcion por la copa en su lugar se occurs as overland flow, shrub control will have little if any influ- mueve hacia el suelo y viaja rapidamente a zonas mas alla de la ence on streamflow. In assessing the potential for shrub control zona radical. Por el contrario, si no hay una conexion subsuper- las to increase streamflow, the runoff generation process should be ficial obvia entre la pendiente de la montana y los canales de la explicitly identified. An improved understanding of the linkages corrientes y que cuando el escurrimiento ocurra sea sobre between shrubs and streamflow on rangelands will require addi- superficie terrestre, el control de arbustos tendra poco, si no es tional research on (1) hillslope hydrologic processes and how que ninguna, influencia en las corrientes. En la evaluacion del these are altered by shrub cover (2) groundwater-surface water potencial del control de arbustos para incrementar las corri- debe ser interactions and (3) hydrologic scale relationships from the patch entes, el proceso de generacion del escurrimiento de las to the hillslope to the landscape levels. explicitamente identificado. Un mejor entendimiento conexiones entre los arbustos y las corrientes en los pastizales requerira de investigacion adicional sobre: (1) los procesos hidrologicos en las pendientes de las montanas y como estos son range hydrology, runoff, shrub control, Key Words: water yield, alterados por la cobertura de arbustos, (2) las interacciones del streamflow, semiarid ecohydrology, agua subterranea con el agua superficial y (3) las relaciQnes a escala hidrologica de nivel parche a nivel montana y hasta nivel In this paper, I review the evidence for whether streamflow can de paisaje. be increased through modification of shrub cover on non-riparian rangelands. The focus is on Texas rangelands, because shrub control is viewed as a viable management option for alleviating the most many of the urgent water supply problems in the State. The planning process designed to ensure that water is used in increasing recent drought-in conjunction with an increasing demand for efficient manner possible and to explore options for water-is focusing attention on water shortages, and any manage- the amount of water available, one of which is shrub control. from Texas ment strategies that may combat these shortages elicit tremen- The perception is widespread that streamflow and therefore water dous interest. The State is in the midst of a comprehensive water watersheds can be significantly augmented, supply substantially increased, through aggressive control of (Prosopis glandulosa Ton. var. glandulosa) and juniper Briske, Urs mesquite The author gratefully acknowledges reviews of this paper by David Buckholtz, Juniperus pinchotii Sudw). For Larry White, Keith Owens, and one anonymous reviewer. (Juniperus ashei Kreuter, John Walker, the San Angelo area of West Manuscript accepted 10 Oct. 01. example, some have argued that in

OF RANGE MANAGEMENT 55(4) July 2002 318 JOURNAL Texas, shrub control could convert now increasing streamflow unless annual pre- Blackburn (1983) conducted a thorough intermittent streams into perennial ones cipitation exceeds 450-500 mm (Hibbert review of the pre-1980 literature on the and dramatically increase the rate at which 1983). linkages between streamflow and shrub water supply reservoirs are filled (UCRA Shrub control as a water management cover on rangelands and concluded that 1998). A similar argument is made regard- tool does warrant serious consideration in much of what had been learned from other ing the Edwards Aquifer recharge area far- Texas, because there are large tracts of semiarid rangelands was not relevant for ther south. In that region, many believe rangeland that are in a relatively high rain- Texas because of the differences in cli- that the spreading juniper communities are fall belt, receiving 600-1,000 mm of pre- mate, soils, and vegetation. In the absence contributing to reduced groundwater cipitation a year. These areas, which were of substantiated data, the major argument recharge and springflow (Wright 1996). once grasslands, are now predominantly for the effectiveness of brush control was Some recent modeling studies support high-density shrublands in which water based on the "Rocky Creek Story" (docu- this notion (UCRA 1998, Bednarz et al. supply is insufficient to meet all of the mented by Kelton [1975]), an anecdote 2001, Wu et al. 2001). One of these-the competing demands. As a means of assess- that has taken on mythic proportions. 1998 UCRA study-was cited as the ing the potential for success of shrub con- Rocky Creek is reported to have dried up major justification for a multi-million-dol- trol in such regions, I have used the limited in the 1930s and to have remained dry lar, state-funded program to subsidize data available to (1) delineate as clearly as until around 1960, when an extensive pro- brush control on the North Concho water- possible the important hydrologic process- gram of brush removal was carried out shed in West Texas. And the more recent es, and (2) identify whether, how, and within the 74,000-acre watershed. Rocky studies will certainly be taken into account under what conditions shrub cover may Creek again began to flow, and has contin- as the State considers whether to allocate modify those processes. The conclusions ued flowing since that time. Beginning in additional funding for brush control aimed are germane not only to Texas rangelands the 1990s, several key field studies have at increasing streamflow in other regions but to other semiarid shrublands as well. been conducted on Texas rangelands that of the State. Given that the programs being considered provide more definitive information about However, results from many field stud- will involve large expenditures of limited the linkages between shrub cover and ies do not fully concur with the predictions public funds, investigation of the scientific water yield. of these modeling studies, particularly for basis for brush control programs is both mesquite-dominated rangelands. It is true timely and important. that in humid landscapes, changes in vege- Relationships between Shrub tation cover-particularly from woody to Cover and Hydrologic Processes herbaceous-can radically alter the water Background cycle (Jackson et al. 2000). For example, in humid areas of Australia the widespread The hydrologic cycle and corresponding The widespread conversion of grass- replacement of Eucalyptus and other deep- water budget are a simple yet powerful lands and savannas to shrublands during rooted woody species by pasture and crop framework for examining how changes in the last 50-100 years has provoked con- species has raised the water table and led to vegetation cover influence water availabil- siderable debate concerning the cause(s) serious salinization problems (Greenwood ity. The linkages between shrub cover and of these changes and given rise to a num- 1992, Walker et al. 1993). A similar rela- the various components of the water bud- ber of investigations. On the basis of sev- tionship between timber harvesting and get are discussed below. eral comprehensive reviews of the litera- streamflow is well documented in many Equation 1 presents a simplified inter- ture, we can conclude that the primary other forest types (Stednick 1996), as is pretation of the water budget, partitioning mechanism behind the increase in shrub the converse: declines in streamflow as a precipitation (the major determinant of the cover has been a dramatic shift in patterns result of afforestation (Trimble et al. 1987, potential for shrub removal to modify of herbivory and fire frequency during this Calder 1990). But in drylands, the corre- streamflow) into (1) evapotranspiration, time, although shifts in climate and CO2 lation between woody cover and stream- (2) runoff, (3) groundwater, and (4) soil concentrations have also been cited as flow is weaker. (Drylands are zones in water: possible factors (Archer 1994, Van Auken which the ratio of precipitation to potential 2000). In Texas, the increase in shrub P=ET+R+G+AS, (1) evapotranspiration is less than 0.65-con- cover has been particularly pronounced for where ditions found in arid, semiarid, and even mesquite (Archer et al. 1988, Archer 1994, P = Precipitation subhumid regions (Middleton and Thomas 1995), ashe, and redberry juniper, espe- ET = Evapotranspiration 1997). In some dryland ecosystems, such cially during the last 50-80 years (Ansley R = Runoff as chaparral woodlands in the southwest- et al. 1995, Fuhlendorf and Smeins 1997, G = Groundwater recharge ern United States, there is unmistakable Smeins et al. 1997, Phillips et al. 2000). z\S = Change in soil water storage evidence that streamflow increases when A logical question is, "Are increases in woody cover is reduced, but in other semi- Evapotranspiration is a process that shrub cover modifying the hydrologic arid environments the linkage is tenuous at includes (1) evaporation from the soil, (2) cycle, and if so, in what way and to what best (Blackburn 1983, Hibbert 1983). How transpiration from the plant, and (3) evap- extent?" In Texas, the perception is wide- streamflow in drylands responds to chang- oration from plant or litter surfaces (com- spread that changes in shrub cover have ing vegetation cover will depend on many monly referred to as interception loss). As led to significant and even dramatic reduc- factors, including precipitation amount shrub cover increases, so too does the tions in the amount of runoff or stream- and pattern, characteristics of the soil, potential for transpiration and/or intercep- flow coming from rangeland watersheds. geology, and vegetation. The amount of tion losses. This issue was examined in detail during precipitation is especially important. In Soil water is the amount of water held in the 1980s, but at that time only a few stud- general, there is no real potential for the soil, which over a period of several ies had been conducted in Texas. years or more is assumed to remain con-

JOURNAL OF RANGE MANAGEMENT 55(4) July 319 stant. Woody vegetation, by virtue of perennially flowing stream is an indication semiarid and subhumid landscapes, it is being more deeply rooted, generally that groundwater flow is important, important to remember that these environ- extracts soil water from greater depths whereas one characterized by ephemeral ments are by definition soil-water-defi- (provided deep water exists) than does or "flashy" flow suggests that either cient, because the evaporative demand is herbaceous vegetation. Soil water that Horton overland flow or shallow subsur- much higher than precipitation. Figure 1, moves beyond the root zone is considered face runoff is the dominant source. in which average monthly potential evapo- to be groundwater recharge, because Woody vegetation may modify the transpiration (Larkin and Bomar 1983) is eventually it will move to an underlying runoff and groundwater recharge compo- compared with average monthly precipita- water body. In semiarid environments, nents of the water budget in myriad ways, tion for the San Angelo area of Texas, flux rates of water moving to groundwater direct and indirect. It may (1) alter soil shows that in this area evaporative are very low, particularly if soils are deep infiltration characteristics, through root demand is about 4 times greater than pre- and of low permeability and/or if aquifers penetration and the addition of organic cipitation. To a large extent, this disparity are located at great depths (Scanlon 1994). matter; (2) preserve soil moisture, through explains why runoff typically accounts for If, on the other hand, the soils are shallow shading and mulching; (3) draw off soil such a small portion of dryland water bud- and the parent material highly perme- moisture, through transpiration or inter- gets-although there are exceptions, as able-as in the Edwards Plateau region of ception; and (4) alter subsurface flow will be discussed later. The consequence is Texas-groundwater recharge may be paths through root activity that leads to the that no matter what the vegetation cover, very rapid (Maclay 1995). formation of macropores (Blackburn 1975, most of the water in a soil-water-deficient Runoff is water that travels from the hill- Seyfried 1991, Breshears et al. 1998, system will be lost through evapotranspi- slope toward the stream channel, a portion Breshears and Barnes 1999, Ludwig et al. ration. This fact alone would suggest that of which (not captured by soils or evapo- 1999, Jackson et al. 2000). But it is removal of shrub cover to minimize evap- rated en route) becomes streamflow. through modification of the evapotranspi- otranspiration in the hope of increasing Runoff travels via a number of pathways, ration component that woody plants most streamflow has a limited chance for suc- including (1) Horton overland flow, (2) influence streamflow. To understand the cess, because most water stored in the soil saturation overland flow, (3) shallow sub- potential for augmentation of streamflow will be either evaporated or used by what- surface flow, and (4) groundwater flow. in these rangelands through removal of ever vegetation is present. A major excep- Horton overland flow, which occurs when mesquite and juniper, therefore, it is tion would be a system in which condi- precipitation intensity exceeds soil infiltra- important to examine not only how these tions allow water to travel very rapidly to tion capacity, is assumed to be the domi- woody plants modify runoff and ground- the stream channel, minimizing opportuni- nant mechanism of streamflow generation water recharge processes, but also how ties for evaporation. for most rangelands, particularly semiarid they might affect evapotranspiration. An important component of total evapo- ones (Dunne 1978). Saturation overland transpiration is interception. Although the flow occurs when soils become saturated. Evapotranspiration available data are somewhat limited, those In more humid environments, soil satura- Environmental characteristics that we do have strongly indicate that intercep- a tion is much higher from juniper than from tion commonly results when rising favor evapotranspiration. When consid- groundwater table brings water to the mesquite rangelands. The greater intercep- sur- ering the process of evapotranspiration in face; this is the primary mechanism for tion capacity of juniper may be attributed variable-source-area runoff (Hornberger et al. 1998). Saturation overland flow may also result from the presence of a shallow impermeable horizon that prevents water from percolating down through the upper Water Balance: San Angelo soil layer. This mechanism has been documented on some rangelands (Lopes and Ffolliott 1993) and likely occurs on many 300 others. Shallow subsurface flow, sometimes referred to as interflow, is that por- 250 -Precipitation tion of runoff that travels laterally through E PET 200 the soil, generally because of some imped- E ing soil horizon. Shallow subsurface flow I- w 150 is more common in humid environments, 0. but it can be important in semiarid envi- 0 100 ronments and can be very rapid, especially D_ when macropores are present in the soil 50 (Wilcox et al. 1997, Newman et al. 1998). Groundwater flow is generally the L L it source for the base flow of a stream (pro- 0 longed flow, not attributable to a specific cX U precipitation event), but probably is not an M onth important pathway for storm flow (streamflow that can be directly attributed to a specific precipitation event) because the Fig. 1. Average monthly precipitation (1969-1998) and potential evapotranspiration pace of groundwater travel is slow. A (Larkin and Bomar,1983) for the San Angelo area.

320 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 to the fact that juniper is an evergreen and water recharge during those 2 years was mm/year, a difference they attribute large- has a high leaf-area density. Working in 70-130 mm greater than if the site had not ly to the much greater interception of the Sonora region of Texas, Thurow and been treated. However, it should be noted water by juniper than by grasses. Hester (1997) found that as much as 70% that by the third year these effects were no In summary, most field studies of of the precipitation is intercepted by the longer measurable because of the regrowth mesquite rangelands, whether based on the juniper canopy and litter layer. Findings of herbaceous and woody vegetation. energy budget or the water budget from other juniper rangelands, although An alternative approach to assessing the approach, have found that eradication of not showing such high percentages, also evaporative demand of shrublands is an mesquite does not lead to increased soil suggest that capture of water by juniper indirect one, the water budget approach, moisture and groundwater recharge, unless canopies can be substantial (Collings in which all the components of the water conditions are such that water can move 1966, Young et al. 1984). Recent work budget except evapotranspiration are mea- rapidly through the herbaceous rooting indicates that the actual percentage of sured directly; evapotranspiration is then zone. For juniper rangelands, however, interception by juniper canopies is highly assumed to be the difference between the both the studies based on the energy bud- dependent on the amount and intensity of sum of these components and the total get approach (Dugas et al. 1998) and those precipitation (Keith Owens, personal com- water budget. For this approach to be based on the water budget approach munication). In contrast, interception by applied successfully on rangelands, (Thurow and Hester 1997) indicate water mesquite canopies is reported to be detailed tracking of soil water is essential. savings resulting from juniper eradication. between 15 and 30% (Desai 1992, Four studies have relied on the water bud- Martinez-Meza and Whitford 1996). For get approach to assess how eradication of Runoff both juniper and mesquite, a much smaller mesquite (3 studies) or juniper (1 study) When evaluating the impact of shrubs percentage of the precipitation is intercept- affects hydrologic processes. on streamflow, it is important to explicitly ed during large, high-intensity storms than The 3 studies in mesquite rangelands did consider which runoff pathway dominates during smaller, low-intensity storms. not all yield similar results. In the earliest in the area being studied. For example, if Measuring evapotranspiration. Evapo- study, Richardson et al. (1979) found that most of the water in rivers and creeks is transpiration at the plant community level in the Blackland Prairie of Texas, where generated from storm flow (either Horton can be measured directly using the Bowen annual precipitation is around 870 mm, overland flow or shallow subsurface flow) Ratio approach, which is based on calcula- deep soil moisture >2 m) increased by ( rather than base flow (groundwater flow), tions of the energy budget (Evett 2000). about 80 mmlyear following eradication of evapotranspiration by shrubs has little Two studies, one of a mesquite rangeland mesquite. In contrast, Carlson et al. (1990) effect on streamflow because water is (Dugas and Mayeux 1991) and the other found that mesquite eradication in the moving through the system too rapidly to of a juniper rangeland (Dugas et al. 1998), Rolling Plains of Texas, where average be transpired. If, on the other hand, base have used Bowen Ratio methodology to annual precipitation is around 640 mm, flow is an important component of the examine changes in community-level had minimal if any influence on soil mois- runoff regime, then there exists the poten- evapotranspiration rates following brush ture-or, by inference, on community- tial for evapotranspiration by vegetation to control. level evapotranspiration-largely because modify streamflow. For first the study, located in the Rolling of the flush of herbaceous growth follow- For flood producing precipitation Plains of Texas, mesquite was removed ing mesquite removal (Heitschmidt and events, shrub cover has relatively little from one site while a second site was left Dowhower 1991). And the third study, by effect on stream discharge. Large precipi- untreated. Dugas and Mayeux concluded Weltz and Blackburn (1995) in south tation events (generally more than that "under circumstances of low grazing Texas (where average annual precipitation 100-120 mm) that result in flood condi- and low runoff potential, honey mesquite is around 700 mm), found little difference tions can overwhelm the capacity of the removal would provide little if any addi- in soil moisture storage or evapotranspira- landscape to store water, regardless of the tional water for off-site uses in the short tion between adjacent mesquite- and extent of tree or shrub cover (Leopold and term." Evapotranspiration was somewhat grassland-dominated communities. The Maddock 1954). This conclusion has been greater from the treated site under dry con- fact that soil moisture increased in the largely borne out by several studies (Ward ditions, but under wet conditions there was Blackland Prairie but not in the other 2 1978, Dunne 1988, Leopold 1997). In no significant difference. The small differ- mesquite rangelands is probably explained other words, the large and relatively infre- ence between the 2 sites was attributed to by the formation, during dry periods at quent flood events that fill many of the the vigorous growth of herbaceous vegeta- this site, of vertical soil cracks that allow rangeland reservoirs would contribute tion following mesquite eradication on the water to move deep into the soil profile. essentially the same quantities of water treated site, a phenomenon noted by many The fourth study was the only detailed whether shrub cover was present or other researchers documenting the effects analysis of water budgets on juniper absent. (This is not to say that rangeland of mesquite control in other areas of Texas rangeland in Texas. From the results, vegetation is unimportant during these (Dahl et al. 1978, Jacoby et al. 1982, Thurow and Hester (1997) concluded that events; but that its major role is protection McDaniel et al. 1982, Bedunah and groundwater recharge could be greatly of the soil resource not modulation of Sosebee 1984, Heitschmidt et al. 1986, increased through removal of all or most flood flow.) Heitschmidt and Dowhower 1991). of the juniper cover. Following complete Horton overland flow. Factors that The second Bowen-Ratio study was in a removal of juniper at the site, they carried contribute to the generation of this type of juniper-dominated rangeland. Dugas et al. out long-term measurements of soil water, flow are high-intensity precipitation (1998) found that brush control did result by means of weighing lysimeters; of and soils having low infiltration capacity, both in significantly lower evapotranspiration canopy and litter interception; and of sur- commonly found in semiarid regions. rates at the community level, but only for face runoff. They calculate that groundwa- Runoff processes in mesquite and juniper 2 years. The authors estimate that ground- ter recharge increased by around 75 rangelands have been explicitly examined

JOURNAL OF RANGE MANAGEMENT 55(4) July 321 only rarely; nevertheless, we can be confi- practices: if surface cover (live or dead Groundwater/surface water interactions dent that Horton overland flow is an vegetation) is encouraged and enhanced, in rangelands have been little investigated, important, if not the dominant mechanism overland flow should be reduced; if sur- partly because of the impression that of runoff generation for the majority of face cover is diminished and the amount groundwater flow is not an important these regions, on the basis of field evi- of bare ground increases, overland flow mechanism for runoff. Hence, there is dence-including the presence of debris may increase. Blackburn (1983), in his much that we do not understand and about dams, signs of channel flow, and the review of the Texas literature, concludes which we can only speculate. For exam- absence of any obvious pathway for rapid that mesquite control either decreases ple, base flow in many cases is probably subsurface flow. My personal observations runoff (by increasing infiltration) or has provided by alluvial aquifers-but by of runoff during flash flooding also con- no effect (Bedunah 1982, Brock et al. what mechanism(s) are these aquifers firm this assertion. 1982, Knight et al. 1983, Franklin 1987). recharged, and at what rates? Does Differences in overland flow between It is probable that many shrub-dominated recharge occur slowly via the hilislope, or grass-dominated and shrub-dominated rangelands, especially former grasslands does it occur quite rapidly, via the stream plots or small watersheds have been docu- and prairies that are now in a degraded channel or other collection point, during a mented for several mesquite and juniper condition, are experiencing greater stream- runoff event? rangelands (Wright et al. 1976, Richardson flow than previously because of overall In many rangeland areas the soils are et al. 1979, Carlson et al. 1990, Thurow lower soil infiltration capacities. In such deep (>1 m), and there is no obvious sub- and Hester 1997, Dugas et al. 1998). degraded environments, we would expect surface connection between them and the Interestingly, Horton overland flow is higher peak flows and "flashier," less sus- stream channel or groundwater aquifer. often reduced following shrub control tained runoff. Under such conditions, little if any water because (1) herbaceous vegetation often Shallow subsurface flow. Shallow sub- moves beyond the root zone. The presence grows vigorously after brush is removed, surface flow on rangelands has received of a calcic, or especially a petrocalcic, and the new growth enhances the infiltra- little attention, but it obviously occurs in horizon is a convincing indicator that the tion capacity of surface soils; and (2) the some areas where soils are shallow and downward flux of water is very small. In increased surface roughness resulting from underlain by highly permeable parent contrast, in landscapes in which soils are the scattered woody debris (and perhaps material-like the Edwards Plateau. It also shallow and the parent material is perme- partly from herbaceous growth) impedes makes up a small portion of streamflow in able or fractured, such as the Edwards overland flow. Both Dugas et al. (1998) the Blackland Prairie of Texas.' In the Plateau, the subsurface connections to and Richardson et al. (1979) report dra- Edwards Plateau region, I have observed groundwater aquifers often allow for rapid matic reductions in Horton overland flow absolutely clear water flowing in stream recharge. Spring flow is common in such following juniper eradication, whereas a channels in the wake of prolonged but regions, as are perennial or intermittent long-term study on small watersheds in low-intensity rains, while at the same time streams. There are numerous anecdotal Sonora showed that removal of juniper there was no evidence of Horton overland reports of spring flow appearing or had little or no effect on surface runoff flow on the hillslopes. Obviously, water increasing after shrub control, and such (Thurow, personal communication). For was traveling a subsurface route-an evidences have been documented for mesquite rangelands, Carlson et al. (1990) occurrence consistent with the presence of juniper rangelands on the Edwards Plateau found that Horton overland flow was shallow soils underlain by permeable or (Wright 1996) and for pinon-juniper lower following mesquite eradication. fractured parent material, which allows watersheds in Utah (McCarthy et al. In contrast, Wright et al. (1976) found water to travel rapidly through the subsur- 1999). That said, it is important to point that Horton overland flow was significant- face to the stream channel or a groundwa- out that although increased spring flow is ly greater for 2-3 years following removal ter body. Many of the areas occupied by vitally important on the local or "on site" of juniper by burning, particularly on steep juniper exhibit such characteristics. Where level, it should not be looked upon as a slopes; presumably it took this much time shallow subsurface flow is rapid, plant way of increasing water supply at larger for the vegetation to completely recover evapotranspiration rates would not directly scales. (in addition, there would be less debris to influence runoff amounts. However, inter- In summary, the mechanism or pathway impede flow following a burn than follow- ception of water by the plant canopy could by which water travels from the hillslope ing mechanical treatment). In another affect those amounts, especially in the to the stream channel to a large extent dic- study, in the Blackland Prairie of Texas, case of juniper (Skau 1964, Young et al. tates the degree to which shrubs may mod- Richardson et al. (1979) observed a 10% 1984, Thurow and Hester 1997). ify streamflow. Modification of the evapo- increase in Horton overland flow after Groundwater flow. Streamflow from transpiration regime will influence stream- eradication of mesquite, which they rangelands is by nature flashy, but for flow only if significant amounts of that attribute to higher soil moisture in the some areas-particularly the more humid streamflow come from subsurface water treated area. ones-base flow does occur and can be sources. Subsurface runoff has been little For regions characterized by Horton important. The presence of base flow or studied in rangelands, even though it is overland flow, then, what we need to be spring flow is an important indicator of the likely to be important-especially in the by examining is how vegetation and land use potential for increasing streamflow higher precipitation zones. Overland flow Base flow is are modifying surface conditions, which manipulating shrub cover. is probably the dominant runoff process are the major determinant of runoff prolonged and indicates relatively slow for most rangelands, but it too has been the subsurface water, which amounts. Depending on those conditions, movement of inadequately examined or quantified. is the potential for augment- shrub control could result in either a means there Where overland flow is the dominant ing flow via shrub removal. decrease or an increase in Horton overland runoff mechanism, modifications of shrub flow. Much will depend on the method of cover will probably have little influence control and the follow-up management 'Clarence Richardson, personal communication. on runoff. In fact, the few rangeland stud-

322 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 ies that have documented overland flow following removal of mesquite or juniper North Cnncho at Ca rlsbe d indicate that if surface disturbance is mini- 3280 mal, herbaceous cover rapidly replaces the km2 shrubs and runoff can actually be lower after shrub removal. 4

Examples from Two Texas E Watersheds E 2 C Additional insights into runoff processes-and thus into the potential for modification of streamflow through manipulation of vegetation-may be gained via analysis U of streamflow hydrographs. Below we dis- 0 200 400 BDd 800 10U0 120Q cuss such analyses from 2 Texas watersheds: the North Concho in West Texas, 1985-1987 (days) a near San Angelo, and the Seco Creek watershed, on the Edwards Plateau in central Texas. The North Concho (lat. 31 ° 35' 33", long. 100° 38' 12") is a comparatively S eco creek-Miller Ranch large watershed (3,280 km2). Average pre- 116 km2 cipitation is around 500 mm/year, whereas average long-term runoff is only about 5 mm/year (i.e., only about 1 % of the water 25 _ budget). From the almost 80-year period of records (USGS historical streamflow 20 data) for the North Concho, it is apparent E that runoff in this area is "flashy" (Fig. 2a E 15 shows runoff for a recent 3-year period). 10 Many of the soils, being high in clay content, have a low infiltration capacity when wet. These soils are moderately deep and often underlain by a caliche layer with no U L obvious subsurface flow pathways. 3 203 403 603 833 1030 1230 Channels that normally transport little or no water will periodically transport very 1965-1967 (days) high flows or even floodwater (Fig. 2a). h Storm-flow runoff (most likely generated as Horton overland flow) makes up a large Fig. 2. Daily runoff for a 3-year period for the North Concho and the Seco River watersheds. percentage of the total runoff. These large Runoff data provided by the United States Geological Survey. flood events are important from a water- supply standpoint, because they are the uation is explained by the low storage evidence, that on rangelands similar to the ones that fill downstream reservoirs. capacity of the soils and the high perme- Seco Creek watershed a reduction of shrub On the Seco Creek Watershed (lat. 29° ability of the underlying parent material. cover has the potential for increasing 34' 23", long. 99° 24' 10"), runoff is also Runoff from a comparable rangeland streamflow and/or groundwater recharge. "flashy" but is sustained for considerably watershed having different soil properties However, on rangelands similar to the longer periods than in the North Concho and parent materials would be much North Concho, where runoff is primarily region. Here, it makes up almost 25% of lower. Horton overland flow, reduction of shrub the total water budget (Fig. 2b). It is likely The strikingly different patterns of cover would likely have little if any influ- that runoff from the Seco Creek watershed runoff from these 2 watersheds highlight ence on streamflow. is generated by multiple processes. Storm the importance of considering specific site flow is rapid and must be accounted for by conditions when attempting to estimate the either Horton overland flow or shallow potential influence of woody vegetation on Summary and Conclusions subsurface flow, both of which have been streamflow. Although runoff from both observed in the region; and groundwater watersheds is dominated by flood events, Mesquite Rangelands base flow from groundwater is an impor- flow is significant. Runoff averaging 25% For most mesquite-dominated uplands of the water budget and reaching more tant component of the water budget in (non-riparian), shrub control is unlikely to 50% in years (e.g., Seco Creek. In contrast, at North Concho than some 1992) affect streamflow significantly, for 4 rea- (Brown et al. 1998) is astoundingly high base flow is insignificant. sons: (1) evaporative demand is high, and for a semiarid watershed. This unusual sit- We can speculate, on the basis of this

JOURNAL OF RANGE MANAGEMENT 55(4) July 323 the herbaceous vegetation that typically cipitation and the amount of potential ly in the case of juniper watershed areas. It grows vigorously following eradication of evapotranspiration diminishes (i.e., the has also been argued that increases in the shrubs uses most of the available soil soil water deficit becomes smaller). shrub cover reduce base flows. But actual water; (2) soils on these sites are typically Hibbert (1983) has proposed as a rule of changes in streamflow following changes deep, effectively isolating the groundwater thumb that no increase in streamflow in shrub cover have yet to be documented; zone from the surface; (3) runoff is gener- should be expected where annual pre- inferences concerning this issue have been ated primarily as Horton overland flow; cipitation is lower than 450 mm/year. made mostly on the basis of measured and (4) runoff is very flashy in nature: Although Hibbert's recommendation changes in evapotranspiration or soil most of it is generated by flood-producing was not based on work in Texas, it has water. If such changes are occurring, we precipitation events, in amounts so over- been commonly applied to Texas condi- should be able to verify them through whelming as to render insignificant the tions (Bednarz et al. 2001). comparison with the historical record. For effects of other factors, such as intercep- 2. Amount of shrub cover. All else being example, trend analysis of long-term tion by vegetation and even soil moisture equal, the clearing of a high-density streamflow should give some indication of storage. stand of shrubs will have a greater effect whether runoff has decreased as shrub on streamflow than will clearing of a co,ver has increased. Juniper Rangelands lower-density stand. Groundwater-surface water interac- 3. subsurface flow charac- tions. Related to the issues of runoff and The available field research data suggest Runoff and teristics. If runoff occurs primarily as vegetation cover is the question of how that there is some potential for increasing overland flow with occasional ground and surface waters interact within streamflow from juniper rangelands. Two Horton flood events, and base flow/groundwater rangeland watersheds-a question espe- studies have indicated that groundwater cially crucial for semiarid landscapes. We recharge will increase following juniper recharge is insignificant, streamflow will be little influenced by woody plant cannot modify one without modifying the removal (Thurow and Hester 1997, Dugas cover. This is probably the case for most other (Jackson et al. 2000). We do not et al. 1998), and 1 study shows increased Texas rangelands-although there are fully understand how alluvial aquifers are spring flow (Wright 1996). As yet, there as the Edwards Plateau recharged (from the stream channel or have been no documented increases in exceptions, such from the hillslope?), nor their role in streamflow as a result of juniper con- region. trol, but the greater potential of these 4. Interception characteristics. In runoff generation. Landscape-scale processes. Our under- rangelands for increased streamflows or juniper rangelands, because the canopy is evergreen and very dense, and litter standing of vegetation and water interac- groundwater recharge is based on two fac- production is high, water losses through tions on a landscape scale is limited. For tors: (1) juniper canopies have a high interception are very high. For mesquite example, where streamflow may be aug- capacity for interception of moisture; and loss via the mented through shrub control, we do not (2) juniper are often found in regions rangelands, interception canopy is probably comparable to inter- know at what scales we would see an where soils are shallow and parent materi- in grasslands. For this rea- effect. In this paper, I have suggested that als are permeable, features conducive to ception loss son, is likely to be any influence of shrub cover on stream- subsurface flow. A recent modeling study removal of juniper flow is likely to be at a small scale and also concluded that increased water yields more effective than removal of mesquite. may not manifest itself at larger scales. (groundwater recharge and/or streamflow) But these scale relationships have yet to be would result from a reduction in juniper documented. cover (Wu et al. 2000). Future Research Finally, apart from any potential effects In those regions where juniper are found Runoff processes at the hillslope scale. on streamflow, there are other reasons- on deep soils and subsurface flow does not For any given rangeland watershed, the and perhaps more compelling ones-to occur, eradication is not likely to increase dominant mechanism by which runoff is as a means of streamflow, for the same reasons noted generated greatly influences that land- practice shrub control restoration of both mesquite and juniper earlier for deep-soil mesquite sites. But scape's streamflow potential, erosion rangelands in Texas. One reason-espe- even in the shallow-soil regions, such potential, and response to land manage- cially with regard to juniper rangelands- increases will occur mainly during wet ment strategies. Relatively few studies on is to prevent thicketization, in the wake of years and at relatively small scales; during rangelands, however, have examined which these rangelands quickly degenerate dry years, and especially during droughts, runoff processes in an explicit and detailed into areas of extremely low productivity, it is doubtful that removal of shrub cover manner. Process-based, hillslope hydrolo- low plant and animal biodiversity, and will affect streamflow. Further, when extra gy studies that couple detailed measure- generally poor wildlife habitat, posing dif- water is generated, storage of that water ments of individual runoff events with ficult management challenges. It has been becomes an issue. To be available for long-term monitoring are required to gain suggested, although not yet demonstrated, water supply, any extra water would have a better understanding of runoff pathways that under these degraded conditions over- to be stored either in a reservoir or as on rangelands, especially at the hillslope land flow is greater, erosion increases, and groundwater. and small catchment level. In New water quality declines. Mexico, we have attempted to implement In other words, brush control can have studies of this type (Wilcox 1994, Wilcox Criteria or Successful Brush positive effects, but for most Texas range- f et al. 1996a, 1996b, Wilcox et al. 1997, Control/Streamflow Augmentation lands increased streamflow is not neces- Newman et al. 1998, Reid et al. 1999). For upland zones, the following factors Influence of shrub cover on runoff sarily one of them. The high soil-water should be considered: deficits, high rates of evapotranspiration, processes. It is commonly assumed that 1. Average amount of precipitation. As weak hillslope-to-stream channel subsur- accelerated erosion and increased overland the amount of precipitation increases, face connections, and predominance of flow accompany thicketization, particular- the difference between the incident pre-

324 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 overland runoff of a "flashy" nature all Brown, D.S., R.N. Slattery, and J.R. Hibbert, A.R. 1983. Water yield improvement limit the possibilities for modifying Gilhousen. 1998. Summary statistics and potential by vegetation management on west- streamflow. The use of brush control in graphical comparisons of historical hydrolog- er rangelands. Water Resour. Bull. the hope of increasing streamflow should ic and water-quality data, Seco Creek 19:375-381. Watershed, south-central be targeted to those areas in which it is Texas. U.S. Geol. Hornberger, G.M., J.P. Raffensperger, P.L. Surv., Austin, Tex. Wilberg, K.N. most likely to work. and Eshleman. 1998. Calder, I.R.1990. Evaporation in the Uplands. Elements of Physical Hydrology. The John John Wiley & Sons, New York, N.Y. Hopkins Univ. Press, Baltimore, Md. Carlson, D.H., T.L. Thurow, R.W. Knight, Jackson, R.B., H.J. Schenk, E.G. Jobbagy, J. References and R.K. Heitschmidt. 1990. Effect of Canadell, G.D. Colello, R.E. Dickinson, honey mesquite on the water balance of C.B. Field, P. Friedlingstein, M. Heimann, Texas Rolling Plains rangeland. J. Range Ansley, R.J., W.E. Pinchak, and D.N. K. Hibbard, D.W. Kicklighter, A. Kleidon, Manage. 43:491-496. R.P. Neilson, W.J. Ueckert. 1995. Changes in redberry juniper Parton, O.E. Sala, and Collings, M.R. 1966. Throughfall for summer M.T. Sykes. 2000. Belowground distributions in northwest Texas. Rangelands conse- thunderstorms in a juniper and pinyon wood- quences 17:49-53. of vegetation change and their treat- land Cibecue Ridge, Arizona. geological sur- ment in models. Ecol. Appl.10:470-483. Archer, S. 1994. Woody plant encroachment vey professional paper 485-B. U.S. Dept. of Jacoby, P.W., Meadors, M.A. into southwestern grasslands and savannas: C.H. Foster, the Interior. and F.S. Hartmann.1982. Honey mesquite Rates, patterns and proximate causes, p. Dahl, B.E., R.E. Sosebee, J.P. Goen, and C.S. control and forage response in Crane County, 13-68, In: M. Vavra, W.A. Laycock and Brumley. 1978. Will mesquite control with Texas. J. Range Manage. 35:424-426. R.D. Pieper (eds.), Ecological Implications 2,4,5,-T enhance production? J. Range Kelton, E. 1975. The story of Rocky Creek. of Livestock Herbivory in the West. Society Manage. 31:129-131. The Practicing Nutr. 9:1-5. for Range Management, Denver, Colo. Desai, A.N. 1992. Interception of precipitation Knight, R.W., W.H. Blackburn, and C.J. Archer, S. 1995. Tree-grass dynamicas in a by mesquite dominated rangelands in the Scifres. 1983. Infiltration rates and sediment prosopis-thornscrub savanna parkland-recon- rolling plains of Texas. M.S. Thesis, Texas production following herbicide/fire brush structing the past and predicting the future. A&M Univ., College Station, Tex. treatments. J. Range Manage. 36:154-157. Ecosci. 2:83-99. Dugas, W.A. and H.S. Mayeux. 1991. Larkin, T.J. and G.W. Bomar.1983. Climatic Archer, S., C.J. Scifres, C.R. Bassham, and Evaporation from rangeland with and with- Atlas of Texas. Texas Dept. of Water Resour. R. Maggio.1988. Autogenic succession in a out honey mesquite. J. Range Manage. 36. Leopold, L.B. 1997. Water, Rivers and Creeks. subtropical savanna: conversion of grassland Dugas, W.A., R.A. Hicks, and P. Wright. Univ. Sci. Books, Sausalito, Calif. to thorn woodland. Ecol. Mono. 58:111-127. 1998. Effect of Removal of Juniperus ashei Leopold, L.B. and T. Maddock. 1954. The Bednarz, S.T., T. Dybala, R.S. Muttiah, W. on Evapotranspiration and Runoff in the Flood Control Controversy; Big Dams, Little Rosenthal, and W.A. Dugas. 2001. Seco Creek Watershed. Water Brush/water yield feasibility studies. Resour. Res. Dams, and Land Management. The Ronald 34:1499-1506. Press Company, New York, N.Y. Blackland Res. Center, Temple, Tex. Dunne, T. 1978. Chapter 7. Field studies of Lopes, V.L. and P.F. Ffolliott. 1993. Bedunah, J.D. 1982. Influence of some vege- hillslope flow processes, p. 227-293, Sediment rating curves for a clearcut tation manipulation practices on the bio- Hillslope Hydrology. John Wiley and Sons, Ponderosa pine watershed in northern hydrological state of a depleted deep hard- New York, N.Y. Arizona. Water Resour. Bull. 29:369-382. land range site. Ph.D. Thesis, Texas Tech Dunne, T. 1988. Geomorphological contribu- Ludwig, J.A., D.J. Tongway, and S.G. Univ., Lubbock Tex. tions to flood control planning, 421-438, Marsden. 1999. Stripes, strands or stipples: Bedunah, J.D. and R.E. Sosebee. 1984. p. In. V.R. Baker, R.C. Kochel and P.C. Patton modelling the influence of three landscape Forage response of a mesquite buffalograss (eds.), Flood Geomorphology. John Wiley banding patterns on resource capture and community following range rehabilitation. J. and Sons, New York, N.Y. productivity in semi-arid woodlands, Range Manage. 37:483-487. Evett, S.R. 2000. Energy and water balances at Australia. Catena 37:257-273. Blackburn, W.H. 1975. Factors influencing soil-plant-atmosphere interfaces, p. Maclay, R.W. 1995. Geology and Hydrology infiltration and sediment production of semi- A.129-A.182, In: M.E. Sumner (ed.), of the Edwards Aquifer in the San Antonio arid rangelands in Nevada. Water Resour. Handbook. of Soil Science. CRC Press, New Area, Texas. Water-Resour. Invest. Rep. 95- Res. 11:929-937. York, N.Y. 4186. U.S. Geol. Surv., Austin, Tex. Blackburn, W.H. 1983. Influence of brush Franklin, J.P. 1987. Consumptive water use Martinez-Meza, E. and W.G. Whitford. control on hydrologic characteristics of range by mesquite and grass communities in north 1996. Stemflow, throughfall and channeliza- watersheds, p. 73-88, Brush Manage. Symp., central Texas. M.S. Thesis, Texas A&M tion of stemflow by roots in three Chihuahuan Soc. for Range Manage. Meeting, Albuquerque, Univer., College Station, Tex. Desert shrubs. J. Arid Environ. 32:271-287. N.M. Fuhlendorf, S.D. and F.E. Smeins. 1997. McCarthy, F.J., J.P. Dobrowolski, and P. D.D. 1999. Breshears, and F.J. Barnes. Long-term vegetation dynamics mediated by Figures. 1999. Ground water source areas Interrelationships between plant functional herbivores, weather and fire in a juniperus- and flow paths to springs rejuvenated by types and soil moisture heterogeneity for quercus savanna. J. Veg. Sci. 8:819-828. juniper removal at Johnson Pass, Utah Olsen semiarid landscapes within the grassland/for- Greenwood, E.A.N. 1992. Deforestation, DS, Potyondy JP. Wildl. Hydro., Proceed.. p. est continuum: a unified conceptual model. revegetation, water balance, and climate: an 5. Landscape Ecol.14:465-478. optimistic path through the plausible, imprac- McDaniel, K.C., J.H. Brock, and R.H. Haas. Breshears, D.D., J.W. Nyhan, C.E. Heil, and ticable, and controversial. Advances in 1982. Changes in vegetation and grazing B.P. Wilcox. 1998. Effects of woody plants Bioclimatology 1:89-154. capacity following honey mesquite control. J. on microclimate in a semiarid woodland: soil Heitschmidt, R.K. and S.L. Dowhower.1991. Range Manage. 35:551-557. temperature and evaporation in canopy and Herbage response following control of honey Middleton, N.J. and D.S.G. Thomas. 1997. intercanopy patches. Intl. J. Plant Sci. mesquite within single tree lysimeters. J. World Atlas of Desertification. Edward 159:1010-1017. Range Manage. 44:144-149. R.H. Arnold, London. Brock, J.H., W.H. Blackburn, and Heitschmidt, R.K., R.D. Schultz, and C.J. Newman, B.D., A.R. Campbell, and B.P. Haas. 1982. Infiltration and sediment pro- Scifres. 1986. Herbaceous biomass dynamics Wilcox. 1998. Lateral subsurface flow path- duction on a deep hardland range site in and net primary production following chemi- ways in a semiarid ponderosa pine hsllslope. North Central Texas. J. Range Manage. cal control of honey mesquite. J. Range Water Resour. Res. 34:3485-3496. 35:195-198. Manage. 39:67-71.

JOURNAL OF RANGE MANAGEMENT 55(4) July 325 Phillips, R.A., D.N. Ueckert, and C.B. Scott. Stednick, J.D. 1996. Monitoring the Effects of Wilcox, B.P., J. Pitlick, C.D. Allen, and D.W. 2000. Long-term changes in Redberry Timber Harvest on Annual Water Yield. J. Davenport. 1996a. Runoff and erosion from Juniper canopy cover in western Texas Pubi. Hydro.176:79-95. a rapidly eroding pinyon-juniper hillslope, pp. No R-8. Angelo State Univ.; Manage., Thurow, T.L. and J.W. Hester.1997. How an 61-71, In: M.G. Anderson and S.M. Brooks Instruction and Res. Center, San Angelo, increase or a reduction in juniper cover alters (eds.), Advances in Hillslope Processes. John Tex. rangeland hydrology, p. 9-22, Juniper Symp. Wiley & Sons, New York, N.Y. D. Reid K. D. B. P. Wilcox D. D. Breshears Texas A&M Univ., San Angelo, Tex. B.P., B.D. Newman, Brandes, , , , , K. Reid. 1997. and L. MacDonald. 1999. Runoff and ero- S.W., F.H. Weirich, and B.L. Hoag. Davenport, and Runoff from a semiarid ponderosa pine hills- sion for vegetation patch types in a pinon- 1987. Reforestation and the reduction of lope in New Mexico. Water Resour. Res. juniper woodland . Soil Sci . Soc . Amer . J . yield on the Southern Piedmont since 63:1869-1878 . circa 1940 . Water Resour . Res . 23:425-437 . B.P., B.D. Newman, C.D. Allen, K.D. Richardson , C . W ., E . Burnett, and R . W . UCRA . 1998 . Concho River Watershed: D. Brandes, J. Pitlick, and D.W. Bovey. 1979. Hydrologic effects of brush Brush Control Planning, Assessment and Davenport. 1996b. Runoff and erosion on control on Texas rangelands. Trans. Feasibility Study. Upper Colorado River the Pajarito Plateau: observations from the ASAE:315-319. Authority, San Angelo, Tex. field, p. 433-439, Geology of the Los Scanlon, B.R. 1994. Water and heat fluxes in Van Auken, O.W. 2000. Shrub invasions of Alamos-Jemez Mountains Region. desert soils 1. field studies. Water Resour. North American semiarid grasslands. Ann. Wright, H.A., F.M. Churchill, and W.C. Res. 30:709-719. Rev. Ecol. & Systematics 31:197-215. Stevens. 1976. Effect of prescribed burning Seyfried, M.S. 1991. Infiltration patterns from Walker, J., F. Bullen, and B.G. Williams. on sediment, water yield, and water quality simulated rainfall on a semiarid rangeland 1993. Ecohydrological changes in the from dozed juniper lands in central Texas. J. soil. Soil Sci. Soc. Amer. J. 55:1726-1734. Murray-Darling Basin. I. The number of Range Manage. 29:294-298. Skau, C.M. 1964. Interception, thoughfall, and trees cleared over tow centruies. J. Appl. Wright, P.N.1996. Spring enhancement in the stemflow in Utah and Alligator juniper cover Ecol. 30:265-273. Seco Creek water quality demonstration pro- types of northern Arizona. Forest Sci. Ward, R.C. 1978. Floods, A Geographical ject. Annual Project Rep.. Seco Creek Water 10:283-287. Perspective. John Wiley and Sons, New Quality Demonstration Project. X.B., E.J. Redeker, and T.L. Thurow. Smeins F.E. S.D. Fuhlendorf and C.A. , N.Y. , , , Vegetation and water yield dynamics Taylor. 1997. Environmental and land use Weltz M.A. and W.H. Blackburn. 1995. , in an Edwards Plateau watershed. J. Range Water budget for south Texas rangelands. J. changes: a long-term perspective , p. 54:98-105. 1 . 3-1 . 21 , Juniper Symposium . Vol . Tech Range Manage . . Young , R.A. Evans, and D.A. Eash. . . . . in inter- Rep. 97-1 Texas A&M Univ ., San Ang elo , Wilcox, B P 1994 and erosion Stem flow on western juniper Tex . canopy zones of pinyon-juniper woodlands . (Juniperus occidentalis) trees. Weed. Sci J. Range Manage. 47:285-295. 32:320-327.

326 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 J. Range Manage. 55: 327-335 July 2002 Early weaning and length of supplementation effects on beef calves

A.J. PORDOMINGO

Author is Research Scientist of National Institute of Agricultural Research, INTA Anguil, CC. 11, Anguil (6326), La Pampa, Argentina.

Abstract Resumen

Early weaning of calves can improve reproduction of beef El destete precoz mejora la eficiencia reproductiva del rodeo y cows, and would be of no detriment to calf growth if the diet is no serfa detrimental del crecimiento del ternero si la dieta es adequate. Digestible energy intake could be limiting, however, adecuada. Sin embargo, cuando los terneros son destetados a when calves are weaned on forage. Performance of calves weaned forraje, el consumo de energia digestible podria ser limitante del at 70 days (early weaning) and 172 days (normally weaning) of crecimiento. En este estudio se comparo la performance de age were compared. Calves from 2 locations in Argentina, terneros destetados a los 70 dias (destete precoz) o a los 172 dias Anguil, and Chacharramendi, were distributed in 14 groups and de vida (destete normal). Terneros provenientes de 2 localidades half were weaned. Early-weaned calves were pen fed a 50% con- de la provincia de La Pampa en Argentina, Anguil y centrate-50% alfalfa hay diet for 12 days, followed by grazing on Chacharramendi, fueron distribuidos en 14 grupos y la mitad se alfalfa for 150 days. During the first 90 days on pasture, early- destetaron. Los terneros destetados precozmente fueron alimen- weaned calves were group supplemented (1.2 % body weight tados a corral con una dieta 50% concentrado -50% heno de alfalfa durante 12 dias pasaron luego a pastorear alfalfa por (BW, DM basis). Calves from Anguil did not differ (P = 0.068) in y average daily gain (ADG). In contrast, early-weaned calves from 150 dias mas. Durante los primeros 90 dias sobre la pastura, Chacharramendi gained faster than normally-weaned calves (P < recibieron un suplemento en grupo (1,2% de peso vivo en base (P 0.01). In a second experiment, 108 calves were classified into 3 seca). Los terneros provenientes de Anguil no difirieron = 0.06) en aumento de peso (ADP). En los age groups on day 0 of trial (AGE1 = 109 ± 2.2 days of age, contraste, terneros AGE2 91 ± 1.6 days of age and AGE3 75 ± 3.8 days of age). provenientes de Chacharramendi destetados precozmente = = (P Two thirds of the calves were weaned (early-weaned calves) the aumentaron a un ritmo mayor < 0,01) que los destetados nor- same day and the remaining third was returned to their dams malmente. En un segundo experimento, 108 terneros fueron (normally-weaned calves). Two feeding treatments were imposed clasificados en 3 grupos de edad el dia 0 del ensayo (EDAD1 = 109 ± 2,2 dias, EDAD2 91 ± 1,6 dias EDAD3 75 ± 3.8 dias). on the early-weaned calves: S15 = supplement during the first 15 = y = Dos tratamientos de suplementacion fueron impuestos sobre los days on pasture (1% body weight, DM basis), and S45 = supplement during the first 45 days. Early-weaned calves grazed on an terneros destetados precozmente: 515 = suplementacion durante alfalfa pasture for 136 days. Calves from the normally-weaned 15 dias sobre pastura (1% del peso vivo en base seca) y 545 = group remained with their mothers until weaning onto pasture suplementacion durante 45 dias. Los terneros de destete precoz on day 87 of the study. Normally-weaned calves were the heaviest pastorearon sobre alfalfa durante 136 dias. Los terneros del (P < 0.05) at the end of trial. Differences in body weight between grupo de destete normal se mantuvieron con sus madres hasty el early weaning ages increased as the supplementation period dia 87 del ensayo, momento en el que fueron destetados y decreased. Calves that were weaned at 75 days of age and fed trasladados a una pastura de alfalfa (sin suplementacion). Los supplement for only 15 days had the lowest (P < 0.05) overall terneros destetados normalmente fueron los mas pesados al ADG and final body weight. Overall results suggested that early finalizar el experimento. las diferencias de peso entre edades al weaning favors reproduction of thin cows, and early-weaned momento del destete precoz se incrementaron con el periodo de calves can be placed on good-quality pasture with no detriment suplementacion. Los terneros que fueron destetados a los 75 dias (P of growth if energy supplement is provided. de vida y suplementados por solo 15 dias tuvieron los menores < 0,05) aumentos de peso y pesos mas bajos al finalizar el ensayo. Los resultados de ambas experiencias sugirieron que el destete precoz favorece la reproduccion de vacas delgadas y que los Key Words: Calf growth, calf diet, calf supplementation, beef terneros destetados precozmente no sufririan un retraso de su cattle crecimiento si se ofrece un forraje de alta calidad y se provee un suplemento energetico.

This research was funded by National Institute of Agricultural Research. INTA Early weaning of calves can improve reproduction rates of thin Anguil, La Pampa. Author wishes to thank Carlos Urquiza and his personnel for cows (Laster et al. 1973, Lusby and Wetteman 1980, Lusby and assistance during field data collection. Appreciation is also extended to Associate Parra 1981, Lusby et al. 1981, 1990, Peterson et al. 1987), and Editor Dr. Elaine Grings for her invaluable assistance in reviewing and editing this allow for increases in stocking rate (Monje et al. 1978, Hofer et manuscript. Manuscript accepted 15 Sept. 01. al. 1984, Kugler et al. 1997b). Lusby and Wetteman (1980), and Gill et al. (1993) demonstrated that weaning as early as 55 days

JOURNAL OF RANGE MANAGEMENT 55(4) July 327 of age is not detrimental to calf growth. 0 of trial: Anguil calves = 70.3 ± 2.82, were prepared for reading following the Moreover, early-weaned calves on a high- Chacharramendi calves = 70.4 ± 2.73 days methodology described by Holechek et al. concentrate diet can gain weight faster of age), all calves were weighed and (1982) and botanical composition was than normally weaned calves before wean- assigned to groups of 12 calves each, with estimated according to Sparks and ing (Myers et al. 1999a, 1999b) or even on 6 groups in Anguil and 8 groups in Malechek (1968). Frequency of occur- a finishing program after weaning Chacharramendi. Groups were made as rence of each species was converted to rel- (Fluharty et al. 2000). Most of this work, homogeneous as possible in weight and ative density and used to calculate propor- however, was done using feedlot diets fol- age of calves within location. Age and tions in the diet (Holechek and Gross lowed by a feedlot program. Little body condition of cows were also account- 1982). Species classes were identified as research has been done with early-weaned ed for in making groups as homogeneous cool-season perennial grasses, warm-sea- calves and high forage diets. The greater as possible prior to final allocation of son perennial grasses, shrubs, forbs and the roughage content of the early-weaned calves to treatments. Groups were consid- annual grasses. About a week after each calf diet, the lower total energy intake and ered experimental units and were then fecal collection, hand-clipped samples rate of gain. Hofer et al. (1984) and Monje assigned to either early weaning or nor- were collected of species that contributed et al. (1993) showed that early-weaned mally-weaning treatments at each location. more than 5 % to the pooled sample or calves should be placed on good quality Body condition of cows was recorded were the single contributor to a forage pasture and provided supplemental energy by individual palpation and observation, class. Prior to hand-clipping, observations if good gains are desired. This research according to the 9-point body-condition on plant parts harvested by the cows were was conducted: a) to evaluate performance score (BCS) scale: 1 = severely emaciated made to sample similar fractions. The col- of early-weaned calves on pasture with to 9 = very obese. At both locations, cows lected material was oven dried at 60° C for limited amounts of supplementation, and BCS were determined on days -63, -33, 0, 48 hours, ground in a Wiley mill (1-mm b) to study the interaction between age at 42, 72, and 102 of the trial at both loca- screen) and stored for later chemical weaning and supplementation period on tions. Day -63 was day 7 of the calving. analysis. Analyses performed were crude average daily gain of calves. The impact period. At the time about 60% of the cows protein (CP = Kjeldahl N * 6.25) (AOAC of early weaning on body condition score had given birth. Cows were exposed to 1990), lignin (Goering and Van Soest changes of the dams grazing rangeland at Angus bulls (3 and 4 bulls for Anguil and 1970) and in vitro DM digestibility 2 locations was also evaluated. Chacharramendi, respectively) during an (IVDMD) (Tilley and Terry 1963). 84-day period beginning on day 3 of the Calves assigned to the normally-weaning trial. Cows were pregnancy checked by treatment were returned to their dams on rectal palpation 75 days after the end of day 0. Calves assigned to the early wean- Materials and Methods breeding. ing treatment were assembled in a drylot in To evaluate diet quality at the 2 loca- Anguil and had ad libitum access to a 50% tions, fecal samples were collected from concentrate - 50% alfalfa hay diet for 12 Experiment 1 20 cows at random from each herd at the days (Table 1). The diet was formulated Performance of early-weaned calves same time of body condition scoring from according to previous research (Hofer et al. weaned at 70 days of age was compared October to March. Immediately after col- 1984, Monje et al. 1993) to meet require- with calves weaned at the normal age of lection, samples were dried at 60° C for 48 ments of young calves for an ADG of 500 180 days. Additionally, the impact of early hours and ground in a Wiley mill to pass a g, and to stimulate rumen development. weaning on cow body condition score was 1-mm screen. Similar aliquots in weight After this period, calves were placed on evaluated. This trial started on the second from each fecal sample were taken and alfalfa pasture at Anguil and limit fed a week of December and took place during composited into a pooled sample for each supplemental feed. A 40-ha alfalfa pasture the spring of 1995, and summer and fall of location and sampling time. Micro-histo- was divided into 14 paddocks with electric 1996, at INTA (National Institute for logical analysis was performed on the fences, one for each group. Paddocks were Agriculture Research) Anguil Experiment pooled sample to determine botanical allotted randomly to early-weaned and nor- Station. Calves came from of 2 different composition and relative proportions of mally-weaned groups from each location. locations: Anguil, located in the sub- main species and species classes. Samples After the 12-day pen feeding period, early- humid temperate area of central Argentina (Latitude 36° 30' S Longitude 63° 59W, Table 1. Composition of feed and forages provided to early-weaned calves. and 165 m above sea level ; mean annual rainfall = 608 ± 154 mm, Roberto et al. Experiment 1 1996), and Chacharramendi, located in the semiarid temperate area of the country CP1 NDF ADF forage (Latitude 37° 22' S Longitude 65° 49' W (%) (%) (%) kg DM 1) DM ha 1) and 242 m above sea level; mean annual Concentrate feed 18.2 15.0 Alfalfa hay 17.5 rainfall = 465±143 mm, Roberto et al. 50.3 1996). Mixed supplement3 18.0 19.2 Medium-frame 5-to-7-year old Angus Alfalfa pasture December 1995 17.8 cows that had calved within a 14-day peri- January 1996 15.1 55.1 od, were selected from the herds at Anguil February 1996 16.6 52.0 and Chacharramendi. Seventy-two cows * were selected at Anguil and 96 cows at 1CP = crude protein (Kjeldhal N 6.25), NDF = Neutral detergent fiber, ADF = Acid detergent fiber, ME = Metabolizable energy (NRC 1996). All nutrients are expressed on DM basis. Chacharramendi. At both locations, cows ZCommercial feed based on: oat grain, sorghum grain, corn, sunflower meal and soybean meal, mineral salt, Ca and P grazed on native grassland rganic sources, and a vitamin- micro mineral premix. with no supple- 3 mental feed. At about 70 days of age (day Mixed feed composition: 40% oats, 40% corn, 18% soybean meal, 1 % vitamin-mineral premix.

328 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 weaned calves were placed in their cone- Experiment 2 initiation of the experiment in 3 weight sponding alfalfa pasture until the end of This trial took place at INTA Anguil groups, with 4 calves each. These groups trial (day 162). Experiment Station. The objective of the were defined as experimental units. Early-weaned calves were supplemented study was to determine the effects of age Early-weaned calves were fed a diet of daily during the first 90 days on pasture. at early weaning and length of supplemen- 50% chopped alfalfa hay and 50% concen- Supplement was offered at 1.2% of aver- tation period on calf performance. One trate for 12 days (Table 2), followed by age body weight on DM basis within hundred and eight medium-frame Angus grazing of alfalfa pasture for 136 days group. Supplement quantity was adjusted calves (70 to 105 days of age) were classi- with supplementation according to treat- every 15 days based on actual BW or cal- fied into 3 age groups on day 0 of trial. ments. A 40-ha pasture was divided in 27 culated BW assuming an ADG of 500 g. Age groups were defined as: AGE1= 111 paddocks (1.45 ha each). Paddocks were For the first 10 days on pasture, the sup- + 2.2 days of age (oldest), AGE2 = 91 ± assigned randomly to each experimental plement was the same as that fed in drylot. 1.6 days of age (intermediate), and AGE3 unit. Supplemental feed (Table 2) was the Afterwards, a concentrate mixture based = 75 ± 3.8 days of age (youngest). Two same concentrate as in pens, offered at 1% on oat grain, corn and soybean meal was thirds of the calves from each age group of BW (on DM basis) daily at 1100 hours, fed (Table 1). The supplement was offered were weaned(early-weaned calves) the calculated on the group BW average. once daily at 1100 hours. A mixture of same day and the remaining third was Immediately after weaning, normally- 50% salt and 50% bone meal was present returned to the dams for later weaned calves were in feeders at all times. weaning distributed in the pre- (normally-weaned calves). After a pen- viously assigned weight Calves on the normally-weaning treat- block within age feeding period of 12 days, early-weaned group and moved to ment were weaned on day 102 of trial the corresponding calves grazed on an alfalfa pasture for 136 alfalfa paddock. (172 ± 2.73 days of age), which also coin- Paddocks had been days. Two supplementation cided with the last day of supplementation periods were grazed by non-experimental steers before imposed on the early-weaned of early-weaned calves. Normally weaned calves: 515 arrival of calves to remove accumulated supplement during calves were placed on pasture in assigned = the first 15 days on mature forage, encourage regrowth and pasture (1% paddocks, which had been grazed by non- of body weight, DM basis), maintain forage quality. Calves from nor- experimental steers to remove accumulat- and 545 = supplement during the first 45 mally-weaned group remained on pasture ed mature forage and promote growth of days on pasture. Supplementation of 515 for 49 days (day 148 of trial - end of the high-quality forage, similar to forage and 545 calves were concluded on day 27 study), and received no supplemental feed being grazed by early-weaned calves. and 57, respectively. at any time. Groups grazed simultaneously for an addi- Calves from the normally-weaned All calves were weighed and weights tional 60-day period. group, remained with their mothers on recorded on days 0, 27, 57, 87, 117, and All calves were weighed at the begin- native rangeland until weaning onto pas- 148 of trial, comprising 5 periods. ning of the experiment and on days 42, 72, ture on day 87 of the study. At that time, Procedures for animal weighing, and alfal- 102, 132, and 162 of trial, equivalent to age groups averaged 196, 178, and 162 fa pasture and feed sampling were the days 30, 60, 90, 120, and 150 on pasture days of age for age groups, AGE1, AGE2 same as described in Experiment 1. for early-weaned calves. First body weight and AGE3, respectively. Comparisons of measurements were taken 5 hours after performance of calves weaned at less than Statistical Analysis weaning, but all other BW were recorded 100 days of age with calves weaned at Data of each experiment were statisti- after a 17-hour fast without access to feed ages ranging from 5 to 7 months were of cally analyzed using GLM procedures of and water. interest for inferences to an ample array of SAS (1991) for repeated measures Beginning on day 10 of the study, for- cow-calf programs common in our region. ANOVA. Calf group served as experimen- age DM availability of alfalfa was deter- The design assumed supplementation tal units in each trial. Factor means were mined every 30 days. Five 1-m2 quadrants periods (515 and 545) and nursing as 3 separated using LSD (predicted difference per paddock were hand-clipped to a 5-cm feeding alternatives crossed by 3 age method, SAS 1991) when significant (P < height and weighed to determine yield. groups. The combination of 3 age groups 0.05) F tests were observed. Body weight One sub-sample was taken to determine x 3 feeding alternatives resulted in 9 treat- and ADG of Experiment 1 were analyzed DM (AOAC 1990), while another was ments with 12 calves each. Calves within under a randomized complete block reserved for chemical analysis. Samples of each treatment were blocked by weight at design, with weaning treatment as main feed offered in pens during the 12-day pen feeding period was taken every 4 days and Table 2. Composition of feed and forages provided to early-weaned calves. samples of the supplement offered on pasture were taken every 15 days. These sam- Experiment 2 ples were oven dried at 60° C to determine DM content and kept for later chemical CP' NDF ADF analysis. Forage samples from each pad- kg DM"') dock were composited within sampling Concentrate feed 18.2 15.0 period. Composites were analyzed for CP Alfalfa hay 17.2 48.2 (AOAC 1990), neutral detergent fiber Alfalfa pasture (NDF) (Robertson and Van Soest 1981), December 1996 18.2 January and acid detergent fiber (ADF) 1997 16.1 55.0 (Goering February 1997 16.3 47.3 and Van Soest 1970) and in vitro DM March 1997 17.8 43.1 digestibility (IVDMD) (Tilley and Terry CP = crude protein, NDF = Neutral detergent fiber, ADF = Acid detergent fiber, ME = calculated metabolizable energy 1963). Metabolizable energy (ME) was NRC 1996). All nutrients are expressed on DM basis. calculated from IVDMD estimates (NRC Commercial feed based on: oat grain, sorghum grain, corn, sunflower meal and soybean meal. Included also: mineral 1996) salt, Ca and P organic sources, and a vitamin- micro mineral premix.

JOURNAL OF RANGE MANAGEMENT 55(4) July 329 effect, location, and period in the sub-plot Table 3. Average daily gain (g animal'') of Angus calves weaned at 70 (EW) or 172 (NW) days of (repeated measures factor). Treatment age from 2 locations'. effect, treatment x location, calf group(treatment), period, treatment x peri- Day of trial EW calves' NW od and the 3-way interaction (treatment x Animal-' ----- location x period) were included in the ------Anguil------cnnbA cnnbA 0 to 42 model. One-way ANOVA was performed 592bA 601bA 42 to 72 by location and period if an interaction 657cA 72 to 102 632A weaning treatment x location x period was 528aA 389aB l02 to 132 527aA noted (P < 0.05). Effects of diet shifts 132 to 162 616'' were also assessed by ANOVA of period SE3 7.9 18.6 607A 613A using GLM procedure of SAS O to 102 effect 527A 102 to 162 502B (1991). The model included period, calf 577A 572A group(period), location and period x loca- O to 162 tion. an interaction of period by location Chacharramendi ------If 592bA O to 42 371bB was detected (P < 0.05), interaction means 602bA 42 to 72 364bB were reported. 337aB 72 to 102 632A 2 as a com- aA 336aB Experiment was designed 102 to 132 54 540aA pletely randomized design with a 3 x 3 132 to 162 477cB factorial arrangement of treatments (AGE SE 7.6 7.1 607A 359E in the O to 102 x FA) in the main plot and period 540A 102 to 162 4078 sub-plot (repeated measures). Factors 582A 377E AGE, FA and the AGE x FA interaction, O to 162 period, and the 3-way interaction (FA x treatment x location x period interaction was detected (P < 0.001); therefore treatment means are reported by location and period. x were included in the AGE Period) 3SE = Standard error for weaning effect within period and location. model. Period effects were analyzed with- SE = Standard error for period effect within weaning treatment and location. in columns within weaning treatment and location with different lowercase superscripts differ (P < 0.05). in FA by ANOVA with GLM of SAS with ABMeans Means in rows with different uppercase superscript differ (P < 0.05). a completely randomized design. Interaction means were reported if a meaningful interaction was detected (P < Anguil prised the diet of Anguil cows during 0.05). Cows from Anguil were in good body spring and summer. Most of the native condition at early weaning time and forage available was perennial grasses. remained in good condition until all calves Crude protein and IVDMD data indicated Results and Discussion were weaned (Table 5). However, differ- that quality would have not been a limiting ences in BCS in favor of cows from the factor for milk production and body con- early weaning treatment were noted (P < dition retention on the cows, and it did not Experiment 1 0.05) as time progressed. Pregnancy rate imposed major restrictions to the calves' At initiation of the experiment, age of was 97% and 92% for cows from early forage intake. cows was very similar between treatments weaning and normal weaning treatments, No differences between treatments were within location (Anguil = 5.96 ± 0.745 respectively. Table 6 shows composition detected (P = 0.33) in ADG of calves for years, (P = 0.875); Chacharramendi = 6.01 and quality of the plant classes that com- the first 102 days of trial (Table 3). ± 0.736 years; P = 0.890). Body condition of cows was also similar (Anguil = 5.42 ± 0.576, P = 0.710; Chacharramendi = 3.85 Table 4. Live weight (kg) of calves weaned at 70 (EW) or 172 (NW) days of age from 2 locations'. ± 0.405, P = 0.6154). Calf body weight did not differ either (Anguil = 77.1 ± 0.11 Day of trial EW NW SE2 kg, P = 0.492; Chacharramendi = 77.6 ± ------(kg)------0.12 kg, P = 0.766) among weaning treat------Anguil------ments within location at day 0. 0 77.1 77.2 Supply of alfalfa forage was adequate to 42 102.2 101.9 ensure that both forage quality and quanti- 72 120.2 120.4 102 139.0 139.7 (Table 1). A ty did not restrict calf growth 132 154.8 151.4** significant interaction (P < 0.01) was 162 170.6 169.8 detected between location and weaning Chacharramendi ------treatment for calf performance, therefore, 0 77.6 77.7 weaning effects are reported by location 42 102.5 93.2** (Tables 3 and 4). Performance of early- 72 120.6 104.2** weaned calves from Anguil and 102 139.5 114.3** 124.4** Chacharramendi herd was similar (P = 132 155.7 162 171.9 138.7** 0.89) from day 10 to day 102 of trial when IA (P by loca- the supplementation was being fed on pas- treatment x location x period interaction was detected < 0.001); therefore treatment means are reported tion and period ture (Table 3). ZSE = Standard error ** Means in rows differ (P < 0.01).

330 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Table 5. Body condition scores (BCS)1 of cows with calves weaned at 70 (EW) or 172 (NW) days of weaning. Gains of early-weaned calves age at 2 locations, Anguil and Chacharramendi. did not improve in this period compared with the previous one (Table 3). This Day after Day of EW NW SE2 group expressed some compensatory calving trial3 growth in the last 30-day period of the ------Anguil------study, but not as great as in Anguil. This 1 -63 5.67 increase did not compensate for the 30 -33 5.53 reduced rate of growth that took place up 63 0 5.42 105 42 5.47 to weaning. Over the study, ADG of early- 135 72 5.58a weaned calves was 54% greater than gain 165 102 5.61a of normally-weaned calves (Table 3). At ------Chacharramendi------the end of trial, early-weaned calves were

1 -63 5.07 33.2 kg heavier (P < 0.01) than normally- 30 -33 4.89 weaned ones (Table 4). 63 0 4.40 Results from our study would indicate 105 42 3.83 that early-weaned calves grazing good 135 72 4.52a quality pastures and supplemented a 165 102 5.05a for 90-day period would grow at a rate similar Body condition score: 1= severely emaciated, 2= emaciated, 3 = very thin, 4= thin, 5 = moderate, 6 = good, 7 = very to normally-weaned calves if cows can ood, 8 = obese, 9 = very obese. SE = Standard error maintain good body condition through lac- 3Day 0 of trial were dates December 4th and 5th for locations Anguil and Chacharramendi, respectively. tation. Moreover, ADG of early-weaned (P Means in rows with different superscript differ < 0.05). calves could be superior to ADG of normally-weaned calves if cows were under Likewise, calf treatment groups from of the annual rainfall) (Roberto et al. nutritional stress. Fernandez and Zuccari Anguil did not differ (P = 0.48) in BW at 1996), Spring 1995 was unusually dry. The (1996), in a similar environment and day 0 or days 42, 72, and 102 (Table 4). accumulated rainfall at the ranch for weaning program, reported no differences On day 102, supplemental feeding of Spring 1995 and Summer 1996 was 58% in weight gain of Angus calves weaned early-weaned calves was concluded and (202 mm) of the 80-year average (345 mm; from first calf heifers at 60 days of age normally-weaned calves were weaned. Roberto et al. 1996). The Chacharramendi and supplemented on pasture during 60 Comparisons across periods within wean- cow herd lost 0.7 units of body condition days, compared with calves weaned at 6 ing treatment indicated that diet switch (from 5.1 ± 0.2 to 4.4 ± 0.56) from calving months of age. Daily gains reported were resulted in reduction (P < 0.01) of gain for to day 0 of trial (Table 5). Early weaning 679 g animal' for early-weaned vs 670 g the following 30-day period in both groups allowed the cows to recuperate body con- animal' for normally-weaned calves. In a (Table 3). Feeding normally-weaned dition during summer. Cows that remained similar study, Fernandez et al. (1997) calves after weaning on alfalfa forage with their calves continued to lose condi- reported ADG of 646 and 644 g animal"' without supplemental feed, had a greater tion up to weaning (day 165 after calving). for early-weaned and normally-weaned depressive effect (P = 0.016) on ADG than Pregnancy results reflected the body condi- calves. supplement elimination for early-weaned tion differences (91% and 44% for cows In Oklahoma, Lusby and Wettemann calves grazed on the same forage. on early weaning and normal weaning (1980) reported similar rates of growth up Normally-weaned calves appeared to treatments, respectively.) to 7 months of age for normally-weaned express compensatory growth the follow- Although not measured directly, a calves and calves weaned at 56 kg body ing 30 days, showing greater (P = 0.015) reduced forage availability and quality weight (6 to 8 weeks of age). Drylot-fed ADG than early-weaned calves. Overall, would have restricted milk production and early-weaned calves gained weight at a performance of early-weaned and normal- calf's feed intake, which would have been rate of 667 g day'. ly-weaned calves was similar (P = 0.168). detrimental for calf growth. Microhisto- In a more restrictive environment than logy on fecal samples from parturition to ours (Province of Rio Negro, Argentina) Chacharramendi the normal weaning time pointed out an and a similar feeding program, Kugler et Weaning treatment groups from increased presence of shrubs in the diet al. (1997a) reported greater ADG for 69- day old early-weaned calves compared Chacharramendi were similar (P = 0.77) in during the late spring and summer months with normally-weaned calves (745 vs 580 average body weight of calves at early (Table 6), which could indicate a sizable animal'). After weaning, the early- weaning time (Table 4). Early-weaned reduction of grass availability. Although g weaned calf group was supplemented dur- calves, however, gained more (P < 0.01) shrubs provided a greater proportion of ing 60 days on pasture. than normally-weaned calves over the crude protein than perennial grasses, In contrast, Sciotti et al. (1996) reported lower ADG for study (Table 3). At normal weaning time lignin content also increased and early-weaned calves supplemented for 30 (day 102 of trial; 172 days of age), early- digestibility decreased. Compared with the days on pasture compared with normally- weaned calves averaged 25 kg heavier than diet of the herd at Anguil, the cow herd at weaned calves (578 vs 635 animal"'). normally-weaned calves (Table 4). The Chacharramendi had a lower quality for- g Working in a more humid ADG of early-weaned calves was almost age available from calving and throughout area of Argentina, Hidalgo et al. (1996) reported twice (P < 0.01) that of normally-weaned the study. better performance of normally-weaned calves during this period (Table 3). Normally-weaned calves from Cha- calves than early-weaned calves (900 vs The area suffered a 5-month drought charramendi suffered a diet adjustment 493 day-'). during spring and summer. Although effect as did normally-weaned calves from g spring is statistically the rainy season (54% Anguil during the 30-day period after

JOURNAL OF RANGE MANAGEMENT 55(4) July 331 Table 6 . Botanical composition!, and quality of forage classes and estimated diet of beef cows at 2 locations.

Chacharramendi Anguil Spring Summer Spring Summer Oct Nov Dec Jan Feb Mar Oct Nov Dec Jan Feb Mar ------Plantclass proportions (%)------CSPGRS2 35 20 10 13 11 8 55 56 54 48 40. 46 WSPGRS 17 26 28 22 27 32 18 24 35 33 35 30 ANGRS 4 3 - - - 3 15 11 7 9 7 12 Shrubs 38 46 55 61 58 54 10 7 3 7 11 9 Forbs 6 5 7 4 4 3 2 3 1 3 7 3

------Crudeprotein(%)------CSPGRS 11 7 6 5 5 6 15 13 11 8 8 9 WSPGRS 10 11 9 8 8 7 10 12 12 9 9 9 ANGRS 12 11 - - - 10 14 15 13 11 12 11 Shrubs 16 14 11 10 11 10 15 16 14 12 12 13 Forbs 16 17 13 13 12 16 18 16 14 12 13 15

------Lignin(%)------CSPGRS 9 12 14 11 15 15 8 8 10 11 12 12 WSPGRS 8 8 10 11 13 14 7 9 11 10 13 10 ANGRS 5 4 - - - 6 3 4 5 6 5 4 Shrubs 13 15 16 18 17 19 12 15 14 15 14 17 Forbs 3 5 6 7 7 9 2 4 4 5 6 3

CSPGRS 58 57 45 42 37 44 64 59 58 52 53 55 WSPGRS 59 59 52 50 48 48 64 63 63 58 57 55 ANGRS 68 65 - - - 62 70 65 68 62 64 65 Shrubs 61 55 45 39 41 44 63 56 55 54 52 54 Forbs 62 53 54 50 47 43 69 67 62 57 58 60 ------Qualityof diets(%)------Crude protein 13 12 10 9 10 9 14 13 12 9 9 10 IVDMD 60 57 47 42 43 46 65 61 60 55 55 56 Estimated by microhistology of fecal samples 3CSPGRS = Cool-season perennial grasses; WSPGRS = Warm-season perennial grasses; ANGRS = Annual grasses Kjeldahl N x 6.25 51n vitro dry matter digestibility Calculated from class proportion and quality data

Experiment 2 (P > 0.43) over the pre-weaning period Moreover, the youngest group had greater Available biomass in alfalfa pastures (accumulated periods 1, 2, and 3). (P < 0.05) ADG, compared with the other was greater than 2200 kg DM ha"1 Before normal weaning time, ADG was 2, and the greatest ADG of the study, like- throughout the 148-day study period and usually greater (P < 0.01) in normally- ly expressing compensatory growth. calf intake should not have been restricted. weaned calves compared to early-weaned Likewise, ADG of these calves was Table 2 shows nutrient composition of calves, within each period and age group greater (P < 0.01) than ADG of early- feeds and forage used in the study. The (Table 8). Differences became larger as weaned calves within all age groups in this normally-weaned calves were heaviest (P days progressed and early-weaned calves period. Over the study, no differences (P = < 0.01) at the end of trial, compared with were deprived of supplemental feed (peri- 0.58) in ADG were detected between age early-weaned ones, within age groups ods 1, 2, and 3). During the 30-day period groups for normally-weaned calves. (Table 7), and differences increased as age after weaning (period 4), normally-weaned Age at early weaning date (day 0 of at weaning decreased. Calves from the calves showed lower (P < 0.01) ADG, trial) correlated highly with body weight (r normal weaning treatment were 15 and 13 compared with the previous period. In this = 0.97). The youngest calves were 16 days kg heavier in the oldest group (AGE 1), 18 period, the youngest group of normally- younger and 15 kg lighter than those in the and 15 kg in the intermediate (AGE2), and weaned calves had lower (P < 0.05) ADG intermediate group. In turn, the intermedi- 31 and 23 kg in the youngest (AGE3), than the oldest and the intermediate age ate group was 18 days younger and 20 kg compared with the 15-day supplemented group, possibly showing a detrimental lighter than the oldest. Body weight differ- and 45-day supplemented early-weaned effect of weaning 162-day old calves onto ences among age groups noted at early calves, respectively. Within the normal a 100% pasture diet. Average daily gain of weaning time remained significant (P < weaning treatment, a decrease in ADG normally-weaned calves within age group 0.05) for each sampling period and across with decreasing age was detected (P < in period 4 did not differ (P > 0.28) from the study. Within the supplementation 0.05) in period 1 (Table 8). Gains, howev- gain of 15-day supplemented and 45-day treatments of early-weaned calves, the er, did not differ (P > 0.28) between age supplemented early-weaned calves. heaviest calves at early weaning time were groups in periods 2 and 3. Despite the Thirty days later (period 5), normally- also the heaviest at the end of trial (P < cited initial effect, ADG of all age groups weaned calves showed increased (P < 0.05). The shortest supplementation period of normally-weaned calves did not differ 0.05) ADG compared with earlier periods. magnified ADG differences due to age.

332 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Table 7. Weight of Angus calves early weaned (EW) at 3 different ages and supplemented on pas- weight differences, however, between sup- ture for 15 (S15) or 45 (S45) days, compared with weight of calves of corresponding ages at early plementation treatments were not detected weaning time and normally weaned (NW)1'2. yet (P > 0.34) for the oldest calves. Forty five-day supplemented early-weaned Age at Days of EW calves calves calves 5 (P early weaning3 trial S15 were and 8 kg heavier < 0.05) than 15-day supplemented early-weaned Age at initiation of the study(days)------calves in the intermediate and youngest AGE1 0 108 AGE2 0 92 groups, respectively (Table 7). AGE3 0 78 After 30 days on forage diet (period 3), SEs 1.8 3.1 the oldest and intermediate age groups of 15-day supplemented early-weaned calves AGE1 0 115 maintained (P > 0.24) ADG compared AGE2 0 94 with period 2. The youngest group AGE3 0 79 (AGE3) increased (P < 0.05) ADG during SE 1.8 2.0 this period. In the same period, supple- AGE1 27 132 mental feeding of calves in the 45-day AGE2 27 110 AGE3 27 94 supplementation treatment (S45) was SE 1.8 2.2 stopped, and ADG of this treatment 147a AGE1 57 decreased (P < 0.05) compared with peri- 123a AGE2 57 od 2. The switch 104a of diets may have been AGE3 57 the main factor responsible for this ADG SE 1.8 2.3 162a depression. During this period, ADG AGE1 87 of 137a AGE2 87 15-day supplemented early-weaned calves 116a AGE3 87 from the oldest and intermediate age SE 1.8 2.4 groups were greater (P < 0.05) than ADG 178a AGE1 117 of 45-day supplemented 153a calves (Table 8). AGE2 117 129a On day 87, body weights for calves from AGE3 117 the 15 and 45-day supplementation treat- SE 2.0 2.6 194a ments within the oldest and the intermedi- AGE 1 148 170a AGE2 148 ate age groups, did not differ (P > 0.17), 143a b AGE3 148 but weights between supplementation SE 2.2 2.8 treatments were different (P < 0.05) for 1Calves of the NW treatment were kept with the dams until day 87 of trial; NW calves reached the ages 196, 178, and the youngest group. 162 days, for AGE1, AGE2 and AGE3, respectively. In period 4, ADG of 45-day supple- 2A weaning treatment x age group x period interaction was detected (P < 0.001); therefore, interaction means are reported. mented early-weaned calves of all age 3Age groups at early weaning date. AGE1=111 ± 2.2 days of age; AGE2 = 91 ± 1.6 days of age, and AGE3 = 75 ± 3.8 groups and 15-day supplemented early- days of age. weaned calves from the intermediate and SE = Standard error for diet effect. the youngest groups increased (P < 0.05), SSE = Standard error for age group effect. Weight differences between age groups within period and diet treatment are all significant (P < 0.05). compared to gains achieved in period 3. abcMeans in rows with different superscripts differ (P < 0.05). Both supplementation treatments (S 15 and S45) had similar (P > 0.15) ADG within Feeding supplement for only 15 days ADG were not expected because diets were age groups during this period. Similarly, a detrimentally affected the youngest group the same for all early-weaned calves dur- month later (period 5) no differences were at early weaning. After 148 days, the 15- ing the pen feeding period and the first 15 detected (P > 0.43) between supplementa- day and 45-day supplemented calves from days on pasture. After day 15 on pasture, tion treatments of early-weaned calves the oldest early-weaned group (AGE 1) early-weaned calves of the 15-day supple- within age group. Over the 148-day peri- were 23 and 24 kg heavier (P < 0.05) than mentation treatment received no supple- od, no differences in BW were detected (P calves from corresponding groups of the mental feed. During the following period, > 0.22) between supplementation treat- intermediate age at early weaning (period 2) all age groups of this treatment ments for the oldest and the intermediate (AGE2). No supplementation effect was (S15 calves) showed lower (P < 0.05) age groups (Table 7). Differences, howev- detected (P > 0.34) within these 2 age ADG compared with period 1. Group er, were evident in the youngest group (P groups. Differences, however, between the AGE3 x S 1 S reached the lowest rate of <0.01). oldest (AGE 1) and youngest (AGE3) gain of this trial in this period. The switch Overall ADG increased (P < 0.05) with calves were greater (P < 0.05) in the 15- of diet to 100% forage may explain this weaning age within the 15-day supple- day than in the 45-day supplementation effect. During the same period, 45-day mentation treatment for early-weaned treatment (52 and 45 kg, respectively). supplemented early-weaned calves (S45) calves. A similar trend was noted for the Body weight and ADG did not differ (P maintained similar (P > 0.25) ADG to 45-day supplementation treatment, > 0.26) between supplementation treatments period 1. This response would be expected although no differences in overall ADG imposed on early-weaned calves within because diet did not change in this treat- were detected (P = 0.16) between the old- each age group for period 1. For early- ment. Consequently, ADG of 15-day sup- est and the intermediate (111 and 91 days weaned calves, period 1 consisted of 12 plemented early-weaned calves were of age at weaning) groups in this treat- days on pen-fed diet plus 15 days on pasture lower (P < 0.05) than gains of 45-day sup- ment. Gain increased (P < 0.05) also with with supplemental feed. Differences in plemented early-weaned calves. Body length of supplementation period within

JOURNAL OF RANGE MANAGEMENT 55(4) July 333 Table 8. Average daily gain (ADG, g animal') of Angus calves early weaned (EW) at 3 different most affected under feeding programs that ages and supplemented on pasture for 15 (S15) or 45 (S45) days, compared with ADG of calves use a supplementation period on pasture normally weaned (NW)l. shorter than 45 days. Shortening of supplementation period Age at EW calves NW calves on pasture to 15 days after weaning can early weaning3 S l5 S45 have negative effects on performance. The (g animal') effect, however, would be less dramatic if ------Periodl6------calves are 90-day old or older at weaning 647aAi 641aAW AGE1 time. If calves are younger, supplementa- AGE2 590aBv 597aBW 552aCi` 550aCv tion during at least 45 days may be neces- AGES of SE5 10.3 9.9 sary to achieve an acceptable rate growth. More research is needed to devel- AGE! 485aAe 638bAijj cAer op improved options that combine ade- 449aBe AGE2 594bBW quate calf growth and reduced costs with- 352aCa 552bCij` cAq AGE3 out compromising future calf perfor- SE 6.9 8.1 mance. Questions such as calf age and AGE! 526aAO? 458bAe weight, forage quality, and supplement 441bAe cAe AGE2 463aBe level interactions remain to be addressed 404aBe AGE3 398aCX in future research on early weaning. SE 11.2 6.6

517aAX AGE1 520aAX 524aAX 517aA?. Cited AGE2 Literature 453aBX AGE3 431aBy SE 10.5 11.6 AOAC. 1990. Official methods of Analysis (15`h 532aAX Ed.). Assoc. of Official Analytical Chem., AGEI 541 aAA. 535aAX Washington, DC. AGE2 529 aAX Fernandez, G.D. and A.E. Zucari. 1996. 440aBy 444 AGE3 aBX6 Efecto del destete precoz sobre la perfor- SE 13.1 19.9 de Overall ------mance reproductiva en vaquillonas 540aA 557bA Rev. Arg. Prod. Anim. AGE1 primera paricion. 534bA AGE2 511aB 16(1):49. 432aC AGE3 479bB Fernandez, G.D., A.E. Zucari, and E.L. SE 6.7 8.7 Anton. 1997. Efecto de la edad del destete ------SE for periodcomparisons7 ------sobre la performance de vacas de refugo y AGE1 12.5 14.0 ternero. Rev. Arg. Prod. Anim. 17(1):279. AGE2 6.8 11.9 Fluharty, F.L., S.C. Loerch, T.B. Turner, AGE3 11.6 10.3 S.J. Moeller, and G.D. Lowe. 2000. Effects

of weaning age and diet on growth and car- C laC vesof the NW treatment were kept with the dams until day 87 of trial. Normally weaned calves were 196,178, and 162 days of age for groups AGE!, AGE2 and AGE3, respectively, at weaning. cass characteristics in steers. J. Anim. Sci. 3A weaning treatment x age group x period interaction was detected (P < 0.001); therefore, interactive means are reported. 78:1759-1767. Age groups at early-weaning date (day 0 of trial): AGE1=111 ± 2.2 days of age; AGE2 = 91 ± 1.6 days of age, and Gill, D.R., F. N. Owens, M.C. King, and AGE3 = 75 ± 3.8 days of age. H.G. Dolezal. 1993. Body composition of 4SE Standard error for diet effect. = grazing or feedlot steers differing in age and 5SE = Standard error for age group effect. Period 1= ADG between day 0 and 27 of trial, Period 2 = between day 27 and 57; Period 3 = between day 57 and 87; background. Okla. Agr. Exp. Sta. Res. Rep. Period 4 = between day 87 and 117; Period 5 = between day 117 and 148. pp. 993:185-190. 7SE = Standard error for period effect within age group and diet. Goering, H.K. and P.J. Van Soest. 1970. means with different lowercase superscripts differ (P < 0.05). fiber analyses (apparatus, reagents, ABRow (P Forage Column means within period with different uppercase superscripts differ < 0.05). Agr. yq'gColumn means within age group across periods with different superscripts differ (P < 0.05). procedures, and some applications). Handb. 379. ARS, USDA, Washington, D.C. Hidalgo, L.G., S. Callejas, M.A. Cauhepe, and across weaning ages. Within early Conclusions and M.J. Otero. 1996. Efecto del destete weaning treatments, calves weaned at the precoz sobre la condicion corporal y la youngest age (75 days of age) were the prenez en vacas multiparas. Rev. Arg. Prod. Our study provides evidence that early group most affected by the diets imposed, Anim. 16(1):36. weaning at 2 to 3 months of age, with and those in this age group receiving sup- Hofer, C.C., J.J. Bruno, and A.R. Monje. proper feeding, is a feasible technique to 1984. Comportamiento de terneros desteta- plement for only 15 days experienced the reduce herd requirements without affect- dos a los 60 dias de edad. Manejo del destete lowest overall rate of growth. ing calf performance. Moreover, in cases metdo de crianza. EEA INTA C. Del on compared performance of y Reports of feed restrictions for the cows and Uruguay. Prod. Anim. Inf. Tec. 1:126-132. early-weaned calves exposed to limited it could favor not only the cow, but Holechek, J.L. and B.D. Gross. 1982. supplementation programs are scarce. calves, also the calf. Evaluation of diet selection in ruminants. Early weaning on a 100% forage diet com- Low-cost feeding programs can success- Functional Ecol. 2:15-22. pared to a 1.1 kg day l animal' of supple- Holechek, J.L., M. Vavra, and R. D. Pieper. fully be implemented, however, calf mental feed was reported by Simeone et 1982. Botanical determination of range her- requirements should not be underestimat- al. (1997). Supplemented early-weaned bivore diets: a review. J. Range Manage. ed. Results from this study indicate that 70 35:309-315. calves gained 310 day' more than non- g to 80 day-old weaned calves would be the supplemented ones.

334 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Kugler, N.M., H. Giorgetti, G. Cecchi, G. Monje, A.R., C.C. Hofer, and I.O. Galli. Roberto, Z.E., G. Casagrande, and E.F. Rodriguez, y 0. Montenegro. 1997a. 1978. Destete precoz y creep-feeding. Efecto Viglizzo, 1996. Lluvias en la pampa central. Modulos de cria: Destete precoz vs conven- sobre la receptividad. AAPA Prod. Anim. Tendencias y variaciones del siglo. Cambio cional. Jornadas de cria en campos de monte. 6:391-396. climatico y agricultura sustentable en la region EEA Valle Inferior de Rio Negro, IDEVI- Monje, A.R., C.C. Hofer, and I.O. Galli. pampeana. Ed. INTA PEI 2:3-25. INTA p. 52-53. 1993. Destete precoz: Efecto sobre los vien- Robertson, J.B. and P.J. Van Soest. 1981. Kugler, N.M., H. Giorgetti, G. Cecchi, G. The detergent system analysis and its tres, manejo de los terneros e impacto de la appli- Rodriguez, y 0. Montenegro. 1997b. cations to human foods: In: W.P.T. James tecnica sobre los sistemas Destete precoz e incremento en la carga ani- de produccion. and 0. Theander (Ed.) The Analysis of de mal. 1. Efecto sobre los vientres. Rev. Arg. Memorias. Jornada difusion tecnica: Dietary Fiber. pp 123-158. Marcell Dekker, Prod. Anim. 17(1):269. Destete precoz en cria vacuna. EEA INTA C. New York. Laster, D.B., H.A. Climp, and K.E. Gregory. Del Uruguay. pp.13-38. SAS. 1991. Statistics Analysis System User's 1973. Effects of early weaning on paspartum Myers, S.E., D.B. Faulkner, F.A. Ireland, Guide. SAS Institute Inc. Cary. N.C. reproduction of cows. J. Anim. Sci. and, D.F. Parrett. 1999a. Comparison of Sciotti, A.E., J. Carrillo, L.M. Melucci and 36:734-740. three weaning ages on cow-calf performance A. Cano.1996. Efecto del destete precoz en Lusby, K.S. and A.A. Parra.1981. Effects of and steer carcass traits. J. Anim. Sci. vacas primiparas y de ultima paricion sobre early weaning on calf performance and on 77:323-329. los pesos ganancias de peso de los terneros reproduction in mature cows. Okla. Agr. y Myers, S.E., D.B. Faulkner, F.A. Ireland, y sus madres. Rev. Arg. Prod. Anim. Exp. Sta. Rep. MP 108:64-68. 16(1):30. Lusby, K.S. and R.P. Wettemann. 1980. L.L. Berger, and D.F. Parrett. 1999b. Production systems comparing early weaning Simeone, A., A.I. Trujillo, G. Cordoba, J. Effects of early weaning calves from first Gil, M. Rodriguez, A. Bejerez, R. on heifer performance. to normal weaning with or without creep calf heifers calf and Zanoniani, F. Baccino, and M. Umpierrez. Okla. Agr. Exp. Sta. Res. Rep. MP feeding for beef steers. J. Anim. Sci. 1997. Efecto del destete precoz de dos sis- 107:55-58. 77:300-310. y Lusby, K.S., R.P. Wettemann, and E.J. temas de alimentacion post-destete sobre la National Research Council. 1996. Nutrient ganancia de peso de terneros Hereford hasta Turman. 1981. Effects of early weaning requirements of beef cattle. Seventh Revised calves from first-calf heifers on calf and los 15 meses de edad. Rev. Arg. Prod. Anim. Ed. Washington D.C.: National Academy 17(1):58. performance. J. Anim. Sci. 53:1193- heifer Press. 1197. Sparks, D.R. And J.C. Malechek. 1968. T.B. K.M. M. Lusby, K.S., D.R. Gill, D.M. Anderson, T.L. Peterson, O.W., Turner, Irvin, Estimating percentage dry weight in diets Gardner, and H.G. Dolezal. 1990. Limit E. Davis, H. W. Newland, and W. R. using a microscopic technique. J. Range feeding vs full feeding high concentrate diets Harvey. 1987. Cow and calf performance Manage. 21:264-265. to early-weaned calves- effects on perfor- and economic considerations of early wean- Tilley, J.M.A. and R.A. Terry. 1963. A two- mance to slaughter. Okla. Agr. Exp. Sta. Res. ing of fall-born beef calves. J. Anim. Sci. stage technique for in vitro digestion of for- Rep. MP 129:128-134. 64:15-22. age crops. J. Brit. Grassl. Soc. 18:104.

JOURNAL OF RANGE MANAGEMENT 55(4) July 335 J. Range Manage. 55: 336-340 July 2002 Density and reproductive success of Florida grasshopper sparrows following fire

MICHAEL F. DELANY, STEPHEN B. LINDA, BILL PRANTY, AND DUSTIN W. PERKINS

Authors are Wildlife Biologist and Biometrician, Florida Fish and Wildlife Conservation Commission, Wildlife Research Laboratory, Gainesville Fla. 32601; Avian Project Leader, Audubon of Florida, Tampa, Fla. 33619; and Ph.D. candidate, University ofMaine, Orono, Me. 04684.

Abstract Resumen

Information on the response of the endangered Florida Se necesita informacion de la respuesta de la especie en peligro grasshopper sparrow (Ammodramus savannarum floridanus de extincion "Florida grasshopper sparrow" (Ammodramus Mearns) to range management, especially prescribed fire, is savannarum floridanus Mearns) al manejo del pastizal, especial- needed to determine conservation strategies. Intensive manage- mente al fuego prescrito, para determinar estrategias de conser- ment of grasslands for cattle grazing and conversion of grassland vacion. El manejo intensivo de los pastizales para apacentamien- to other agricultural use is considered the greatest threat to the to de ganado y la conversion del pastizal a otros usos agricolas se sparrow. Territory spot-mapping and estimates of reproductive consideran como la mayor amenaza para el "Sparrow". De 1997 success were examined in relation to time post-burn in managed a 1999 el maneo de manchas territoriales y estimaciones de exito cattle pastures at Avon Park Air Force Range, Highlands reproductivo se examinaron en relacion al tiempo despues de la County, Florida from 1997-1999. We tested the hypothesis that quema, to cual se realizo en potreros manejados para ganado en sparrow density and reproductive success did not depend on time la Base Aerea Avon Park, en el condado Highlands, Florida. following fire. Contrary to previous work, there was no evidence Probamos la hipotesis de que la densidad de "Sparrow" y el that Florida grasshopper sparrow territory density depended on exito reproductivo no depende del tiempo despues de la quema. years post-burn (P = 0.842). The probability of reproductive suc- Contrario a trabajos previos, no hubo evidencia de que la densi- cess was significantly higher 0.5 year post-burn than at 1.5 years dad territorial del "Florida grasshopper sparrow" dependa de post-burn (P < 0.05) and 2.5 years post-burn (P < 0.01). No other los anos despues de la quema (P = 0.842). La probabilidad de significant differences were observed among years post-burn (P exito reproductivo fue significativamente mayor 0.5 anos que 1.5 > 0.27 for each pairwise comparison between years). No simple anos despues de la quema (P < 0.05) y que 2.5 anos despues del trend or highly significant polynomial relationship between fuego (P < 0.01). No se observaron otras diferencias significativas reproductive success and territory density was indicated (P > entre anos despues de la quema (P > 0.27 para cada comparacion 0.41). Compared to other subspecies, Florida grasshopper spar- de pares entre anos). No se detecto (P > 0.41) una tendencia sim- rows exhibited relatively low density (0.22 territories/ha or less) ple o una tendencia polynomial altamente significativa entre el and reproductive success (20%). Our results suggest increased exito reproductivo y la densidad territorial. Comparado con reproductive success at a population level 0.5 year following fire, otras subespecies, el "Florida grasshopper sparrows"mostro una and did not suggest an association between territory density and densidad territorial relativamente baja (0.22 territorios/ha o individual reproductive success. Additional information is need- menos) y un bajo exito reproductivo (20%). Nuestros resultados ed on the effects of seasonality of fire on Florida grasshopper sugieren un mayor exito reproductivo, a nivel de poblacion, 0.5 sparrows. anos despues del fuego y no sugieren una asociacion entre la densidad territorial y el exito reproductivo individual. Se necesita informacion adicional sobre los efectos de la estacionalidad del Key Words: Ammodramus savannarum floridanus, Florida fuego en el "Florida grasshopper sparrows". prairie, prescribed fire

Declines in the abundance of some grassland birds in North south-central, dry prairie region of the state and was classified as America are attributed to habitat loss induced by changes in land endangered because of its restricted distribution, loss of habitat, management (Peterjohn and Sauer 1999). Grasshopper sparrows and population decline (U.S. Fish and Wildlife Service 1988). (Ammodramus savannarum Gmelin) are locally distributed in low Breeding aggregations are known from only 6 locations and fewer to medium height grasslands throughout most of temperate North than 1,000 birds may exist (Delany et al. 1999). Information on America, from Mexico to Ecuador, and the West Indies. Breeding the sparrow's response to range management is needed before bird surveys evince a decline in some populations (Vickery 1996). conservation strategies for the subspecies can be fully assessed The Florida subspecies (A. s. floridanus Mearns) is endemic to the (U.S. Fish and Wildlife Service 1988). An examination of structural characteristics of occupied and abandoned territories indicated that Florida grasshopper sparrows This study was supported by Cooperative Agreement No. 14-45-0009-154, Research work order 175, funded by the U.S. Air Force through the Florida cannot adapt to habitat perturbations that remove potential nest Cooperative Fish and Wildlife Research Unit. sites (bunch grasses and dwarfed shrubs) and bare-ground forag- Manuscript accepted 24 Oct 01. ing areas (Delany and Linda 1994). Intensive management of

336 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 grasslands for cattle grazing and conver- Rupr.), bluestems (Andropogon spp.), yel- veys. Observations of sparrow locations sion of grassland to other agricultural use low-eyed grass (Xyris spp.), saw palmetto and reproductive success were recorded in is the greatest threat to the sparrow (U.S. (Serenoa repens W. Bartram, Small), reference to the grid markers. Surveys Fish and Wildlife Service 1988). dwarf live-oak (Quercus minima Sarg.), were conducted between sunrise and 1000 Improved pastures are created and main- and shrubs; and included scattered cypress hours in the absence of rain and at wind tained by mechanical clearing and planting (Taxodium sp.) domes and small (4 ha or velocities less than 11 km/hour. Plots were bahia grass (Paspalum sp.), pangola grass less) hypericum (Hypericum spp.) ponds. sampled 8-13 times each year at about 10 (Digitaria sp.), American joint vetch The study area was bordered by slash pine day intervals from 22 March-25 August, (Aeshynomene americanus L.), and clover (Pinus elliottii Engelm.) plantations, open- 1997-1999. Based on maximum territory (Trifolium spp.) (Milleson et al. 1980, canopy longleaf pine (P. palustris Mill.), size, male sparrows observed on different Delany and Linda 1994). Prairies also are introduced pastures (plowed and planted days more than 203 m apart were consid- plowed and planted with bahia grass for with non-native grasses), and freshwater ered separate individuals (Walsh et al. sod production. Shriver and Vickery marsh. Cattle grazed the study area at 1 1995), and those exhibiting breeding (1999) estimated only 19% (156,000 ha) animal per 8.7-28.3 ha with grazing peri- behavior for at least 4 weeks were consid- of the original dry prairie remained in cen- ods of 21 days or less followed by longer ered territorial (Vickery et al. 1992a). tral peninsular Florida. periods of exclusion. Forage utilization Locations of male sparrows were digitized The Florida grasshopper sparrow has during a grazing period was light. Four and territories mapped with ArcView GIS benefitted from prescribed burning at prairie pastures (165-324 ha) were burned software (ESRI 1990). Density was con- some locations. Prescribed fire at 2-3 year with head fires (burned with the wind) on sidered the number of territories per ha intervals during the winter dormant season 3-year rotations between December and and included only those that were at least (December-March) is the most common mid-March to enhance forage production 50% within a plot. A territorial male was practice used to improve prairie pastures and manage habitat for Florida grasshop- deemed to have reproduced successfully if for cattle grazing in Florida (Lewis 1964, per sparrows. Ordnance-ignited wildfires fledglings were observed or adults carry- Abrahamson and Hartnett 1990). Frequent occur year-round. ing food for at least 9 days were observed burning also maintains vegetation in a (according to Vickery et al. 1992a). The sparse, early successional stage preferred detectability of sparrows on the ground by the sparrow (Delany et al. 1985). Methods was probably related to time post-burn and Increased densities of Florida grasshopper corresponding differences in vegetation sparrows associated with time post-burn Sparrow Density and Reproductive density, however, adults carrying food (Walsh et al. 1995) may be related to Success perched on shrubs or fences and were easi- changes in prairie physiognomy following ly observed in any post-burn stage. The density of Florida grasshopper spar- Time fire and have important management post-burn was recorded for each row territories was determined using stan- plot and implications. However, population density dard spot-mapping techniques (Inter- categorized into 0.5, 1.0, 1.5, or 2.5 year estimates without demographic data may classes (Table 1). national Bird Census Committee 1970) on not be a reliable indicator of habitat quali- 12 study plots (8.5 to 20.0 ha, x =12.8 ha) ty (Van Home 1983). We examined the density and reproductive success of at least 50 m from non-prairie habitats and Data Analysis separated by at least 100 m. Plots were the For territory density, the experimental Florida grasshopper sparrows in relation to same as those used by Shriven and Vickery design was considered to be an incomplete time post-bum to assess the effects of pre- (2001). Plots were marked with 50 m grids blocks design, in which the blocks were scribed burning, and tested the hypothesis of labeled wired flags. Spot-mapping peri- plots and the treatment was years post- that territory density and reproductive suc- ods of about 1 hour per plot were made by burn. A search for the Box-Cox variance cess did not depend on time following fire. walking to each grid marker, with direc- stabilizing power transformation indicated tion alternated between successive sur- that the square-root transformation Study Area Table 1. Experimental design used to examine Florida grasshopper sparrow territory density and reproductive success following fire, Avon Park Air Force Range, Highlands County Fla., The study area was a 700-ha prairie 1997-1999. The year in which a given years post-burn treatment occurred in a given plot is indi- (27°38.08'N, 81°19.30'W) on the U.S. Air cated. Force Avon Park Range in Highlands County, Fla. The climate is humid sub- Years post-burn tropical (Chen and Gerber 1990). Mean Plot number 0.5 1.01 annual temperature is 22°C with monthly 1 1998 1997, 1999 mean temperatures ranging from 17°C in 2 1999 1997 January to 28°C in August. Mean annual 3 1997 1999 rainfall is 1,370 mm, with most (60%) pre- 4 1997 1998 cipitation associated with convective thun- 5 1997 1998 derstorms and tropical systems occurring 6 1999 1997 7 1997 1999 from June to September (Carter et al. 1989). The topography is nearly level and 9 1999 the sandy soils (Malabar and Oldsmar series) are poorly drained (Carter et al. 1997 1998 1989). The grassland was dominated by 1998 1997, 1999 - wiregrass (Aristida beyrichiana Tnin. and 1Bumed during an ordnance-ignited wildfire on 29 June 1998.

JOURNAL OF RANGE MANAGEMENT 55(4) July 337 induced homoscedasticity. Therefore, it Table 2. Predicted values of Florida grasshopper sparrow territories per ha at years post-burn was assumed that the number of territories (back-transformed scale), Avon Park Air Force Range, Fla., 1997-1999. followed a Poisson distribution, and gen- eralized linear mixed model (GLMM) P-value for Ho: pair of years-post burn CI limits means are equal methodology, as can be implemented in Years 95% post-burn Mean SE' 1.5 2.5 the NLMIXED procedure (SAS Inst., Inc. 1999), was used. The Poisson error distri- 0.5 0.20 0.050 1.0 0.22 0.099 bution and log link were specified for the 1.5 0.17 0.043 number of territories per plot, and the off- 2.5 0.16 0.053 set was specified as the log of plot size. 1Obtained by application of the delta method (Seber 1982). The GLMM methodology, implemented in PROC NLMIXED in the SAS System (SAS Inst. Inc. 1999), also was used to (Walsh et al. 1995, Shriver and Vickery gible in the fit of the GLMM for reproduc- (P analyze reproductive success. A logit link 2001). Walsh et al. (1995) speculated that tive success = 1.00). Based on the sim- and binomial error distribution were speci- density changed because young of the year ple GLM, Florida grasshopper sparrow fied for the number of territories with suc- birds dispersed to more suitable, recently reproductive success depended on years (P cessful males for a given plot. Fixed years burned locations. Increased shrub and saw post-burn = 0.0036). The probability of post-bum effects and random plot-specific palmetto growth with the exclusion of fire reproductive success was higher 0.5 year (P effects were included in the linear predic- may eventually allow vegetation to reach a post-burn than at 1.5 years post-bum < tor. If random plot-specific effects dense successional stage unusable by 0.05) and 2.5 years post-burn (P < 0.01). appeared negligible, then the simple GLM Florida grasshopper sparrows (Delany et No other pairwise comparisons between (P was fitted in PROC GENMOD (SAS Inst. al. 1985). Florida grasshopper sparrow years post-bum were significant > 0.27 3, Inc.1999). If the global test for years post- density (0.22 territories/ha or less) was for each pairwise comparison) (Table burn was significant then tests of differ- lower than most ranges reported for other Fig. 2). Shriver and Vickery (2001) also ences in counts between pairs of years subspecies (0.21-1.30 territories/ha, in found increased reproductive success for post-burn levels were performed; if ran- grasslands) (reviewed in Vickery 1996). Florida grasshopper sparrows on our study dom plot effects appeared negligible, then The sparrow is at the edge of the species' area in plots 0.5 year following fire. Fishers's exact test was performed for range and may therefore exhibit relatively Johnson and Temple (1986) reported higher each pair of years post-bum levels. low abundance (Curnutt et al. 1996). reproductive success for grasshopper spar- To graphically examine the possibly The plot variance component was negli- rows (A. s. pratensis Vieillot) in recently nonlinear relationship between reproductive success and territory density, residuals from the analysis of reproductive success 0 were plotted against the residuals from the n=12 n=3 n=15 n=6 analysis of density, and a nonparametric fitted. To L( smoothed curve (LOESS) was 0 statistically test for a relationship between 0 reproductive success and density, hierar- T chical fixed effects terms for a quintic trend in density were added to the linear .c: oCo predictor in the GLM for reproductive ci) success, and Type 1 hypothesis tests of the 0 M O fixed effects were performed. a + + N N Results and Discussion co 0ai We found no evidence that Florida grasshopper sparrow territory density depended on years post-burn up to 2.5 years (X2 = 0.830, DF = 3, P = 0.842 for a Wald test of the years post-burn effect) (Table 2, Fig. 1). Anecdotal accounts indicate higher densities of Florida grasshop- 0.5 1 1.5 2.5 per sparrows at recently burned (within Years post - burn 2.5 years) locations (Smith 1968, Delany (n and Cox 1986). Studies that have quanti- Fig. 1. Boxplots of Florida grasshopper sparrow density = number of plots), Avon Park fied habitat selection in relation to time Air Force Range, Highlands County, Fla., 1997-1999. Study plot and year, and thus possible correlation among observations, were ignored. The bottom and top edges of the box post-burn have found Florida grasshopper plots are located at the sample 25th and 75th percentiles; the center horizontal line is locat- sparrow densities ranged from 0.05-0.75 ed at the sample median; the plus symbol is the sample mean; the vertical lines extend to territories/ha during the first year follow- the range of the data, to a distance of at most 1.5 interquartile ranges from the box; and ing fire, and decreased to 0.01-0.18 terri- values more extreme than 1.5 interquartile ranges from the box are denoted by circles. tories/ha during the subsequent 1.5 years Widths of boxes are proportional to the square-root of the number of observations.

338 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Table 3. Proportion of Florida grasshopper sparrow territories with reproductive success at years density terms in the model) versus residu- post-burn, Avon Park Air Force Range, Fla., 1997-1999. als from the analysis of density indicated oscillation about zero (Fig. 3). Therefore, Years 95% CI limits P-value for pairwise Fisher's exact test no simple trend or highly significant poly- post-burn n Estimate 1.0 nomial relationship between an individ- 0.5 32 0.38 0.278 ual's reproductive success and territory 1.0 4 0.00 density was indicated. Our results support 1.5 35 0.14 contentions that population density may 2.5 13 0.00 not be a reliable indicator of habitat quality (Van Home 1983), or the reproductive burned (1 year post-burn) grasslands in than nest diameter (about 12.3 cm) within success of emberizine sparrows (Vickery Minnesota. The overall reproductive suc- more open patches (at least 4 m in diame- et al. 1992b). cess during our study was only 20% (17 of ter) were important factors influencing 84 males successfully reproduced). Using Florida grasshopper sparrow nest site the same index, Delisle and Savidge (1996) selection on the study area (Delany and Conclusions and Management estimated 52% reproductive success for Linda 1998a). Changes in vegetation den- Implications grasshopper sparrows in Nebraska, and sity and bare-ground cover among nest Vickery et. al. (1992a) estimated 40-50% sites corresponded with an increase in Our study encompassed the entire burn reproductive success in Maine. plant material with time following fire cycle at this location and results suggest Higher reproductive success in pastures (Delany and Linda 1998b). Optimal com- increased reproductive success of Florida 0.5 year post-burn may be related to struc- binations of vegetation density for nest grasshopper sparrows at a population level tural characteristics of potential nest sites shielding and bare-ground cover for nest 0.5 year following fire. However, our available for this ground-nesting bird. access, predator distraction displays, and results should be viewed with caution. Features of the nest site may have impor- foraging may occur 0.5 year after fire. Data were obtained from only one popula- tant implications for reproductive success Type 1 hypothesis tests of linear, qua- tion, and there was some confounding by providing concealment from predators dratic, cubic, and quintic terms in density between calendar year and years post- (Martin 1993, but see Vickery et al. in a GLM for reproductive success were burn. The 2.5 years post-burn treatment 1992c) and ameliorating the microclimate nonsignificant (P > 0.41 for each term). occurred only in 1998 and 1999, and the at nests (With and Webb 1993). Small Inspection of the plot of residuals from the 1.0 years post-burn treatment occurred clumps of dense vegetation slightly larger analysis of reproductive success (without only in 1999. Also, the sample size for the

N O

> O U O

_-J- -0.10 -0.05 0.00 0.05 0.10 0.15 Years post-burn Density residual

Fig. 2. Proportion of Florida grasshopper sparrow territories with Fig. 3. Plot of residuals from the analysis of Florida grasshopper reproductive success at years post-burn, Avon Park Air Force sparrow reproductive success versus residuals from the analysis of Range, Highlands County, Fla., 1997-1999. Plus symbols indicate territory density, Avon Park Air Force Range, Highlands County, empirical proportions, grey bars indicate exact 95% confidence Fla., 1997-1999. Curve indicates nonparametric smooth fit intervals for proportions, and curved line indicates the fit of a (LOESS, with span = 3/4). smoothed curve (the cubic smoothing spline) to the empirical proportions.

JOURNAL OF RANGE MANAGEMENT 55(4) July 339 1.0 year post-burn treatment included only Delany, M.F. and J.A. Cox. 1986. Florida Shriver, W.G., P.D. Vickery, and S.A. 3 plots. grasshopper sparrow breeding distribution and Hedges. 1996. Effects of summer burns on Appropriate habitat management for abundance in 1984. Fla. Field Nat. 14:100-104. Florida grasshopper sparrows. Fla. Field Nat. Delany, M.F. and S.B. Linda. 1994. 24:68-73. most grasshopper sparrow populations Characteristics of occupied and abandoned Shriver, W.G., P.D. Vickery, and D.W. appears to be the maintenance of grass- Florida grasshopper sparrow territories. Fla. Perkins. 1999. The effects of summer burns lands that are frequently burned and Field Nat. 22:106-109. on breeding Florida grasshopper and devoid of forest edges (Johnson and Delany, M.F. and S.B. Linda. 1998a. Bachman's sparrows. Stud. in Avian Biol. Temple 1986, Delisle and Savidge 1996, Characteristics of Florida grasshopper sparrow 19:144-148. Vickery 1996). Prescribed burning during nests. Wilson Bull. 110:136-139. Smith, R.L. 1968. Grasshopper sparrow. p. Delany, M.F. and S.B. Linda. 1998b. Nesting the winter (December-March) at 2-3 year 725-745 In: Life histories of North American habitat of Florida grasshopper sparrows at cardinals, grosbeaks, buntings, towhees, intervals seems to maintain suitable habi- Avon Park Air Force Range. Fla. Field Nat. finches, sparrows and allies, A. C. Bent (ed.). tat for the Florida subspecies (Delany and 26:33-39. Pt. 2. U.S. Natl. Mus. Bull. 237. Washington Cox 1986). Winter burning also avoids Delany, M.F., H.M. Stevenson, and R. D. C. potential nest loss from fires during the McCracken. 1985. Distribution, abundance, U.S. Fish and Wildlife Service. 1988. Recovery late-March through July breeding season. and habitat of Florida grasshopper sparrows. plan for Florida grasshopper sparrow. U.S. J. Wildl. Mgt. 49:626-631. However, winter fires may alter the com- Fish and Wildlife Service, Atlanta, Ga. Delany, M.F., P.B. Walsh, B. Pranty, and U.S. Fish and Wildlife Service. 1999. Dry position of Florida's dry prairies by D.W. Perkins. 1999. A previously unknown increasing woody biomass and reducing Prairie p. 279-339 In: South Florida multi- population of Florida grasshopper sparrows on species recovery plan. U.S. Fish and Wildlife Avon Park Air Force Range. Fla. Field Nat. native grasses and forbs (U.S. Fish and Service, Atlanta, Ga. 27:52-56. Wildlife Service 1999). Wiregrass normal- Van Home, B. 1983. Density as a misleading Delisle, J.M. and J.A. Savidge. 1996. Repro- ly does not flower unless burned during indicator of habitat quality. J. Wildl. Mgt. ductive success of grasshopper sparrows in 47:893-901. late-spring or summer, and some prairie relation to edge. Prairie Nat. 28:107-113. forbs such as deer tongue (Carphephorus ESRI. 1990. ArcView 3.0 GIS software. Vickery, P.D. 1996. Grasshopper sparrow paniculatus Gmel.) and yellow bachelor's Environmental Systems Research Institute, (Ammodramus savannarum). In: The birds of button (Polygala rugelii Chapm.) flower Inc. Redlands, Calif. North America, number 239. A. Poole and F. International Bird Census Committee. 1970. Gill (eds.). Acad Natural Sci., Philadelphia, more conspicuously (Abrahamson 1984). Penn., and Amer. Ornith. Union, Washington, Florida's prairie ecosystem has probably An international standard for a mapping method in bird census work recommended by D.C. been maintained by a frequent fire cycle (1-4 the International Bird Census Committee. Vickery, P.D., M. Hunter, Jr., and J.V. Wells. years) caused by summer (May-September) Aud. Field Notes 24:722-726. 1992a. Use of a new reproductive index to thunderstorms (Abrahamson and Hartnett Johnson, R.G. and S.A. Temple. 1986. evaluate relationship between habitat quality 1990, Chen and Gerber 1990). Prescribed Assessing habitat quality for birds nesting in and breeding success. Auk 109:697-705. P.D., J.V. Wells. bums that mimic a natural fire regime may fragmented tallgrass prairies. p. 245-249 In: J. Vickery, M. Hunter, Jr., and 1992b. Is density an indicator of breeding suc- better sustain this pyrogenic plant communi- Verner, M. L. Morrison, and C. J. Ralphs (eds.) Wildlife 2000: Modeling habitat rela- cess? Auk 109:706-710. ty. Florida grasshopper sparrows are proba- tionships of terrestrial vertebrates. Univ. Wise. Vickery, P.D., M.L. Hunter, Jr., and J.V. bly adapted to summer fires, and may exhibit Press, Madison, Wise. Wells. 1992c. Evidence of incidental nest pre- increased densities and a prolonged breeding Lewis, C.E. 1964. Forage response to month of dation and its effects on nests of threatened season following such bums (Shriver et al. burning. U.S. For. Serv. Res. Note 5E-35, grassland birds. Oikos 63:281-288. 1996, Shriver et al. 1999). Studies of Florida Asheville, N.C. Walsh, P.B., D.A. Darrow, and J.G. Dyess. grasshopper sparrow abundance and pro- Martin, T.E. 1993. Nest predation and nest 1995. Habitat selection by Florida grasshop- sites, new perspectives on old patterns. per sparrows in response to fire. Proc. Annu. ductivity are needed to determine the BioSci. 43:523-532. Conf. Southeast. Assoc. Fish and Wildl. effects of the seasonality of fire. Milleson, J.F., R.L. Goodrick, and J.A. Van Agencies 49:340-347. Arman. 1980. Plant communities of the With, K.A., and D.R. Webb. 1993. Microcli- Kissimmee River Valley. South Fla. Water mate of ground nests: relative importance of Literature Cited Mgt. Dist. Tech. Pub. 80-7. radiative cover and wind breaks for three Peterjohn, B.G. and J.R. Sauer. 1999. grassland species. Condor 95:401-413. Population status of North American grass- Abrahamson, W.G. 1984. Species response to land birds from the North American breeding fire on the Florida Lake Wales Ridge. Amer. bird survey, 1966-1996. Stud. in Avian Biol. J. Bot. 71:35-43. 19:27-44. Abrahamson, W.G. and D.C. Hartnett. 1990. SAS Inst., Inc. 1999. SAS OnlineDocTM Pine Flatwoods and Dry Prairies. p. 103-150 Version 7-1. SAS Institute Inc., Cary, NC. In: R. L. Myers and J. J. Ewel (eds.) Seber, G.A.F. 1982. The estimation of animal Ecosystems of Florida. Univ. Central Fla. abundance and related parameters. Charles Press, Orlando, Fla. Griffin and Co., London and High Wycombe. Carter, L.J., D. Lewis, L. Crockett, and J. Shriver, W.G. and P.D. Vickery. 1999. Aerial Vega.1989. Soil survey of Highlands County, assessment of potential Florida grasshopper Florida. USDA Soil Conserv. Serv., Sebring, sparrow habitat: conservation in a fragmented Fla. landscape. Fla. Field Nat. 27:1-9. Chen, E. and J.F. Gerber. 1990. Climate. p. Shriver, W.G. and P.D. Vickery. 2001. 11-13 In: R. L. Myers and J. J. Ewel (eds.) Response of breeding Florida grasshopper and Ecosystems of Florida. Univ. Central Fla. sparrows to winter Press, Orlando, Fla. Bachman's prescribed burning. J. Wildl. Mgt. 65:470-475. Curnutt, J.L., S.L. Pimm, and B.A. Maurer. 1996. Population variability of sparrows in space and time. Oikos 76:131-144.

340 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 J. Range Manage. 55: 341-349 July 2002 Large ungulate habitat preference in Chobe National Park, Botswana

UYAPO J. OMPHILE AND JEFF POWELL

Authors are Lecturer, Botswana College of Agriculture, Gaborone; and Professor of Rangeland Livestock-Wildlife Relations, Wyoming Agricultural Experiment Station, Laramie, Wyo, USA.

Abstract Resumen

Both large ungulates and wildlife tourists tend to concentrate Durante la estacion seca los grandes ungulados y los turistas de along the Chobe River in Chobe National Park, Botswana, dur- la fauna silvestre tienden a concentrarse a to largo del rio Chobe ing the dry season causing concern for wildlife habitat and the en el Parque Nacional Chobe de Botswana, causando preocu- recreational experience for wildlife viewers. Therefore, ground pacion por el habitat y la experiencia recreacional de los obser- reconnaissance inventory data of 5 most common large ungulates vadores de fauna silvestre. Por to tanto, cada dos meses durante were collected during the early morning and late afternoon hours un periodo de 24 meses, se colectaron en el Parque Nacional along tourist routes in 5 different habitat types every second Chobe datos de un inventario de reconocimiento terrestre de los month for a period of 24 months in Chobe National Park to 5 grandes ungulados mas comunes. La colecta de datos se realizo determine their relative, seasonal habitat preference and avail- temprano en la manana y en la tarde a to largo de rutas de turis- ability for viewing by vehicular tourists. A total of 909 herds: tas en 5 diferentes tipos de habitat para determinar la preferen- greater kudu (Tragelaphus strepsiceros), 285; impala (Aepyceros cia estacional relativa de habitat y la disponibilidad para ser melampus), 209; elephant (Loxodonta africana), 200; giraffe visto por turistas en vehiculos. Un total de 909 manadas, 285 de (Giraffa camelopardalis), 138; and buffalo (Syncerus caffra), 77 Gran kudu (Tragelaphus strepsiceros), 209 de Impala (Aepyceros were observed during the 760 observation periods. The average melampus), 200 de elefante (Loxodonta africana), 138 de jirafa; frequency of observation of a herd of 1 or more of these 5 ungu- (Giraffa camelopardalis), 77 de bufalo (Syncerus caffra), se obser- lates per habitat type was alkali flats, 3.42; floodplain grassland, varon durante 760 periodos de observacion. La frecuencia 2.67; shrub savanna, 2.29; tree savanna, 1.04; and woodland, promedio de observacion de una manada de uno o mas de estos 5 0.30. This order of frequency of observation is highly correlated ungulados por tipo de habitat fue de 3.42 para las llanuras alcali- with nearness to the Chobe River, the major water source during nas, 2.67 para los pastizales inundables, 2.29 para la savana con the dry season. Elephant and giraffe were more wide-ranging arbustos, 1.04 para la savana con arboles y 0.30 para los than buffalo, impala, and kudu. During the dry season, all ani- bosques. Este orden de frecuencia de observacion esta altamente mals were seen more often on the floodplain grassland in the correlacionada con la cercania del rio Chobe, la mayor fuente de afternoon than in the morning. Giraffe were never seen in any agua durante la epoca seca. El elefante y la girafa tuvieron un habitat type in December, and impala were never seen in the rango mas amplio de dispersion que el bufalo, el impala y el woodland in any month. Our data confirm that tour operators kudu. Durante la epoca seca todos los animales fueron vistos mas interested primarily in providing their guests with a view of the a menudo en los pastizales inundables durante la tarde que greatest numbers of animals in a limited period of time are justi- durante la manana. Las jirafas nunca fueron vistas en diciembre fied in congregating along the Chobe River during the dry sea- en ninguno de los habitats y los impalas no fueron vistos en el son. However, as in most public wildlife reserves, Chobe National bosque en ninguno de los meses. Nuestros datos confirman que Park management is faced with the decision of how best to opti- los operadores de recorridos interesados principalmente en mize the biological needs of Park animals and their habitat with proveer a sus invitados una vista con el mayor numero de ani- the economic and recreational desires of Park users. males en un periodo de tiempo limitado son justificados en con- gregase a to largo del rio Chobe durante la epoca seca. Sin embargo, como en la mayoria de las reservas publicas de fauna Key Words: Southern Africa, recreationists, buffalo, ele- silvestre, el manejo del Parque Nacional Chobe, se encara con la phant, giraffe, impala, kudu decision de como optimizar mejor las necesidades biologicas de los animales del parque y su habitat con los deseos economicos y recreacionales de los usuarios del parque. Chobe National Park is one of Botswana's major tourist attrac- tions, primarily because of the variety of large animals present. In monetary terms, McNeeley (1988) estimated that a lion in Amboseli National Park, Kenya is worth $27,000 per year in for 108,700 tourist-days or 1.21 days/tourist (pers. commun., tourist attraction, and each elephant herd is worth $610,000 per Botswana Dep. National Parks and Wildlife). Tourists included year. In 1999, about 89,540 tourists visited Chobe National Park 27% who drove their own vehicles and 73% guided by commercial tour operators. Most (56%) of the private visitors were from Southern Africa, whereas guided visitors were primarily (81.3%) Research was funded in part by the Kellogg Foundation. from countries outside Africa. Private visitors provided 40% of Manuscript accepted 24 Sept. 01.

JOURNAL OF RANGE MANAGEMENT 55(4) July 341 the total 21,833 vehicle-days with an aver- mainly between November and April. dent from tree stumps and bark-bearing age of 3.5 people per vehicle; tour opera- Precipitation at Kasane averages 690 mm fire scars. The area tends towards a forest tors provided 60% of the vehicle-days (Child and Von Richter 1968) with a to the east between Kasane and Lesoma, with an average of 6.0 people per vehicle. 70-80% probability of exceeding 500 mm stretches south from the crest of the savan- The majority of visitors tour Chobe in any one year (Field 1978). The remain- na (sandridge) to the edge of the Kakulani National Park during the dry, winter sea- ing months of the year are dry, and the plains around Ngwezumba River and from son (pers. commun., Botswana Dept. degree of dryness increases progressively Kasane, southwest to Kachikau. The soils National Parks and Wildlife) because the from June to October. Maximum and min- (FGU undated) are predominantly ferralic temperature is cooler, the risk of malaria is imum average monthly temperatures range arenosols to eutric arenosols, deep to very lower, and the loss of leaves from decidu- from 36 to 19° C, but decline to winter deep and excessively drained. Locally, ous trees and shrubs during the dry season minima in June and July when the range is these soils are interspersed with a variety enhances wildlife viewing (pers. observ.). between 26 and 9 C. Temperatures reach a of loamy sands on alluvial deposits in the In addition, nearly all large animals tend maximum in October when the average valleys. to concentrate near the Chobe River along maximum over 24 years was 35° C and Savanna soils are consistently sandy and the northern edge of the Park at this time the minimum was 19° C (Child and Von lack interruptions except on rocky out- because other water sources are limited. Richter 1968). crops where larger rock boulders or loose The concentration of animals along the Topography.-Northeastern Botswana, gravel replace the sand. The habitat northern edge of the Park during the dry like the rest of the country, is relatively flat abruptly gives way either to the Chobe season attracts wildlife-viewing tourists, terrain characterized by low elevation floodplains, alkali flats, or the shrub thus apparently leaving much of the Park ranging from 100 to 1,000 m (Field 1978). savanna to the north and west, and the under-utilized by both animals and tourists Within the park, however, elevations only Kgalagadi woodland to the south. at this time. The Botswana Department of range from 910 to 1,050 m (FGU undated). Although logging was reported (Miller National Parks and Wildlife is concerned Soils and vegetation.-The Park is eco- 1939), and signs of that activity are still about the "over-crowding" of animals and logically divided into 5 zones; the Chobe evident, especially to the north, the indus- tourists along the northern edge of the floodplain grassland in the north, Mababe try does not appear to have significantly Park, and therefore, the objective of this depression and Savuti channel and marsh affected the tree population. study was to document the numbers of in the southwest, the Pan area into which The shrub savanna, an extension of the herds of 5 common ungulate species in the Ngwezumba River empties, and the tree savanna, but with an ecologically dis- different habitat types and in different sea- Kgalagadi woodland which covers most of tinct characteristic resulting from exces- sons as "seen by vehicular tourists". The 5 the Park. The study site includes both sive logging during the turn of the century species included 2 browsers, giraffe Kgalagadi woodland and Chobe flood- (Child 1968) and excessive use of the (Giraffa camelopardalis) and greater kudu plains. Within the woodland, 2 additional woody vegetation by elephants, is located (Tragelaphus strepsiceros), 2 mixed feed- areas (i.e., tree savanna and shrub savan- east of the sand ridge and alkali flats and ers, elephant (Loxodonta Africana) and na) were identified for the purposes of the west of the Sedudu valley. The soils are impala (Aepyceros melampus), and 1 graz- study. These areas, together with a third generally sandy, but stretches of harder er, buffalo (Syncerus caffra) to reflect the habitat type, the alkali flats, were described loamy soil, especially around the few range of large ungulate herbivory in the (Selous 1881, Henry 1966) and referred to pans, are common. study area. (Child 1968) as variations within a broader The alkali flats is an open area, though vegetation type. Therefore, we considered some portions, especially those within the 5 areas (i.e., floodplain grassland, alkali sandy soils catena, have a closed canopy Materials and Methods flats, shrub savanna, tree savanna, and of tall brush. This area, believed to be an Kgalagadi woodland) as 5 ecologically dis- elevated floodplain and named "Puku" Study Area tinct habitat types affecting ungulate flats (Selous 1881) is oblong in shape and narrow Location and general description.-The behavior and ecology in this study. about 9-13 km long, but rather wide). study area is located in the northern part of The total area of study is about 575 km2 (0.5-1.2 km Chobe National Park, Botswana. The Park or about 5% of the total Park area. Within covers an area of 11,000 km2 and lies the study area, the habitat type composi- Observational Methods tion is as follows: woodland (55.6%), tree between longitude 24° and 25° E and lati- The 5 animal species selected for study grassland tude 18° and 19° S (Child 1968). It is par- savanna (23.7%), floodplain were observed in each of the 5 habitat tially drained by the Chobe River which (11.1%), shrub savanna (5.2%), and alkali types delineated in the study area. forms the boundary between northeastern flats (4.2%). Reconnaissance tours (Dekker et al. 1996) grasslands, an Botswana and eastern Carprivi Strip of The Chobe floodplain were conducted during the first 3 hours of integral part of the Chobe River system, is Namibia. The river meanders through an daylight and the last 3 hours before dark a 90 km2 bor- extensive floodplain which stretches east linear area of approximately every other day for a minimum of 5 and a including the river. The almost unbroken to the Zambezi River in dering and part of maximum of 7 different days per month Zambia and abruptly gives way to a sand soils differ, but are characterized by calcic during both wet and dry seasons from ridge composed of a basaltic core overlain gleysols and eutric arenosols, deep to very October 1993 through August 1995. by the Kgalagadi sand to the south (Child deep, poorly to imperfectly drained, very Observational data recorded for each her- and Von Richter 1968). The interior of the dark sand clay to clay interspersed by bivore included habitat type, location Park is drained by the ephemeral Savuti riverine alluvials on dead river courses within habitat type, and distance (m) from and Ngwezumba rivers and numerous (FGU undated). observer to animal. intermittent streams. The Kgalagadi woodland is an open The line transect method (Mugangu et as evi- Climate.-Rainfall is seasonal, falling woodland, probably fire-induced al. 1995) was used for data recording. The

342 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 route of each reconnaissance tour was lim- per habitat type each observation period. 1980) using years as replications to deter- ited by vehicular accessibility. However, The recording team included a driver mine the level of significance among dif- all habitat types were sampled during each and 2 observers in the back of a pickup ferences in habitat type preference for dif- observation period, and the route order truck. Each observer recorded individual- ferent times of day and months by differ- was randomized by and within habitat ly, and the 2 observations were averaged ent herbivore species. Where significant types to obtain an estimate of the preferred for each kind of herbivore. Due to differ- (P < 0.05) differences were observed, habitat types based on relative numbers of ences in vegetation cover density, the Tukey's HSD was used to separate the herds of different species of animals sight- sighting distance (Dekker et al. 1996) var- means. All differences discussed are sta- ed in each habitat type. Animals seen ied among the different vegetation types. tistically significant at the 5% level unless along habitat type ecotones were not The initial sighting distance was disre- otherwise indicated. recorded, and only those sighted in a habi- garded because of the emphasis on the tat type that was currently being moni- "tourist viewing perspective". In other tored were considered valid observations. words, the objective was to determine how Results The total number of habitat preference many herds of a species a tourist would samples (observation means) per species see from a vehicle along routes of vehicu- Weather was: 2 years x 6 months/year x 5 habitat lar access in different habitat types in Annual precipitation during the 2 wet- types/month x 5-7 days/month x 2 times Chobe National Park. Not all species were to-dry years from October, 1993 through /day = 760 observation periods. The num- observed in all habitat types or in all September, 1995 was about equal but 20% ber of observations of herds of an herbivore observation periods. However, if more less in both years than the long-term aver- in each habitat type was adjusted by a fac- than one herd of the same species was age. The early wet season rains in the first tor to compensate for the different distances observed at different locations in the same year (PY94; 10/93-9/94) were about aver- traveled in different habitat types. The dis- habitat type during the same observation age, but there were only limited rains after tances traveled each observation period in period, each herd was counted as a sepa- February 1994. Early wet season precipi- the different habitat types were alkali flats rate observation. tation in PY95 (10/94-9/95) was less than (28.4 km), floodplain grassland (23.4 km), The general linear model (GLM) proce- that in PY94; however, above average shrub savanna (21.3 km), tree savanna dure (SAS 1985) was used for a split (time rainfall in February 1995 brought the (40.7 km), and woodland (30.5 km) for a of day)- split (habitat type) - plot (month) cumulative total up to equal that of PY94. total of 144 km or an average of 28.8 km experimental design (Steel and Torrie The net effects were to create normal plant

Table 1. Numbers (mean ± SE) of buffalo herds observed per time period during different months, times of day, and in different habitat types in Chobe National Park, Botswana. (N = 6/month/time period/habitat type). Means for habitat types for the same time or month followed by a different letter are statistically different at the 5% level of probability. Significant (P < 0.05) differences between means for different times within habitat type and month are indicated by `*'.

Habitat Type Month Time2 AF Year4 AM 0.17a ± 0.04 0.02 PM 0.lOb±0.03 0.43*a±0.08 0.04 0.01 TMean5 0.13b ± 0.03 0.22a ± 0.04 0.01

Dec. AM 0.25±0.13 ±0.11 PM 0.09 ± 0.09 0.21 ± 0.14 0.0 0.0 0.0 TMean 0.17 ± 0.08 0.15 ± 0.09

Feb. AM 0.07 ± 0.07 0.0 ± 0.0 PM 0.16±0.11 0.0±0.0 TMean 0.11 ± 0.06 0.0 ± 0.0 ±0.05

Apr. AM 0.23 ± 0.12 0.0 ± 0.0 ±0.04 PM 0.16±0.11 0.28*±0.15 0.0 ±0.04 TMean 0.20 ± 0.08 0.14 ± 0.08 ±0.03

June AM 0.25 * ± 0.13 0.11 ± 0.08 PM 0.0±0.0 0.62*±0.19 TMean 0.13 ± 0.07 0.31 ± 0.11 ±0.08

Aug. AM 0.09 ± 0.09 0.0 ± 0.0 PM 0.18±0.12 0.90*±0.34 0.0 + 0.0 0.0 TMean 0.14b ± 0.08 0.45a ± 0.19

Oct. AM 0.15 ± 0.10 0.0 ± 0.0 PM 0.0±0.0 0.62*±0.17 0.0±0.0 0.05 0.31a ± 0.10 zAF = alkali flats; FG = floodplain grassland; SS = shrub savanna; TS = tree savanna; WL = woodland. AM =1 3 hours after daylight; PM =1 3 hours before dark. 4HTMean = Mean for all habitat types per time or month. Year = Mean for all months per time and per habitat type. 5TMean = Mean for both AM and PM per habitat type or month.

JOURNAL OF RANGE MANAGEMENT 55(4) July 343 growth during the early and mid wet sea- Buffalo There was no significant difference in son of PY94 with an early onset of the dry The average number of herds of buffalo buffalo sightings in different habitat types season. In PY95, early and mid season for all observation periods was about 0.11 during any month of the wet season growth in PY95 was less than average, but and varied only from 0.08 to 0.16 for the (December-April) or the early part (June) the above average late wet season rains average of any month (Table 1). However, of the dry season, but buffalo sightings delayed the onset of the dry season. there were differences between habitat were greater in the floodplain grassland in types and between times of day. the mid (August) and late (October) dry Relative Frequency of Observation For the year, the greatest number of season than in other habitat types. of Herds herds of buffalo were observed in the A total of 909 herds of animals (kudu, floodplain grassland and the lowest num- Elephant 285; impala, 209; elephant, 200; giraffe, ber in the woodland habitat type. This var- The overall average number of herds of 138; and buffalo, 77) were observed over ied with time of day. During the morning, elephant observed per observation period the 760 observation periods for an average buffalo were observed most often in the was 0.42 (Table 2). The numbers of herds of 1.20 herds per observation period and tree savanna and alkali flats and least often one herd of some species for every 15.5 sighted were similar during the wet sea- in the floodplain grassland and woodland. son, then tended to increase and peak at km of route traveled. The average, adjusted Except for June, buffalo were never (i.e., for different size habitat types) num- the end of the dry season. The annual ber of herds per observation period per observed in the woodland. During the average elephant sightings per habitat type habitat type for all animals combined was afternoon, buffalo were observed most were in descending order as follows: as follows: alkali flats (3.42), floodplain often in the floodplain grassland. Buffalo floodplain grassland (0.80) > alkali flats grassland (2.67), shrub savanna (2.29), tree sightings were much more numerous in (0.53) >_ shrub savanna (0.41) >_ tree savanna (1.04), and woodland (0.30). the tree savanna in the morning than the savanna (0.27) > woodland (0.11). In the The total number of observations for all afternoon and much more numerous in the morning, elephant sightings were about herds was similar in PY94 (473) in PY95 floodplain grassland during afternoon than equal and more numerous in the alkali (436) and for the wet (427) and dry sea- morning. These differences occurred, flats, shrub, and tree savannas than in the sons (482) seasons. However, there were however, only in the late wet season floodplain grassland or woodland, but in large differences in the numbers of obser- (April) and all of the dry season (June- the afternoon elephant sightings were vations of herds due to kind of animal, October). much greater in the floodplain grassland month, habitat type, and time of day.

Table 2. Numbers (mean ± SE) of elephant herds observed per time period during different months, times of day, and in different habitat types in Chobe National Park, Botswana. (N = 6/month/time/habitat type). Means for habitat types for the same time or month followed by a different letter are statistically different at the 5% level of probability. Significant (P < 0.05) differences between means for different times within habitat type and month are indicated by `*'.

Habitat Type' Month Time2 AF Year4 AM 0.61 a ± 0.09 0.07 PM 0.45b ± 0.08 1.44*a ± 0.18 0.07 03 ±05 TMean5 0.53b ± 0.06 0.80a ± 0.11 0.06 ±04 d ± 03

Dec. AM 1.10±0.26 10 PM 0.76±0.25 0.21 ±0.21 0.25 0.13 0.08 ± 09 TMean 0.93a±0.18 O. l Oc ± 0.10 0.18 0.11

Feb. AM 0.39 ± 0.18 0.0 ± 0.29 0.16 0.12 09 PM 0.58±0.21 0.10±0.10 0.25 0.05 TMean 0.49ab ± 0.14 0.05d ± 0.05

Apr. AM 0.31±0.18 PM 0.55 ± 0.19 1.70* ± 0.41 ±0.07 TMean 0.43ab ± 0.13 0.85a ± 0.26 0.05 07

June AM 0.34 ± 0.14 0.0 ± 0.0 PM 0.51 ± 0.15 1.74* ± 0.35 TMean 0.42b ± 0.10 0.87a ± 0.25

Aug. AM 0.46 ± 0.25 0.11 ± 0.10 0.08 PM 0.28 ± 0.20 2.35* ± 0.42 TMean 0.37b ± 0.16 1.23a ± 0.32 0.08 0.09

Oct. AM 1.01* ± 0.24 0.33 0.07 ± 0.07 2.55* ± 0.51 TMean 0.54b±0.15 1.67a±0.34 0.03 10

AF = alkali flats; FG = floodplain grassland; SS = shrub savanna; TS = tree savanna; WL = woodland. 3AM =1sr 3 hours after daylight; PM =1sr 3 hours before dark. HTMean = Mean for all habitat types per time or month. Year = Mean for all months per time and per habitat type. TMean = Mean for both AM and PM per habitat type or month.

344 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Table 3. Numbers (mean ± SE) of giraffe herds observed per time period during different months, times of day, and in different habitat types in Chobe National (N Park, Botswana. = 6/month/time period/habitat type). Means for habitat types for the same time or month followed by a different letter are statistically different at the 5% level of probability. Significant (P < 0.05) differences between means for different times within habitat type and month are indicated by `*'.

Habitat Type' Month Time AF FG Year4 AM 0.41*ab ± 0.07 0.02 0.09 0.06 0.06 03 PM 0.24b ± 0.05 0.10 0.05 0.03 TMean 5 ± 0.32a 0.04 0.31a ± 0.06 0.06 0.03 0.02 02 Dec. TMean6

Feb. AM 0.62±0.18 PM 0.31 ±0.18 0.0±0.0 ±0.22 ±0.12 TMean ± 0.47a 0.13 O.Ob ± 0.0 0.17 ab ± 0.08 0.05 05

Apr. AM 0.47 ± 23 0.0 0.24 0.13 ±07 PM 0.23 ± 12 0.85* ± 0.22 0.10 ± 0.05 ± TMean 0.35a± 11 0.43a±0.14 0.13 0.07 0.04 ±05 June AM 0.51 ± 0.23 0.0 ± 0.20 0.14 0.08 ± 08 PM 0.34±0.14 1.23*±0.34 0.06 ± TMean 0.42ab ± 0.14 0.62a ± 0.21 0.13 0.08

Aug. AM 0.28 ± 0.14 0.11 ± 0.27 0.20 0.09 ± 0.09 PM 0.46±0.16 1.01*±0.32 0.0 ± 0.0 TMean 0.37a ± 0.11 0.56a ± 0.19 0.16 0.11

Oct. AM 0.51* ± 0.14 0.09 ± 0.23 0.16 0.07 ± 0.07 PM 0.15±0.10 0.53*±0.21 0.05 ± TMean 0.33a ± 0.09 0.31a ± 0.12 0.13 0.08 0.03 0.04 AF = alkali flats; FG = floodplain grassland; SS = shrub savanna; TS = tree savanna; WL = woodland. 3AM = 1° 3 hours after daylight; PM =1sr 3 hours before dark. HTMean = Mean for all habitat types per time or month. SYear = Mean for all months per time and per habitat type. TMean = Mean for both AM and PM per habitat type or month. 6No giraffe were seen anywhere in the study area in December.

than anywhere else. Differences due to December, when giraffe were never seen the number of impala sightings during dif- time of day were greatest for the tree savanna anywhere in the study area. ferent months was very similar. and woodland (i.e., the 2 habitat types far- On an annual basis, giraffe were most Impala preference for the alkali flats thest from the Chobe River) and the flood- often seen and seen equally in the alkali was about twice that for any other habitat plain grassland (i.e., closest to the river). flats, floodplain grassland, and shrub type, although they were seen most often Elephant sightings were relatively low savanna, less often in the tree savanna, and in the afternoon on the floodplain grass- in the floodplain grassland during the least often in the woodland. However, as land. Impala were also seen about twice as early and mid wet season, but were greater with the elephants, giraffe sightings were often in the shrub savanna in the morning in the floodplain grassland than in any higher in the morning in the alkali flats and as in the afternoon and more often in the other habitat type thereafter. By the end of shrub savanna than elsewhere and higher alkali flats in the morning than in the the dry season in October, elephant sight- in the afternoon in the floodplain grassland afternoon. Impala were never seen in the ings in the floodplain grassland were 3 than elsewhere. Giraffe were never seen in floodplain grasslands in the morning dur- times greater than the overall habitat type the woodland in the afternoon. ing any month, nor were impala ever seen average. However, elephants were never The major difference in habitat prefer- in the woodland at either time of day or sighted in the floodplain grassland in the ence occurred in February (mid wet sea- during any month of the year. The prefer- morning until the middle of the dry season. son) and in June (beginning of the dry sea- ence of impala for the alkali flats in the Conversely, elephant sightings were son). Giraffe preference for the floodplain morning and the floodplain grassland in greater in the morning than in the after- grassland tended to increase from the afternoon became more pronounced as noon in the tree savanna and woodland February to June and then decrease the seasons changed from wet to dry. until late in the dry season. At this time, through October with the greatest differ- Impala preference for the alkali flats was elephants tended to concentrate in the alka- ence in the afternoon. much higher in the early-mid wet season li flats in the morning and move out onto and dry season than in the late wet season the floodplain grasslands in the afternoon. Impala (i.e., April), and this low number of sight- The overall average number of impala ings in April on the alkali flats was the pri- Giraffe sightings was 0.53 per observation period mary reason for the low monthly average The average number of giraffe sightings (Table 4). Unlike the sightings for other in April for all habitat types. Impala prefer- for all months, habitat types, and times of animals, impala sightings were significant- ence for the floodplain grassland increased day was 0.23 (Table 3). This number was ly lower in April at the end of the wet sea- from low in the early-mid wet season to relatively similar for all months except son than during other months. Otherwise, high in the mid-late dry season.

JOURNAL OF RANGE MANAGEMENT 55(4) July 345 Table 4. Numbers (mean ± SE) of impala herds observed per time period during different months, times of day, and in different habitat types in different letter Chobe National Park, Botswana. (N = 6/month/time/habitat type). Means for habitat types for the same time or month followed by a are statistically different at the 5% level of probability. Significant (P <0.05) differences between means for different times within habitat type and month are indicated by `*'.

Habitat Type' Month Time2 AF WL Year4 AM 1.45*a ± 0.10 0.0 0.12 03 04 PM 1.09a±0.13 1.35*a±0.17 06 TMean 1.27a ± 0.08 0.67b ± 0.10 0.08 02 04

Dec. AM 1.78±0.18 12 PM 1.86±0.37 0.21±0.21 TMean 1.82a ± 0.20 0. l Oc ± 0.10 0.24 0.03

0.12 Feb. AM 1.67 ± 0.20 0.0 ± 0.16 0.09 PM 1.81±0.36 0.19±0.19 0.12 TMean 1.74a ± 0.20 0.1 Oc ± 0.10 0.20 0.06 0.08

Apr. AM 0.70 ± 0.21 0.0 ± 0.28 0.09 08 PM 0.62 ± 0.22 1.04* ± 0.37 0.32 0.05 ± TMean 0.66a ± 0.13 0.52ab ± 0.21 0.21 0.06

June AM 1.52 ± 0.27 0.0 ± 0.25 0.10 PM 1.35 ± 0.23 1.33* ± 0.39 0.23 ± 0.06 ± 13 TMean 1.44a ± 0.10 0.67b ± 0.23 0.18 0.06

Aug. AM 1.48* ± 0.29 0.0 ± 0.32 0.09 PM 0.65±0.25 3.02*±0.19 17 TMean 1.06a ± 0.21 1.51a ± 0.34 0.19 0.04 10

Oct. AM 1.59* ± 0.28 0.0 ± 22 0.05 ± 10 PM 0.29±0.17 2.37*±0.40 14 TMean 0.94a ± 0.20 1.19a ± 0.30 15 0.03 ab ± 09 ;AF alkali flats; FG floodplain grassland; SS shrub savanna; TS = tree savanna; WL = woodland. = lst = = 3AM = 3 hours after daylight; PM =1st 3 hours before dark. HTMean = Mean for all habitat types per time or month. Year = Mean for all months per time and per habitat type. Impala were never observed in the Woodland habitat type at any time during the study. 6TMean = Mean for both AM and PM per habitat type or month.

Conversely, impala preference for the grasslands (0.00 in February during flood- ences due to time of day in the number of shrub savanna tended to be higher in the ing - 1.10 during the late dry season). By sightings only occurred in April through early-mid wet season than later in the year. mid-late dry season, the number of kudu August, and the number of sightings was sightings was very similar on the alkali always higher in the afternoon. The most grasslands, and shrub consistent effect of time of day on the Kudu flats, floodplain savanna - all 3 habitat types next to the number of kudu sightings occurred in the The overall average number of kudu Chobe River. tree savanna where kudu were observed sightings was 0.66 per observation period An interaction between habitat type and more often in the morning during all (Table the highest number for all 5 ani- 5), time of day also influenced the number of months of observation except December. mals observed. The average number of sightings of kudu; more kudu herds were sightings per month was relatively consis- seen in the morning than in the afternoon tent (i.e., 0.57-0.76). The number of sight- in the shrub savanna and woodland where- Discussion ings in the morning and afternoon were as more kudu herds were seen in the after- generally similar each month except for noon than in the morning on the flood- October when more herds of kudu for all The sighting of relatively numerous plain grasslands. habitat types were seen in the morning herds during February on the alkali flats Differences in the average number of than in the afternoon. may have resulted from a prey behavior sightings due to time of day also varied There were relatively large differences pattern. Peak growth of both herbaceous with month of observation within habitat in the average number of sightings of kudu and woody vegetation in all habitat types type. The number of kudu sightings on the per habitat type, ranging from a high of meant that visibility (Montfort 1973) was alkali flats in the morning and afternoon 1.17 per observation period on the alkali generally poor during the middle of the was similar in February and June, slightly flats down to 0.14 in the woodland habitat wet season. This is particularly important greater in the afternoon than in the morn- number of sightings per to those species, such as impala and kudu, type. The average ing in December and August, much month was relatively consistent on the which easily fall prey to predators greater in the afternoon in April (i.e., late tree savanna, and woodland, (Pienaar 1963, Kingdon 1982). Buffalo alkali flats, wet season), and much greater in the but varied more widely in the shrub savan- fleeing from lion attack (Mills et al. 1995) morning in October (i.e., late dry season). na and most widely on the floodplain also usually regrouped on the alkali flats On the floodplain grasslands, large differ-

346 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Table 5. Numbers (mean ± SE) of kudu herds observed per time period during different months, times of day, and in different habitat types in Chobe National Park, Botswana. (N = 6/month/time/habitat type). Means for habitat types for the same time or month followed by a different letter are statistically different at the 5% level of probability. Significant (P < 0.05) differences between means for different times within habitat type and month are indicated by `*'.

Habitat Typei Month Time2 AF FG

Year4 AM 1.13a5 ± 0.11 0.09 0.05 0.05 * PM 1.20a ± 0.12 1.05 a ± 0.16 0.03 0.05 TMeans 1.17a ± 0.08 0.67c ± 0.10 0.10 0.05

Dec. AM 1.Olab ± 0.31 0.10 0.42 0.21 PM 1.69a±0.29 0.21b±0.21 0.26 0.16 0.11

Feb. AM 1.23a ± 0.19 0.0 0.34 0.15 0.12 10 PM 1.17a±0.38 0.Oc±0.0 0.23 0.13 0.10 TMean 1.20a ± 0.20 O.Oc ± 0.0

Apr. AM 0.78a ± 0.20 0.0 0.25 0.18 PM 1.64*a ± 0.25 1.70*a ± 0.36 0 TMean 1.21a±0..18 0.85ab ± 0.24 0.15 0.11 0.59±0.08

June AM 1.Ola ± 0.25 PM 1.10ab ± 0.23 1.44*a ± 0.40 0.31 0.08 0 13 TMean 1.06a±0.17 0.82ab ± 0.24 0.26 0.12 0.64±0.09

Aug. AM 0.74b±0.31 0.44 0.24 0.14 0.14 PM 1.38a ± 0.37 1.90*a ± 0.48 TMean 1.06a ± 0.25 1.06a ± 0.31 0.29 0.15

Oct. AM 1.88*a ± 0.26 0.36 0.37 0.23 PM 0.36bc ± 0.17 1.14a± 0.40 0.34 0.07 TMean 1.12a±0.21 1.10a ± 0.26

'AE c alkali flats; FG = floodplain grassland; SS shrub savanna; IS tree savanna; WL woodland. lst 1st = = = 3AM = 3 hours after daylight; PM = 3 hours before dark. HTMean = Mean for all habitat types per time or month. 4Year = Mean for all months per time and per habitat type. TMean = Mean for both AM and PM per habitat type or month.

because they had a better view of the sur- mals generally have smaller home ranges tory herds travel up to 200 km to reach rounding cover. with regard to the water sources (Estes water in the dry season (Verlinden and The relatively high numbers of herds 1991, Leuthold 1978) during the dry sea- Gavor 1998). during October occurred when available son and are easier to locate. The few The low frequency of sighting of buffalo forage mineral supply was low (Bell 1970, ephemeral pans in the alkali flats, tree herds in all habitats may have resulted Western 1975, McNaughton 1990), and savanna, and shrub savanna may meet the from simply fewer herds with larger num- many herds frequented the alkali flats to water needs of many animals during the bers per herd (Omphile 1997) or from utilize the mineral licks (Klaus et al. 1998) rainy season causing them to disperse their feeding habit described by Kingdon located in this habitat type. However, the widely during this season (Jarman 1979, (1982) as "economic feeding" which availability of evergreen woody plants, Leuthold 1978, Pellew 1984). involves short feeding bouts and longer especially Capparis tomentosa Lam., may The woodland habitat type had the low- rests. Buffalo only graze for an average of have also attracted browsers, such as est frequency of observation although it 9 hours in the 24-hour period, 5 or 6 hours kudu. had the most available herbage. Early of which are at night. This pattern of feed- The high frequency of sighting herbi- sprouting of both herbaceous and woody ing allows only 3 to 4 hours of day-time vores on the floodplains during the dry plants from dry season fires were common activity, rendering it difficult to sight the season marks the importance of the Chobe (Omphile 1997). This habitat type, despite animals during day. Winterbach and River as the major contributing source of being the greatest distance from the Chobe Bothma (1998), however, noted that buffa- water for all animals in the study area. River, is still within the home ranges of all lo in the South African `Sourveld' (a sub- Some species, such as kudu and giraffe, the study herbivores except the impala optimal habitat for buffalo) showed the which obtain much of their water needs (Kingdon 1982). Elephants can cover dis- same circadian rhythm as other ruminants from green leaves (Innis 1958, Owen- tances in excess of 40 km on their daily in the area: a few long feeding periods, Smith 1979) during the wet season, fre- foraging bouts (Douglas-Hamilton 1972). followed by a few long ruminating and quent the floodplain grasslands during dry However, if the animals obtain sufficient resting periods. Regardless, nighttime season. Other water-dependent species forage from other habitat types closer to grazing was favored during the warmer like buffalo (Kingdon 1982, Mugangu et the river, traveling to the woodland would months, and all drinking occurred during al. 1995), elephant (Laws 1970), and be a waste of energy. Resident elephant the day. impala (Estes 1991) also use the flood- herds remain close to permanent water While the Chobe study area provides a plains during the dry season. Also, ani- sources in northern Botswana, but migra- variety of habitat types, it appears the loca-

JOURNAL OF RANGE MANAGEMENT 55(4) July 347 tion of the habitat with regard to water is phant during the dry season. of animals in a limited period of time are an important element affecting elephant The use of the shrub savanna is proba- currently justified in congregating on the distribution as more bands were frequently bly facilitated by the complementary floodplain grassland in the afternoons dur- sighted on the floodplains and the adjacent heavy use of this habitat type by ele- ing the dry season. However, a more wide- alkali flats. Water availability and use of phants. Heavy use over time shortens the spread distribution of water and animal salt licks (Henshaw and Ayeni 1971, Klaus heights of most dominant browse species herds during the dry season should provide et al. 1998) may also contribute to the pref- and reduces canopy cover. However, the for a lower frequency of tourists seeing erence of these habitat types during the dry reduction of woody plant species does not other tourist vehicles and, perhaps, a higher season. Elephants were seen picking and extend into the adjoining tree savanna quality of wildlife-viewing experience for eating soil balls in both the floodplains and which is too densely vegetated and the soil the Park visitors. As in most public wildlife the alkali flats during the dry season. In . too sandy and loose to make good impala reserves, Chobe National Park management addition to providing the major source of habitat. During the dry season, however, is faced with the decision of how best to water most of the year, the floodplains also more herds of impala were recorded in the optimize the biological needs of the ani- provide some green grass during the dry tree savanna when visibility was greater mals and their habitat with the economic season. Given the low levels of nitrogen and forage supply elsewhere within impala and recreational desires of Park users. and phosphorus observed in elephant dung home range had declined. Even then, during the dry season (Omphile 1997), it is impala appeared to concentrate on north- not surprising that elephants spent more ern fringes of this habitat type adjacent to Literature Cited time where the chance of maximizing the alkali flats and the shrub savanna nutrient intake was greater. Most kudu sightings on the alkali flats, Bell, R.H.V. 1970. The use of herb layer by Despite the woodland being the least including those seen on the riverine eco- grazing ungulates in the Serengeti. p.111-123. preferred by giraffe, the habitat was still tone, were recorded during the dry season In: A. Watson (ed.). Animal populations in within their 12 km radius home range when the need for the richest (Fritz et al. relation to their food resources. Blackwell, (Pellew 1984). It was likely that availabili- 1996, Owen-Smith 1979), but not London. ty of adequate good forage in the other necessarily the most varied vegetation, was Child, G. 1968. An ecological survey of northeast- habitats within closer proximity to the relatively high. While the tree and shrub em Botswana. F.A.O. Report T.A. 2563. United savannas offer more species of woody Nations, Rome. Chobe River discouraged frequent use of Child, G. and W. Von Richter. 1968. the distant habitats like the woodlands. The plants than the alkali flats, the quality of Observations on ecology and behavior of alkali flats provided an abundance of an these species in the dry season has linchwe, puku, and waterbuck along the Chobe evergreen woody species Capparis tomen- declined. However, kudu herds were seen River, Botswana. Z. Sangetierk. 34:275-295. tosa Lam., the late dry season blossoming in both habitats quite frequently during the Dekker, B., N. van Rooyen, and J.D. Securinega virosa Roxb., Combretum rainy season indicating wet season prefer- Bothma.1996. Habitat partitioning by ungu- mossambicense Klotzsch, and the late ence for these habitats. Kudu sighted on lates on a game ranch in the Mopani veld. S. senescenting Croton megalobostrys. Both floodplain grassland were either salting or Afr. J. Wildl. Res. 26117-122. the shrub savanna and tree savanna provide drinking from the river; almost none were Douglas-Hamilton, I. 1972. On the ecology of the African elephant. The the widest choice of browse species in seen feeding as browse was limited to areas and behaviour elephants of Lake Manyara. D. Phil. Thesis, outside this habitat. In Zambia, kudu rarely addition to being closer to the river than Oxford Univ. the woodlands. The floodplains provide entered the Kafue River floodplain during Estes, R.D. 1991. The behavior guide to water as well as salt licks and adjacent any season (Sheppe and Osborne 1971). African mammals. Univ. Calif. Press, Los riverine woody species, most of which are Angeles, Calif.. green throughout the dry season. Field, D. 1978. Basic ecology for range man- Impala habitat preference in the Chobe Management Implications agement in Botswana. Min. of Agr., National Park reflects certain aspects of Gaborone. this antelope's special requirements which FGU. Undated. Chobe National Park - a The greater frequency of sighting herbi- description of the resource. (Botswana Dept. often produce an irregular and clumped vores on the floodplain grassland and distribution. Impala are ecotone species Nat. Parks and Wildl. Mimeo,.) adjacent alkali flats and shrub savanna as Fritz, H., M. de Garine-Wichatitsky, and G. preferring light woodland with sparse opposed to the more distant habitat types Letessier. 1996. Habitat use by sympatric undergrowth and grassland of low to of the tree savanna and woodland implies wild and domestic herbivores in an African medium height (Estes 1991). Also impor- a disproportionate use of the Kasane- savanna woodland: The influence of cattle tant are soils with free drainage, firm foot- Ngoma section of the Chobe National spatial behaviour. J. Appl. Ecol. 33 589-598. ing, level ground and free water. The 3 Park. Animal concentrations in this area Henshaw, J. and J. Ayeni. 1971. Some aspects of mineral licks in most preferred habitat types in this study tend to increase competition for limited of big game utilization Yankari Game Reserve, Nigeria. E. Afr. (alkali flats, floodplains grassland and the ungulates with similar diets forage among Wildl. J. 9:73-82. shrub savanna) each offer at least 1 of (Omphile 1997). This disproportionate use report on the these requirements. Henry, F.W.T. 1966. Enumeration can lead to underutilization of some habitat Chobe main forest. Min. of Agr., Gaborone. a The alkali flats also offer variety of types at the expense of others, a condition Innis, A. 1958. The behavior of the giraffe short grasses and scattered evergreens that may be driving the public and some (Giraffa camelopardalis) in the eastern which impala appear to relish during the special interest groups to suggest that the Transvaal. Proc. Zool. Soc. Lond. 35:129-146. dry season. The floodplain is a source of riverfront area of the park is over-utilized Jarman M.V. 1979. Impala social behavior. water as well as short to medium height and, therefore, unattractive to tourists. Beihefte Z. Tierpsychol. 21:1-92. grasses whose dominance appears to be Our data confirm that tour operators Kingdon, J. 1982. East. African mammals: an of evolution in Africa. Vol. 3 part C facilitated (McNaughton 1976, Perrin and primarily in providing their atlas interested (Bovids). Academic Press, New York. Brereton-Stiles 1999) by buffalo, and ele- guests with a view of the greatest numbers

348 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Klaus, G., C. Klaus-Hugi, and B. Schmid. Montfort, A. 1973. Notes sur l'ecologie et le SAS. 1985. SAS user's guide. Version 6 ed. 1998. Geophagy by large mammals at natural comportement des oribis (Qurebia qurebi, SAS Institute Inc. Raleigh, N. C.. licks in the rain forest of the Dzanga National Zimmermann 1783). Terre el Vie. 28:169, Selous, F.C. 1881. A hunter's wanderings in Park, Central African Republic. J. Trop. 208. Africa. London. Ecol.14 829-839. Mugangu, T.E., M.L. Hunter, and J.R. Sheppe, W. and T. Osborne.1971. Patterns of Laws, R.M. 1970. Elephants as agents of habi- Gilbert.1995. Food, water, and predation: A use of a flood plain by Zambian mammals. tat and landscape change in East Africa. study of habitat selection by buffalo in Ecol. Monogr. 41:179-205. Virunga Oikos 21:1-15. National Park, Zaire. Mammalia 59 Steel, R.G.D. and J.H. Torrie.1980. Principles 349-362. Leuthold, B.M. 1978. Day time activity pat- and procedures of statistics. 2nd ed. McGraw- tern of gerenuk and giraffe in Omphile, U.J.1997. Seasonal diets, diet quali- Tsavo National Hill Book Co., Inc. New York N. Y . ty and habitat preference , Park, Kenya. E. Afr. Wildl. J. 16:129-141. of large, savanna Verlinden, A. and I.K.N. Gavor. 1998. ungulates in the Chobe National Park, McNaughton, S.J. 1976. Serengeti migratory Satellite tracking of elephants in northern wildebeest: Botswana. Ph.D. Dissertation, Univ. Wyo., facilitation of energy flow by Botswana. Afr . J . Ecol . . Laramie, Wyo. 36105-116 grazing. Science 191:92-94. Western D. 1975. Water availability Owen-Smith, N. 1979. Assessing the foraging , and its McNaughton, S.J. 1990. Mineral nutrition and influence on the structure and dynamics a efficiency of a large herbivore, the kudu. S. of seasonal movements of African migratory savanna . . Afr. J. Wildl. Res. 9:102-110. large mammal community. E Afr ungulates. Nature 345:6113-6115. Wildl . . . Pienaar, U. de V. 1963. Large mammals of the J 13:265-286 McNeeley, J.A. 1988. Economics and biologi- Winterbach H.E.K . and .D . . . Krueger National Park - their distribution and , J Bothma 1998 cal diversity: developing and using economic Activity patterns of the Cape buffalo Syncerus present day status. Koedoe 6:1-37 . incentives to conserve biological resources. Pellew, R.A. 1984. The feeding ecology of a caffer the Willem Pretorius Game Int. Union Conserv. Nature Nat. Resources, selective browser, the giraffe (Giraffa Reserve , Free State . S . Afr . J . Wildl . Res . Gland, Switzerland. camelopardalis). J. Zool. Lond. 202:57-81. 28:73-81. Miller, O.B. 1939. The mukusi forests of Perrin, M.R. and R. Brereton-Stiles. 1999. Bechuanaland. Emp. For. J.18:193-201. Habitat use and feeding behaviour of the buf- Mills, M.G.L., H.C. Biggs, and I.J. Whyte. falo and the white rhinoceros in the 1995. The relationship between rainfall, lion Hluhluwe-Umfolozi Game Reserve. S. Afr. predation and population trends in African J. Wildl. Res. 29:72-80. herbivores. Wildl. Res. 22 75-87.

JOURNAL OF RANGE MANAGEMENT 55(4) July 349 J. Range Manage. 55: 350-359 July 2002 Plains larkspur (Delphinium geyeri) grazing by cattle in Wyoming

JAMES A. PFISTER, DALE R. GARDNER, BRYAN L. STEGELMEIER, ANTHONY P. KNIGHT, JAMES W. WAGGONER, JR., AND JEFFERY 0. HALL

Pfister, Gardner and Stegelmeier are Rangeland Scientist, Research Chemist, and Veterinary Pathologist, respectively, USDA-ARS Poisonous Plant Research Lab., 1150 E. 1400 N., Logan, Ut. 84341; Knight is Dept. Head, Veterinary Clinical Science, Colorado State Univ., Ft. Collins, Cob; Waggoner is Associate Professor, Dept. of Renewable Resources, University of Wyoming, Laramie, Wyo.; Hall is Veterinary Toxicologist, Dept. of Animal, Dairy, and Vet. Sci., Utah State University, Logan, Ut. 84322. Pfster's email [email protected].

Abstract Resumen

Plains larkspur (Delphinium geyeri Greene) is a major cause of El "Plains larkspur" (Delphinium geyeri Greene) es la princi- cattle deaths in the northern Great Plains of Wyoming and pal causa de muerte de ganado en las Grandes Planicies del Colorado. We examined the amount and timing of larkspur norte de Wyoming y Colorado. Examinamos la cantidad y tiem- ingestion by grazing cattle in relation to larkspur phenology, po de ingestion de "Larkspur" por el ganado en apacentamiento nutrient concentrations, and weather conditions. Four summer en relation a la fenologia del "Larkspur", la concentracion de grazing trials were conducted near Cheyenne (1996 and 1997) nutrientes y las condiciones climaticas. Se condujeron cuatro and Laramie, Wyo. (1998 and 1999). All trials began when plains ensayos de apacentamiento de verano cerca de Cheyenne (1996 y larkspur was vegetative or in the early bud stage. In the first 2 1997) y Laramie (1998 y 1999). Todos los ensayos iniciaron cuan- studies, 6 yearling heifers grazed from 3 May to 4 August 1996; do el "Plains larkspur"estaba en etapa vegetativa o inicios de the same animals plus 5 cow-calf pairs grazed from 13 May to 10 brotacion de yemas. En los primeros dos estudios, 6 vaquillas de August 1997. During both 1996 and 1997, cattle ate 0.5 to 1 % of un ano apacentaron del 3 de mayo al 4 de agosto de 1996; los bites as larkspur during May, then consumption decreased to mismos animales mas 5 pares de vaca-becerro apacentaron del nearly 0 during the remainder of both summers. When eaten, 13 de mayo al 10 de agosto de 1997. En Mayo de ambos aiios, larkspur was typically consumed during cool, foggy weather con- 1996 y 1997, del 0.5 a 1 % de las mordidas del ganado fueron de ditions. In the last 2 studies, 6 cow-calf pairs grazed near "Larkspur", despues, en el resto del verano, el consumo decrecio Laramie, Wyo., from 13 May to 30 June 1998, and 6 different a casi cero. Cuando el "Larkspur" fue consumido tipicamente cow-calf pairs grazed from 2 June to 20 July 1999. Cattle ate ocurrio cuando el clima era frio y neblinoso. En los ultimos dos substantial amounts of plains larkspur (herd average -3%) dur- estudios 6 pares de vaca-becerro apacentaron cerca de Laramie, ing the vegetative and bud stages from mid-May into early June, Wyoming del 13 de mayo al 30 de junio de 1998 y 6 pares de 1998. Cattle may have eaten more larkspur during 1998 because vaca-becerro diferentes apacentaron del 2 de junio al 20 de julio drought reduced spring availability of green grass. Consumption de 1999. Durante las etapas vegetativa a inicio de brotacion de of larkspur was negatively related (r2 = 0.43) to daily tempera- yemas, de mediados de mayo a inicios de junio de 1998, el gana- ture in 1998, but not during 1999. During 1999 cattle ate essen- do comio cantidades substanciales de "Plains larkspur" (prome- tially no plains larkspur during the vegetative and bud stages, dio del hato -3%). El ganado pudo haber comido mas but ate larkspur (herd average -5%) during the flower and pod "Larkspur"durante 1998 porque la sequia redujo la disponibili- stages when larkspur plants were beginning to desiccate and dad de forraje verde en primavera. En 1998, el consumo de ambient temperatures were above average. This series of trials "Larkspur" fue relacionado negativamente (r2 = 0.43) con la indicates that it will be difficult to predict plains larkspur con- temperatura diaria, pero esto no sucedio en 1999. Durante 1999 sumption based on larkspur growth patterns or weather. el ganado esencialmente no comio "Plains larkspur" durante las Although cattle sometimes increase plains larkspur consumption etapas vegetativas a inicio de brotacion de yemas, pero comio when temperatures are cooler than normal, this pattern is not "Larkspur" durante los estados de floracion y fructificacion consistent enough to serve as a basis for management recommen- (promedio del Nato -5%) cuando las plantas de "larkspur" dations. comenzaron a desecarse y las temperaturas ambiente estuvieron arriba del promedio. Esta serie de estudios indica que basados en los patrones de crecimiento o clima sera dificil predecir el consumo de "Plains larkspur". Aunque el ganado algunas veces Key Words: toxic plants, diet selection, cattle grazing, methylly- incrementa el consumo de "Plains larkspur" cuando las temper- caconitine, alkaloids aturas son mas frias de to normal, este patron no es to suficiente- mente consistente para servir como base para recomendaciones The authors wish to thank Kermit Price, Dirk Salmon, Dr. Mike Smith, Dr. de manejo. Mike Ralphs, Lee Woolsey, Jeff Salmon, Al Maciulis, Rex Probst, Danny Hansen, Jason Dormady, and Rick Shea for assistance with various aspects of the study. We appreciate the generous cooperation provided by Jerry Johnson and Mike McGraw, Belvoir Ranch, Cheyenne, Wyoming, and Andy and Mark Johnson, Johnson Ranch, Laramie, Wyoming. Manuscript accepted 24 Oct. 2001.

350 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Plains larkspur (Delphinium geyeri ground to pass a 1-mm screen in a during all active grazing periods until Greene) is a major cause of cattle poison- Cyclone grinder. Grass and larkspur sam- about 1900 hours, when cattle were placed ing in the northern Great Plains of ples were composited for each week, and in a corral for the night. Bites were catego- Wyoming and Colorado (Chesnut 1898, then analyzed for neutral detergent fiber rized as grasses, other forbs, and larkspur Beath 1919, 1925). The perennial larkspur (NDF) using dacron bags in a modifica- bud, flower, pod, leaf, leaf and stem, or typically begins growth in spring before tion of the Van Soest et al. (1991) method. whole plant. We defined an individual bite grasses and other forbs have begun Plant tissue was extracted with boiling as a single cropping motion, always indi- growth, and thus may offer green forage neutral detergent using filter bags in a cated by a head jerk, often accompanied while grasses are dormant (Chesnut 1898). batch fiber analyzer (ANKOM, Fairport, by a visible sweep of the tongue, and inde- Plains larkspur contains numerous diter- N.Y.). An elemental analyzer (NA 2100 pendent of chewing motions. penoid alkaloids, but the principal toxins Protein Nitrogen Analyzer, ThermoQuest Weather variables (temperature, baro- are methyllycaconitine (MLA) and nudi- Italia S.p.A., Milan, Italy) was used to metric pressure, relative humidity, wind cauline (Manners et al. 1995). determine nitrogen content, and crude pro- speed and direction, precipitation) were No grazing studies have examined cattle tein content was calculated (AOAC 1990). monitored on site during the grazing stud- consumption of plains larkspur. The Grass and larkspur samples were also ies using an automated weather station objective of this study was to evaluate the analyzed for micronutrient concentrations (Campbell Scientific, Logan, Ut.). amount and timing of larkspur consump- using Inductively Coupled Plasma Mass Climatology information was supplement- tion by cattle in relation to its phenology, Spectroscopy (ICP-MS). Plant tissue was ed using data from the Cheyenne airport toxicity, and other potentially important digested via a modification of EPA located 18 km from the study site. variables such as weather. method 3050 (Smith 1994). Digestions were performed in screw-cap teflon tubes, Laramie 1998 and 1999 using of ground material in 10 0.5 g plant Two trials were conducted on the Materials and Methods ml of trace metal grade nitric acid at 90° C Laramie Plains about 10 km southwest of for 2 hours. The plant digests were diluted Laramie, Wyo. (41 ° 17.19' N 105° 43.83' 18.3 ml Cheyenne 1996 and 1997 with of ultrapure water to 5% W; 2,225 m elevation). Larkspur density, Two trials were conducted during 1996 nitric acid prior to analysis. Standard phenology, and alkaloid concentration curves for each element consisted of 5 and 1997 on the Belvoir Ranch about 15 were determined as detailed above. Forage ° concentrations between 10 and 2,500 km west of Cheyenne, Wyo. (41 5.40' N ng availability and nutrient status was deter- ml 1. Quality control checks and standard 105° 0.31' W; 2,026 m elevation). Western mined by clipping weekly and handling the curves were analyzed every 5 samples. wheatgrass (Pascopyrum smithii [Rydb.] samples as noted above, except that grass A moderate stocking rate was A. Love) and blue grama (Bouteloua gra- used dur- samples for nutrient analysis were inadver- ing all trials. Six yearling heifers (305 cilis [H.B.K.] Lag. ex Steud.) were the kg) tently discarded in 1998. Forbs were gen- grazed from 3 May to 4 August 1996 on dominant constituents of the vegetation on erally a minor component in the vegetation the 15 ha pasture. Heifers had not been the mixed grass prairie site; plains lark- each year, except for 1999. Because of a spur was a major forb. previously exposed to larkspur. The trial tremendous growth of Drummond's began when plains larkspur was 3 to 6 cm Forage availability was determined at the milkvetch (Astragalus drummondii Dougl. tall (i.e., vegetative). The trial was divided beginning and end of the grazing trials by ex. Hook.) on the site during 1999, we into 4 periods corresponding to larkspur clipping 30, 0.25m2 quadrats placed along 2 sampled milkvetch and other forbs in addi- phenology: vegetative (3 to 16 May), veg- pace transects through the pasture. Clipped tion to grasses and larkspur. We separated etative/bud (17 May to 7 June), bud/early samples were placed into paper bags, and dry and green grass in 1998, but did not in flower (8 June to 11 July), and flower/pod oven-dried at 60° C to a constant weight. 1999 as drought and grazing the previous (12 July to 4 August). During 1997, 5 Density of plains larkspur was determined of year greatly reduced the amount of residual the same heifers (then 2 years old) grazed by counting the number of larkspur plants dry grass on the study site. site, 5 in 300, 0.25 m2 quadrats placed systemati- the study plus an additional cow- During 1998, 6 cow-calf pairs grazed on no cally along 3 pace transects. calf pairs that had previous experience the 16 ha pasture from 13 May to 30 June. Larkspur phenology was assessed by with plains larkspur. Cattle grazed from Based on larkspur phenology, the 48-day marking 30 plants just outside the study 13 May to 10 Aug 1997, and the study trial was divided into 4 periods corre- pasture, and weekly observations were pasture was enlarged to 24 ha. The trial sponding to vegetative/early bud (14 to 25 made on each plant. Larkspur plants were was divided into 4 periods corresponding May), bud stage (26 May to 6 June), late collected weekly for toxic alkaloid and to larkspur growth: vegetative (13 to 24 bud/early flower with most plants green (7 nutrient analysis (see below). Twenty lark- May), vegetative/bud (26 May to 25 June), to 18 June), and late bud/flower with most spur plants at each phenological stage flower (26 June to 10 July), and late plants becoming desiccated (19 to 30 were harvested and composited by growth flower/pod (11 July to 10 August). Trace June). During 1999 the pasture was stage. These weekly samples were frozen mineral blocks were offered free choice enlarged to 20 ha, and was grazed by 6 in plastic bags, freeze-dried, ground during this and all subsequent trials. different cow-calf pairs from 2 June to 20 through a 1-mm screen, then analyzed for Daily bite counts were used to deter- July. Based on larkspur phenology in concentrations of toxic alkaloids by elec- mine animal diets (Pfister et al. 1997b). 1999, the trial was divided into 3 periods: (2 trospray mass spectrometry (Gardner et al. Beginning at 0700 hours every day, indi- vegetative/early bud June to 11 June); 1999). vidual animals were focally sampled bud/early flower (12 June to 2 July); (3 Grass samples for nutrient analysis were (Altmann 1974) in a predetermined ran- flower/early pod July to 20 July). As far as we know, larkspur was a novel food for clipped weekly from five, 0.25 m2 plots in dom order. Each animal was observed in all animals at the beginning of each sum- 5 representative areas of the pasture, oven turn for 5 minutes. After all animals had mer's trial near Laramie. During 1998 and dried at 40° C to a constant weight, then been observed, the process was repeated 1999, bite counts began at dawn (0500 to

JOURNAL OF RANGE MANAGEMENT 55(4) July 351 0520 hours) each day and continued dur- Cheyenne Temperature Laramie Temperature ing active grazing periods until about 1800 hours. The cattle were placed into a corral 20 18 each evening and group-fed 600 g head' of a mix of rolled barley and melenges- 16 terol acetate (550 mg cow-' day-') to pre- 14 vent estrus and avoid interference from 12 bulls grazing the adjoining pasture. 10 Weather variables were measured using 8 an automated weather station as noted pre- 6 viously. Climatology information was 4 supplemented using data from the Laramie 2 airport located 7 km east of the study site. Apr May Jun Jul Aug Apr May Jun Jul

Statistical Analysis Cheyenne Precipitation Laramie Precipitation 120 100 Bite count data were analyzed using the 88-year average SAS 110 90 GLM procedure of (1998). Each year 1998 was analyzed separately, with animals 100 80 ®- 1999 90 considered as blocks, and periods as treat- 70 80 ments, with repeated measurements over 60 days within periods. Stepwise multiple 70 50 regression was used to examine relation- 60 50 ships between larkspur consumption (the 40 40 30 dependent variable) and weather and for- 30 20 age nutrient variables. Larkspur consump- 20 tion in relation to weather was examined 10 10 in 12 and 24 hour blocks. Mineral values 0 0 for plants were averaged by weeks to pro- Apr May Jun Jul Aug Apr May Jun Jul vide a monthly mean. Regression analysis Month Month to relate forage nutrient (i.e., mineral, NDF, and crude protein) concentrations Fig. 1. Temperature (monthly average in degrees C) and precipitation (monthly total in mm) were done using weekly averages for lark- during spring and summer grazing trials for Cheyenne (1996 and 1997) and Laramie, spur consumption. No statistical analysis Wyo. (1998 and 1999). was done on weekly protein, NDF, or alkaloid concentrations from the compos- Larkspur Density, Phenology and Cattle Diets ited samples. Alkaloid Concentration Cattle ate about 1% of bites as larkspur Larkspur density was 2.4 and 2.2 plants during May 1996, then consumption 2 Results m during 1996 and 1997, respectively. decreased to nearly 0 in June. Cattle ate All larkspur plants were vegetative when essentially no larkspur during the last 2 the grazing study began in early May each months of the study during 1996 (data not Cheyenne 1996 and 1997 year (Fig. 2). During 1996 larkspur growth shown), and periods did not differ (P> Weather was diverse, with some plants maturing 0.10). Over the entire 1996 trial, grasses Average daily temperatures were close while others were emerging. Most larkspur and other forbs comprised 91.7 and 8.0% to normal during 1996 and from May to plants were fully mature or dead by late of bites, respectively. The only day of August 1997, but were colder than normal July, 1996. Growth was initially slower substantial larkspur consumption was 17 during April 1997 (Fig. 1). Summer pre- and more uniform during 1997, but in late May 1996, a hot (maximum temperature cipitation was near or exceeded normal June larkspur plants quickly began flower- 30° C) and windy day. Two cows ate up to 1). amounts during both years (Fig. ing. These flowering plants were slow to 30% of their bites as larkspur leaves and Summer forage production is highly corre- produce pods, and many larkspur plants stems during several grazing bouts, and lated with precipitation during March, senesced during mid-to-late July, 1997. one was visibly intoxicated during the April and early May (Rauzi 1964, Hart Plains larkspur plants during 1996 were afternoon. Visible signs included stagger- and Samuel 1985), and during both years precipitation during those months was highly toxic (>6 mg toxic alkaloid g' plant ing gait, muscular tremors, and periodic close to normal. dry matter) early in the vegetative stage sternal recumbency. (Table 2), and remained moderately toxic Larkspur consumption was very low (< (3-6 mg g-') until mid July when they 0.2% of bites) during all periods in 1997, Forage Availability were in the late flower or pod stages of and periods did not differ (P > 0.10) On 2 About 80% of the available forage was growth. Larkspur plants showed similar cold, damp days (19 and 21 May 1997) 2 grass during the 2 trials near Cheyenne, patterns during 1997, except that vegeta- cows ate > 15% of bites as larkspur in the with plains larkspur comprising about 6 to tive plants were not as high in alkaloid morning, but showed no signs of intoxica- 12% of the standing crop (Table 1). In concentrations as in 1996, and concentration. Nonetheless, there was no relation- general, forbs other than larkspur were a tions remained moderately toxic (> 3 mg ship P > 0.1) between weather variables minor component of the standing crop in g') even when plants were largely senes- and larkspur consumption. Cattle diets the pasture. cent in early August (Fig. 2). were primarily grasses (93.5%) and other forbs (> 6%) during the 1997 trial.

352 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 tares were cooler than normal until July l Table 1. Standing crop (kg ha + SE) during plains larkspur grazing trials during summer, 1996 1999, while precipitation was at or slightly and 1997 (Cheyenne), and 1998 and 1999 (Laramie), Wyo. above normal in April and May 1999, then falling below normal levels in June and Forage Class Sample Period July 1999. Begin End 3 May-4 August 1996 ------(kgha )------Grass 663 ± 96 511 ± 63 Forage availability Forbs 47 ± 15 100 ± 25 Forage growth was delayed by drought Larkspur 47±11 68±12 in spring, 1998. Grasses made up more than 50% of the standing crop, and most 12 May-10 August 1997 of the grasses had begun green growth in Grass 322±38 mid-May Forbs 36±10 90±54 1998 (Table 1). During 1999, Larkspur 49 ± 14 63 ± 28 forage growth was delayed by colder than normal weather even though precipitation 13 May-30 June 1998 was adequate, and the 1999 grazing trial Dry grass 37±5 15±6 started almost 3 weeks later than did the Green grass 125±9 157±11 1998 trial at the same location. When the Forbs 97±11 38±6 study Larkspur 60±13 3±2 began, most grasses had initiated green growth, as had Drummond's

1 June-20 July 1999 milkvetch. Grass 178 ± 14 16 Forbs 28± 17 17±5 Larkspur 89 ± 18 63 ± 20 Larkspur Density, Phenology, and Milkvetch (Astragalus drummondii) 32 ± 12 8±3 Alkaloid Concentration Larkspur density averaged 1.7 and 1.9 plants m 2 during 1998 and 1999, respec- Nutrient Content of Forage increased to above normal amounts during tively. The drought during 1998 retarded We found no relationships between June 1998 (Fig. 1). During 1999, tempera- the development of larkspur plants, as nutrient or mineral content of the forage and larkspur consumption. Plains larkspur Table 2. Toxic alkaloid concentrationa (mg toxic alkaloid g'1 plant dry matter) of plains larkspur averaged more than 13% crude protein (Delphinium geyeri) during summer grazing studies near Cheyenne, Wyo. during 1996 and 1997. during 1996, while grasses averaged 9% crude protein, but by the end of the study 1996 1997 both plant types contained about 8% crude Phenological Date Toxic alkaloid alkaloid protein. Conversely, grasses contained Stage Concentration Concentration more than 66% NDF, whereas larkspur --(mg g 1)-- --(mg g 1)-- averaged 34% NDF (Fig. 3). During 1997 Vegetative 3 May 7.0 May larkspur averaged 12% crude protein com- 10 May 5.1 21 May pared to 10% for grasses, but grasses and l7 May 4.9 28 May 24 May larkspur had similar crude protein concen- 3.9 4 June 31 May 4.1 trations after May 1997. Grasses had NDF 7 June 3.3 concentrations of 61% while larkspur 14 June 3.2 averaged 34% NDF (Fig. 3). 28 June 2.1 Both larkspur and grasses were highest 5 July 1.4 in mineral concentrations when immature, Bud 31 May . and concentrations decreased with maturi- 3.9 June 7 June 3.2 11 June ty (Table 3). Larkspurs generally exceeded 14 June 3.8 17 June grass as a source of calcium, phosphorus, 21 June 3.6 25 June potassium, and magnesium. Both larkspur 28 June 3.5 1 July and grasses were low in copper and sodi- S July 3.0 um, and had marginally low concentrations of zinc and manganese. Cobalt, Flower 14 June 3.6 July 21 June 3.4 9 July molybdenum, iron and selenium concen- 28 June 3.3 16 July trations were adequate during each month 5 July 3.5 (data not shown for Fe and Se). 12 July 2.8 10 July 2.8 26 July 1.8 Laramie 1998 and 1999 Weather Pod 12 July 4.0 July Average ambient temperatures were at 19 July 2.6 31 July or above normal during spring, 1998, but 26 July 2.4 6 August 2 August 2.1 June 1998 was cooler than normal (Fig. 1). aLakkspur Conversely, precipitation was lower than samples were composites of twenty individual plants. Toxic alkaloid concentrations indicate the amount of MSAL-type alkaloids, of which methyllycaconitine (MLA) is the dominant alkaloid. Concentrations of nudicauline, a normal during April and May, 1998, but potent alkaloid in plains larkspurs, were very low.

JOURNAL OF RANGE MANAGEMENT 55(4) July 353 1996 consumption during 1998, but not 1999, was negatively related (r2 = 0.43) to maximum and minimum daily temperature. H The regression equation relating temperature to larkspur consumption was y =12.4 - 0.16x - 0.09z, where y is larkspur consumption, x is minimum daily temperature (degree C) and z is maximum daily temperature. Larkspur averaged < 0.01, 0.9 and 4.8% of bites (P < 0.05) for the 3 periods in 1999 corresponding to early, mid-, and late-maturity for larkspur, respectively (Fig. 5). Grasses comprised 96, 88, and 82% of bites during periods 1 to 3, respectively, during 1999. Cattle ate substantial amounts (7 to 10% of bites) of Drummond's milkvetch during periods 2 and 3, whereas 1997 consumption of other forbs (exclusive of I larkspur and milkvetch) did not exceed 5% 100 100 100 100 100 77 77 77 77 50 47 20 7 of bites in any period.

80 Nutrient Content of Forage Plains larkspur averaged 12% and 32% so crude protein and NDF, respectively, during 1998 (Fig. 3). During 1999 larkspur ao averaged 9.5% crude protein compared to 9.8% for grasses. The crude protein con- 20 tent of plains larkspur was negatively related to weekly larkspur consumption (r2 0I = 0.51) during 1999 as cattle ate more >% > > C larkspur when plants were maturing (Fig. 5). Grasses had NDF concentrations of N 0) t0 N 0 0 M 0 '- N r N M 59% while larkspur averaged 33% NDF Dates during 1999. Drummond's milkvetch and other forbs had higher crude protein con- Vegetative Bud Flower Pod ® ® centrations and similar NDF concentra- Fig. 2. Proportions (% of marked plants) of plains larkspur (Delphinium geyeri) in various tions compared to the other forage types. phenological growth stages during summer, 1996 and 1997, near Cheyenne, Wyo. Thirty All measured mineral elements except plants were marked during each year. The number on top of each bar represents the per- Zn increased in plains larkspur from June centage of marked plants still alive when the observations were taken. to July, 1998 (Table 5). Larkspur plants increased mineral concentrations greatly most died before flowering (Fig. 4). In Cattle Diets (except for Zn) from the June to July, addition, larkspur plants appeared to be Cattle averaged almost 2% of bites as 1998 sampling periods. None of the differ- shorter than normal although we did not plains larkspur during 1998 (Fig. 5) and ent forage types contained adequate Na measure height. Larkspur growth was periods differed (P < 0.05). Cattle ate concentrations for lactating cattle (NRC delayed during May 1999 from colder about 1% of bites as larkspur during the 1996); except for forbs, all forage types than normal temperatures, and most plants vegetative and early bud stage (14-25 were also inadequate in Mn concentration. were vegetative or in the bud stage when May), and this increased to about 3% of Grasses were low in both P and Zn in June the trial began in early June. Dry condi- bites during the bud stage (26 May-6 and July 1999. Furthermore, weekly lark- tions during June and July also hampered June), decreased to about 2% during the spur consumption during both 1998 and the development of larkspur plants, and late bud and early flower stages (7-18 1999 was related to Zn concentration in most dried out in early July without pro- June), and further declined to < 1% during plains larkspur (r2 = 0.58 and 0.79, respec- ducing pods. the flower stage (19-30 June). Grasses tively). No other mineral variables were Plains larkspur plants were moderately comprised about 95% of the diet with related to larkspur consumption (P > high (> 5 mg g') in toxic alkaloid concen- other forbs comprising about 3% of bites. 0.15). tration at the beginning of the trials in both During a 2-week period from about 21 1998 and 1999. Toxic alkaloid concentra- May to 2 June, cattle often had grazing tions remained relatively stable (about 3 to bouts where larkspur exceeded 15-20% of Discussion 4 mg g') throughout both grazing trials bites, and numerous animals were visibly (Table 4). Drummond's milkvetch con- intoxicated (i.e., staggering and sternal Consumption of Plains Larkspur tained very low concentrations (< 0.1%) No animals were fatally recumbency). Cattle ate very little plains larkspur dur- of the indolizidine alkaloid, swainsonine intoxicated during the study. Larkspur (Gardner, unpublished data). ing the grazing trials in Cheyenne except

354 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Table 3. Mineral concentration (mg kgl) of plains larkspur and grasses during summer, 1996-1997 near Cheyenne, Wyo.

June July August Elementa Larkspur Grass 1996

Ca 16895.5 2959.7 K 37255.1 11488.2 Mg 2760.5 565.7 Na 25.2 22.0 Cu 9.5 3.7 Co 0.08 0.07 Mn 43.9 31.0 Mo 1.2 1.4 P 3377.0 1877.9 Zn 26.9 16.9

1997

Ca 17566.1 3374.1 K 37086.8 13593.9 Mg 2971.7 683.0 Na 151.7 14.9 Cu 7.8 4.3 Co 0.07 0.07 Mn 34.7 28.2 Mo 0.6 1.6 P 3334.9 2103.6 Zn 22.3 34.3 aRequirements 1) (mg kg of mature beef cow during early lactation: Ca - 3400; K - 8000; Mg - 2000; Na - 1000; Cu - 10; Co - 0.1;Mn - 40; Mo - <5; P - 2400; Zn - 30. Source: NRC 1996) during brief periods in May of each year. have suggested that cattle eat plains lark- larkspur in Laramie during early July Those brief periods were (with one major spur during flowering, but consumption 1999, when grass was relatively abundant exception on 17 May 1996) characterized was essentially 0 during flowering at the and availability of the most preferred forb, by cool, foggy weather conditions. Even Cheyenne site. The study was not Drummond's milkvetch, was greatly so, cattle were very inconsistent and ate no designed to compare larkspur consump- reduced by grazing. larkspur on many damp, cool days. The tion at Cheyenne and Laramie, and any In previous work we have shown that very low consumption of plains larkspur comparisons are confounded over time. alkaloid (Pfister et al. 1996) or carbohy- in the Cheyenne grazing trials was unex- Even so, availability of grasses was higher drate (Ralphs, unpublished data) concen- pected because plains larkspur populations at Cheyenne than at Laramie, and this may trations have little impact on consumption were abundant and livestock losses are have influenced cattle consumption of of tall larkspurs by cattle. In some sum- common in this area (A.P. Knight, person- plains larkspur. In contrast, cattle ate most mers cattle have eaten low amounts of tall al communication). Nonetheless, reported livestock losses were few during the Table 4. Toxic alkaloid concentrationa (mg toxic alkaloid g't plant dry matter) of plains larkspur spring and summer of both 1996 and (Delphinium geyeri) during summer grazing studies near Laramie, Wyo. during 1998 and 1999. 1997. We are aware of fewer than 10 reported cattle deaths during both sum- 1998 1999 mers in the Cheyenne area. Larkspur was Phenological Date Toxic alkaloid alkaloid not highly toxic during most of both sum- Stage Concentration Concentration mers (i.e., < 4 mg g' of toxic alkaloid), --(mg g- )-- --(mg g )-- and we surmise that consumption of lark- Vegetative 16 May 5.4 June spur in our grazing trials mirrored trends 23 May 4.7 12 June in the local cattle herds. 30 May 4.1 We have no plausible explanation for Bud S June 3.9 June the low consumption during the 2 sum- 13 June 3.6 12 June mers in Cheyenne. The Cheyenne site had 20 June 4.2 18 June abundant grass each year, and most of the cattle diets were composed of grasses. Flower 27 June 4.3 June Local ranchers believe that cattle tend to 25 June 4.5 3 July 4.5 eat more plains larkspur early in the season when plains larkspur has initiated Pod 9 July 3.5 growth, but grasses lag behind in green 16 July 3.8 aLarkspur growth. Even though we began the studies samples were composites of twenty individual plants. Toxic alkaloid concentrations indicate the amount of in May each year when plains larkspur MSAL-type alkaloids, of which methyllycaconitine (MLA) is the dominant alkaloid. Concentrations of nudicauline, a was only 3-6 cm tall, cattle ate little or no potent alkaloid in plains larkspurs, were very low. Larkspur plants did not produce pods during the drought of 1998. larkspur during these times. Ranchers also

JOURNAL OF RANGE MANAGEMENT 55(4) July 355 30 1998 -.- Plains Larkspur 25 . 0, Grasses 1 20

70 0 0 60 0 0.0 0. 0.0 .0 ,0.0.0 0 U- z C C C C i --- Plains Larkspur C Ca .0 Grasses 7 7 7 10 Cl C0 Cl 0) CD M 0 Cl Cl ) ) a®y7o), m7?y 79 ?7 ?e 9 77 7 S ? S 7 7s 1999 Cheyenne Dates 1996 Cheyenne Dates 1997 I 100 100 100 57 53 30 30 100 I -- Plains Larkspur 25 25 0 Grasses ® - Drummond locoweed I 20 20 . ® . Other forbs

15 15 I

10 10

5 5 40 I

0 0 20 80 80 70 70 0z 60 -S Plains Larkspur 60 7 50 50 7 7 7 C) C0 40 40 Cl r r 30 30 Dates 20 20 -- Plains Larkspur . 0 Grasses Vegetative Bud Flower Pod 10 -.- Drummond locoweed ® ® ® . .5. 0 Other forbs 7o.y ? d8y Fig. 4. Proportions (% of marked plants) of plains lark- Laramie Dates 1998 Laramie Dates 1999 spur (Delphinium geyeri) in various phenological growth stages during summer, 1998 and 1999, near Fig. 3. Crude protein and NDF (% of dry matter) concentrations in weekly com- Laramie, Wyoming. Thirty plants were marked dur- posited forage samples from grazing studies in Cheyenne and Laramie, Wyo. ing each year. The number on top of each bar repre- during the summers of 1996, 1997, 1998, and 1999. sents the percentage of marked plants still alive when the observations were taken. larkspurs (Pfister et al. 1999), and we have ended. Cattle may have eaten more lark- during cooler and wetter periods (Pfister et attributed these patterns to summer spur in response to the reduced availability al. 1988, Ralphs et al. 1994). Even though drought (Pfister and Manners 1991, 1995). of green grass early in the grazing season. the early summer of 1999 was colder than Unlike tall larkspurs, there was no indica- On the contrary, during 1999 cattle ate normal, we did not find a relationship tion that consumption of plains larkspur essentially no plains larkspur during the between temperature and larkspur con- was reduced during drought except that vegetative and bud stages, but ate most sumption, as cattle ate little larkspur dur- limited moisture at Laramie in 1998 larkspur during the flower and pod stages. ing the cooler period. Most larkspur con- reduced plains larkspur phytomass and The study pasture had adequate spring sumption occurred during early July when hence availability to grazing animals. (i.e., April and May) moisture during larkspur plants were beginning to dry out Different larkspur consumption patterns 1999, but received lower than normal pre- and average daily temperatures were at or emerged in the Laramie grazing studies. cipitation during June and July. above normal. During 1998, cattle ate substantial Cattle had somewhat different daily eat- Larkspur consumption averaged over amounts of plains larkspur during the veg- ing patterns during the 2 summer trials in groups of animals does not clearly show etative and bud stages from mid-May into Laramie. Cattle ate larkspur at various trends by individual animals (Pfister et al. early June. The drought during spring times throughout the day in 1998, but par- 1997a). We consolidated larkspur selec- 1998, severely retarded growth of both ticularly at times when cooler tempera- tion data for each animal from 3 consecu- larkspur and grasses; plains larkspur never tures prevailed. Damp, cold, and fog tive observation periods, termed here as a recovered from this lack of initial moisture seemed to be associated with plains lark- grazing bout, when larkspur consumption and much of the shorter-than-normal larkspur consumption for reasons that are not was highest during each day. Individual spur in the pasture had either been grazed clear. We have noted in other studies that maxima clearly highlight the periods of or was desiccated and dead when the study cattle avidly eat other larkspur species greatest consumption in Laramie during

356 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Table 5. Mineral concentration (mg kg 1) of plains larkspur and grasses during summer, 1998-1999 near Laramie, Wyoming.

June July Element Larkspur Other forbs forbs 1998

Ca 2661.8 10825.7 K 4555.2 11393.9 Mg 521.7 2419.6 Na 14.7 68.1 Cu 3.1 13.8 Co 0.38 2.83 Mn 4.4 14.0 Mo 0.4 1.5 P 586.9 1626.0 Zn 54.7 36.6

1999

Ca 15430.9 11662.6 K 26502.5 25222.7 Mg 3427.1 3389.5 Na 236.6 49.7 Cu 15.0 25.7 Co 1.40 2.71 Mn 20.3 51.7 Mo 1.9 2.1 P 3131.9 2963.0 Zn 29.7 37.3

1998 and 1999 (Fig. 5). Larkspur con- Larkspur Consumption and Forage ratio. It is possible that Ca-deficient cows sumption by individual animals often Nutrient Concentration might eat more Ca-rich larkspur, but it exceeded 20% of their diet during these One commonly recommended means to doesn't appear that grasses in this region grazing bouts, and in some cases exceeded reduce larkspur losses is through mineral are deficient in Ca during early to mid- 50% of bites. We have shown in other supplementation (Hughes 1941, Logan summer. work that cattle consume tall larkspur in a 1973, Knowles 1974). Anecdotal accounts As expected, the forage in all 4 years cyclic fashion, with days of higher con- suggest that dicalcium phosphate is often was deficient in Na. Both grasses and lark- sumption followed by a detoxification supplemented in hopes of reducing larkspur were deficient in Cu (i.e, < 10 mg kg') period and lower consumption for 1 or 2 spur consumption by cattle. Larkspur is an at the Cheyenne site during both summers, days (Pfister et al. 1997a). Cattle in the excellent source of Ca, generally far but Cu concentrations were adequate in Laramie grazing studies followed a similar exceeding the 3,000 mg kg-' requirement both grasses and larkspur during 1999 in cyclic pattern. It was particularly notewor- for lactating animals; grasses also often Laramie. Zinc was the only mineral ele- thy that larkspur consumption sufficient to exceeded this required amount of Ca. ment in larkspur that was related to con- cause temporary intoxication (i.e., short The forage P requirement for early lac- sumption, and this relationship was term paralysis and recumbency) was usu- tating cows is 1900 mg kg' (forage dry observed during both 1998 and 1999 in ally followed by very low consumption for wt., NRC 1996), and larkspur often Laramie. Zinc concentrations were gener- 1 or 2 days. exceeded this amount whereas grasses ally inadequate (< 30 mg kg-') in forage We periodically monitored calves for were sometimes deficient (Tables 3 and from Cheyenne; except for grasses, for- diet selection for 2 or 3 minutes for each 5). Cattle ate substantial amounts of lark- ages in Laramie contained adequate Zn observation, although total grazing activity spur in late May and early June 1998 concentrations. Because other forages in by calves was sporadic and often of short when plains larkspur had low P concentra- Laramie also contained adequate zinc con- duration. Calves in all the grazing studies tions. During a period of major larkspur centrations, it is not clear if the association appeared to graze larkspur in concert with consumption in July 1999, grasses and between zinc concentration and larkspur their adult cohorts. Calves ate essentially larkspur both met or exceeded P require- consumption is biologically significant. no larkspur in Cheyenne, whereas con- ments. Plains larkspur was generally higher in sumption in the Laramie trials was some- The ideal Ca:P ratio is 1:1 but not protein than were grasses early in the times very high (> 50% of bites) for brief greater than 7:1 (NRC 1996); the ratio in growing season, but differences were usu- periods. We speculate that consumption as larkspur was 9.3, 7.5, 5.6 and 5.6 in 1996 ally minor after larkspur matured. Crude calves predisposes these individuals to to 1999, respectively. The Ca:P ratio in protein concentration of larkspur in greater acceptability of larkspur in future grasses was generally 1.5 to 1.7 (Tables 3 Laramie during 1999 was negatively relat- years, as noted by studies on other plants and 5). Cattle ate little or no larkspur when ed to larkspur consumption because cattle (Distel and Provenza 1991, Distel et al. the Ca:P ratio was highest in 1996 and ate more larkspur when larkspur was 1996), but we have not tested this notion. 1997, suggesting that high Ca concentra- maturing and nitrogen concentrations were tions in larkspur do not promote increased declining. We doubt if crude protein con- consumption, at least in cows that are centration had any real effect on larkspur grazing grasses with an adequate Ca:P consumption.

JOURNAL OF RANGE MANAGEMENT 55(4) July 357 larkspur consumption is typically not 70 steady during the day, but cattle often 65 4 -- Mean Larkspur Consumption have grazing bouts when much of the diet 35 -g- Maximum Larkspur Consumption consists of larkspur. Larkspur during this 30 period averaged about 4 to 5 mg toxic alkaloid g' dry wt. (Table 4), thus cattle were ingesting about 8.1 to 13.5 g day' of alkaloid. A toxic, but not fatal, dose of tall larkspur alkaloids in cattle is 21 mg kg"' body weight (Pfister et al. 1994). For cattle weighing 500 kg, the 21 mg kg-' value indicates a toxic alkaloid dose of about 10.5 g. From these calculations it appears

C C C that plains larkspur has similar or perhaps 7 slightly greater toxicity than tall larkspurs. M) 0 1A o 0) 0) 0) N N M N Interestingly, we did not observe as many Vegetative/early bud bud bud/early flower late bud/flower cases of overt intoxication during 1999 as 1998 Date and Larkspur Phenology during 1998, even though cattle ate more larkspur at times during 1999. Toxic alkaloid concentrations were similar, and differences in eating patterns may have been responsible. As previously noted, during 1998 cattle ate larkspur steadily throughout the day, whereas consumption in 1999 was characterized by more rapid consumption during fewer grazing bouts. Both sets of experimental animals were lactating, but during 1998 the cows were in a lower body condition than during 1999. Lower body condition may alter the absorption

C C and elimination patterns of toxic alkaloids 7 7 (Lopez and Launchbaugh, unpublished N ti N N h N ti N ti t- r N N T r data), thus rendering thin animals more Vegetative/bud bud/early flower flower/early pod susceptible to intoxication. 1999 Date and Larkspur Phenology

Fig. 5. Larkspur consumption (% of bites) by cattle during spring and summer, 1998 to Conclusions 1999, near Laramie, Wyo. Mean larkspur consumption reflects mean consumption by all experimental cows. Maxima reflect larkspur consumption data that were consolidated for each experimental animal from 3 consecutive observation periods, termed here as a graz- This series of trials suggests that it will ing bout, when larkspur consumption was highest during each day. Only the single highest be difficult to predict plains larkspur con- daily maximum from among all grazing animals (i.e., one animal) is shown. sumption based on larkspur growth patterns, nutrient concentrations, or weather. Results for protein and NDF are typical dose of plains larkspur. Grazing cattle As with tall larkspurs, cattle sometimes of comparisons between forbs and grasses often displayed signs of larkspur toxicosis increased plains larkspur consumption (Cook and Harris 1968, Abouguendia during May and early June 1998, and early when temperatures were cooler than nor- 1998, Perez Corona et al. 1998). Taken July 1999, as on many occasions cattle mal. Nonetheless, this pattern was not together, there was little evidence that were temporarily paralyzed and unable to consistent and therefore will not serve as a consumption (or lack thereof) of larkspur walk or stand. We did not treat any intoxi- basis for management recommendations was driven by mineral or nutrient concen- cated animals, although at times we were that can reduce risk and losses. Currently, trations. As with tall larkspurs, we recom- uncertain if the affected cattle would sur- our best management recommendation is mend that livestock producers make sup- vive (none died). Affected cattle averaged to ensure that other desirable forage production is ample when graze in plementation decisions based on forage 5% of total daily bites (n = 24, x = 5.1, SD cattle deficiencies and cattle nutrient require- = 2.6) as plains larkspur on days when plains larkspur-infested pastures. Extra ments, but not as a basis for altering con- overt toxicity was noted. Cattle had total caution may be warranted if cattle are in sumption of larkspur by cattle (Pfister and biting rates of about 45 bites minute'. low body condition. Additionally, if losses Manners 1991, 1995). Individual larkspur plants weighed 1.5 to 2 persist from year-to-year, the flat terrain in many areas may provide an opportunity to g plant"' (dry wt.) and cattle generally ingested most or all of one larkspur plant use tractor-mounted herbicidal control to Toxicity of Plains Larkspur reduce dense larkspur patches. Whitson et No toxicology studies have established bite' (Pfister unpublished data). Thus, on a typical day ('-10 hours of actual grazing al. (1992) has shown that herbicidal con- the threshold of toxicity for plains larkspur trol of plains larkspurs can effectively in cattle. Therefore we estimated the time) during this period cattle ingested about 1,350 bites reduce dense populations. amount of larkspur that constituted a toxic of larkspur day' or 2.0 to 2.7 kg of larkspur day"'. We note that

358 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Literature Cited Hughes, H.R. 1941. Larkspur poisoning pre- Pfister, J.A., G.D. Manners, M.H. Ralphs, ventative. San Juan District, Rocky Mountain Z.X. Hong, and M.A. Lane. 1988. Effects Regional Bull., USDA Forest Service. of phenology, site and rumen fill on tall lark- Abouguendia, Z. 1998. Seasonal pattern of Knowles, K. 1974. An evaluation of larkspur spur consumption by cattle. J. Range nutrient content of native plants: an poisoning in cattle and the trampling damage Manage. 41:509-514. overview. The Grazing Gazette, Vol. 7, that occurs during grazing on a summer Pfister, J.A., F.D. Provenza, G.D. Manners, No.l, Grazing and Pasture Tech. Prog., range in eastern Idaho. M.S. Thesis, Univ. of D.R. Gardner, and M.H. Ralphs. 1997a. Regina, Sask., Can. Idaho, Moscow, Ida. Tall larkspur ingestion: can cattle regulate Altmann, J. 1974. Observational study of Logan, J.R.1973. Evaluation of a specifically- intake below toxic levels? J. Chem. Ecol. behavior: sampling methods. Behav. formulated supplement for the prevention of 23:759-777. 49:227-267. larkspur poisoning in cattle. M.S. thesis. Pfister, J.A., M.H. Ralphs, G.D. Manners, AOAC. 1990. Official Methods of Analysis Utah State Univ., Logan, Ut. D.R. Gardner, K.W. Price, and L.F. (15th Ed.). Assoc. of Official Analytical Manners, G.D., K.E. Panter, and S.W. James. 1997b. Early season grazing of tall Chemists, Washington, D.C. Pelletier. 1995. Structure-activity relation- larkspur- (Delphinium spp.) infested range- Beath, O.A. 1919. The chemical examination ships of norditerpenoid alkaloids occurring in land. J. Range Manage. 50:391-398. of three species of larkspurs. Univ. Wyoming toxic larkspur (Delphinium) species. J. Nat. Pfister, J.A., D.R. Gardner, K.E. Panter, Agr. Exp. Sta. Bull. 120. Prod. 58:863-869. G.D. Manners, M.H. Ralphs, B.L. Beath, O.A. 1925. Chemical examination of NRC. 1996. National Research Council. Stegelmeier, and T.K. Schoch. 1999. three Delphiniums. Univ. Wyoming Agr. Nutrient requirements of beef cattle. 7th Larkspur (Delphinium spp.) poisoning in Exp. Sta. Bull. 143. revised ed. National Academy Press, livestock. J. Nat. Toxins 8:81-94 Chesnut, V.K. 1898. Thirty poisonous plants Washington, D.C. Ralphs, M.H., D.T. Jensen, J.A. Pfister, D.B. of the United States. USDA Farmers Bull. Perez Corona, M.E., B.R.V. de Aldana, B.G. Nielsen, and L.F. James. 1994. Storms 86. Criado, and A.G. Ciudad.1998. Variations influence cattle to graze larkspur: an observa- Cook, C.W. and L.E. Harris. 1968. Nutritive in nutritional quality and biomass production tion. J. Range Manage. 47:275-278. value of seasonal ranges. Utah Agr. Exp. Sta. of semiarid grasslands. J. Range Manage. Rauzi, F. 1964. Late-spring herbage produc- Bull. 472. 51:570-576. tion on shortgrass rangeland. J. Range Distel, R.A. and F.D. Provenza. 1991. Pfister, J.A. and G.D. Manners. 1991. Manage. 17:210-212. Experience early in life affects voluntary Mineral supplementation of cattle grazing SAS. 1998. SAS/STAT guide for personal intake of blackbrush by goats. J. Chem. Ecol. larkspur-infested rangeland during drought. computers, Ver. 7. SAS Institute, Cary, N.C. 17:431-449. J. Range Manage. 44:105-111. Smith, R. 1994. Handbook of Environmental Distel, R.A., J.J. Villalba, H.E Laborde, and Pfister, J.A. and G.D. Manners. 1995. Effects Analysis, 2nd Ed. Genium Publishing Co., M.A. Burgos. 1996. Persistence of the of carbachol administration in cattle grazing Schenectady, N.Y. effects of early experience on consumption tall larkspur-infested range. J. Range Van Soest, P.J., J.B. Robertson, and B.A. of low-quality roughage by sheep. J. Anim. Manage. 48:343-349. Lewis. 1991. Methods for dietary fiber, neu- Sci. 74:965-968. Pfister, J.A., K.E. Panter, and G.D. tral detergent fiber, and non-starch polysac- Gardner, D.R., K.E. Panter, J.A. Pfister, and Manners. 1994. Effective dose in cattle of charides in relation to animal nutrition. J. A.P. Knight. 1999. Analysis of toxic toxic alkaloids from tall larkspur Dairy Sci. 74: 3583-3597. norditerpenoid alkaloids by electrospray, (Delphinium barbeyi). Vet. Human Toxicol. Whitson, T.D., W.R. Tatman, and R.J. atmospheric pressure chemical ionization, 36:10-11. Swearingen.1992. Control of geyer larkspur and sequential tandem mass spectrometry. J. Pfister, J.A., G.D. Manners, D.R. Gardner, (Delphinium geyeri Greene) at two growth Agr. Food Chem. 47:5049-5058. K.W. Price, and M.H. Ralphs. 1996. stages with various herbicides. West. Sec. Hart, R.H. and M.J. Samuel. 1985. Influence of alkaloid concentration on Weed Sci. Soc. Res. Prog. Rep. I 64-65. Precipitation, soils and herbage production acceptability of tall larkspur (Delphinium on southeastern Wyoming range sites. J. spp.) to cattle and sheep. J. Chem. Ecol. Range Manage. 38:522-525. 22:1147-1168.

JOURNAL OF RANGE MANAGEMENT 55(4) July 359 J. Range Manage. 55: 360-366 July 2002 Elk and cattle forage use under a specialized grazing system

LACEY E. HALSTEAD1, LARRY D. HOWERY, GEORGE B. RUYLE, PAUL R. KRAUSMAN, AND ROBERT J. STEIDL

Authors are former Graduate Research Assistant, Associate Professor and Range Management Specialist, Professor and Range Management Specialist, Professor, and Assistant Professor, School of Renewable Natural Resources, The University of Arizona, Tucson, Ariz. 85721. 'Current address: The Nature Conservancy of Texas, P.O. Box 1440, San Antonio, Tex. 78295-1440.

Abstract Resumen

The Walker Basin Allotment grazing system in central Arizona El sistema de asignacion del apacentamiento de la cuenca is designed to allocate resource use under elk (Cervus elaphus L.) Walker, en la parte central de Arizona, esta disenado para asig- and cattle (Bos taurus L.) grazing. The grazing system was nar el use de los recursos bajo el apacentamiento de alces designed to promote biologically acceptable levels of forage use (Cervus elaphus L.) y ganado bovino (Bos taurus L.). El sistema on the half of the allotment scheduled for cattle grazing and to de apacentamiento se diseiio para promover niveles biologica- rest the other half by attracting elk to pastures recently grazed mente aceptables de utilizacion de forraje en la mitad del terreno by cattle. The objectives of our 2-year study were to determine programado para el ganado bovino y descansar la otra mitad whether the grazing system facilitated proper forage use as mediante la atraccion de los alces a potreros recenn apacentados defined by recent forage use and residual stubble height guide- por bovinos. Los objetivos de nuestro estudio, de 2 aiios de lines (i.e., 30 to 40% use and an 8- to 10-cm stubble height) and duracion, fueron determinar si el sistema de apacentamiento whether the system rested one half of the allotment from elk and facilito el use adecuado del forraje tal como to definen los lin- cattle grazing. Mean (± SEM) total elk and cattle forage use for eamientos recientes de use de forraje y altura del rastrojo (esto western wheatgrass (Pascopyrum smithii Rydb.), the key forage es, 30 a 40% de use y 8 a 10 cm de altura del restrojo) y si el sis- species, was 32 and 61 % ± 7 in 1997 and 1998, respectively; cor- tema descanso del alce y bovinos una de las mitades de la asi- responding mean (± SEM) stubble heights were 11 and 10 cm ± gnacion. La media (± EEM) de use total de forraje por alces y 0.6. Mean total cattle and elk forage use in 1998 (61%) exceeded bovinos para el "Western wheatgrass" (Pascopyrum smithii the 30 to 40% use guidelines. However, mean end-of-year stubble Rydb.), la especie cave, fue 32 y 61% ± 7% en 1997 y 1998 height was never below 10 cm. The grazing system did not pro- respectivamente, y la medias correspondientes (± EEM) para la vide half the allotment with complete rest; elk used all study pas atura del rastrojo fueron 11 y 10 cm ± 0.6. En 1998, la media tures. Elk use was higher in pastures with heavier tree cover and total de use de forraje por alces y bovinos (61%) excedio el 30 a steeper terrain in both years, regardless of where cattle grazing 40% de use establecido en los lineamientos. Sin embargo, la occurred. Elk grazing patterns were apparently more dependent media de altura del rastrojo al final del aiio nunca fue menor a on tree cover and topography than any changes in forage caused 10 cm. El sistema de apacentamiento no proveyo un descanso by the grazing system. completo para la mitad del terreno, el alce use todos los potreros bajo estudio. El use por el alce fue mayor en potreros con una densa cobertura de arboles de terreno con pendiente, sin Key Words: Arizona, deferred grazing, paired-plot, rest-rota- y importar donde ocurrio el apacentamiento de los tion, stubble height, western wheatgrass bovinos. Los patrones de apacentamiento del alce aparentemente fueron mas dependientes de la cobertura de arboles y la topografia del ter- Specialized grazing systems have been developed that use cat- reno que cualquier cambio en el forraje causado por el sistema tle grazing as a tool to alter forage characteristics and, thereby, de apacentamiento. modify elk (Cervus elaphus L.) distribution. Such grazing systems often include elements of rest-rotation, deferment, or both (Vavra and Sheehy 1996). Improvements in forage availability, (Anderson and Scherzinger 1975, Alt et al. 1992, Wisdom and palatability, production, and/or animal performance have been Thomas 1996). These systems may attract elk to areas recently suggested as advantages to using specialized grazing systems grazed by cattle (Gordon 1988, Alt et al. 1992, Frisina 1992), or they may produce little or no change in forage production, nutritional Research was funded by a grant from The University of Arizona Agriculture content, or herbivore distribution (Lacey and Van Poolen Experiment Station. We acknowledge the Coconino National Forest and the 1981, Yeo et al. 1993, Wambolt et al. 1997). Arizona Game and Fish Department for their support of this project. We would The Walker Basin Allotment grazing system in central Arizona like to thank Dr. Lamar Smith and 3 anonymous reviewers for reviewing earlier combines elements of rest-rotation and deferment, employing drafts of this paper. We thank Matt Barnes, Alex Connley, Tom DeLiberto, Vicki seasonal cattle grazing on half the allotment each year (United Gempko, Arlo Halstead, Elizabeth Howery, Dan Koepke, and Dave Womack for assistance in the field and laboratory. States Forest Service [USFS], unpublished Walker Basin Manuscript accepted 19 Sept. 01. Allotment Management Plan 1990, Camp Verde, Ariz.). The

360 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 objectives of our 2-year study were to Description of Grazing System and hemionus Rafinesque), and white-tailed determine whether the grazing system pro- Herbivores deer (Odocoileus virginianus Rafinesque) moted target levels of forage use and The Walker Basin Allotment grazing are thought to use insignificant amounts of residual stubble height (Holechek et al. system is designed to promote biologically western wheatgrass compared with elk and 1998), and whether the system rested half acceptable levels of forage use by cattle, cattle due to lower animal numbers or the allotment from large ungulate her- elk, and other wild herbivores and to rest dietary differences (Thomas Britt, bivory each year. half of the allotment by attracting elk and Regional Supervisor, Arizona Game and other wild ungulates to pastures recently Fish Department, personal communica- grazed by cattle (USFS, unpublished tion, Vavra et al. 1989). Thus, our research Materials And Methods Walker Basin Allotment Management focused on elk and cattle impacts on west- Plan 1990, Camp Verde, Ariz.). Hereford ern wheatgrass. Study Area and Hereford cross-bred (Saler, Braford, The Walker Basin Allotment (111 ° 40' Brahman, Angus, Gelbvieh) cow-calf General Sampling Procedures 40" W, 34° 38' 30" N) is comprised of pairs were used in this experiment and The study was conducted from March about 31,000 ha of USFS rangeland. The handled following the University of 1997 to October 1998. We estimated for- Walker Basin Allotment's 57 pastures are Arizona's Animal Care Protocol # 96-118. age use by elk and cattle and residual stub- divided into 2 main sections that run west During 1997, 413 cow-calf pairs were ble height during 2 growing seasons. The to east along an increasing elevational gra- used (mean body weight = 476 kg ± 62 southern pastures (3 and 4) were grazed by dient. The Walker Basin Allotment has 3 SD), while 450 cow-calf pairs were used cattle in 1997 and the northern pastures (1 seasonal ranges: winter (1,220 to 1,524 m in 1998 (mean body weight = 457 kg ± 87 and 2) were grazed by cattle in 1998 as elevation), transitional (1,525 to 1,982 m SD). Each year, cattle were moved up the required by the USFS grazing manage- elevation), and summer (1,983 to 2,287 m elevational gradient through pastures on ment plan (Fig. 1). Average stocking rates elevation). Our study area (about 6,750 one half of the allotment during the grow- for cattle, calculated using definition `a' of ha) consisted of 4 pastures (850 to 2,100 ing season (about 14 days/pasture) and the Glossary Revision Special Committee ha) on the transitional range (Fig. 1) moved down the elevational gradient on the report (1989), were 5.1 and 6.9 ha/AUM because local resource managers identified same half during the dormant season (about during 1997 and 1998, respectively (Fig. this range as having high potential for elk 5 days/grazed pasture). The other half of 1). All pastures were accessible to elk dur- and cattle competition (personal commu- the allotment was rested from cattle grazing ing the entire study. nication, USFS, Arizona Game and Fish the entire year (i.e., rested pastures). Paired-plot and stubble height tech- Department, and ranch managers). Seasonal cattle grazing is used to increase niques (Interagency Technical Reference Western wheatgrass (Pascopyrum palatable regrowth, which is hypothesized 1996) were used to evaluate grazing of smithii Rydb.) has been identified by the to attract elk to pastures recently grazed by western wheatgrass in 12 randomly select- USFS as the key herbaceous forage cattle, thereby resting the other half of the ed 3-ha sampling areas (3/pasture). Each species in the study pastures (USFS, allotment from cattle and elk herbivory. sampling area was located >_ 0.4 km from unpublished Walker Basin Allotment According to the Arizona Game and well-traveled roads, fences, and water and Management Plan 1990, Camp Verde, Fish Department, elk is the main wild her- > 0.3 km from adjacent sampling areas. Arizona). Other plant species included bivore species that significantly impacts Sampling areas contained 6 paired-plot blue grama (Bouteloua gracilis H.B.K.), western wheatgrass, the key forage species units. Each paired-plot unit consisted of 1 sideoats grama (Bouteloua curtipendula used in this study. Jack rabbits (Lepus cal- protected macroplot (1.7-m2 grazing Michx.), downy brome (Bromus tectorum ifornicus Gray), pronghorn (Antilocapra exclosure), and two, 1.7-m2 unprotected L.), buckwheat (Eriogonum spp.), snake- americana Ord), mule deer (Odocoileus macroplots, for a total of 72 protected and weed (Gutierrezia sarothrae Pursh.), squawbush (Rhus trilobata Nutt.), emory NT oak (Quercus emoryi Ton), pinyon pine (Pinus edulis Engelm.) and juniper (Juniperus osteosperma Torr.). Soils are predominantly classified as Vertic Pasture 1 Pasture 2 Haplustalfs (Terrestrial Ecosystems Grazed by cattle 1998 Grazed by cattle 1998 Survey of the Coconino National Forest (2100 ha) (1900 ha) 1992). The growing season (frost-free days) on the transitional range extends from March to October. Average yearly precipitation is 33 cm, occurring primarily from December to Pasture 3 Pasture 4 February and July to September (National Grazed by cattle 1997 Grazed by cattle 1997 Oceanic and Atmospheric Administration (850 ha) (1900 ha) 1997). Winter precipitation occurs primarily as snow. The northern pastures (1 and 2) were near an ephemeral drainage and contained more pinyon pine and juniper cover than the southern pastures (3 and 4). Fig. 1. Study pasture layout on the Walker Basin Allotment, central Arizona. Cattle grazed southern pastures (3 and 4) JunJJul. and Nov. of 1997, and northern pastures (1 and 2) Jun./Jul. and Nov. of 1998. Elk had access to study pastures year-round.

JOURNAL OF RANGE MANAGEMENT 55(4) July 361 144 unprotected plots in the study area Table 1. Mean relative and total western wheatgrass forage use (%) under 2 treatments (elk graz- (i.e., 72 paired-plot units). Two unprotect- ing and elk/cattle grazing) across 3 sampling periods on the Walker Basin Allotment, central (SEM 7). ed macroplots, rather than the traditional Arizona, 1997/1998 = 1, were matched with each protected Sampling period' macroplot to account for possible patch grazing by cattle and elk (Klingman et al. Before cattle After cattle End of Year/Grazing animal2 Pasture grazing season 1943, Grelen 1967). Unprotected and pro------(%)------tected macroplots were matched based on similar ocular estimates of western wheat- 1997 grass phytomass. Each sampling area also Elk 1 and 2 28 cattle 3 and 4 14 contained a 400-m stubble height transect Elk and 1998 protected and unprotected located between Elk and cattle 1 and 2 28 macroplots and >_ 10 m from protected Elk 3 and 4 13 macroplots. 'Before cattle grazing = relative forage use, early/mid-Jun.; After cattle grazing = relative forage use, mid-Jun./early To avoid attracting animals to protected Jul.; End of growing season, or 3 to 4 months after cattle exited grazed pastures = total forage use, mid-Oct. plots within sampling areas, each protect- Year and grazing animal main effects were significant and are reported in the Results section and Figs. 2a,2b. ed plot was >- 100 m from the others. Protected and unprotected plots were >- 50 did not measure dormant season grazing ate 6 sampling areas for each sampling m apart, and unprotected plots within a because of snow depth. period (Table 1). paired-plot unit were >- 10 m apart. To In 1998, we altered our paired-plot sam- minimize bias due to enhanced growth Paired-plot Sampling pling procedure slightly to address the within protected plots, new paired-plot For each sampling period, we clipped high range of variability of 1997 use esti- March each year within units were established in western wheatgrass from 2 randomly mates between paired-plot units (Owensby 1969). Before initially estab- selected paired-plot units in each sampling sampling areas (SE range = 0-35). For the pre- we lishing paired-plot units, we tested area within each pasture for a total of 24 each paired-plot unit in 1998, ocularly cision of ocular matching by clipping, dry- paired-plot samples. All plots were matched three, 0.25-m2 subplots within the ing, and weighing western wheatgrass clipped to ground level. During 1997, we protected macroplot to three, 0.25-m2 sub- 29 pairs 0.25-m2 circular plots that 2 from of used a 0.25-m2 circular frame to clip 4 plots within each of the unprotected were later used to clip subsamples within subplots within each 1.7-m2 macroplot. macroplots, creating 3 individually paired-plot units. A paired t-test revealed We averaged the 4 subplot dry weights to matched subsamples in each paired-plot no difference between paired test plots (P unit. Percentage use for a matched sub- obtain 1 mean protected weight for each > 0.2, a = 0.05). The same observer protected macroplot. For the 2 correspond- sample was the ratio of dry weights always established paired-plot units. ing unprotected macroplots, we averaged clipped from the 2 unprotected 0.25-m2 We estimated forage use and stubble the corresponding the 8 subplot dry weights (4 subplots x 2 subplots (averaged) and height in each pasture in 3 sampling peri- 0.25-m2 protected subplot. Percentage use macroplots) to obtain 1 mean unprotected 1) immediately before cattle entered ods: weight. Percentage use for a paired-plot for a paired-plot unit was the mean of the 2) grazed pastures, immediately after cat- unit was the ratio of the mean unprotected 3 matched subsample use estimates. and 3) at the tle exited grazed pastures, and protected weights. Negative utilization Means for sampling areas and pastures end of the growing season, about 3-4 values from paired-plot units were zeroed were calculated as described in 1997. We cattle exited grazed pastures months after (Werner and Urness 1998). Mean use for a calculated western wheatgrass standing 2 3; 1 and 2). 1) (Fig. and Fig. Tables sampling area was calculated from the 2 crop dry weight (kg ha from protected grazed Measurements made before cattle randomly selected paired-plot units. Mean macroplots each sampling period during relative elk use (early/mid-June) estimated use for each pasture treatment (grazed or both years of the study. in Measurements made after all pastures. rested) was calculated across the appropri- cattle grazed (mid-June/early July) estimated relative cattle and/or elk use, while measurements made at the end of the growing season (mid-October) estimated Table 2. Mean western wheatgrass residual stubble height (cm) under 2 treatments (elk grazing and elk/cattle grazing) across 3 sampling periods on the Walker Basin Allotment, central total forage use for the entire growing sea- Arizona, 1997/1998 (SEM = 0.6). son (cattle and/or elk). Relative use describes the amount of forage consumed Sampling period' or destroyed up to a certain time during Before cattle After cattle End of the growing season but prior to peak Year/Grazing animal2 Pasture grazing season standing crop (e.g., June or July; Frost et ------(cm)------al. 1994). Total forage use is "the proportion of current year's forage consumed or 1997 Elk 1 and 2 12 by measured destroyed grazing animals," Elk and cattle 3 and 4 13 at the end of the growing season (Glossary

Revision Special Committee 1989). Cattle 1 grazed study pastures about 5 days in mid- Elk and November (dormant season), and elk cattle grazing = relative forage use, early/mid-Jun.; After cattle grazing = relative forage use, mid-Jun./early migrating to winter ranges likely used Jul.; End of growing season, or 3 to 4 months after cattle exited grazed pastures = total forage use, mid-Oct. in the Results section and Figs. 3a,3b. study pastures at this time. However, we Year and grazing animal main effects were significant and are reported

362 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 60 year interaction (P > 0.9). On average, elk Cattle and Elk had used about 20% (relative use) of west- Elk ern wheatgrass in pastures 50 scheduled for grazing and rest before cattle were released into the study area (Fig. 2A). In rested pastures (elk use), forage use measured after cattle grazing, and at the end of the growing season, averaged 25% (relative use) and 24% (total use), respectively, while corresponding use in grazed pastures (cattle and elk use) averaged 48% (relative use) and 46% (total use). Total 10 forage use averaged 22 percentage points higher in grazed pastures (46%) than in 0 rested pastures (24%) at the end of the 60- growing season. Relative forage use attributable to elk averaged about 20% before cattle arrived 50 --1 01997 in the study area both years of the study T 11998 B (Fig. 2B). However, forage use measured after cattle grazing and at the end of the growing season averaged 10 and 16 percentage points higher in 1998 than in 1997, respectively.

Residual Stubble Height Residual stubble height also was influenced by type of grazing animal (Fig. 3A; 10 F2 19 = 3.6; P = 0.06) and year (Fig. 3B; F2 19 = 32.2; P < 0.0001), and there was 0 no type of animal x year interaction (P > Before cattle grazing After cattle grazing End of growing season 0.8). Stubble height averaged 14 cm in Sampling period* pastures scheduled for grazing and rest before cattle arrived at the study area (Fig. Fig. 2. Grazing animal (A) and year (B) effects on relative and total western wheatgrass for- 3A). In grazed pastures (cattle and elk age use (%) across 3 sampling periods on the Walker Basin Allotment, central Arizona use), mean stubble height decreased to 12 * (SEM = 5). Sampling periods were: Before cattle grazing = relative forage use, early/mid- cm after cattle grazing and to 10 cm at the Jun.; After cattle grazing = relative forage use, mid-Jun./early Jul.; End of growing season, end of the growing season, compared with or 3 to 4 months after cattle exited grazed pastures = total forage use, mid-Oct. 13 and 12 cm in rested pastures (elk use), respectively, during those same sampling Stubble Height Sampling Statistical Analyses periods. Stubble heights averaged 2 cm Stubble height sampling methodology We employed a completely randomized lower in grazed pastures (10 cm) than in was the same in both years. For each sam- design experiment with 2 x 2 factorial rested pastures (12 cm) by the end of the pling period, the average height of 60 indi- arrangement of treatments. Because we growing season. vidual western wheatgrass plants was measured forage use of western wheat- In 1997, stubble heights, measured measured (Interagency Technical Reference grass and residual stubble height repeated- before cattle grazing (12 cm) and after cat- 1996). Individual plants (grazed and ly in sampling areas, we used a multivari- tle grazing (10 cm), averaged 4 and 5 cm ungrazed) were measured as encountered at ate repeated measures analysis of variance lower than respective 1998 stubble heights approximate 3-m intervals along a transect. to test the effects of grazing animal (elk vs (i.e., 16 and 15 cm; Fig. 3B). However, An individual western wheatgrass plant elk and cattle), year (1997 vs 1998), and stubble height averaged about the same by was defined as a turf of vegetation occu- grazing animal x year interaction. We the end of the growing season for both pying a circle at least 5 cm in diameter. applied an arc-sine transformation to for- years of the study (i.e., 12 and 11 cm in Height of the extended green leaf area of age-use data prior to analysis (Steel and 1997 and 1998, respectively). each plant was measured to the nearest 0.5 Torrie 1980). cm after gently placing a ruler in the mid- Elk and Cattle Grazing Patterns dle of the turf circle. When a plant was Because it is important to discuss both grazed unevenly, we ocularly estimated Results relative and total forage use in relation to the average height of the 5-cm turf area. corresponding stubble heights across sam- Mean residual stubble height for each Forage Use pling periods, we tabulated results by year sampling area was from the 60 calculated Forage use was influenced by type of and grazing treatment (Tables I and 2). plants, and mean residual stubble height Higher use levels occurred after cattle had grazing animal (Fig. 2A; F2,19 = 6.1; P = for pastures was calculated as described occupied grazed pastures (cattle and elk 0.009) and year (Fig. 2B; F2 19 = 2.9; P = for paired-plot estimates. 0.08) and there was no type of animal x use), which typically yielded lower stub-

JOURNAL OF RANGE MANAGEMENT 55(4) July 363 16 Cattle and Elk Discussion Elk 14 i A Level of Forage Use I Whether the Walker Basin Allotment grazing system provided "proper" forage use depended on the metric used. For example, mean 1998 total use in grazed pastures (61%, Table 1) exceeded levels recommended for semi-arid ranges (e.g., 30-40%, Holechek et al. 1998); however, residual stubble height for western wheat- 4 grass (Table 2) never fell below the 8-10 cm (3-4 in) minimum recommended by 2 the same authors (Holechek et al. 1998). In rested pastures, mean total elk use (22% in 1997 and 26% in 1998) and end-of-grow-

18 ing-season stubble height (12 cm during both 1997 and 1998) were also within rec- 16 ommended limits (Holechek et al. 1998). B Adhering strictly to utilization or stub- 14 ble height guidelines while ignoring 12 effects of intensity, frequency, and season use yields an incomplete picture 10 of of grazing impacts (Zhang and Romo 1995, 8 Smith 1998). For example, cattle grazing in each pasture occurred only 19 days/year 6 (intensity, frequency) and after seed set or 4 during dormancy, when western wheatgrass is less vulnerable to herbivory (sea- 2 son of use) (Smith 1998). Pastures grazed

0 by cattle 1 year were rested from cattle the Before cattle grazing After cattle grazing End of growing season following year (frequency) and received Sampling period" relatively light elk use (intensity). Because total use in rested pastures was Fig. 3. Grazing animal (A) and year (B) effects on western wheatgrass stubble height (cm nearly the same in June as in October across 3 sampling periods on the Walker Basin Allotment, central Arizona (SEM = 0.4). (Table 1), we concluded that most elk use Sampling periods were: Before cattle grazing = relative forage use, early/mid-Jun.; After occurred early in the growing season when cattle grazing = relative forage use, mid-Jun./early Jul.; End of growing season, or 3 to 4 western wheatgrass was in the vegetative months after cattle exited grazed pastures total forage use, mid-Oct. = stage (season of use). Perennial grasses tend to be more tolerant of grazing during ble heights compared with rested pastures Table 2), whereas estimates of total forage the vegetative stage compared to the 1) much higher in grazed (elk use) during both years of the study use (Table were reproductive "boot" stage because meris- 1 As expected, mean for- 1998 (61% ± 7) than in (Tables and 2). pastures during tematic tissue is relatively closer to ground age use in was higher after 1997 (32% ± 7). Higher stubble grazed pastures overall (Frost et al. 1994, Smith 1998). with higher cattle grazing (37 and 58% relative forage height in 1998 was correlated The sampling technique used may also use during 1997 and 1998, respectively), mean western wheatgrass standing crop influence the results obtained. One disad- the (32 (i.e., 277 vs. 471 kg ha -1 in 1997 and 1998, and at the end of growing season vantage of the paired-plot technique is that SEM 32). Higher produc- and 61% total forage use during 1997 and respectively, = use estimates are imprecise unless sample cattle graz- tion levels in 1998 allowed 1998, respectively), than before apparently size is very large (Klingman et al. 1943). higher total forage use levels than in 1997, ing (14 and 28% relative forage use during An attempt in 1998 to decrease paired-plot 1). to similar end-of-growing- 1997 and 1998, respectively; Table and contributed unit variability within sampling areas by Corresponding mean stubble heights in season stubble heights both years. individually matching subsamples was grazed pastures were lower after cattle In rested pastures, mean relative and ineffective (Bork and Werner 1999). grazing (10 and 14 cm during 1997 and total elk use remained relatively light and Additionally, Bork and Werner (1999) 1998, respectively), and at the end of the similar across the 3 sampling periods both suggested that the standard practice of growing season (11 and 10 cm during years (i.e., range = 22-26% during 1997; zeroing negative utilization values when Table 1). 1997 and 1998, respectively), than before range = 13-26% during 1998; using the paired-plot technique could con- grazing and 16 cm during 1997 mean stubble heights in cattle (13 Corresponding tribute to overestimation of forage use on and 1998, respectively; Table 2). rested pastures were also fairly constant spatially heterogeneous ranges. Negative Interestingly, stubble height at the end of across sampling periods during 1997 (i.e., use values from paired-plot units were the growing season was about the same 12, 10, and 12 cm, respectively), but zeroed following Werner and Urness regardless of graz- showed a declining trend during 1998 (i.e., both years of the study (1998). Zeroed paired-plot data in our ing range 10-12 cm; 17, 16, and 12 cm, respectively; Table 2). treatment (i.e., = study were 9 percentage points higher than

364 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 non-zeroed paired-plot data (Halstead tures 1 and 2 compared to pastures 3 and 4 Elk and other wild ungulates have the 1998); thus, overall results and interpreta- both years (Table 2). Pastures 1 and 2 were freedom to choose foraging sites that best tion would not have changed appreciably near a drainage thought to be used by meet their physiological and behavioral had non-zeroed use data been used as sug- migrating elk and had more pinyon and needs. Managers designing dual or multi- gested by Bork and Werner (1999). juniper cover than pastures 3 and 4. Thus, species grazing systems should consider The paired-plot technique has also been elk use patterns were apparently more how intensity and season of livestock shown to overestimate forage production related to preferences for protective cover, grazing, in combination with local level and use due to enhanced moisture condi- topography, and traditional migration pat- habitat characteristics and animals' behav- tions within protected plots (i.e., the terns than the grazing system's hypothe- ioral habits, might influence spatial and "microclimate effect"; Grelen 1967, sized effect on forage regrowth and palata- temporal movements of wild ungulates Owensby 1969, Halstead 1998). The bility (Skovlin 1982, Wallace and (Cook et al. 1998, Unsworth et al. 1998). microclimate effect may be magnified Krausman 1987, Vavra 1992, Dyke et al. when birds and small mammals "fertilize" 1998). protected plots or when precipitation and Literature Cited the resulting production are above average (Owensby 1969). Stubble height and for- Conclusions and Management Alt, K. L., M. R. Frisina, and F. J. King. age production were higher in 1998; how- Implications 1992. Coordinated management of elk and ever, 1998 total use in grazed pastures cattle, a perspective-Wall Creek Wildlife exceeded 1997 use levels despite slightly Management Area. Rangelands Our study provided local knowledge of 14:12-15. lower stocking rates in 1998 (i.e., 5.1 vs Anderson, E. W. and R. J. Scherzinger. 6.9 ha/AUM in 1997 and 1998, respective- cattle-elk forage use and stubble height 1975. Improving quality of winter forage for Finally, rougher topography in the patterns in central Arizona, which is an elk by cattle grazing. J. Range Manage. ly). essential step for managers northern pastures (1 and 2) compared to first to evalu- 28:120-125. ate whether management practices are the southern pastures (3 and 4) may have Bork, E. W. and S. J. Werner. 1999. achieving resource objectives (McIntosh Viewpoint: Implications limited cattle distribution in 1998, con- of spatial variability and Krausman 1982, O'Neil 1985, Cook for estimating forage use. J. Range Manage. tributing to the higher use levels and the et al. 1998). Forage use data provides a 52:151-156. seasonal decline in stubble heights mea- general idea of the level and patterns of Cook, J. G., L. L. Irwin, L. D. Bryant, R. A. sured that year. large ungulate herbivory (Miller et al. Riggs, and J. W. Thomas. 1998. Relations 1994, Smith 1998). Stubble height data of forest cover and condition of elk: a test of Elk and Cattle Grazing Patterns complements forage use and other moni- the thermal cover hypothesis in summer and winter. Wildl. Monogr. 62:1-61. The grazing system did not provide half toring data because it is correlated with Dyke, F. G., W. C. Klein, and S. T. Stewart. of the study pastures with complete rest erosion protection, soil moisture retention, 1998. Long-term range fidelity in Rocky because elk used all pastures both years of forage regrowth potential, and small ani- Mountain elk. J. Wildl. Manage. 62:1020-1035. the study. In 1997, relative and total elk mal and insect habitat (Hall and Frisina, M. R. 1992. Elk habitat use within a use in rested pastures remained slightly Lindenmuth 1998). Although it appears rest-rotation grazing system. Rangelands below the recommended 30 to 40% range that the Walker Basin Allotment grazing 14:93-96. during all 3 sampling periods (Table 1). In system promotes proper levels of residual Frost, W. E., E. L. Smith, and P. R. Ogden. 1998, the same trend in rested pastures vegetation, a more comprehensive range- 1994. Utilization guidelines. Rangelands 16: held except there was relatively less elk land monitoring program is needed to con- 256-259. use before cattle grazing (13%). A decline firm or refute this assertion (Laycock Glossary Revision Special Committee. 1989. A glossary of terms used in range manage- in 1998 stubble height for both grazed (10 1998, Smith 1998). Rangeland trend data (e.g., plant species composition, herba- ment. 3rd ed. Soc. Range Manage., Denver, ± 0.6) and rested (12 ± 0.6) pastures at the Colo. ceous and shrub cover, vegetation struc- end of the growing season suggested that Gordon, I. J. 1988. Facilitation of red deer some elk use occurred in the study area ture) would help determine long-term sus- grazing by cattle and its impact on red deer after cattle grazing (Table 2). However, tainability of forage use or residual stubble performance. J. Appl. Ecol. 25:1-9. similar forage use estimates made after height levels. Grelen, H. E. 1967. Comparison of cage meth- cattle grazing, and at the end of the grow- Because elk foraging on the Walker ods for determining utilization on pine- ing season during both 1997 and 1998, Basin Allotment was evidently affected bluestem range. J. Range Manage. 20:94-96. suggested that elk did not return to the more by elk habitat preferences than the Hall, F. C. and R. Lindenmuth. 1998. Developing and management study area in high enough numbers to sig- grazing system, further field investigations achieving objectives on National Forest System lands, nificantly increase total forage use in are needed. For example, the relationship between elk and cattle forage use and tem- p. 47-49. In: Stubble height and utilization either grazed or rested pastures from June measurements: uses and misuses. Agr. Exp. to October (Table 1). We therefore con- poral and spatial distribution on the Walker Basin Allotment may differ at Sta. Oregon State Univ., Corvallis, Ore. clude that forage manipulation by cattle Halstead, L. E. 1998. Monitoring elk and cat- higher or lower elevations in did not attract elk to cattle-grazed pastures and result tle forage utilization under a specialized within the same year. We also conclude different elk use patterns. Manipulative grazing system in Arizona. M.S. Thesis, The that elk use was not influenced by previ- experiments at different elevations are Univ. of Arizona, Tucson, Ariz. to ous year's cattle grazing, because elk use needed determine whether habitat Holechek, J., R. D. Pieper, and C. H. Herbel. changes induced by the grazing system prior to the arrival of cattle was about twice 1998. Range management, principles and (i.e., forage quality and quantity) exert practices. 3rd ed. Prentice Hall, Upper as high in pastures 1 and 2 as in pastures 3 more influence on elk at these elevations Saddle River, N.J. and 4 both years, regardless of grazing than other habitat elements like cover and treatment (Table 1). Corresponding residual topography (Vavra and Sheehy 1996, stubble heights were slightly lower in pas- Cook et a1,1998).

JOURNAL OF RANGE MANAGEMENT 55(4) July 365 Interagency Technical Reference. 1996. O'Neil, J. 1985. Management strategies and Vavra, M., M. McInnis, and D. Sheehy. Utilization studies and residual measure- future research on elk in Arizona, p. 35-38. 1989. Implications of dietary overlap to man- ments. BLM National Appl. Resource Sci. In: G. W. Workman (ed.), Proc. of Symp. on agement of free-ranging large herbivores. Center. Cooperative Exten. Serv., USFS, Western Elk Management. Utah State Univ. Proc. Western Sect.Amer. Soc. of Anim. Sci. NRCS, BLM. Agr. Exp. Sta., Logan, Ut. 40:489-495. Klingman, D. L., S. R. Miles, and G. 0. Owensby, C. E. 1969. Effect of cages on Wallace, M. C. and P. R. Krausman. 1987. Mott. 1943. The cage method for determin- herbage yield in true prairie vegetation. J. Elk, mule deer, and cattle habitats in central ing consumption and yield of pasture Range Manage. 22:131-132. Arizona. J. Range Manage. 40:81-83. herbage. J. Amer. Soc. Agron. 35:739-746. Skovlin, J. M.1982. Habitat requirements and Wambolt, C. L., M. R. Frisina, K. S. Lacey, J. R. and H. W. Van Poolen. 1981. evaluations, p. 369-413. In: Thomas, J. W. Douglass, and H. W. Sherwood. 1997. Comparison of herbage production on mod- and D. E. Toweill (eds.), Elk of North Grazing effects on nutritional quality of blue- erately grazed and ungrazed western ranges. America, ecology and management. bunch wheatgrass for elk. J. Range Manage. J. Range Manage. 34: 210-212. Stackpole Books, Harrisburg, Penn. 50:503-506. Laycock, W. A. 1998. Variation in utilization Smith, E. L. 1998. Seasonal effects on the Werner, S. J. and P. J. Urness.1998. Elk for- estimates caused by differences among meth- measurement and interpretation of utiliza- age utilization within rested units of rest- ods, years, and observers, p. 17-24. In: tion, p. 9-16. In: Stubble height and utiliza- rotation grazing systems. J. Range Manage. Stubble height and utilization measurements: tion measurements: uses and misuses. Agr. 51:14-18. uses and misuses. Agr. Exp. Sta. Oregon Exp. Sta. Oregon State Univ., Corvallis, Ore. Wisdom, M. J. and J. W. Thomas. 1996. Elk, State Univ., Corvallis, Ore. Steel, R. G. D. and J. H. Torrie. 1980. p. 157-182. In: P. R. Krausman (ed.), McIntosh, B. J. and P. R. Krausman. 1982. Principles and procedures of statistics, a bio- Rangeland wildlife. Soc. Range Manage., Elk and mule deer distributions after a cattle metrical approach. 2nd ed. McGraw Hill Denver, Colo. introduction in northern Arizona, p. 545-552. Book Co., New York, N.Y. Yeo, J. M. Peek, W. T. Wittinger, and In: J. M. Peek and P. D. Dalke (eds.), Proc. J. J., C. T. Kvale.1993. Influence of rest-rotation 10 Wildlife-Livestock Relationships Terrestrial Ecosystems Survey of the elk habitat Symposium Univ. of Idaho Forestry, Wild!. Coconino National Forest. 1992. cattle grazing on mule deer and Idaho. J. Range Manage. and Range Exp. Sta., Moscow, Ida. USDA/USFS Southwestern Region. use in east-central Miller, W. H., J. G. Brock, and J. Horsley. Unsworth, J. W., L. Kuck, E. 0. Garton, and 46:245-250. 1994. Elk-cattle interaction in central B. R. Butterfield.1998. Elk habitat selection Zhang, J. and J. T. Romo. 1995. Impacts of Arizona. Arizona Game and Fish Dep. Tech. on the Clearwater National Forest, Idaho. J. defoliation on tiller production and survival Rep. Phoenix, Ariz. Wildl. Manage. 62:1255-1263. in northern wheatgrass. J. Range Manage. National Oceanic and Atmospheric Vavra, M. 1992. Livestock and big game rela- 48:115-120. Administration. 1997. Climatological data tionships. Rangelands 14:57-59. annual summary Arizona. Vol. 101(13), Vavra, M. and D. Sheehy. 1996. Improving ISSN 0145-0387. Natl. Environmental elk habitat characteristics with livestock Satellite, Data and Info. Service, Natl. grazing. Rangelands 18:182-185. Climatic Data Ctr., Asheville, N.C.

366 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 J. Range Manage. 55: 367-373 July 2002 Sediment movement and filtration in a riparian meadow following cattle use

R.R. McELDOWNEY, M. FLENNIKEN, G.W. FRASIER, M.J. TRLICA, AND W.C. LEININGER

Authors are Riparian/Wetland Ecologist, SAIL, 8100 Shaffer Parkway, Suite 100, Littleton, Colo. 80127, Natural Resource Specialist, Larimer County Open Lands, Loveland, Colo. 80537, Research Hydraulic Engineer (Retired), USDA, Agriculture Research Service, Crops Research Lab, Fort Collins, Colo. 80523, and Professors, Rangeland Ecosystem Science Department, Colorado State University, Fort Collins, Colo. 80523.

Abstract Resumen

Improper livestock grazing practices in western U.S. riparian Las practicas inadecuadas de apacentamiento de ganado en areas may reduce the nutrient and pollutant removal function of areas riberenas del oeste de E.U.A pueden reducir la funcion riparian communities, resulting in degradation of surface water removedora de nutrientes y contaminantes de las comunidades quality. Short duration-high intensity cattle use in 3 x 10 m plots riberenas, resultando en la degradation de la calidad del agua was evaluated in a montane riparian meadow in northern superficial. En parcelas de 3 x 10 m en una pradera riberena Colorado to quantify livestock effects on sediment movement and montana del norte de Colorado se evaluo el sistema de apacen- filtration under simulated rainfall (400 mm hour 1) plus overland tamiento de corta duration - alta intensidad para cuantificar los flow (25 mm hour') conditions. Four treatments: 1) control, 2) efectos del ganado en el movimiento de sedimento y filtration mowed to 10 cm stubble height, 3) trampled by cattle, and 4) cattle bajo lluvia simulada (100 mm hora1) mas condiciones de flujo grazed plus trampled (grazed) were evaluated. Sixty kg of sedi- superficial (s25 mm horal). Se evaluaron cuatro tratamientos: l) ment was introduced to overland flow in each plot. Sediment control, 2) segado a 10 cm de altura del rastrojo, 3) pisoteado por movement was evaluated using sediment traps positioned in ganado y 4) apacentado por ganado mas pisoteado (apacentado). microchannels and on vegetation islands at 5 distances downslope En el flujo superficial de cada parcela se introdujeron 60 kg de from the upper end of the plots and by sediment front advance- sedimento. El movimiento de sedimento se evaluo usando tram- ment. Most sediment deposition occurred within the first meter pas de sedimento posicionadas en microcanales a islas de veg- downslope from application. About 90% of the applied sediment etacion en 5 distancias pendiente abajo a partir de la parte supe- was filtered from runoff within 10 m in the control and mowed rior de la parcela y por el avance del frente de sedimento. La treatments, while approximately 84 and 77% of the applied sedi- mayoria de la deposition de sedimento ocurrio dentro del primer ment was trapped in the trampled and grazed treatment plots, metro pendiente abajo del punto de aplicacion. En los tratamien- respectively. The primary variables that influenced sediment fil- tos control y segado aproximadamente 90% del sedimento apli- tration were stem density and surface random roughness. Stem cado se filtro del escurrimiento dentro de 10 m, mientras aproxi- density was the most influential variable that affected sediment fil- madamente de 84 a 77% del sedimento aplicado fue atrapado en tration. Cattle grazing reduced the stem density by 40%. las parcelas de los tratamientos pisoteado y apacentado respecti- Monitoring of stem density should aid land managers in regulating vamente. Las principales variables que influenciaron la fil- cattle use of riparian communities and facilitate the protection of tracion de sedimento fueron la densidad de tallos y la rugosidad surface water quality from sediment in overland flow. aleatoria de la superfcie. La densidad de tabs fue la variable de mayor influencia en la filtration de sedimento. El apacentamien- to de ganado redujo la densidad de tallos en 40%. El monitoreo Key Words: Vegetation filter strip, stem density, rainfall simula- de la densidad de tallos debe ayudar a los manejadores de tierras tion, NPS pollution, grazing, trampling en regular el use de las comunidades riberenas por el ganado y facilitar la protection de la calidad de agua superficial del sedi- Sediment movement from upland areas is typically associated mento en el flujo superficial. with intense precipitation events (Edwards and Owens 1991, Larson et al. 1997). Excessive sediment loads in runoff water can cause radical changes in streambed morphology, loss of aquatic Practice to remove sediment and other types of non-point source habitat, reduction in storage capacity of reservoirs, and loss of (NPS) pollution from overland runoff (Dillaha 1989, Magette et aesthetic value (Novotny and Olem 1994). al. 1989). The efficiency of VFS in trapping sediment depends on Vegetated filter strips (VFS) are used as a Best Management the microtopography, vegetation cover and density, slope, length of the buffer strip, and the type and degree of surface disturbance (NRC 1993, Landry and Thurow 1997). Because of their location Research was funded by the USDA-CSREES Rangeland Research Grants adjacent to surface waters, riparian areas often function as VFS to Program, USDA-Agricultural Research Service, and the Colorado State University trap sediment derived from upland sources before it reaches the ltu Authors wish to thank Dennis Mueller Maxine Agricu ral Experim?nt Station: stream (Cooper et al. 1987, Osborne and Kovacic 1993, Hairsine Cottrell, and Elizabeth Nibarger for their assistance with the data collection. Manuscript accepted 20 Oct. 01. 1996, Pearce et al. 1998b). Though livestock use has been one of

JOURNAL OF RANGE MANAGEMENT 55(4) July 367 the most important modifiers (i.e. soil Control: Undisturbed plots with natur- Vegetation compaction, defoliation, and physical al vegetation approximately Vegetation density (stems m 2) was damage to vegetation) of western United 20 cm tall. determined by class (i.e., grasses, forbs, States riparian areas (Elmore and Beschta Mowed: Plots were mowed to a uni- sedges, and Baltic rush) at ground level 1987), the effects of cattle use on sediment form 10 cm stubble height, using a 100 cm2 quadrat. Stems within trapping ability of riparian filter strips with the clippings removed. each quadrat were counted at 10 random have not been examined. The objectives of Trampled: Three muzzled 320 kg locations within the lower 213 of each plot this study were to determine the effects of heifers (stocking rate 200 before all treatments and following the cattle trampling and grazing on vegetation Animal Unit Days (AUD)/ha) cattle grazing and trampling treatments. and soil surface variables, and to evaluate were placed on these plots for Stem density was assumed to remain con- their effects on sediment movement and 8 hours (1630 to 2030 hours stant on the control and mowed plots; filtration in a montane riparian meadow and 0430 to 0830 hours). therefore, stem density was only sampled under simulated rainfall and overland flow Grazed plus trampled: Three 320 kg prior to treatments on these plots. conditions. heifers (stocking rate - 200 Vegetation was clipped to ground level AUD/ha) were taken off feed in 1/8 m2 circular quadrats and bagged to for 12 hours and then placed estimate aboveground biomass. Five bio- Methods and Materials on these plots for a total of 8 mass samples were clipped at random hours, which coincided with locations in each plot before and after their normal feeding times of grazing, after mowing, and in the control Study Site 1630 to 2030 hours and 0430 plot. The trampled treatment was only Research was conducted in the summer to 0830 hours. sampled prior to treatment. Samples were (June-August) of 1997, in the Roosevelt The mowed treatment represented dried in a forced air oven at 50° C for 72 National Forest approximately 80 km canopy reduction without hoof action. The hours prior to weighing. northwest of Fort Collins, Colo. at an ele- trampled treatment was used to examine vation of 2,500 m, in a riparian meadow the hoof action of cattle without grazing Rainfall Simulations adjacent to Sheep Creek (40° 56' 58" N of vegetation, while the grazed removal Two simulation runs, approximately 24 Lat., 105° 40' 51" W Long.). Dominant the combined treatment represented hours apart, were made on each plot pair; vegetation in the meadow included shrub- effects of canopy removal and hoof action. by cinquefoil (Potentilla fruticosa L.), one before (pre-conditioning) and one Kentucky bluegrass (Poa pratensis L.), after the treatments were applied. Rainfall tufted hairgrass (Deschampsia caespitosa Plot Characterization was simulated at an approximate rate of (L.) Beauv.), small-winged sedge (Carex Cover, soil microtopography (i.e., ran- 100 mm hour"'. This amount of simulation microptera Mack.), water sedge (C. dom roughness), vegetation stem density, water applied was almost double that used aquatilus Wahl), Baltic rush (Juncus balti- and aboveground biomass were assessed in previous studies at Sheep Creek by cus Willd.), and various forb species such to determine their effects on sediment Fernald (1997) and Pearce et al. (1998a), as marsh marigold (Caltha leptosepala movement and deposition. Whenever pos- and was chosen, not to duplicate natural D.C.) and yarrow (Achillea lanulosa L.). sible, a catwalk was used above the plots storm events, but to provide adequate flow Soils within the Sheep Creek riparian to minimize surface disturbance. to move sediment. Actual rainfall applica- area are primarily Fluvaquents. These tion rate of the simulator was measured sandy to clay loam soils have a water table Cover with a 20 cm diameter rain gage equipped that is commonly at a depth of less than 30 Cover by class (i.e., litter, bare ground, with a bubble gage recorder positioned cm at some time during the spring and forbs, grasses, and shrubs) at the soil sur- between the paired plots. Distribution of measured with summer. A highly organic (> 7%) `A' hori- face was determined with a 10 point pin total simulated rainfall was zon, up to 20 cm thick, covered the study frame before treatment (adapted from 6 small plastic rain gages randomly locat- (Frasier et al. area (USDA 1980, Frasier et al. 1998). Hofmann et al. 1983, Devaurs and Gifford ed within each plot border 1998). Sixteen, 3 x 10 m plots, established dur- 1984). Ten random pin frames (i.e., 100 ing previous experiments at Sheep Creek points) within each plot were taken. The Overland flow during pre- and post- (Fernald 1997, Pearce et al. 1998a), were plots were sampled again after the post- treatment runs was simulated by spraying used in this study. All plots had been treatment rainfall simulation. water at a rate equivelent to 25 mm hour"' mowed to a 10 cm stubble height the pre- over the entire plot onto a 3.0 x 0.6 m tilt- ed (6% slope) tray at the upper edge of vious 2 summers. Paired plots (3 m apart) Random Roughness were delineated with 15 cm high steel edg- each plot. Actual application rate was Surface microtopography was measured ing placed 6 cm into the soil and oriented determined for each simulation run by vol- with a 100 point (10 x 9.6 cm grid) eleva- perpendicularly to Sheep Creek. A umetric sampling of the spray bar nozzles. tion pin table (0.6 x 2 m), positioned `Swanson type' large boom rain- Runoff was measured at the bottom of rotating lengthwise across the center of each plot fall simulator was used to simulate an each plot with a calibrated flume attached (Linse et al. 2001). Pin height (mm) above intense rainstorm over the plots (Swanson to a pressure transducer bubble gage the table was measured with a digital 1965, Laflen et al. 1991, Simanton et al. recorder. Discharge was recorded at 1 min caliper. The standard deviation of all 100 1991). intervals. pin elevations was used as an index of surface random roughness (Hairsine et al. Treatments 1992, Frasier et al. 1998). Pre-treatment Rainfall Simulation One of the following treatments was Plots were preconditioned for 1 hour applied to each plot: with simulated rainfall and overland flow, 24 hours prior to treatment. Once steady-

368 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 state (equilibrium) runoff was achieved, a Table 1. Responses of selected variables for 4 treatments applied to a montane riparian meadow. dye tracer was applied as a line source Similar letters following means in a row indicate non-significant differences (P > 0.10). across the plots at 2 locations (3 and 6 m downslope from the top of the plots) to Treatment delineate microchannel flow paths and rate Variable Control Mowed of water front advancement down the Plot slope (o) 3.8a 3.6a plots. Flow channels were delineated with Litter cover O 63a 65a spray paint for placement of sediment Stem density (# stems m 2) 5,300ab 5,500a traps before the post-treatment simulation Random roughness (mm) 13.Oa 11.9a event. Aboveground biomass (kg ha') 2,300a 1,700a - Coarse particle movement (mm) 726b 733b Sediment filtration* ha 1) 19,600a 19,800a Post-treatment Rainfall Simulation (kg 20,000 kg ha sediment applied to all plots. Following treatment, rainfall plus over- of land flow was again simulated over paired plots. When equilibrium runoff was Sediment Movement Results and Discussion achieved, sediment of known particle size Sediment transport and deposition was was introduced to overland flow on each determined using 5 rows of 4 sediment Cover traps at 1, 1.5, 2.0, 2.5, and 3.0 m downs- plot at a rate of 3 kg every 2 min. A total Cover (live + litter) was greater than lope from the top of each plot. Two traps of 60 kg (20,000 kg ha 1) of sediment all plots. Litter was the dominant in each row were placed within 96% on was applied. This rate was selected based ground cover (61 to 65%) for all treat- on previous overland flow studies at Sheep microchannels and 2 in vegetation `islands' between microchannels. ments, with no differences among treat- Creek where 30 kg (10,000 kg ha"') of ments (Table 1). Ground cover classes Sediment traps were plastic cups, 6.2 cm sediment was applied (Pearce et al. 1998a, such as sedges, moss, and forbs also were diameter at the top and 7.5 cm tall. The 1998b). Reported levels of sediment that not affected by treatments. The large traps were placed with their tops flush might reach a riparian area in overland amount of litter likely reduced soil com- with the soil surface. Sediment caught in flow are usually considerably less (5,000 paction and alterations to the soil surface the 2 channel traps or the 2 island traps kg ha') than the amount used in this study that may have been caused by hoof action within a row was composited by location (Buckhouse and Mattison 1980, Buck- of the heifers. house and Gaither 1982). The introduced and was later vacuum filtered, dried and sediment was white and mixed by weight weighed. The maximum distance that from 2 commercial ground silica products, coarse particles traveled was visually esti- Random Roughness 60% SIL-CO-SIL® 250 and 40% MIN-U- mated and measured from the top of each Random roughness of the soil surface 1). SIL®5, with a particle size distribution of plot after the rainfall simulation. was similar among treatments (Table 31.6% very fine sand (> 50 um), 41.8% In North Dakota, Hofmann and Ries differ- silt (2-50 um) and 26.6% clay (< 2 um). Experimental Design (1991) determined there were no ences in surface roughness among their This sediment had a higher portion of The 4 treatments were organized in a grazed and ungrazed treatments. Their fines (< 50 um ) than the soils on the randomized complete block design with 4 index of surface roughness was the vari- uplands in the Sheep Creek area and thus, blocks that were on an aridity gradient ance from 10 pin elevations spaced 50 mm would be more easily moved by the water perpendicular to the stream. A repeated apart. Frasier et al. (1998) found no differ- (Pearce et al. 1998a). measures analysis of variance was used to ences in surface random roughness among A total of 15 runoff samples were col- determine differences among treatments stubble height treatments for riparian com- lected at the base of each plot for sediment and between sediment trap position (in or munities, and concluded that surface ran- yield determination. Beginning with the out of microchannels) and distance downs- dom roughness was very difficult to ade- onset of equilibrium runoff, eleven, 950 lope. Sediment trap data were transformed quately characterize. As several authors ml grab samples of runoff were taken at 4 to log scale before analysis, but results 10 have suggested (Linse 1992, Femald 1997, min intervals. When the simulator was are reported as actual values. Analysis of Frasier et al. 1998, Pearce et al. 1998b), shut off, discharge decreased, and the covariance was used to determine relation- the connectivity of microtopographical remaining 4 runoff samples were 500 ml ships among treatments, vegetation char- features may be more important for water each. Each runoff sample was filtered acteristics, soil microtopography, and flow routing than is random roughness itself. through pre-weighed high wet strength 15 parameters with coarse particle sediment cm diameter filter paper with pore diame- movement. Sediment yield was analyzed ters of 1 µm. The samples were dried in a to determine treatment effects (P <_ 0.10). Vegetation forced draft oven at 50° C for 4.5 hours Multiple regression models were con- Cattle grazing decreased stem density and re-weighed. Sediment concentration structed for sediment trap data and sedi- by 40% compared to the control (Table 1). was determined for each sample and these ment yield by treatment using a forward Specifically, forb stem densities were data were correlated with discharge mea- selection process, and for coarse particle decreased by 50% on both trampled and surements to arrive at an estimated total movement using a backward selection grazed plus trampled treatments, as com- sediment yield for each plot within each process. Significance was accepted at P < pared to the control. Grass stem densities treatment. Only negligible amounts of 0.10. Data are presented as main effects were reduced approximately 40% on the organic residues were collected in runoff unless there was a significant interaction. grazed plots compared to the control. The samples (McEldowney 1999). Data analyses were done using SAS® for grazed (plus trampled) treatment had the Windows® (SAS®1996). lowest average stem density, because cattle removed stems by both pulling up

JOURNAL OF RANGE MANAGEMENT 55(4) July 369 Grazed Trampled Mowed Treatment Control 1.5 2 2.5 3 Distance Downs lope (m)

Fig. 1. Sediment deposition in traps positioned within microchannels. Comparison of distances done separately within each treatment. Values within the same position with the same letter are not significantly different (P > 0.10). stems with grazing and through hoof shear vegetation and flow characteristics. Cattle overland flow transport capacity that (Kauffman et al. 1983, Abdel-Magid et al. grazing reduced vegetation barriers to sed- resulted from increased flow velocity and 1987). These differences imply that, under iment movement that were present on the depth in microchannels than in interspace these very wet soil conditions, the relative trampled, mowed, and control treatments. areas (Rogers and Schumm 1991, impact of hoofs is less on changing plant With less sediment trapped within the veg- Flenniken 1999). More sediment was stem density than the actual animal graz- etation, there was more sediment available caught in island traps at the 1 m distance ing. The reduction in stem density follow- to be transported and deposited in the than at 2.5 or 3 m distances (Fig. 2). The ing the cattle treatments subsequently microchannels. Additionally, with less presence of sediment in island traps may impacted microchannel, runoff, and sedi- vegetative cover, sediment was subjected have resulted from particle movement into ment movement characteristics (Flenniken to raindrop splash and transport phenome- traps from raindrop splash detachment, et al. 2001). na into the channel traps (Thurow et al. and from overland flow in areas where the Aboveground biomass was approxi- 1986). Microchannel sinuosity was also flow was deep enough to submerge traps mately 60% less on grazed plots than on decreased on the grazed plots (Flenniken on vegetation islands. Vegetated filter control plots. However, there were no sig- et al. 2001), which may have increased the strips have been shown to become less nificant differences between mowed and transport capacity of sediments in chan- effective once overland flow depth grazed plots for aboveground biomass nels to the traps within the first meter of exceeds vegetation height (Pearce et al. (Table 1). Forage utilization of grazed the grazed plots. 1997). If vegetation height is greatly plots ranged from 12 to greater than 75%, On the mowed and trampled plots most decreased by cattle grazing and trampling, with a mean utilization level of 54%. of the sediment in the microchannels was then overland flow depths may exceed the trapped in the top 1-2 m. On the trampled height of the vegetation. Sediment Traps plots more sediment was trapped within 1 1.5 m distances downslope Analysis of sediment trap data indicated the first m to Coarse Particle Movement than at 2.5 m and 3 m positions (Fig. 1). a significant 3-way interaction among Greater downslope coarse particle mowed plots there was more sediment treatments, position (in channels vs. on On (sand) movement occurred on the grazed islands), and distance downslope. On in the microchannel traps in the first meter treatment than on the trampled, mowed, or than at 2, 2.5, or 3 m downslope. In con- grazed plots, there was almost 4 times the control treatments (Table 1). However, amount of sediment in channel traps trast, there were no significant differences there were no significant differences in in the amounts of sediment caught in traps placed 1 m downslope than at the next dis- sand movement among these 3 other treat- positioned in microchannels at any distance (1.5 m) downslope (Fig. 1). The ments. Sand moved 78% farther downs- large difference between the amount of tance downslope for the control plots. lope from the upper edge of grazed plots sediment caught within this half meter dis- The general trend that more sediment as compared with control plots. Since in in tance may have resulted from insufficient was caught traps positioned coarse particle movement downslope on in energy in microchannel flow to transport microchannels than islands for each grazed plots was greater than on trampled treatment was expected since microchan- sediment beyond 1 m downslope. plots, this indicated that grazing removal car- The greater quantity of sediment trapped nels were topographically lower and of canopy cover and stems was more ried the majority of flow (Fig. 1 and 2). in channels in the first meter of the grazed important than the effect of trampling plots resulted from differences in both The greater mass of sediments that moved alone. Variables that most influenced the in microchannels was a result of increased

370 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Treatment

Distance Downslope (m)

Fig. 2. Sediment deposition in traps positioned on islands between microchannels. Comparison of distances done separately within each treatment. Values within the same position with the same letter are not significantly different (P > 0.10). downslope movement of coarse sediment among treatments was accounted for by stress was reduced to 11-38% of control were stem density and microchannel flow stem density alone. Surface random plot values. This allowed more sediment velocity (Table 2). roughness values were poor estimators of to be transported on clipped plots. Prosser sediment yield when used alone. However, et al. hypothesized that 2 flow regimes Sediment Filtration when used in conjunction with stem densi- existed once herbaceous vegetation was improved Sediment filtration in the grazed treat- ty, random roughness our ability overwhelmed by water; a slower flow by 14% ment was significantly lower than in the to predict sediment yield (Table through vegetation stems with a more trampled, mowed, or control treatments, 2). Both of these factors exert friction on rapid laminar flow above them. This shear which were not different from one another overland flow. stress partitioning suggested that on a The interaction between discharge and densely grassed surface, more than 90% (Table 1). Simulated trampling studies of have been conducted elsewhere to evalu- bed friction directly influences sediment the resistance to flow was caused by plant transport; as discharge increases, the ener- al. In ate the effects of trampling on sediment stems (Prosser et 1995). a small plot to yield. Packer (1953) found that increased gy available carry sediment also typi- study in Arizona, Abrahams et al. (1994) trampling generally caused increased soil cally increases. Several studies have determined that plant stems and litter the importance of friction created loss. However, simulated trampling had no shown cover were the primary variables that cre- by stem density as it affects overland flow. ated resistance effect on soil loss when initial ground to overland flow on grass- an 2 3 cover was over 90%. In another simulated Fernald (1997) found that effective land plots clipped to a height of to cm. trampling experiment, Dadkhah and friction subfactor, that was derived from Results from this present experiment Gifford (1980) showed that grass cover stem width to stem spacing ratio, better supported findings of Pearce et al. (1997, estimated discharge than did plot averaged 1998a, 1998b) was not was the most important variable affecting that stubble height sediment yield, regardless of trampling friction subfactors or microchannel effec- a good indicator for sediment filtration in level. They recommended the maintenance tive friction subfactors based on ground a montane riparian community. Pearce et and al. of 50% ground cover for watershed pro- cover class, soil texture, standing (1998a, 1998b) determined that sedi- plant biomass. A field flume study con- ment yield neither decreased nor increased tection for sites with loamy soil and 15% by et al. as vegetation slopes. ducted Prosser (1995) showed stubble height increased, as Stem density was the primary variable that critical shear stress was reduced by water flow depth was less than the height clipped a 10 cm that affected total sediment yield in the half when grass was to of the stubble. They concluded that accu- present study. Sixty four percent of the stubble height, but that rates of sediment rate prediction of sediment filtration from measured variability in sediment yield transport were still very low. With com- overland flow required the consideration plete removal of plant stems, critical shear of a combination of vegetation and soil surface variables such as stem density, Table 2. Predictive models for sediment movement and sediment yield in a montane riparian forb cover, amount of bare ground, silt meadow. content of soil, clay content of introduced sediment, amount of runoff, and slope. Dependent variable Independent variables and regression equation r Variables that were considered in this Coarse particle movement present study, but had little effect on sedi- downslope (mm) 0.015 (stems m 2) + 9175.86 (flow velocity m sec') ment yield or became insignificant when Sediment yield (kg ha') 9778.1- 0.998 (stems m 2) -171.14 surface roughness index (mm) 0.78 considered simultaneously with stem den-

JOURNAL OF RANGE MANAGEMENT 55(4) July 371 sity, were bare ground, litter cover, moss drop impact sufficiently to nullify the ments as compared with control and cover, microchannel sinuosity, drainage effect of rainfall intensity on sediment re- mowed treatments where sediment filtra- density, and rainfall intensity. Because entrainment (Hairsine et al. 1992). Moss tion was about 90%. These filtration val- both the amount and type of ground cover (1988) found that raindrop impact on tur- ues were slightly less than those found by among treatments were similar, and proba- bulent flow was minor and had little effect Pearce et al. (1998b) for the same mon- bly caused analogous turbulent flows at on sediment transport. He found that rain- tane community. In that study, both the the soil-water interface, the influence of drop impact was most important in re- rainfall intensity (85 mm hr') and the ground cover on sediment yield was simi- entrainment of bedload size (0.2 mm) par- amount of sediment that was added to the lar for all 4 treatments. There can be no ticles. Because of the relatively low ener- plots (30 kg) were both significantly lower doubt of the importance of ground cover gy required to transport fine particles such than in this present study. to sediment transport, but when ground as silt and clay (suspended load), raindrop The density of herbaceous plant stems cover is similar among treatments the impact would probably not affect their was the most important variable that influence of other variables, such as stem transport. affected sediment filtration in this present density, is dominant. study. As stem density increased, so did Increased microchannel sinuosity can sediment filtration. Cattle grazing not only also influence sediment yield (Thurow et Conclusions reduced the height of vegetation, but stem al. 1988, Spaeth et al. 1994, Prosser et al. density was decreased as well. This caused 1995) by reducing flow velocity, thus movement as stubble height alone to be a poor predictor reducing sediment transport capacity. In a Studies where sediment differences of sediment filtration in areas grazed by companion study to ours, short duration- affected by cattle activity and in sediment movement in channels versus cattle. Friction provided by more dense high intensity cattle trampling resulted in a plant stems on both the mowed and con- reduction of microchannel sinuosity islands could not be found. However, there are studies that have shown that veg- trol treatments attenuated overland flow (Flenniken et al. 2001). This reduction, velocities and increased sediment deposi- however, did not explain as much variabil- etation height was less influential than canopy cover in reducing downslope sedi- tion (Flenniken et al. 2001). In addition, ity in sediment yield as did stem density. use of the standard deviation for surface At the watershed scale, drainage density ment movement (Pearce et al. 1998a). Down slope coarse particle movement did random roughness quantification also and sediment yield are typically thought to improved the prediction model for sedi- be negatively correlated with vegetaion not differ between control and mowed treatments in this present study. These ment yield. cover and positively correlated with maxi- Under the extreme conditions of this mum precipitation events (Abrahams results agreed with Pearce et al. (1998a) who determined that there were no differ- study (i.e., intense rainfall, high sediment 1972, Dunne and Leopold 1978). At the loading, and high stocking rate of cattle), scale of this present study, cattle trampling ences in sand movement between 10 cm stubble height and natural vegetation stem densities of approximately 5,000 and grazing resulted in a decrease in 2 height (about 22 cm) treatments in a stems m were effective in trapping 90% drainage density (Flenniken 1999) and an of the sediment applied. In areas where increase in sediment yield (McEldowney grass/sedge community. Treatment, positioning of traps in chan- riparian filtration of NPS pollution is a 1999). A few larger microchannels with primary management goal, monitoring of greater flows should result in increased nels or islands, and distance downslope were all significant factors that affected stem density could be an important tool in sediment transport. A good correlation the decision making process. Results of between sediment yield and drainage den- sediment deposition in traps. Sediment traps were used in this study as indicators this research supported the findings of sity, however, was not found. The other studies (Lowrance et al. 1986, decrease in drainage density on the cattle of sediment transport, not sediment retention. Therefore, the greater mass of sedi- Cooper et al. 1987, Welsch 1991, Daniels disturbed plots may indicate a homoge- and Gilliam 1996, Hairsine 1996, Pearce nization effect by cattle on spatial variabil- ment caught in traps on grazed plots posi- et al. 1998b) that indicated that riparian ity of the vegetation. Additional research tioned in channels at 1 m downslope compared with the other 4 distances reflected filtration could be effective for reducing is needed to determine spatial variability NPS pollution to surface waters. In addi- of vegetation in riparian areas as it affects greater transport of sediment on these plots. Lower vegetation stem density and tion, our findings indicated that properly sediment filtration. managed cattle grazing along the outer Canopy removal by cattle or by mowing aboveground biomass on grazed plots reduced the ability of these plots to retain edges of montane riparian meadows might may have permitted more raindrops to be compatible with sediment filtration strike the sediment laden overland flow, sediment and allowed transport of sediment into channel traps. Microchannels in goals. and thereby facilitate the movement of The importance of stem density in this sediment downslope. Rainfall intensity riparian areas act as conduits to transport sediment downslope. The straightening montane riparian community for sediment over the very narrow range experienced in filtration indicates that stem density data this study had little effect on total sedi- and reduction in density of these microchannels that resulted from cattle might be useful in modeling sediment ment yield. However, splash resulting transport. Furthermore, this research has from raindrop impact could have trans- grazing caused deeper flow depths and increased flow velocities. These straighter demonstrated the necessity to consider ported sediment laden drops short dis- riparian areas separately from adjacent tances as and deeper microchannels then had greater (both up- or down-hill), such uplands. If this is done, improved esti- into the island sediment cups. Most of the sediment transport capacity and sediment movement downslope (Flenniken et al. mates of sediment delivery to a stream or plot area was covered with a layer of lake should result. water during equilibrium runoff that 2001). ranged in depth from 14 to 36 mm Sediment filtration was less in both the (Flenniken 1999), and likely reduced rain- grazed (77%) and trampling (84%) treat-

372 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Literature Cited Hairsine, P.B. 1996. Comparing grass filter Osborne, L.L. and D.A. Kovacic. 1993. strips and near-natural riparian zones for Riparian vegetated buffer strips in water trapping sediment and sorbed nutrients. p. quality restoration and stream management. Abdel-Magid, A.H., M.J. Trlica, and R.H. 109. In: Paper presented at ASSSI and Freshwater Bio. 29:243-250. Hart. 1987. Soil and vegetation responses to NZSSS National Soils Conference, July 1996 Packer, P.E.1953. Effects of trampling distur- simulated trampling. J. Range Manage. - Oral paper. bance on watershed condition, runoff, and 40:303-306. Hairsine, P.B., C.J. Moran, and C.W. Rose. erosion. J. Forestry 51:28-31. Abrahams, A.D.1972. Drainage densities and 1992. Recent developments regarding the Pearce, R.A., M.J. Trlica, W.C. Leininger, sediment yields in Eastern Australia. Aust. influence of soil surface characteristics on D.E. Mergen, and G. Frasier. 1998a. Geo. Studies 10:19-41. overland flow and erosion. Aust. J. Soil Res. Sediment movement through riparian vegeta- Abrahams, A.D., A.J. Parsons, and J. 30:249-264. tion under simulated rainfall and overland Wainwright. 1994. Resistance to overland Hofmann, L. and R.E. Ries. 1991. flow. J. Range Manage. 51:301-308. flow on semiarid grassland and shrubland Relationship of soil and plant characteristics Pearce, R.A., M.J. Trlica, W.C. Leininger, hillslopes, Walnut Gulch, southern Arizona. to erosion and runoff on pasture and range. J. J.L. Smith, and G.W. Frasier. 1997. J. Hydrol.156:431-446. Soil and Water Cons. 46:143-147. Efficiency of grass buffer strips and vegeta- Buckhouse, J.C. and R.E. Gaither. 1982. Hofmann, L., R.E. Ries, and J.E. Gilley. tion height on sediment filtration in laborato- Potential sediment production within vegeta- 1983. Relationship of runoff and soil loss to ry rainfall simulations. J. Environ. Qual. tive communities in Oregon's Blue ground cover of native and reclaimed grazing 26:139-144. Mountains. J. Soil and Water Cons. land. Agron. J. 75:599-602. Pearce, R.A., G.W. Frasier, M.J. Trlica, 37:120-122. Kauffman, J.B., W.C. Krueger, and M. W.C. Leininger, J.D. Stednick, and J.L. Buckhouse, J.C. and J.L. Mattison. 1980. Vavra. 1983. Effects of late season cattle Smith. 1998b. Sediment filtration in a mon- Potential soil erosion of selected habitat grazing on riparian plant communities. J. tane riparian zone under simulated rainfall types in the High Desert Region of central Range Manage. 36:685-690. and overland flow. J. Range Manage. Oregon. J. Range Manage. 33:282-285. Laflen, J.M., L.J. Elliot, J.R. Simanton, C.S. 51:309-314. Cooper, J.R., J.W. Gilliam, R.B. Daniels, Holzhey, and K.D. Kohl. 1991. WEPP Soil Prosser, I.P., W.E. Dietrich, and J. and W.P. Robarge. 1987. Riparian areas as erodibility experiments for rangeland and Stevenson. 1995. Flow resistance and sedi- filters for agricultural sediment. Soil Sci. cropland soils. J. Soil and Water Cons. ment transport by concentrated flow in a Soc. Amer. J. 51:416-420. 46:39-44. grassland valley. Geomorphology 13:71-86. Dadkhah, M. and G.F. Gifford. 1980. Landry, M.S. and T.L. Thurow. 1997. Rogers, R.D. and S.A. Schumm. 1991. The Influence of vegetation, rock cover, and Function and design of vegetation filter effect of sparse vegetation cover on erosion trampling on infiltration rates and sediment strips: An annotated bibliography. Texas and sediment yield. J. Hydrol. 123:19-24. production. Water Resour. Bull. 16:979-986. State Soil and Water Conservation Board SAS.1996. SAS System for Windows Version Daniels, R.B. and J.W. Gilliam. 1996. Bull. No. 97-1. Temple, Tex. 6.12. Cary, N.C. Sediment and chemical load reduction by Larson, W.E., M.J. Lindstrom, and T.E. Simanton, J.R., M.A. Weltz, and H.D. grass and riparian filters. Soil Sci. Soc. Schumacher. 1997. The role of severe Larsen. 1991. Rangeland experiments to Amer. J. 60:246-251. storms in soil erosion: a problem needing parameterize the water erosion prediction Devaurs, M. and G.F. Gifford. 1984. consideration. J. Soil and Water Cons. project model: vegetation canopy effects. J. Variability of infiltration within large runoff 52:90-95. Range Manage. 44:276-282. plots on rangelands. J. Range Manage. Linse, S.J. 1992. The influence of ground Spaeth, K.E., M.A. Weltz, H.D. Fox, and 37:523-528. cover on upland range erosion. M.S. Thesis. F.B. Pierson. 1994. Spatial pattern analysis Dillaha, T.A. 1989. Water quality impacts of Univ. Wyoming. Laramie, Wyo. 98 p. of sagebrush vegetation and potential influ- vegetative filter strips. Paper No. 89-2043. Linse, S.J., D.E. Mergen, J.L. Smith, and ences on hydrology and erosion. pp. 35-50. Am. Soc. Agr. Eng. St. Joseph, Mich. 9 p. M.J. Trlica. 2001. Upland erosion under a In: W.H. Blackburn, F.B. Pierson, G.E. Dunne, T. and L.B. Leopold. 1978. Water in simulated most damaging storm. J. Range Schumann, and R. Zartman (eds.) Environmental Planning. W.H. Freeman and Manage. 54:356-361. Variability in Rangeland Water Erosion Co. New York. 818 p. Lowrance, R., J.K. Sharpe, and J.M. Processes. SSSA Special Pub. No. 38. Edwards, W.M. and L.B. Owens.1991. Large Sheridan.1986. Long-term sediment deposi- Madison, Wisc. storm effects on total soil erosion. J. Soil and tion in the riparian zone of a coastal plain Swanson, N.P. 1965. Rotating-boom rainfall Water Cons. 46:75-78. watershed. J. Soil and Water Cons. simulator. Trans. Amer. Soc. Agr. Eng. Elmore, W. and R.L. Beschta.1987. Riparian 41:266-271. 8:71-72. areas: perceptions in management. Range- Magette, W.L., R.B. Brinsfield, R.E. Palmer, Thurow, T.L., W.H. Blackburn, and C.A. lands 9:260-266. and J.D. Wood. 1989. Nutrient and sedi- Taylor, Jr. 1986. Hydrologic characteristics Fernald, A.G. 1997. Microtopographic flow ment removal by vegetated filter strips. of vegetation types as affected by livestock paths and riparian surface hydrology. Ph.D. Trans. Am. Soc. Agr. Eng. 32:663-667. grazing systems, Edwards Plateau, Texas. J. Diss. Colorado State Univ. Fort Collins, McEldowney, R.R. 1999. Montane riparian sur- Range. Manage. 39:505-509. Cob. 136 p. face hydrology and sediment filtration as affect- Thurow, T.L., W.H. Blackburn, and C.A. Flenniken, M. 1999. Flow characteristics and ed by cattle disturbance. M.S. Thesis. Colorado Taylor, Jr. 1988. Infiltration and interrill sediment movement in a montane riparian State Univ., Fort Collins, Cob. 163 p. erosion responses to selected livestock graz- ecosystem. M.S. Thesis. Colorado State Moss, A.J. 1988. The effects of flow-velocity ing strategies, Edwards Plateau, Texas. J. Univ. Fort Collins, Colo. 99 p. variations on rain-driven transportation and Range Manage. 41:296-302. Flenniken, M., R. McEldowney, G.W. the role of rain impact in the movement of USDA, Soil Conservation Service and Forest Frasier, W.C. Leininger, and M.J. Trlica. solids. Aust. J. Soil Res. 26:443-450. Service. 1980. Soil survey report. Larimer 2001. Runoff response in a montane riparian National Research Council (NRC).1993. Soil County, CO. U.S. Govt. Print. Off. area following cattle use. J. Range Manage. and Water Quality: An Agenda for 239-812148. Washington, D.C. 54:567-574. Agriculture. National Academy Press. Welsch, D.J. 1991. Riparian Forest Buffers: Frasier, G.W., M.J. Trlica, W.C. Leininger, Washington, D.C. 516 p. Function and Design for Protection and R.A. Pearce, and A. Fernald.1998. Runoff Novotny, V. and H. Olem. 1994. Water Enhancement of Water Resources. USDA from simulated rainfall in 2 montane riparian Quality: Prevention, Identification, and Forest Serv. NA-PR-07-91. Radnor, Penn. communities. J. Range Manage. 51:315-322. Management of Diffuse Pollution. Van 24p. Nostrand Reinhold. New York. 1054 p.

JOURNAL OF RANGE MANAGEMENT 55(4) July 373 J. Range Manage. 55: 374-382 July 2002 Calibrating fecal NIBS equations for predicting botanical composition of diets

JOHN W. WALKER, SCOTT D. McCOY, KAREN L. LAUNCHBAUGH, MERRITA J. FRAKER, AND JEFF POWELL

Authors are Resident Director of research, Texas Agricultural Experiment Station, San Angelo, Tex. 76901; Research Technician, USDA U. S. Sheep Experiment Station, Dubois, Ida. 83423; Assistant Professor, University of Idaho, Moscow, Ida, 83844; Research Associate, Montana State University, Bozeman, Mont. 59717; and Professor, University of Wyoming, Laramie Wyo. 82071.

Abstract Resumen

E1 objetivo de este estudio fue investigar el use de la The objectives of this study were to investigate the use of near Espectroscopia de Reflectancia en el Infrarrojo Cercano (NIRS) spectroscopy (NIRS) of fecal samples for predicting the infrared en muestras de heces fecales para predecir el porcentaje de (Artemisia tridentata Nutt. percentage of mountain big sagebrush Artemisa Grande de la Montana (Artemisia tridentata ssp. ssp. vaseyana Beetle) in sheep diets and to quantify the (Rydb) vaseyana Beetle) en dietas de borregos para cuantificar limitations of using NIRS of fecal samples to predict diet compo- [Rydb.] y las limitaciones del use de Espectroscopia de Reflectancia en el sition. material from a sheep feeding trial with known lev- Fecal Infrarrojo Cercano en muestras fecales para predecir la com- els of sagebrush and several background forages was used to posicion de la dieta. Se usaron heces fecales de borregos bajo develop fecal NIRS calibration equations validated with fecal dietas con niveles conocidos de artemisa varios forrajes com- material from 2 other sheep feeding trials with known levels of y plementarios para calibrar las ecuaciones de Espectroscopia de sagebrush in the diets. The 1996 calibration trial varied the level Reflectancia en el Infrarrojo Cercano, las cuales se validaron con of sagebrush, alfalfa, and grass hay in the diets. The 1998 trial material fecal de otros dos ensayos nutricionales en borregos con compared frozen to air-dried sagebrush. The Wyoming trial was niveles conocidos de artemisa en las dietas. En el ensayo de cali- a metabolism study using frozen sagebrush. Trials used different bracion de 1996 los niveles de artemisa, alfalfa paja de pasto en levels of sagebrush varying from 0 to 30% of the diet in incre- y las dietas variaron. En el ensayo de 1998 se comparo artemisa ments of 4 to 10 percentage points. Internal validation of the congelada artemisa secada al aire. El ensayo de Wyoming fue 1996 trial with a subset of the samples not used for calibration y un estudio de metabolismo en el que se use artemisa congelada. showed that when predicted samples are from the same popula- En los ensayos se usaron niveles de10 al 30% de artemisa en la tion as the calibration samples, this procedure can accurately dieta con incrementos de 4 a 10 puntos porcentuales. La vali- predict percent sagebrush (R2 0.96, SEP =1.6). However, when = dacion interna del ensayo de 1996 con un subgrupo de muestras predicted samples were from a different population than calibra- no usadas para la calibracion mostro que cuando las muestras tion samples, accuracy was much less, but precision was not predecidas eran de la misma poblacion que las muestras de cali- affected greatly. Low accuracy was caused by a compression of bracion, este procedimiento puede predecir correctamente el the range of data in the predicted values compared to the refer- porcentaje de artemisa (R2 0.96, SEP =1.6). Sin embargo cuan- ence values, and the predicted sagebrush levels in the diet should = do las muestras predecidas provenian de una poblacion diferente be considered to represent an interval scale of measurement. a las muestras de calibracion, la exactitud fue mucho menor, Modified partial least squares regression resulted in better cali- pero la precision no se afecto grandemente. La baja exactitud fue bration than stepwise regression, and calibration data sets with causada por una compresion del rango de datos en los valores only high, low, and no sagebrush resulted in calibrations almost predecidos cuando se compararon con los valores de referencia, as good as data sets with several intermediate levels of sagebrush. los niveles predecidos de artemisa en la dieta deben ser consid- High values of the H statistic were related to low precision but y erados para representar un intervalo en la escala de medicion. did not affect the accuracy of predictions. We believe the interval La regresion linear o procedimiento PLS (partial least-squares scale of measurement will contain sufficient information for the regression) modificada resulto en una mejor calibracion que la purpose of addressing many questions on rangelands. regresion multiple (stepwise regression), y la calibracion de los grupos de datos con niveles altos, bajos y sin artemisa resultaron en calibraciones casi tan buenas como los grupos de datos con Key Words: Artemisia, diet selection, sagebrush, sheep varios niveles intermedios de artemisa. Niveles altos de estadisti- ca-H (H statistic) estuvieron relacionados a una baja precision, pero no afectaron la exactitud de las predicciones. Creemos que Since the initial report of using near infrared spectroscopy el intervalo en la escala de medicion que representa la (NIRS) of fecal material to predict nutritional parameters of graz- Espectroscopia de Reflectancia en el Infrarrojo Cercano en la ing animals (Coleman et al. 1989), the use of this procedure has prediccion de la composicion de la dieta contiene suficiente infor- expanded rapidly. The NIRS of feces can accurately predict macion para responder muchas preguntas en relacion a dietary crude protein, digestible organic matter (Lyons and Stuth praderas.

Manuscript accepted 16 Sept. 2001.

374 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 1992, Leite and Stuth 1995), and botanical calibrations based on botanically pure The objective of this study was to deter- composition (Walker et al. 1998) when the samples were as acceptable as calibrations mine if NIRS of fecal samples could pre- validation samples are a subset of the cali- developed from mixtures. Volesky and dict the percentage of mountain big sage- bration population. Currently, NIRS of Coleman (1996) showed that the addition brush (Artemisia tridentata Nutt. ssp. fecal material is used to predict dietary of botanically pure samples to esophageal vaseyana (Rydb) Beetle) in sheep diets. composition of samples whose relation- extrusa calibration data sets was benefi- Four specific objectives were evaluated: 1) ship to the calibration population is not cial. However, pure diets of many of the stepwise regression was compared to mod- well known (Snowder et al. 2001, USDA- plant species of interest to the study of ified partial least squares regression; 2) NRCS 1999), even though Coleman et al. dietary habits of ruminants cannot be fed calibrations based on maximum, mini- (1995) warned that extrapolation beyond because of problems with toxicosis at high mum, and no sagebrush in the diets were the conditions represented in the calibra- levels of intake. Nonetheless, it may be compared to calibrations that included tion samples is risky. possible to predict botanical composition intermediate levels of sagebrush; 3) the Because the material being predicted of diets with just 2 levels of the compo- ability of calibrations based on diets com- (i.e., the diet) is not available for standard nent of interest, which would provide a posed of dried sagebrush to predict diets laboratory analysis, the use of NIRS to great improvement in the design of cali- from frozen sagebrush were investigated; predict dietary parameters of grazing ani- bration trials by reducing the number of and 4) the relationship of H with the preci- mals presents unique challenges compared samples required for calibration. sion and accuracy of predictions was to other agricultural applications of this Development of calibration equations investigated. technique. In other applications of NIRS using stepwise regression techniques (with technology, if samples are spectrally dif- a limited number of individual wave- ferent from the calibration population, or lengths) has not been compared with mul- Methods if the predicted value is outside the expect- tivariate procedures using information ed range, the true reference value of the from all wavelengths. Multivariate proce- This research was conducted at the sample can be determined by standard lab- dures use a data reduction technique such USDA, U.S. Sheep Experiment Station oratory procedures. as principal component analysis (PCA), located in Dubois, Ida. (112°10'W, One of the statistics used in NIRS analy- which reduces the original data to sets of 44°21'N). Fecal materials from a feeding sis to determine if a sample is from the scores that have 2 properties 1) the vari- trial with sheep fed known levels of sage- same spectral population as the samples ance of the first set of scores is maximized brush with several background diets were that the calibration equation was devel- and successive sets of scores explain suc- used to develop NIRS fecal calibration oped from is called H (Shenk 1989). The cessively less of the variation in the origi- equations. Equations were validated with H statistics is the distance of the sample nal data; and 2) the sets of scores are samples from 2 independent feeding trials from the population centroid (i.e., the pop- uncorrelated to each other. This is accom- with known levels of sagebrush with a sin- ulation mean in multi-dimensional space). plished by multiplying the original log 1/R gle background forage. Because the com- The H statistics is standardized to have data by a weighting coefficient (i.e., eigen- position of the base diet did not vary in the unit variance so it represents the number vector) and summing these values across validation trials, calibration equations for of standard deviations that a sample is all wavelengths to create a score for each sagebrush were identical to equations for from the multi-dimensional population sample. This is analogous to the coeffi- other components, and samples from these mean. Shenk (1989) suggested that if the cients in a multiple regression equation trials could not be used to develop unam- H of a predicted sample is greater than 3 transforming a set of independent vari- biguous calibration equations. standard deviates from the average spec- ables to a single dependent variable, i.e., trum of the calibration samples, the results predicted Y variable. The principle com- 1996 Feeding Trial should be reported with caution. Whereas ponent scores produced can then be used The 1996 feeding trial consisted of feed- the use of H for typical NIRS agricultural in multiple regression to predict the value ing varying levels of mountain big sage- applications was investigated by Shenk of the dependent variable (i.e., botanical brush, early bloom alfalfa hay (Medicago (1989) and Shenk and Westerhaus composition in the current study). In this sativa L.) and a grass hay composed of (1991 a,1991 most previous research on study we used modified partial least b), smooth brome (Bromus inermis Leyss.) using fecal NIRS equations did not exam- squares regression, which is similar to and timothy (Phleum pratense L.). The ine the relationship of H to prediction PCA regression except that information sagebrush was collected in late September accuracy (Coleman et al. 1989, Lyons and from the dependent variable is included in 1996 and air-dried in the shade (maximum Stuth 1992, Coleman et al. 1995, Leite and the transformation so that the resultant daily temperature 21 to 30° C). After dry- et al. scores have maximum correlation with the Stuth 1995, and Coleman 1999). ing, sagebrush branches were flailed The NIRS technique can be calibrated to dependent variable and the residuals against a wire panel over a large tub to predict botanical composition of clipped obtained after each factor is calculated are remove leaves and small current year's forages composited to create samples of standardized (divided by the mean residual stems. plant mixtures (Coleman et al. 1985, value) before calculating the next factor. The flailed sagebrush was chopped Pitman et al. 1991, Petersen et al. 1987, Pitman et al. (1991) reported there were using a small flail-type garden shredder; Moore et al. 1990). The calibration sam- no consistent differences between modi- in these studies were created by com- fied partial least squares regression and ples 1Protocols for feeding trials were approved by posting mixed samples with known modified stepwise regression when pre- Animal Care and Use Committee of the U.S. Sheep amounts of the different botanical species. dicting the botanical composition of tropi- Experiment Station and the University of Idaho # Coleman et al. (1990) demonstrated that cal grass-legume mixes. 8028).

JOURNAL OF RANGE MANAGEMENT 55(4) July 375 other forages were ground through a ham- twice daily, and a sub-sample of feces was points in a moving average). Scatter cor- mer mill with a 2.5 cm screen. Diets were collected. Afternoon samples were com- rection was done with the standard normal mixed to contain 0, 4, 8, 12, 16, and 24% posited with collections made the follow- variance and detrend procedure (Barnes et sagebrush with a base diet of hays in the ing morning and dried in a forced air oven al. 1989). following alfalfa/grass proportions: 0:100, at 60° C. 20:80, 40:60, 60:40, 80:20, and 100:0 for Data Analyses a total of 36 different diets. Molasses was Wyoming Feeding Trial Before further analyses, data from the added to the diets (5% DM) during the The Wyoming trial used 16 Rambouillet 1996 and 1998 trials were individually mixing process to cause feed particles to wether lambs (28 to 41 kg body weight) examined to identify and remove outlier adhere together and reduce sorting when fed diets containing mixtures of hand-har- samples. To eliminate outliers, separate eaten. vested current year's growth of mountain calibration equations were done for the Diets were fed to 36 mature, white- big sagebrush leaves and native grass hay 1996 and 1998 trials. Samples that had an faced, western range ewes housed in indi- (Ngugi et al. 1995). Sagebrush leaves H larger than, 3 or a residual "t" statistic vidual pens. One animal was fed each diet. were harvested in September from the greater than 3 were eliminated. This Diets were fed at 0800 and 1600 hours; western edge of the Medicine Bow Range, resulted in the elimination of 6 and 10 refusals (orts) were removed at 0800 Carbon County, Wyo., and stored in samples from the 1996 and 1998 data sets, hours. The trial consisted of a 6-day adap- sealed plastic bags in a freezer until fed. respectively, or approximately 5% of the tation period and 4-day collection period. Four diets in the following proportions of samples from each of the data sets. Fecal samples were collected from the grass hay:sagebrush: 100:0, 90:10, 80:20, Calibration equations were evaluated for floor of the pens at 1600 and 0800 hours and 70:30 were fed. Four animals received usefulness based on validation statistics and dried in a forced air oven at 60°C. each diet. The trial consisted of a 9-day for samples that were not used for calibra- adjustment period followed by a 6-day tion. Calibration refers to the development 1998 Feeding Trial collection period when all urine and feces of a regression equation using the reflect- The 1998 feeding trial consisted of vary- were collected. Fecal samples were com- ed energy (log 1/R) of the near infrared ing levels of mountain big sagebrush posited by animal for the 6-day collection spectra as independent variables to predict stored either air-dried or frozen before the period and ground through a 1-mm screen the botanical component (i.e., dependent study (Fraker 1999). Sagebrush for this of a Wiley mill. Samples were reground variable or reference value). Validation trial was current season's growth of vege- through a cyclone mill before being refers to the ability of the calibration equa- tative stems (no flowering stalks) collected scanned to collect NIRS reflectance. tion to predict the reference value of sam- the last week of August 1997. Half of the Details of this trial were presented by ples not part of the data set used in the sagebrush was air-dried (maximum tem- Ngugi et al. (1995). development of the calibration equation. perature 30 to 33°C) and stored in plastic Validation samples are referred to as bags; half was frozen promptly after col- Equation Development `internal' if they are a subset of a uniform lection and stored in a freezer. Diets were All fecal samples were ground in a group of samples used to develop a cali- 0, 8, mixed to contain 16, and 24% of cyclone mill to pass a 1-mm screen, bration equation and independent if they either dry or frozen sagebrush with a base packed into sample cells with a near- were from a set of samples from a differ- diet of 1:1 alfalfa/grass (grass was a mix- infrared transparent quartz cover glass, ent trial or treatment. The statistics evalu- ture or smooth brome and timothy hay) for and scanned 32 times using a NIR ated included standard error of prediction a total of 7 different diets. Molasses was Systems, Inc. (Silver Spring, Md.) model (SEP), standard error of prediction cor- added to the diets (5% DM) during the 6500, scanning reflectance monochroma- rected for bias (SEPc), coefficient of mixing process to cause feed to adhere tor. Reflected energy (log 1/R) was mea- determination (R2), slope, and bias. together and reduce sorting. sured, averaged over the 32 scans and These statistics indicate the relationship An intake and digestion trial was con- recorded at 2-nm intervals from 400 to between the predicted percentage sage- ducted in January 1998 using 6 to 7- 2,500 nm. brush in the diets and the percentage sage- month-old, white-faced crossbred lambs Calibration equations were developed brush that was fed. The coefficient of with an average weight of 47 kg. The ani- using stored NIRS spectra from fecal sam- determination is the amount of variation in mals were penned individually and 5 ani- ples as the independent variables and per- sagebrush fed that is explained by fecal mals were assigned to each of the 7 diets cent sagebrush fed in the diets as the NIRS predictions and is an indicator of the for a total of 35 animals (34 wethers and 1 dependent reference data. Scanned sam- precision of the calibration equation. ewe). Animals were fitted with fecal bags ples were daily individual animal feces in Slope and bias are estimates of the accura- to determine total fecal production. Feed the 1996 and 1998 trials, whereas for the cy of the calibration equation; in general was offered in excess from 0800 to 1100 Wyoming trial, they were individual ani- accuracy decreases as the deviation from 1 hours and from 1300 to 1600 hours. Every mal feces composited across 6-days. and 0, respectively, increases. However, 30 min, while feed was offered, feed Before calibration, each spectrum was slopes may not always be directly compa- bunks were checked and more feed was rable because of the differences in the Y transformed with a (1,8,8) derivative. The added if needed. This procedure limited first number in parenthesis is the order of intercept. The SEP and SEPc statistics are the potential for sorting and minimized the the derivative, the second number is the estimates of error in units that are related amount of refusal. The trial consisted of a gap (number of data points over which the to the unit of measurement; in this study 5-day adaptation period and a 6-day col- derivative is calculated), and the third the percentage sagebrush in the diet. The lection period. Fecal bags were emptied number is the smooth (number of data SEP includes error that results from poor

376 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 precision (i.e., lack of fit) and systematic Table 1. Comparison of stepwise to modified partial least squares regression equations on valida- error as a result of slope and bias deviation statistics for calibrations developed from the 1996 trial on days 1 and 3. tions from 1 and 0. The SEPc has much of He the systematic error (lack of accuracy) Validation data set R2 Slopea removed and is an indicator of precision. 96 2&4 Stepwise Objective 1: This objective compared 0.92 1.00 MPLS 0.96 1.01 stepwise (SW) and modified partial least 98 Dry squares (MPLS) calibrations. Fecal sam- Stepwise 0.66 ples from days 1 and 3 of the 1996 trial MPLS 0.93 1.26 (96 were used to develop aSlope l&3) calibration slope of the line between reference and predicted values (indicator of accuracy). eBias = equations using either SW, or MPLS = systematic error (i.e., mean of differences, and an indicator of accuracy). regressions and validated with either days cSEP = standard error of prediction (error caused by both lack of precision and lack of accuracy). dSEP(c) standard error of prediction corrected for bias (indicator of precision). 2 and 4 (96 2&4) eAverage= from the same trial (i.e., H = standardized Mahalanobis distance (mean spectral distance of validation data set from centroid of calibra- internal validation), or the 1998 Dry sam- tion data set in multiples of standard deviates). ples (i.e., independent validation). The 1998 Dry samples were used for independent validation because they most closely of dried forages represents a simpler tion data sets (Table 1; Fig. 1). Predictions duplicated the procedures used for the method for conducting trials. Calibrations for the internal validation set (96, 2&4) feeding trial from which samples for cali- were developed using all samples from the showed that MPLS compared to SW bration were produced. The best calibra- 1996 trial. Frozen samples from 1998 and slightly increased the precision of predictions as by tion method was determined by comparing Wyoming samples were used for valida- indicated the higher R2 and lower SEP(c), but not the accuracy validation statistics and then was used for tion and dried samples from 1998 served of predictions (i.e., bias and slope). The value of all further analyses. as a control. MPLS compared to SW calibrations for Objective 2: This objective examined Objective 4: The effect of the value of H improving precision were more apparent the effect of eliminating diets with inter- on the precision and accuracy of predic- for the independent validation of the 1998 mediate levels of sagebrush on the preci- tions was evaluated for the 1998 Frozen and Wyoming data sets. The value of H is Dry samples. In this comparison R2 and sion and accuracy of resultant calibration SEP(c) showed much improvement in equations. The purpose of this objective the number of standard deviates that a particular sample is from the mean of the MPLS compared to SW, the similar bias was to determine if the efficiency of feed- for the 2 predictions showed that accuracy ing trials for developing fecal NIRS cali- population of samples that were used to develop the calibration equation. Thus it was about equal. The improved precision bration equations could be improved by of the MPLS calibration compared to the reducing the number of levels fed without represents the spectral similarity between the predicted samples and the calibration SW calibration indicates that the wave- affecting the precision or accuracy of pre- lengths selected by the samples. The simple correlation between stepwise procedure dictions. The MPLS calibrations were for the 1996 calibration data set did not H and the absolute value of the residual developed for all 1996 samples and com- include all of the relevant wavelengths for was calculated to determine if the preci- pared to calibrations from 3 subsets: HiLo the independent sion of predictions (i.e., difference 1998 Dry validation data All contained all samples with 0, 4, and set. The average H value was higher for between actual and predited values) was 24% sagebrush; HiLo Mixed contained the MPLS predictions of the 1998 Dry related to the spectral samples with 60:40 or 40:60 proportions closeness of the samples compared to the SW prediction of to the of alfalfa and grass as the background and sample calibration population. Each these samples. The higher H statistic is a 0, 4 and 24% sagebrush; HiLo Pure con- data set was divided in half based on the result of the use of all wavelengths in the median tained samples with 100% alfalfa or 100% value of H in the data set. The pre- MPLS equation compared to only 5 wave- grass and 0, 4, and 24% sagebrush. The dictions were compared between the 2 lengths in the SW equation. This apparent diets with no sagebrush were included in subsets to determine the effect of H on increase in spectral distance between inde- the calibration data sets because, in prac- accuracy of predictions. If the slope of the pendent predictions and the mean calibra- tice, treatments with only the base diets line relating the predicted and actual levels tion spectra, and the assumption that it require relatively less labor to conduct of sagebrush in the diets was the same for reduces the robustness of the resultant because the target plant species does not both of the subsets of data, it would indi- equations, caused most previous fecal need to be hand-collected. Thus the addi- cate that accuracy was not affected by the NIRS calibrations to use SW equations. H) tional level can be added to the calibration spectral distance (i.e., between the Because of the improved precision of pre- data set at a relatively low cost. The cali- sample and the calibration population. dictions using the MPLS equation, this brations for this objective were validated calibration procedure was used for subsequent analysis. using the 1998 Dry samples. Results and Discussion Objectives 3: The purpose of this analy- Objective 2: The effect of calibrations sis was to evaluate the ability of calibra- based on the 1996 trial after removing the tion equations developed from feeding tri- Objective 1: Comparison of stepwise samples with intermediate levels of sage- als using target forages that were dried to (SW) and modified partial least squares brush (i.e., 8, 12, and 16%) on precision predict diets from feeding trials where the (MPLS) calibrations for days 1 and 3 of and accuracy for predictions of the 1998 target forage was kept frozen until fed. the 1996 trial showed that calibration Dry samples, is shown in Table 2 and The latter case more closely represents a using MPLS improved predictions for Figure 2. Reducing the number of samples normal grazing situation, whereas the use both the internal and independent valida- reduced the accuracy of predictions, but

JOURNAL OF RANGE MANAGEMENT 55(4) July 377 mixtures compared to the 100% alfalfa or 100% grass background mixtures. 2 5 - A As the number of samples in the calibration data set was reduced, the average H of 20 the validation data decreased, even though the validation data set was the same for 15 the 4 different calibration data sets. This would normally be interpreted to mean 10 -J that the validation data were more spectrally similar to the calibration data. 5. However, in this instance it indicates that as the number of samples in the calibration 0 data set decreased the variance of the mul- tidimensional centroid increased and because H is standardized to have unit variance the distance of the predicted population from the calibration population is -10 -5 0 5 10 15 20 25 30 smaller in terms of multiples of the unit variance. Comparison of the MPLS validation statistics for 1998 Dry samples between 30 - a) Tables 1 and 2 also demonstrates the LL of size of calibration data set on the 25 effect average value of H. In this instance the R2 the same for 20 slope and are essentially calibrations based on days 1 and 3 of the 1) 15 1996 samples (Table or all days (Table 2), but the average H statistic increased 10 from 2.7 to 3.8 when all of the samples were used in the calibration data set. This analysis indicates that increased efficiency of experimental design can be obtained by reducing the number of intermediate levels and that there is an advan- -5 tage of using spectrally distinct backgrounds, i.e., pure rather than mixed background diets.

.5 0 5 10 15 20 25 30 Objective 3: The use of calibrations based on dried samples to predict diets of Predicted % Sagebrush animals eating an independent source of frozen forage is shown in Table 3 and O SW MPL - 1:1 Line Figure 3. Compared to the 1998 Dry samples, the validation statistics for the 1998 Frozen samples showed that the slope for Fig. 1. Comparison of stepwise regression (SW) and modified partial least squares regression the frozen samples (1.43) was greater than (MPLS) predictions of percent sagebrush fed using 1996 feeding trial days 1 and 3 for cali- that for the dried samples (1.27), resulting bration and either internal validation with 1996 days 2 and 4 for validation (A) or indepen- in a larger bias and consequently a larger dent validation using 1998 Dry samples for validation (B). SEP. The SEP(c) was increased only slightly for the frozen samples, and the R2 this reduction was less than the propor- (HiLo Mixed) background or 100% alfalfa was similar for the dried (0.94) and frozen tional decrease in number of samples. For or 100% grass (HiLo Pure) as a back- (0.93) predictions. These statistics, in con- instance the HiLo ALL subset represented ground further reduced the precision and junction with the increased deviation from a 50% reduction in the number of samples accuracy of predictions. Validation statis- unity of the slope, indicate that using an but only a 3% reduction in the R2 between tics from calibrations based on HiLo equation based upon dried forages from actual and predicted diets, a slope with a Mixed samples showed that this subset the independent 96 trial to predict percent- 4% greater deviation from unity, a 27% resulted in a much higher bias and a slope age 98 frozen sagebrush in the diet, increase in the SEP, and a 16% increase in with a greater deviation from unity. The decreased the accuracy of predictions with SEP(c). reduced performance of HiLo Mixed cali- little effect on the precision. Further reducing the calibration data brations presumably was caused by a lack Validation using the Wyoming samples sets to 23 samples by using only samples of unique spectral information between the was much worse than the 1998 Frozen from diets with mixed alfalfa and grass 60:40 and 40:60 alfalfa:grass background samples, but was similar to the 1998

378 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Table 2. Effect of reducing the number of intermediate treatment levels in the Calibration data set ferent treatment groups or individuals on validation statistics for 1998 Dry samples. within a group of animals. But it could not na be used to determine how much of a plant Calibration data set R2 Average Hf species of interest should be fed to * 96 A11 137 0.94 3.8 increase their level of consumption to a HiLo Allh 69 0.91 2.2 desired level. HiLo Mixed` 23 0.78 2.3 Objective 4: For typical agricultural HiLo Pure' 23 0.82 1.3 aNumber applications, the size of H (i.e., the stan- of samples in calibration data set. dardized spectral distance of the predicted bSlope = slope of the line between reference and predicted values (indicator of accuracy). `Bias systematic error (i.e., mean of differences, and an indicator of accuracy). sample aSEP = from the calibration data spectral standard error of prediction (error caused by both lack of precision and lack of accuracy). eSEP(c)= mean) normally is used to determine if a = standard error of prediction corrected for bias (indicator of precision). rAverage H = standardized Mahalanobis distance (mean spectral distance of validation data set from centroid of calibra- sample is from the same spectral population data set in multiples of standard deviates). tion as the samples used to develop cali- g96 All = all samples. bration equations. This analysis investi- hHLLo All = all samples with 0, 4, or 24% sagebrush. 'HiLo Mixed = samples with background matrix of 40:60 or 60:40 alfalfa and grass hay with 0, 4, or 24% sagebrush. gates the relationship between H and the HiLo Pure = samples with background matrix of 100% alfalfa or 100% grass hay with 0, 4, or 24% sagebrush. precision and accuracy of predictions *Exceeded value recommended by Shenk et al. (1989). when calibration equations from the 1996 trial were used to predict the 1998 Frozen Frozen in that most of the error was a interval level of measurement for the fecal or Wyoming diets. The value of H was result of reduced accuracy. The low accu- NIRS predicted values, an increase from positively correlated to the absolute value racy was caused primarily by the high 0.6 to 10.9% is about twice as much as an of the residual (Fig. 4), which indicates slope (2.63), which indicates that the increase from 6.3 to 10.9%, but it is not that prediction error increased as H range of the predicted values was about appropriate to state that 6.3% is about 10 increased. However, when the data sets one-third the range of the actual values. times larger than 0.6%. In practice this were split into two groups based on the The compressed range of data in these means that fecal NIRS predictions could median value of H, the slope (P > 0.30) predicted data sets indicates that when the be used to estimate the difference in and intercept (P > 0.30) of the predictions predicted samples are from a different botanical composition of the diets of dif- for the 2 groups were not affected by the population than the calibration samples, the ratio scale of measurement for the reference value of the samples may be reduced to an interval scale for the predict- 30 , ed values. Measurement scale refers to the different properties (i.e., relationships and 2 operations) of the numbers that can be applied to the measurements. The 4 levels of measurements in order of increasing amount of information that can be determined from them are 1) nominal, 2) ordinal, 3) interval, and 4) ratio. Nominal 10 measurements name the attribute; ordinal measurements can be rank ordered; for interval measurements the distance 0 a) between attributes has meaning; and ratio LL U measurements have an absolute zero and a meaningful ratio can be constructed. The -5 effect of different scales of measurements on the information that can be extracted -10 from measured attributes can be demon- -15 -10 -5 0 5 10 15 20 25 30 35 strated using the reference values of sagebrush fed, which are ratio scale measure- Predicted % Sagebrush ments, and the fecal NIRS predicted values, which are interval scale measure- OAII HiLo All A HiLo Mixed ments. The mean predicted values for the 8, 16 and 24% diets in the 1998 Frozen XHiLo Pure -1:1 Line data were 0.6, 6.3 and 10.9%, respectively. data it is appro- For the reference ratio Fig. 2. A comparison of modified partial least squares calibrations using reduced subsets of priate to state that the difference between calibration samples from the 1996 trial was used to predict samples from animals fed dry 8 and 24% is twice as large as the differ- sagebrush in the 1998 trial. (All = all samples; HiLo = 24, 4, and 0% sagebrush with all ence between 8 and 16%. It is also appro- backgrounds; Mixed = 24, 4, and 0% sagebrush with 60:40 and 40:60 alfalfa:grass back- priate to state that 16% is 2 times larger grounds; and Pure = 24, 4, and 0% sagebrush with 100% alfalfa or 100% grass back- and 24% is 3 times larger than 8%. At the grounds).

JOURNAL OF RANGE MANAGEMENT 55(4) July 379 Table 3. Validation statistics for predicting fresh samples using calibration from 1996 samples with coefficients of determination between 1998 Dry samples included for comparison. actual and predicted values, consistently

He accounted for 90% or more of the varia- Validation data set R2 Slopea Biasb SEPc SEP(c)d Average tion. The consistently high precision but 1998 Dry 0.94 1.27 4.8 5.6 2.7 3.8 * low accuracy may indicate that the wave- * 1998 Frozen 0.93 1.43 6.8 7.5 3.1 4.1 lengths that best predicted the calibration * Wyoming 0.9 2.63 14.2 16 . 7.7 12.8 aSlope set were not the optimal set for the valida- = slope of the line between reference and predicted values (indicator of accuracy). tion data sets. However, the wavelengths bBias = systematic error (i.e., mean of differences, and an indicator of accuracy). CSEP = standard error of prediction (error caused by both lack of precision and lack of accuracy). from the calibration data sets apparently dSEP(c) = standard error of prediction corrected for bias (indicator of precision). covaried closely with the appropriate CAverage H = standardized Mahalanobis distance (mean spectral distance of validation data set from centroid of calibration data set in multiples of standard deviates). wavelengths for the independent valida- Exceeded value recommended by Shenk et al. (1989). tion set. The suspected shift in optimal wavelengths between the calibration and valida- size of H (Fig. 5). Walker et al. (1998) Frozen samples (4.1) and a much lower tion data sets could be caused by the fact reported that when NIRS fecal predictions accuracy (i.e., slope = 2.63 vs. 1.43), but that the 1996 calibration diets used one for percent leafy spurge (Euphorbia esula good precision as shown by the high R2 source of sagebrush for all diets, which L.) in the diet were divided into groups (0.90 vs. 0.93; Table 3). Finally, in both caused low spectral variation in the feces based on the size of H, validation statistics trials the treatments with the highest levels and may be responsible for the poor accu- were similar between the groups. of sagebrush had the highest H values, racy of prediction of independent samples. The positive relationship between H and which may have implications on the Use of a single source of sagebrush means magnitude of the residual and the lack of source of this spectral variation. that all diets regardless of the proportion effect of H on the relationship between The results of this study show that when of sagebrush in them had the same spectral actual and predicted percent sagebrush in predicted samples are from the same pop- information relative to sagebrush. Had the diets show that for these data, the ulation as the calibration samples, such as more sources, e.g., locations, season and value of H is positively related to preci- the internal validation of the 1996 trial years of sagebrush been used the resultant sion but not to accuracy. This interpreta- (Table 1), NIRS of feces can accurately equation would have been more global. tion of the relationship of H to the preci- predict botanical composition of diets. However, animal variation between tri- sion and accuracy of predictions is also However, when predicted samples are als also may have been an important con- supported by comparison of the average H from a different population, accuracy is tribution to the lack of accuracy. Mature for the 1998 Frozen and Wyoming sam- low and the predicted values should be ewes were used for the 1996 calibration ples to their validation statistics. The considered to represent an interval scale of trial, whereas approximately 6 month old Wyoming data set had a much higher measurement. Of interest is that the preci- whether lambs were used for all of the average H (12.8) compared to the 1998 sion of the predictions, as indicated by independent validations. Age, sex, or physiological status of the 35 animal could affect fecal composition via the effect these factors have on intake and 30 digestibility. Furthermore, the amount of sagebrush in the diet affects both intake 25 and digestibility (Ngugi et al. 1995), and w effects are greater for fresh com- 5- these 0 20 pared to dried sagebrush (Fraker 1999). ID Typically 60% or more of ruminant feces (J 15 is composed of metabolic products includ- Co ing bacteria and endogenous waste (Van Soest 1982). The remainder of the feces composed of residual forage matter would consist of undigested plant cell walls (i.e., neutral detergent fiber). 0. ½ Comparison of the slopes of the independent validation sets shows that as the -5 -T- T T effect of treatments on intake and -10 -5 0 5 10 15 20 25 digestibility increased, the accuracy (but not precision) of the predictions Predicted % Sagebrush decreased. In the 1998 trial, frozen diets caused a small decrease of forage intake 098 Dry 98 Frozen L Wyoming -1:1 Line (Fraker 1999), which was associated with an increase in average H and greater deviation from unity of the slope (i.e,1.27 and Fig. 3. Prediction of diets from frozen sagebrush using calibration equations based on diets from dried sagebrush; predictions of dry samples are included for comparison. 1.43 for the dry and frozen diets, respectively).

380 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 30 intake and digestibility was associated with a much larger average H and lower slope than the 1998 predictions. This may 25 indicate that when fecal NIRS is used to 0 D predict the percent of a chemically defended plant in the diet, the spectra on which 20 -I 1998 Frozen Wyoming the calibrations are based may result from 0 r0.58 r _ 0.88 the effect of secondary plant metabolites 15 -I on the digestive process rather than the plant residues in the feces. If this is the 0 0 0 case, the loss of accuracy for independent 10 validations noted in this study may be somewhat lessened when fecal NIRS is 0 used to predict the diets of grazing herbivores because animals normally will limit their intake of chemically defended plants 0 below a level that restricts the rate of

0 5 10 15 20 nutrient capture (Provenza 1995.)

Value of H Conclusions O 98 Frozen UVyoming Relative to the design of future trials for developing fecal NIRS calibration equa- Fig. 4. Relationship between the H value of samples and the absolute value of the residual for tions, several recommendations can be independent validation samples and modified partial least squares regression using all 1996 made from these results. Variation in the samples. source of a target plant species, the background forages, and the animals used for Ngugi et al. (1995) reported a much BW on the 0% sagebrush diet to 9 g/kg the trial should be maximized to the extent greater effect of percent sagebrush in the BW on the 30% sagebrush diet with a con- possible. The composition of diets should diet on intake and digestibility for the comitant decline in apparent dry matter consist of the `high' level of the target Wyoming trial than was reported for the digestibility of from 59 to -3% in this plant that is slightly above the amount 1998 trial. Intake declined from 38 g/kg study. The greater effect of sagebrush on expected to be selected by free grazing herbivores, the `low' level of about 5% of

30 , the target plant, and a diet that consists of only the background forage. Furthermore, 25 we recommend the use of MPLS calibrations to increase the spectral information 20 considered in the calibration equation compared to the limited number of wave-

15 lengths that are used in SW calibrations. When feeding trials are conducted to develop fecal NIRS calibration equations, 43 10 the samples used for calibration rarely can be considered to be from the same popula- a 5 -I tion as the predicted samples. Even if the 0) target plant species is well represented in LL 0 -I the diets, other botanical components normally will not be, and the potential asso- a ciative effects on digestion indicate that

-10 assumptions about the populations being the same would be tenuous. -10 -5 0 5 10 15 20 25 30 35 Furthermore, the H statistic may not Predicted % Sagebrush always indicate if a sample is from the same population as the calibration set. Therefore, until evidence to the contrary is o 98 Frozen, H 4 98 Frozen, H 4 available, predictions of dietary botanical L Wyoming, H 1 X Wyoming, H 1 composition using fecal LAIRS should be considered to represent an interval scale of measurement. This scale of measurement Fig. 5. Effect of size of H on the validation accuracy.

JOURNAL OF RANGE MANAGEMENT 55(4) July 381 is adequate for most problems of interest sheep. M.S. Thesis, Univ. Idaho, Moscow, Van Soest, P.J. 1982. The nutritional ecology on rangelands such as comparing treat- Ida. of the ruminant. 0 & B Books, Corvallis, ment effects on diet preference or measur- Harniss, R.O., D.A. Price, and D.C. Tomlin. Ore. 374 pp. 1975. Number of fistula samples needed for Volesky, J.D. and S.W. Coleman. 1996. in ing the phenotypic variation preference determination of sheep diet on sagebrush- Estimation of botanical composition of for a target plant in a population of ani- grass range. J. Range. Mange. 28:417-419. esophageal extrusa samples using near mals, but would not be sufficient if it is Leite, E.R. and J.W. Stuth.1995. Fecal NIRS infrared reflectance spectroscopy. J. Range necessary to know the actual composition equations to assess diet quality of free-rang- Manage. 49:163-166. of the diet. ing goats. Small Ruminant Res. 15:223-230. Walker, J.W., D.H. Clark, and S.D. McCoy. The reduced information in the interval Lyons, R.K. and J.W. Stuth. 1992. Fecal 1998. Fecal NIRS for predicting percent NIRS equations for predicting diet quality of leafy spurge in diets. J. Range Manage. scale of measurement compared to a ratio free-ranging cattle. J. Range Manage. 51:450-455. scale of measurement will be more than 45:238-244. offset by the ability to greatly increase Moore, K.J., C.A. Roberts, and J.O. Fritz. sample size by using fecal NIRS com- 1990. Indirect estimation of botanical compared to other methods for determining position of alfalfa-smooth bromegrass mix- botanical composition of diets. For tures. Agron. J. 82:287-290. Ngugi, R.K., Hinds, and Powell.1995. instance, Harniss et al. (1975) reported F.C. J. Mountain big sagebrush browse decreases that for sheep grazing sagebrush-bunch- dry matter intake, digestibility, and nutritive grass rangelands, 150 fistula diet determi- quality of sheep diets. J. Range Manage. nations would be necessary to estimate the 48:487-492. percent of plant species that comprise 20% Petersen, J.C., F.E. Barton, II, W.R. of the diet to within ±10% of the mean Windham, and C.S. Hoveland. 1987. with a 95% probability. Such sampling Botanical composition definition of tall fescue-white clover mixtures by near infrared intensity exceeds the practical capacity of reflectance spectroscopy. Crop Sci. any other available method for determin- 27:1077-1080. ing botanical composition of diets except Pitman, W.D., C.K. Piacitelli, G.E. Aiken, fecal NIRS. and F.E. Barton II. 1991. Botanical composition of tropical grass-legume pastures estimated with near-infrared reflectance spec- Literature Cited troscopy. Agron. J. 83:103-107. Provenza. F.D. 1995. Postingestive feedback as an elementary determinant of food prefer- Barnes, R.J., M.S. Dhanoa, and S. J. Lister. ence and intake in ruminants. J. Range 1989. Standard normal variate transformation Manage. 48:2-17. and detrending of near-infrared diffuse Shenk, J.S. 1989. Monitoring analysis results. reflectance spectra. Appl. Spectroscopy. p. 27-28. In: G.C. Maren, J.S. Shenk, and 43:772-777. F.E. Barton II (eds.). Near Infrared Coleman, S.W., R.E. Barton, II, and R.D. Reflectance Spectroscopy (NIRS): Analysis Meyer. 1985. The use of near-infrared of Forage Quality. USDA Agr. Handb. No. reflectance spectroscopy to predict species 643. composition of forage mixtures. Crop Sci. Shenk, J.S. and M.O. Westerhaus. 1991a. 25:834-837. Population definition, sample selection, and Coleman, S.W., S. Christiansen, and J.S. calibration procedures for near infrared Shenk.1990. Prediction of botanical compo- reflectance spectroscopy. Crop Sci. sition using NIRS calibrations developed 31:469-474. from botanically pure samples. Crop Sci. Shenk, J.S. and M.O. Westerhaus. 1991b. 30:202-207. Population structuring of near infrared spec- Coleman, S.W., H. Lippke, and M. Gill. tra and modified partial least squares regres- 1999. Estimating the nutritive potential of sion. Crop Sci. 31:1548-1555. forages. p. 647-695. In: H.G. Jung and G.C. Shenk, M.O. Westerhaus, and S.M. Vth J.S., Fahey, Jr. (eds.) Proc. Int. Symp. on the Abrams. 1989. Supplement 2. Protocol for Nutrition of Herbivores. San Antonio, Tex. NIRS calibration: Monitoring analysis Coleman, S.W., J.W. Stuth, and J.W. results and recalibration. p. 104-110. In: Holloway. 1989. Vol. 2 p. 881-882. In: G.C. Maren, J.S. Shenk, and F.E. Barton II Monitoring the nutrition of grazing cattle (eds.). Near Infrared Reflectance Spectro- with near-infrared analysis of feces. Proc. scopy (NIRS): Analysis of Forage Quality. XVI Int.. Grassld. Congr. Nice, France. USDA Agr. Handb. No. 643. Coleman, S.W., J.W. Stuth, and J.W. Snowder, G.D., J.W. Walker, K.L. Launch- Holloway. 1995. Prediction of intake by baugh, and L.D. Van Vleck. 2001. Genetic near-infrared spectroscopic analysis of fecal and phenotypic parameters for dietary selec- samples. p. 145-155. In: F.N. Owens, D. tion of mountain big sagebrush (Artemisia Gill, K. Lusby, and T. McCollum (Ed.) tridentata Nutt. ssp. vaseyana [Rydb] Beetle) Symp: Intake in Feedlot Cattle. Oklahoma in Rambouillet sheep. J. Anim. Sci. Exp. Sta. P-942. 79:486-492. Fraker, M.J. 1999. Examination of digestive USDA-NRCS, Grazing Lands Technol. Inst. and behavioral mechanisms to explain indi- 1999. Technical Support for the NIRSINUT- vidual variation in sagebrush consumption by BAL Nutritional Management System.

382 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 J. Range Manage. 55: 383-389 July 2002 Short-term monitoring of rangeland forage conditions with AVHRR imagery

DAVID P. THOMA, DEREK W. BAILEY, DANIEL S. LONG, GERALD A. NIELSEN, MARI P. HENRY, MEAGAN C. BRENEMAN, AND CLIFFORD MONTAGNE

Authors are Research Assistant, Department of Soil, Water and Climate, University of Minnesota, St Paul, Minn.; assistant professors (Bailey and Long), Northern Agricultural Research Center, Montana State University, Havre, Mont.; professors (Nielsen and Montagne) and research assistants (Henry and Breneman), Department of Land Resources and Environmental Sciences, Montana State University, Bozeman, Mont.

Abstract Resumen

A study was conducted to determine the potential of using Se condujo un estudio para determinar el potencial de utilizar Advanced Very High Resolution Radiometer (AVHRR) imagery las imagenes de Radiometria Avanzada de Muy Alta Resolucion to monitor short-term changes in rangeland forage conditions on (RAMAR) para monitorear los cambios a corto plazo en las a regional scale. Forage biomass and nitrogen concentration condiciones de forraje del pastizal a escala regional. La biomasa were estimated at 6 study sites throughout a typical grazing sea- de forraje y las concentraciones de nitrogeno fueron estimadas son (April to October). Study sites were located in northern and en 6 sitios de estudio a to largo de una estacion tipica de apacen- southern Montana in areas classified as foothills grassland and tamiento. (Abril a Octubre). Los sitios de estudio estuvieron shortgrass prairie. Normalized Difference Vegetation Index localizados en el norte y sur de Montana, en areas clasificadas (NDVI) values from AVHRR imagery (1 km pixels) were used to como pastizales de pie de montana y pastizales cortos. Los vat- predict live biomass, dead standing biomass, total biomass, nitro- ores del Indice de Diferencias Normalizadas de Vegetacion gen (N) concentration and standing N. Values of the NDVI were (IDNV) obtenidos a partir de imagenes de RAMAR (pixeles de 1 correlated (r > 0.4, P < 0.01) to live, dead, and total biomass esti- km) se usaron para predecir la biomasa viva, la biomasa muerta mates and standing N, but were not correlated to N concentra- en pie, la biomasa total la concentracion de nitrogeno (N) y el N tion (r = 0.04, P = 0.8). Relationships between NDVI and vegeta- en pie. Los valores de IDNV se correlacionaron (r > 0.4, P < tive attributes were similar (P > 0.05) for all 6 study sites, which 0.01) con las estimaciones de biomasa viva, muerta y total y el N indicates that NDVI could be used to predict forage abundance en pie, pero se correlacionaron con la concentracion de N (r = at multiple locations and at variable dates. Using simple linear 0.04, P = 0.8). Las relaciones entre IDNV y los atrrbutos vegeta- regression, NDVI accounted for 63% of the variation in live and tivos fueron similares (P > 0.05) para los 6 sitios de estudio, to total biomass, 18% of the variation in dead biomass, 66% of the cual indica que los IDNV pudieran ser usados para predecir la variation in standing N, but < 1 % of the variation in N concen- abundancia de forraje en multiples localidades y en fechas vari- tration. The NDVI obtained from AVHRR imagery was a good ables. Con regresion lineal simple, el IDNV explico el 63% de la predictor of forage abundance as measured by live, dead and variacion de la biomasa viva y total, el 18% de la variacion de la total biomass as well as standing N, but it was not related to for- biomasa muerta, el 66% de la variacion del N en pie, pero age quality as measured by N or crude protein concentration. On explico menos del 1 % de la variacion en la concentracion de N. a regional basis, land managers could use AVHRR-NDVI values El IDNV obtenido a partir de imagenes de RAMAR fue bueno to identify areas with high or low levels of forage abundance that para predecir la abundancia de forraje medido como la biomasa may result from factors such as drought, variable precipitation viva, muerta y total, asi como por el N en pie, pero no estuvo patterns, or uneven grazing. relacionado a la calidad del forraje medida como la concentracion de N o proteina cruda. En base regional, los manejadores de tierras pudieran usar los valores de RAMAR-IDNV para Key Words: NDVI, biomass, nitrogen, regional identificar areas con niveles altos o bajos de abundancia de forraje que pueden resultar de factores tales como sequia, patrones variables de precipitacion y apacentamiento desigual. Forage quantity and quality are major determinants of carrying capacity and performance of grazing animals (Valentine 1990). Ground-based methods are not practical for assessing forage sensing research has demonstrated, however, that quantity of quality and quantity over extensive geographic areas. Remote green vegetation may be estimated using vegetation indices derived from multispectral, aerospace imagery (Tueller 1989, Wessman et al. 1995). Vegetation indices, which contrast the Research was funded by the Upper Midwest Aerospace Consortium, Montana Space Grant, Montana State University Mountain Research Center, Montana State high chlorophyll absorption region in the red vs. the high reflec- University Department of Land Resources and Environmental Sciences and the tivity in the near infrared, are assumed to indicate plant photosyn- Montana Agricultural Experiment Station. This manuscript has been assigned thetic activity and aboveground primary production (Tucker Journal Series No. 2000-80, Montana Agricultural Experiment Station, Montana State University, Bozeman. 1979, Wiegand and Richardson 1990). Frequently, red and Manuscript accepted 20 Oct. 01. infrared image bands are used to compute a Normalized

JOURNAL OF RANGE MANAGEMENT 55(4) July 383 Difference Vegetation Index (NDVI), (NDVI) and forage characteristics could that comprised relatively homogenous defined as the ratio of the difference be evaluated in a wide range of conditions vegetation. All study areas comprised 9 between near-infrared and red wave- during the summer grazing season. km2 (3 x 3 pixels of AVHRR) except for lengths to the sum of the 2 (Colwell 1974, Three of the study areas were estab- the site near Gardiner that measured only Tucker et al. 1983, Justice and Hiernaux lished on foothill grassland: 1) 10 km east 1 km2 (1 pixel of AVHRR) Boundaries of 1986). of Livingston, Mont. (45° 41' 48" N 110° this study area were specifically placed so One of the most commonly encountered 26' 05" W), 2) 20 km south of Havre, that the center of the AVHRR pixel was in NDVI image products is that provided by Mont. (Thackeray Ranch, 48° 21' 50" N the geometric center of the site. The extent the NOAA Advanced Very High 109° 35' 00" W), and 3) 30 km southwest of homogeneous vegetation was not as Resolution Radiometer (AVHRR) (Gutman of Chinook, Mont. (Ross Ranch, 48° 15' extensive at the Gardiner location. 1991). The NDVI has been widely investi- 46" N 109° 28' 29" W). Average annual gated for use in predicting the amount of precipitation for these 3 areas varies from Ground Sampling vegetative biomass in rangeland. For exam- 350 to 450 mm (Caprio et al. 1994). Mean standing crop of live and dead ple, Merrill et ai. (1993) derived a vegeta- Dominant species includes rough fescue vegetation (live and dead biomass) and tion index from Landsat multispectral scan- (Festuca scabrella Torr.), bluebunch mean N concentration were estimated in ner data for use as an auxiliary variable for wheatgrass (Pseudoroegneria spicatum early April and on a bi-weekly basis from a linear regression model to predict green (Pursh) A. Love), western wheatgrass May to October at the Gardiner and herbaceous phytomass on rangeland in (Pascopyrum smithii (Rydberg) Love), Livingston study areas. These vegetation Yellowstone National Park. Tucker and Idaho fescue (Festuca idahoensis Elmer), attributes were estimated once per month Vanparet (1985) found a linear relation- and bluegrass (Poa spp.). Hilltops are from May to August at the other 4 study ship between AVHRR-NDVI and end of forested with Douglas fir (Pseudotsuga sites. season dry biomass of grassland in the menziesii var, glauca (Bessin.) Franco) or To maximize sampling efficiency, each African Sahel. Kennedy (1989) discovered ponderosa pine (Pinus ponderosa Dougl.). study area was divided into smaller, rela- that AVHRR-NDVI was strongly correlat- Vegetation along drainage bottoms tively homogenous map units using a ed with biomass and cover of range vege- includes chokecherry (Prunus virginiana stratified sampling approach (Tucker et al. tation over a wide range of soil and topo- L.), western snowberry (Symphoricarpos 1983, Townshend and Justice 1986, graphic conditions in Tunisia. occidentalis Hook), and quaking aspen Tappan et al. 1992). Map units were delin- Land managers, livestock producers, (Populus tremuloides Michx.) (Mueggler eated on 1:24,000 orthophotographs and animal feed suppliers, and others are often and Stewart 1980). topographic maps based on spatial patterns interested in both forage quantity and The 3 remaining study areas were estab- in vegetation, landform, and land use. At 1) quality. For example, a combination of lished in shortgrass prairie: 5 km west least 2 plots (0.5 m2) were randomly forage quantity and quality attributes of Gardiner, Mont. (45° 01' 21" N 110° 41' placed within a mapping unit. Study areas 1) 2) (standing nitrogen, kg N ha was a good 28" W), 30 km northwest of Chinook, included at least 4 sampling units, and predictor of where cattle preferred to graze Mont. (48° 42' 37" N 109° 29' 07" W), and measurements were obtained from a total 3) (Senft et al. 1985). Handheld radiometers 10 km north of Zurich, Mont. (48° 36' of 20 to 40 plots within each study area have been used to estimate nitrogen con- 33" N 108° 57' 26" W). These sites near during each sampling period. Plots were centration of plant tissue (Tucker 1979, Chinook and Zurich are characterized by clipped to ground level and clippings were Plummer 1988). Researchers have not, terraces and glacial till plains with inter- separated in the field into standing live however, examined the potential of mittent, dendritic drainages. Average and dead plant tissue. Clipped plant parts AVHRR for assessing the quality of range annual precipitation is 250 to 300 mm that were green along >_ 50% of their forage. The purpose of this study was to (Caprio et al. 1994). Dominant species length were bagged as live material. Parts investigate relationships between includes blue grama (Bouteloua gracilis with < 50% green along their length were AVHRR-NDVI and measures of forage (H.B.K.) Griffiths), western wheatgrass, separated as dead material. Separations of quality, and describe linear regression and needle-and-thread (Stipa comata Trin. plant material were made without regard models for predicting both quantity and & Rup.). Other grasses include Sandberg to the identity of plant species because the quality of forage in Montana rangelands. bluegrass (Poa sandbergii Vasey), and spatial resolution of AVHRR imagery was upland sedges (Carex spp.) (Mueggler and too coarse for species level detail (Merrill Stewart,1980). Materials and Methods et al. 1993, Kremer and Running 1993). The pixel size of the advanced very high Clipped plant material was oven-dried at resolution radiometer (AVHRR) is 1 x 1 50°C for 48 hours (Richardson et a1.1983, Study Areas km, but the potential error in registration Richardson and Everitt 1992). Total bio- Six study areas were established in 1997 between successive image scenes is up to mass was computed as the sum of the live on grasslands in southern and northern 2 to 3 pixels. Usually the error is less than and dead biomass after oven drying. Montana. Grasslands were classified into 1 pixel. The AVHRR image-to-image reg- Percent nitrogen by dry weight was deter- 2 types: shortgrass prairie and foothill istration using automated correlation tech- mined on ignition with a LECO FP-328 grassland, based on dominant vegetation niques improves geometric accuracy (root combustion furnace (LECO Corp., St. (Mueggler and Stewart 1980). Selection of mean square less than 1 pixel) as com- Joseph, Mich.). study sites in these 2 types of rangeland pared to traditional image-to-map proce- Live standing N was computed as the ensured that relationships between nor- dures (Eidenshink 1992). To enhance dry weight of live biomass multiplied by malized difference vegetation index accuracy in the NDVI calibration effort, percent N in live biomass (Senft et al. all study areas were situated within zones

384 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 1000 16

14 ab ab b 800 ab s z 'cu 12 I rn ab ab ab z 10 U) 600 U) a a CU 8 E z 0 I 400 C a) > 4 J 200 2

0 0

600 1400 a a a a is 500 1200 a a a CM a a 0)1000 - 400 U) a U) 800 E 300 n0 0 600 -o 200 .0 CU 400 0a) 0 100 200

a> >> U) m C J a) C 0) M E Q 0 U LL coO

Fig. 1. Mean values of live biomass, dead biomass, total biomass and standing nitrogen for April or May, June, July, August, and September or October during the 1997 growing season (n = 33). Mean values of these dependent variables are also presented for foothills grassland and shortgrass prairie sites. Standard errors are presented for each mean. Monthly means without common superscripts differ (P < 0.05). (** *** Means for foothill grassland and shortgrass prairie sites differ if superscripts are presented = P < 0.01, = P < 0.001).

1985). Mean values of the biomass (live, can vary from -1.0 to 1, and in this study were pooled together because there were dead, and total), N concentration of live NDVI values for each pixel were scaled so fewer measurements collected in these biomass and standing N were determined that the range was 0 to 200 using the fol- months and they represented similar for- for each sampling event within each map lowing equation. age conditions typical of spring condi- were then spatial- * tions. Similarly, those collected in unit. These mean values NDVI = (NDVIraw + 1) 100 (1) ly weighted by the percentage of the study September and October were pooled area comprised by each map unit. Mean In addition, imagery is geometrically together and represented conditions often projection to biomass, standing N and N concentration registered to a common map observed in autumn. Tukey's studentized Thus, for study area during a given sampling enable image-to-image registration. range test was used for mean separation period were calculated from the spatially land managers, who are mainly interested (Steel and Torrie 1980). We computed the weighted means for each map unit. in the data for vegetation assessment and arithmetic mean from the NDVI values monitoring, are freed from preprocessing from the 9 pixels comprising the study the raw data. (9 km2) considered it representa- Model for Estimating areas and Regression Live biomass, dead biomass, total bio- NDVI value for each study Forage Quantity and Quality tive of the real mass, nitrogen concentration, standing site. The value of a single pixel was used Maximum value composite AVHRR- nitrogen, and NDVI estimates from on the for the study area near Livingston that NDVI images were obtained from the at each site were ground measurements measured 1 km2. USGS EROS Data Center in Sioux Falls, evaluated using a model that contained Estimates of biomass and N content that S. Dak. Maximum value composite site type (foothill or prairie), month were then were regressed on NDVI values. imagery is based on the maximum NDVI (April-May, June, July, August, and Because the study areas were expected to value for each pixel of imageries acquired September-October) and the site type by differ in terms of vegetation and climate, daily over a 2-week period (Holben 1986) month interaction (SAS 1985). Measure- location was evaluated as a fixed effect in and is used to compensate for variable ments collected during April and May PROC GLM of SAS (SAS 1985). This atmosphere conditions. Raw NDVI values

JOURNAL OF RANGE MANAGEMENT 55(4) July 385 variable was used to test if the absolute 3.5 values in vegetation parameters varied among locations at given NDVI values. In Foothill 3.0 other words, did the y-intercepts from the F Prairie regression of vegetation attributes on z 2.5 NDVI vary among locations? In addition, a the interaction between NDVI and loca- ab y tion was included as a test if the slopes of 2.0 the regression of vegetation attributes on y y NDVI varied among locations (Timm and c y Mieczkowski 1997). Therefore, we con- 1.5 -1 I c structed a predictive model where standing live biomass, dead biomass, total biomass, 1.0-i %N of live tissue and standing nitrogen were dependent variables, and NDVI, 0.5 location, and NDVI by location were independent variables. A simple linear regres- 0.0 sion model was also evaluated with NDVI as the only independent variable. 160

Results and Discussion a ab Foothill 150 H ab Prairie Live biomass was greater (P < 0.05) in

August than when first measured in April 140 HI ab and May (Fig. 1). Dead biomass and total b x biomass did not change (P > 0.05) from x p 130 April to September-October. Live bio- Z x mass, dead biomass, total biomass and x x standing nitrogen were greater (P < 0.001) 120 on foothill grassland than on shortgrass H prairie (Fig. 1). This difference in biomass was likely the result of site differences in 110 H precipitation. Foothills grassland typically receives 100 to 250 mm more rainfall than shortgrass prairie. Live biomass in June 100 was greater (P < 0.05) than during August. April June July August September Standing nitrogen in June was greater (P = - May - October 0.05) than during September-October. The interaction between site type and month of Fig. 2. Mean values of nitrogen concentration and NDVI for April or May, June, July, collection was not important (P > 0.1) for August and September or October for foothill grassland and shortgrass prairie sites during the 1997 growing season (n 33). Standard errors are presented for each mean. There was live biomass, dead biomass, total biomass, = an interaction (P < 0.05) between month of measurement and site type, and comparisons of and standing nitrogen. However, there was monthly means were separately made for each site type. Monthly means from foothill an interaction between site type and month grassland (black bars) without common superscripts (a, b, c) differ (P < 0.05). Monthly of collection for plant nitrogen concentra- means from shortgrass prairie sites (gray bars) without common superscripts (x, y) differ (P tion (P = 0.003) and NDVI (P = 0.04), and < 0.05). monthly means from each site type were evaluated separately. For foothill grass- prairie, NDVI tended (P = 0.06) to vary between NDVI and locations were not sig- land sites, nitrogen concentration in among months. Changes in NDVI more nificant (P > 0.05) in the initial regression April-May was greater (P < 0.05) than closely followed the patterns observed in model for all dependent variables. measurements collected in July, August, live biomass and standing nitrogen, which Consequently, data were pooled, and a and September-October (Fig. 2). Nitrogen combines both live biomass and nitrogen simple linear regression model containing concentration in June was greater (P < concentration. We anticipated that NDVI only NDVI was used to analyze the data. 0.05) than measurements collected in would be associated with live biomass and As in this study, Wylie et al. (1995) used a August and September-October. For standing nitrogen because greenness of single relationship between AVHRR- shortgrass prairie, plant nitrogen concentra- plant foliage has been found to be highly NDVI and biomass to assign pastoral tion was greater (P < 0.05) in April-May representative of photosynthetic capacity zones of Niger into 3 classes of rangeland. than other months. For foothill grassland, and efficiency (Richardson et al. 1983, In simple linear regression models, NDVI was greater (P < 0.05) in June than in Tucker et al. 1983, Plummer 1988, NDVI explained 63% of the statistical September-October (Fig. 2). For shortgrass Benedetti and Rossini 1993). variation in both total and live biomass, Effects of location and the interaction 18% in dead biomass, and 66% in stand-

386 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 1600 4.0 Y= 1.37±0.002X 1400 3.5 R2=0.01 P=0.8 1200 3.0 . 1000 2.5 a, . 800 . . 2.0 600 1.5 r '. 400 M . 200 1.0

0 0.5 110 120 130 140 150 160 170 110 .120 130 140 150 160 170

NDVI NDVI

U, 0 110 120 130 140 150 160 170 110 120 130 140 150 160 170

NDVI NDVI

2200 2000 1800 1600 I 1400 1200 1000 800 6001 4001 20010 110 120 130 140 150 160 170

NDVI

Fig. 3. Scatter plots of live, dead and total biomass, nitrogen concentration and standing nitrogen versus NDVI values from AVHRR imagery. Simple regression lines of dependent variables (live biomass, dead biomass, total biomass, nitrogen concentration and standing nitrogen) on NDVI are also presented. ing N (Fig. 3). Poor correlation between levels across sites and sampling times may 463 and 25.0 ± 3.5, respectively, for the NDVI and dead biomass was not surpris- have overwhelmed any effects of nitrogen entire data set (Fig. 3). For total biomass, ing given the sensitivity of this spectral concentration on NDVI. the y-intercept for peak growth was -2896 index for green foliage. Wylie et al. (1995) When the simple linear regression was ± 717 and the slope was 28.1 ± 5.2 (R2 = reported that NDVI accounted for 67 to restricted to the period when range vegeta- 0.81, P = 0.001) as compared to -3694 ± 80% of the variation in biomass in 3 of 4 tion reached peak growth (1 June to 15 635 and 34.8 ± 4.8, respectively, for the years of their study in Niger. Plant nitro- July), model results improved slightly. For entire data set (Fig. 3). This improvement gen concentration was not related (P = live biomass, the y-intercept was -1475 ± is probably due to radiometric dominance 0.8) to NDVI (Fig 3). Differences in 537, and the slope was 15.9 ± 3.9 (R2 = of live vegetation, which tends to be NDVI resulting from variable biomass 0.71, P = 0.005) as compared to -2739 + stronger than that of soil background and

JOURNAL OF RANGE MANAGEMENT 55(4) July 387 dead biomass in that time of year. These imagery (Smith et al. 1997). Box, E.O., B.N. Holben, and V. Kalio.1989. results suggest that peak growth (June and Image processing and the use of NDVI Accuracy of the AVHRR Vegetation Index early July in our study) is a good time rather than absolute reflectance values as a predictor of biomass, primary production and net CO2 flux. Vegetatio 80:71-89 period to use NDVI to predict and com- reduces the chance that temporal variation in Caprio, J.M., D.I. Cooksey, J.S. Jacobsen, pare total annual vegetative growth within atmospheric conditions could result in the G.A. Nielsen, and R.R. Roche. 1994. a region. presumption of a change in forage condi- MAPS Atlas Version 5.0: A land and climate In this study, relationships between tions when in fact none occurred information system EB 125, Montana agri- NDVI and standing live biomass or stand- (Eidenshink 1992). Using the maximum cultural potentials system. Montana State ing N were based on model calibration at 6 NDVI value during a 2-week interval theo- Univ., Bozeman, Mont. study sites with ground data collected retically reduces the effects of cloud-conta- Carneggie, D.M, B.J. Schrumpf, and D.A. Mouat. 1983. Rangeland Applications. throughout the growing season. These minated pixels and theoretically reduces the p. 2329-2359. In: Manual of Remote Sensing models may need to be calibrated periodi- effects of atmospheric conditions associated (2n' ed.) Vol. II, R.N. Colwell (ed.), Amer. cally to account for variation in local cli- with changes in nadir viewing angles Soc. of Photogrammetry, Falls Church, Virg. mate and range vegetation phenology. (Holben 1986). Temporal variability in Colwell, J.E. 1974. Vegetation canopy However, any adjustments should be small reflectance is also accounted for in the reflectance. Remote Sens. Environ. 3:175-183. because regression coefficients (intercept NDVI value because it is based on the rela- Eidenshink, J.C. 1992. The 1990 contermi- and slope of live, dead or total biomass and tive difference in the reflectance of red and nous U.S. AVHRR data set. Photogramme- tric Eng. Remote Sens. 58:808-813. standing nitrogen on NDVI) from the lim- infrared wavelengths rather than the absolute Gutman, G.G. 1991. Vegetation indices from ited peak growth period (June and early values (Colwell 1974, Tucker et al. 1983). AVHRR: An update and future prospects. July) in this study were similar in order of The importance of AVHRR-NDVI lies Remote Sens. Environ. 35:121-136. magnitude to those obtained from the in its ability to provide the temporal resolu- Holben, B.N. 1986. Characteristics of maximum- entire data, and the relationship between tion needed for frequent monitoring of value composite images from temporal AVHRR the dependent vegetative variables and changes in the quality of green vegetation. data. Int. J. Remote Sens. 7:1417-1434. NDVI did not vary among site types.The By exploiting relationships between the Justice, C.O. and P.H.Y Hiernaux. 1986. Monitoring the grasslands of the Sahel using interaction between site types and NDVI NDVI and live biomass or N, standing land NOAA AVHRR data: Niger 1983. Int. J. was not significant in this study. In addi- managers may be able to predict when and Remote Sens. 7:1475-1497. tion, the relationship between total biomass where shortfalls in higher quality vegeta- Kennedy, P. 1989. Monitoring the vegetation and NDVI observed by Kennedy (1989) in tion may occur within extensive manage- of Tunisian grazing lands using the normal- Tunisia was similar to observed in ment units. Furthermore, the predictive ized difference vegetation index. Ambio Montana for standing crops up to 2500 kg ability of standing N by AVHRR data may 18:119-123. ha' when similar regression models were allow managers to predict where livestock Kremer, R.G. and S.W. Running. 1993. Community type differentiation using applied. Intercepts were -3741 ± 1220 and may graze. Senft et al. (1985) reported that NOAA/AVHRR data within a sagebrush- -3693 ± 695 and slopes were 43.5 ± 11.6 cattle grazed in areas with greater standing steppe ecosystem. Remote Sens. Environ. 46: and 34.8 ± 4.8 in the Tunisia and Montana N. Cattle grazing was associated signifi- 311-318. studies, respectively. Confidence intervals cantly more with standing N than with bio- Merrill, E.H., M.K. Bramble-Brodahl, R.W. (95%) of regression coefficients between mass or N concentration. Marrs, and M.S. Boyce. 1993. Estimation NDVI and total biomass overlapped in all We conclude that maximum value com- of green herbaceous phytomass from Landsat MSS data in Yellowstone National Park. J. but 1 year of a 5-year study conducted by posite AVHRR-NDVI imagery could pro- Range Manage. 46:151-157. Wylie et al. (1992). duce reasonable estimates of live biomass Mueggler, W.F. and W.L. Stewart. 1980. Relationships between forage character- and standing N at regional scales, when Grassland and shrubland habitat types of istics and NDVI were linear over a rela- calibrated on ground data from 6 study western Montana. USDA Forest Serv., Gen. tively wide range of vegetative productivi- sites within foothills grassland and short- Tech. Rep. Int-66, Washington, D.C. ty levels (300 to 1300 kg ha' in mean live grass prairie areas of Montana. These Plummer, S.E. 1988. Exploring the relation- biomass) within the 2 major grassland results provided estimates of total forage ships between leaf nitrogen content, biomass and near-infrared/red reflectance Int. J. community types studied. If live biomass quantity and more importantly the quantity ratio. Remote Sens. 9:177-183. falls below 250 kg spectral ha', responses of higher quality forage (live biomass and Richardson, A.J. and J.H. Everitt.1992. Using of bare soil and dead biomass may domi- standing N) that would otherwise have spectral vegetation indices to estimate range- nate and result in underestimation of live been time consuming to obtain with con- land productivity. Geocarto Int. 7: 63-69. biomass (Tucker and Vanparet 1985). If ventional ground surveys alone. However, Richardson, A.J., J.H. Everitt, and H.W. biomass levels are high and plant canopy AVHRR-NDVI was not a good predictor Gausman. 1983. Radiometric estimation of is relatively closed, live biomass may be of forage quality as measured by nitrogen biomass and nitrogen content of Alicia grass. Remote Sens. Environ. overestimated, plant canopy closure is or crude protein concentration. 13:179-184. high SAS. 1985. SAS user's guide: Statistics. SAS (Carneggie et al. 1983, Box et al. Inst., Inc. Cary, N.C. 1989). Therefore, the model should not be Senft, R.L., L.R. Rittenhouse, and R.B. used to extend predictions beyond the Literature Cited Woodmansee.1985. Factors influencing pat- range of original data that were used to terns of cattle grazing behavior on shortgrass establish a linear relationship. Other fac- Benedetti, R. and P. Rossini. 1993. On the use steppe. J. Range Manage. 38: 82-87. tors that negatively influence NDVI-vege- of NDVI profiles as a tool for agricultural Smith, P., S. Kalluri, S. Prince, and R. DeFries. 1997. tation relationships include statistics: The case study of wheat yield esti- The NOAA/NASA Path- atmospheric finder AVHRR 8-KM Land Data Set. attenuation, topographic complexity, and mate and forecast in Emilia Romagna. Remote Sens. Environ. 45:311-326. Photogrammetric Eng. Remote Sens. pixel misregistration in successive 63:12-31.

388 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Steel, R.G.D. and J.H. Torrie. 1980. Tucker, C.J. 1979. Red and photographic Wessman, C.A., D.S. Schimel, S. Archer, Principles and procedures of statistics: A bio- infrared linear combinations for monitoring C.A. Bateson, B.H. Braswell, D.S. Ojima, 2°d metrical approach, Ed. McGraw-Hill vegetation. Remote Sens. Environ. 8:127-150. and W.J. Parton. 1995. New technologies Book Co. New York, N.Y.. Tucker, C.J. and C.L Vanparet. 1985. for remote sensing of ecosystem change in Tappan, G.G., D.J. Tyler, M.E. Wehde, and Satellite remote sensing of total herbaceous rangelands, p. 139-141. In: N. E. West (ed.), D.G. Moore. 1992. Monitoring rangeland biomass in the Senegalese Sahel: 1980-1984. Proc. Fifth Int. Rangeland Congress Soc. dynamics in Senegal with Advanced Very Remote Sens. Environ. 17:233-249. Range Manage. Denver, Colo. High Resolution Radiometer data. Geocarto Tucker, C.J., C. Vanparet, E. Boerwinkel, Wiegand, C.L. and A.J. Richardson. 1990. Int. 1:87-98. and A. Gaston. 1983. Satellite remote sens- Use of spectral vegetation indices to infer Timm, N.H. and T.A. Mieczkowski. 1997. ing of total dry matter production in the leaf area, evapotranspiration, and yield: 1. Univariate and multivariate general models: Senegalese Sahel. Remote Sens. Environ. Rationale. Agron. J. 82:623-629. theory and applications using SAS software. 13:461-474. Wylie, B.K., J.A. Harrington, R.D. Pieper, SAS Institute Inc, Cary, N.C. Tueller, P.T. 1989. Remote sensing technology and 1. Denda. 1992. A satellite-based range assessment system for the Sahel of Africa. Townshend, J.R.G. and C.O. Justice. 1986. for rangeland management applications. J. Geocarta Int. 1:79-85. Analysis of the dynamics of African vegeta- Range Manage. 42:442-453. Wylie, B.K., I. Denda, R.D. Peiper, J.A. tion using the normalized difference vegeta- Vallentine, J.F. 1990. Grazing management. Harrington, B.C. Reed, and G.M. tion index. Int. J. Remote Sens. 7: Academic Press, San Diego, Calif. 1435-1445. Southward. 1995. Satellite-based herbaceous biomass estimates in the pastoral zone of Niger. J. Range Manage. 48:159-164.

granite eoS

Your one-stop supplier for: Pasture and range grasses Alfalfa, shrubs, & wildflowers Custom seed blends Erosion control materials Environmental consulting services Ask for our catalog featuring over 600 native species and named vara etes.

Call or fax our Catalog (801) 768-44,' s'. fax (801) 768-3967 Granite Seed Co., 1697 W. 2100 North,Lehi, UT 84043 www.graniteseed.com

JOURNAL OF RANGE MANAGEMENT 55(4) July 389 J. Range Manage. 55: 390-393 July 2002 Prior feeding practices do not influence locoweed consumption

M.H. RALPHS, G. GREATHOUSE, A.P. KNIGHT, D. DOHERTY, J.D.GRAHAM, B.L. STEGELMEIER, AND L.F. JAMES

Authors are Rangeland Scientist, USDA/ARS Poisonous Plant Lab, Logan Ut 84341; Ranch Manager, CSU Research Foundation; Dept. Head, Clinical Sciences Dept.; former graduate student, Animal Science Dept., Colorado State Univ., Ft. Collins Colo 80523; County Agent, Union County Extension, Clayton NM. 88415; Pathologist, and Research Leader, USDA/ARS Poisonous Plant Lab.

Abstract Resumen

Anecdotal evidence suggests that cattle fed alfalfa hay during Evidencia anecdotica sugiere que el ganado bovino alimenta- the winter are inclined to graze locoweed on spring range. Two do con heno de alfalfa durante el invierno tiende a consumir studies were conducted to compare the influence of feeding alfal- "locoweed" en pasturas naturales durante la primavera. Se fa hay vs grass hay during the winter on subsequent consumption condujeron dos estudios para comparar la influencia de ofrecer of white locoweed (Oxytropis sericea Nutt. ex T&G) in the spring. heno de alfalfa versus heno de gramineas durante el invierno Eight cows were daily fed alfalfa hay (15.2% CP in 1998, % sobre el posterior consumo de "locoweed" blanco (Oxytropis CP in 2000) and 8 cows were daily fed grass hay (10.7% CP in sericea Nutt. Ex T&G) durante la primavera. Ocho vacas 1998, 12i % CP in 2000) plus 20% protein molasses block during fueron alimentadas diariamente con heno de alfalfa (15.2% PC the January-April winter feeding period. Treatment groups en 1998, 17.1 % PC en 2000) y 8 vacas fueron alimentadas diari- grazed in separate pastures (8 ha) on white locoweed-infested amente con heno de gramineas (10.7% PC en 1998, 12.1% en range in May and June in northern Colorado in 1998 and in 2000) mas 20% de un bloque de proteina y melazas durante el northeast New Mexico in 2000. Diets were estimated by bite periodo de alimentacion de invierno que fue de Enero hasta count. There was no difference in locoweed consumption between Abril. Los grupos pastorearon en areas separadas (8 ha) en una the 2 groups (P > 0.22). Cattle grazed locoweed for 5% of diets in pastura natural infestada con "locoweed" blanco en el norte de Colorado and 10% of diets in New Mexico. Feeding alfalfa hay over Colorado durante Mayo y Junio de 1998 y en el noreste de winter did not predispose cattle to graze locoweed in the spring. Nuevo Mexico en 2000. Las dietas fueron estimadas por el. Previous research showed other feeding practices or supplements metdo de contar bocados. No existio diferencia en el consumo do not affect locoweed consumption or poisoning. Prevention of de "locoweed" entre los 2 grupos (P > 0.22). Los bovinos pas- locoweed poisoning requires denying access to locoweed when it is torearon "locoweed" por un 5% de sus dietas en Colorado y relatively more palatable than associated forages. por un 10% de sus dietas en Nuevo Mexico. La oferta de heno de alfalfa durante el invierno no predispuso al ganado para pastorear "locoweed" durante la primavera. Investigacion pre- via muestra que otras practices de alimentacion o suplementos Key Words: Oxytropis sericea, alfalfa hay, grass hay, cattle, poi- no afectan el consumo de "locoweed" o la posibilidad de intoxi- sonous plant cacion con esta planta. La prevencion de la intoxicacion con "locoweed" requiere negar el acceso al "locoweed" cuando este Locoweeds (Astragalus and Oxytropis spp.) are the most es relativamente mas palatable que los forrajes asociados. destructive group of poisonous plants on western rangelands (Kingsbury 1964). Species of Astragalus and Oxytropis contain the toxic alkaloid swainsonine (Molyneux and James 1982) which dence of poisoning. One practice that was thought to influence inhibits essential enzymes in glycoprotein processing, resulting in locoweed consumption was prior feeding of alfalfa during the buildup of unmetabolized hybrid glycoproteins and disruption of winter. The objective of this study was to test the hypothesis that hormone and enzyme synthesis, transport and receptors (see cattle fed alfalfa hay over the winter will graze more white Stegelmeier et al. 1999 for review). Outward effects on animals locoweed (Oxytropis sericea Nutt. ex T&G) on spring range than include reduced fertility of both males and females, abortions, cattle wintered on grass hay. neurological disturbances (ranging from extremes of depression to aggression), and impaired prehension of food and water, resulting in weight loss and eventual starvation (James et al. 1981). Methods Locoweed causes such an insidious poisoning that ranchers and researchers have attempted many practices to reduce the inci- Winter feeding regimen Sixteen, 2-year-old first-calf heifers naive to alfalfa hay and Authors wish to thank Jennifer Vaad and Becky Floyd for assistance in data col- white locoweed were randomly assigned to 2 treatment groups in lection, Joel Vaad in handling the animals, and Jane Chambers in running swainso- winter 1998 at the Colorado State University Research nine analysis. Foundation Maxwell Ranch, 25 km north of Ft. Collins, Colo. Manuscript accapted l Oct. 01. (40° 56.39' N, 105° 15.33W). Animals were handled by methods

390 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 approved by an Institutional Animal and wheatgrass (E, smithii (Rybe.) Gould) dom- grass, and cool-season grasses consisted of Care Use Committee. One group was fed inating the swales. Warm-season grasses patches of western wheatgrass and occa- native meadow grass hay (10.7 % CP, as were minor components: blue grama sional plants of squirreltail. White fed) and supplemented with 20% protein- (Bouteloua gracilis (H.B.K.) Lag. ex locoweed was the dominant forb, with molasses block (0.9 kg cow-' day-'). The Steudel), purple 3-awn (Aristida purpurea fringed sage, and 3 non-toxic Astragalus other group was fed alfalfa hay (15.2% Nutt), and ring muhly (Muhlenbergia tor- species inhabiting the site. The area was CP, as fed). Cattle had free-choice access reyi (Kunth) Hitchc. Ex Bush). White fenced into two, 8-ha pastures using tem- to hay fed in round bales in feeders. Hay locoweed was the dominant forb, with porary electric fence. The groups were was fed from 11 February to 15 May fringed sagewort (Artemisia frigida Willd), randomly allocated to pastures, and rotat- 1998, and all cows calved during this peri- Utah milkvetch (Astragalus utahensis ed to the other pasture on a weekly basis od. Average weight at the beginning of (Torr.)T.&.G), rose pussytoes (Antennaria to minimize pasture bias. Diets were esti- feeding was 375 kg. Cows in the grass microphylla Rydb), geranium (Geranium mated by bite counts as described above. group lost 57 kg, and those in the alfalfa richardsonii Fisch. & Trautv.), and Data from the 2 trials were analyzed group lost 32 kg during the feeding/calv- Penstemon spp. present. separately because location and year were ing period. Mean body condition score Standing crop was estimated at the totally confounded. Percentage of bites of (scale 1-9) at the beginning of the grazing beginning and end of both trials by clip- forage classes were compared between the trial was 3.9, and 4.1 for the grass and ping the forage classes (cool-season grass, two groups in a repeated measures mixed alfalfa-fed groups, respectively. warm-season grass, forbs, and locoweed) model using compound symmetry covari- The cows were used on another in 1 by .25 m quadrats placed systematical- ate structure (SAS Inst. Inc., Cary, NC). locoweed grazing trial in 1999 (Ralphs et ly at 20 step intervals along 2 transects in Cow-within-group was the experimental al. 2001), but were separated into their each pasture (40 plots total). Standing crop unit, group was the main factor and week original groups and fed their respective of forage classes was compared between was the repeated factor. rations (alfalfa vs grass hay) from 13 pastures and beginning and end of each December 1999 to 16 May 2000 at the study separately using analysis of variance. USDA/ARS Poisonous Plant Research Two, 8 ha pastures were fenced with Results facility at Logan, Ut. The cows were temporary electric fence. The 2 groups 11 kg cow"' day-' of the were randomly assigned to a pasture and group-fed about White locoweed comprised 15% of the respective hays from large rectangular were rotated to the other pasture every standing crop in the Colorado grazing trial bales. The orchard grass hay contained week to reduce pasture bias. The cows in 1998, and 38% in the New Mexico trial 12.1% CP (as fed) and alfalfa hay con- were brought into an onsite corral at the in 2000 (Table 1). The greater abundance tamed 17.1% CP (as fed). The cows end of each week and blood samples were of locoweed may have contributed to the weighed 536 kg at the beginning of the taken by jugular venipuncture. Serum was higher consumption of locoweed at New feeding period. The grass-fed group lost separated, frozen and later analyzed for Mexico compared to Colorado (Table 2). 53 kg while the alfalfa-fed group gained swainsonine (Stegelmeier et al. 1995) as a There was no difference in abundance of 63 kg and the condition scores were about measure of intoxication. locoweed between pastures in either trial 4 and 6 respectively. The cows were bred Cattle diets were estimated using a bite (P > 0.10). Locoweed was in the late veg- to calve in July and were in the late stage count technique. Each cow was observed etative stage at the beginning of the of gestation during the spring grazing peri- for 5- minute periods and the number of Colorado trial, and matured to the late od. Only 5 cows from each group were bites of forage classes were counted: cool- flower stage at the end of the trial. available for the 2000 grazing trial season grass, warm-season grass, forbs, Locoweed was more advanced in the 2000 because of early calving and illness. There and locoweed. The cows were observed trial in New Mexico, which began in the was little locoweed available at the during the early morning and evening flower stage, but the flowers aborted and Maxwell Ranch in Colorado because of grazing periods, and each cow was few pods were produced due to drought. severe drought, therefore the cattle were observed at least once during each period. Crude protein content of locoweed was trucked to a site 10 km southwest of Observations were made each day during 15% in 1998 and 12.6% in 2000. Cool Capulin, N.M. (30° 41.25' N, 104° 08.35' the 5 week grazing trial. season grasses such as needle-and-thread, W) where white locoweed was abundant The 2000 grazing trial was conducted prairie junegrass and squirrel tail were abundant on the Colorado site in 1998 and for the spring grazing trial. near Capulin, N.M. and ran from 18 May dominated the standing crop. Blue grama, to 13 June. The site was a short-grass a warm season grass, dominated the site of prairie on a silty clay loam soil. Blue Spring locoweed grazing trials the 2000 study at New Mexico and began grama was the dominant warm-season The 1998 grazing trial on the Maxwell its seasonal growth during during the trial. Ranch in Colorado ran from 15 May to 18 June. The cows were fasted overnight, Table 1. Standing crop of forages classes (kg/ha ± SE) at the beginning and end of grazing trials in

weighed , then turned onto locoweed- 1998 and 2000. infested pasture. The range site was a mixed-grass prairie on shallow gravely Year Time Cool-season Warm-season Forb Locoweed Total sandy loam soil derived from weathered grass grass granite. Dominant cool-season grasses on ------(kg/ha)------the uplands included Needle-and-thread 1998 Begin 422±86* 29 56 (Stipa comata Trin. & Rupr.), prairie june- End 202±47* 209 ± 32 ±57 grass (Koleria macrantha (Ledeb) 2000 Begin 356±59** 31 107 201 Schultes), and squirrel tail (Elymus ely- End 101±18** 576±45** 17 99 139 moides (Raf.) Swezey), with western Difference in standing crop of forage classes at the beginning and end of each study: **P

JOURNAL OF RANGE MANAGEMENT 55(4) July 391 Table 2. Composition of cattle diets estimated by a bite count technique (± SE). alfalfa group very closely (Fig. 2). Swainsonine in the grass group dropped in Year Group Cool-season grass Warm-season grass Forb Locoweed week 3 to a greater extent than the bite ------(% of bites)------counts indicate. Swainsonine is cleared 1998 Alfalfa 79± 1.3 8±0.8 6±0.5 5±0.8 rapidly from the blood (Stegelmeier et al. Grass 80± 1.2 8±0.9 7±0.6 4±0.5 1995). Therefore, the drop in swainsonine 2000 Alfalfa 36 ± 1.5 36±1.4 19±0.9 8±1.0 was due to a decline in locoweed consump- Grass 37 ± 1.6 30±1.3 20±1.0 13±1.4 tion during the latter part of that week, There was no difference between groups in any of the forage classes in either year (P > 0.05). whereas the higher consumption of locoweed at the beginning of the week kept The cool season grasses, western wheat- between swainsonine in the blood and the weekly average consumption elevated. grass and squirreltail, were heavily grazed locoweed consumption (r= 0.79) in the There were no clinical signs of poison- during the 2000 trial. There were no dif- 1998 trial. When the groups were compared ing during either study. Cattle must graze ferences in standing crop of forage classes separately, the alfalfa group had a stronger locoweed for 10 to 25% of their diets for between pastures within the 2 grazing tri- correlation (r = 0.82) compared to the grass 3-4 weeks before signs of poisoning are als (P > 0.05). group (r = 0.67). Swainsonine in the blood apparent. The high levels of locoweed There was no difference in locoweed followed locoweed consumption of the consumption and swainsonine in the blood consumption between treatment groups in either year (P > 0.22). Mean locoweed consumption in the 1998 trial was 5% and 1998 study 4% of bites for the alfalfa and grass hay groups, respectively; and 8.2 % and 13.4%, respectively in the 2000 trial (Table 2). However, there was a group by week interaction in the 1998 trial (P = 0.0001). The grass group started eating locoweed before the alfalfa group during the second week of the trial (9% of bites), but the alfalfa group consumed more locoweed (21% of bites) during the 3rd week (Fig. 1). Locoweed consumption declined during the last 2 weeks of the 1998 trial as locoweed matured and its availability declined. Cattle began consuming locoweed at the beginning of the 2000 trial (10-20% of bites, Fig. 1). Locoweed consumption declined in week 3, but then increased during the last 2 days of the trial. In previous locoweed grazing trials on short-grass prairies in northeastern New Mexico Day (Ralphs et al. 1993, 1997a, 1997b), cattle 2000 study reduced locoweed consumption in late May and early June and ceased grazing it as it matured into the pod stage and warm-season grasses began rapid growth. In another study at the Colorado site, cattle ceased grazing white locoweed as it matured into the mature pod stage in July 1998, but continued to consume its leaves throughout the summer in 1999 (Ralphs et al. 2001). Summer rains were above average in 1999 in Colorado and the leaves of white locoweed remained green and succulent. Cool-season grasses were abundant and dominated cattle diets in 1998 at Colorado (Table 2). Warm season grasses were the dominant forage class at New Mexico, and diets were evenly split between cool-season grass that had matured, and warm-season grasses that were beginning to grow rapidly. Cattle preferred and sought-out Day the low growing, nontoxic Astragalus species and other forbs in the 2000 trial Fig. 1. Daily consumption of locoweed by cattle in the alfalfa and grass - fed groups in the (Table 2). 1998 and 2000 grazing trials (alfalfa group obscured by open circles during first 10 days of There was a moderate correlation the 1998 trial).

392 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Bachman, S.E., M.L. Galyean, G.S. Smith, 25 800 D.M. Hallford, and J.D. Graham. 1992. -.- Loco in diet- Alfalfa Group Early aspects of locoweed toxicosis and eval- -O- Loco in diet- Grass Group 20 Swainsonine- Alfalfa Group uation of a mineral supplement or clinoptilo- f-o- Swainsonine- Grass Group 600 lite as dietary treatments. J. Anim. Sci. 15 70:3125-3132. 400 James, L.F. and K.R. VanKampen. 1974. - Effect of protein and mineral supplementa- 10 tion on potential locoweed poisoning in sheep. - 200 J. Amer Vet. Med. Assoc. 164:1043-1043. 5i James, L.F., W.J. Hartley, and K.R. VanKampen. 1981. Syndromes of Astragalus 0 -0 poisoning in livestock. J. Amer. Vet. Med. Assoc. 178:146-150. Kingsbury, J.M. 1964. Poisonous Plants of the r 0 2 3 4 5 6 United States and Canada. Prentice-Hall, Week Entlewood Cliffs, N.J. Mikus, J.H., G.C. Duff, C.R. Krehbiel, D.M. Hallford, D.A. Walker, J.D. Graham, and Fig. 2. Relationship between locoweed consumed (as estimated by bite counts) and swainso- M.H. Ralphs. 2001. Effects of an estradiol nine level in the blood in the 1998 grazing trail. implant on locoweed consumption, toxicity, and recovery in growing beef steers. Prof. Anim. Sci.17:109-114 during these trials were not sustained for (Richards et al. 1999). This is not to say Molyneux, R.J. and L.F. James. 1982. Loco sufficient time to poison these animals. that mineral supplements should not be intoxication: indolizidine alkaloids of spotted used. Where minerals are deficient in for- locoweed. Sci. 216:190-191. age, supplements should be used to main- Pulsipher, G.D., M.L. Galyean, D.M. Discussion tain animal health, but they are unlikely to Hallford, G.S. Smith, and D.E. Kiehl. prevent locoweed poisoning. 1994. Effects of graded levels of bentonite on There was also concern that implants serum clinical profiles, metabolic hormones, Cattle fed alfalfa hay during the winter which stimulate growth may enhance and serum swainsonine concentrations in did not consume more locoweed in the locoweed poisoning. Implants increase lambs fed locoweed (Oxytropis sericea). J. and Anim. Sci. 72:1561-1569. spring early summer than cattle fed intake and rate of gain, thus animals may grass hay. Despite differences in nutrition- graze less selectively and/or consume Ralphs, M.H., D. Graham, M.L. Galyean, al plane or body condition, there was no and L.F. James. 1997a. Creating aversions more locoweed. On the other hand, estro- to locowed in naive and familiar cattle. J. difference in subsequent locoweed con- gen from implants may provide some neu- sumption. The notion that alfalfa (a highly Range Manage. 50:361-366. rological protection. An estradiol implant Ralphs, M.H., D. Graham, M.L. Galyean, nutritious legume) predisposes cattle to did not cause steers to select more graze locoweed (also a highly nutritious and L.F. James. 1997b. Influence of over- locoweed in a grazing trial, and did not wintering feed regimen on consumption of legume) was unsubstantiated in this study. affect the degree of poisoning or rate of locoweed by steers. J. Range Manage. Other anecdotal evidence suggested that recovery in a locoweed feeding trial 50:250-252. cattle wintered on green winter wheat are (Mikus et al. 2001). Ralphs, M.H., D.Graham, R.J. Molyneux, prone to graze locoweed when turned onto and L.F. James. 1993. Seasonal grazing of shortgrass prairie in the spring, based on locoweeds by cattle in northeastern New the assumption that cattle accustomed to Management Implications Mexico. J. Range Manage. 46:416-420. eating green forage would continue to seek Ralphs, M.H., G. Greathouse, A.P. Knight out green-growing locoweed. A previous and L.F. James. 2001. Cattle preference for grazing trial showed there was no differ- Cattle fed alfalfa during the winter did lambert locoweed over white locoweed. J. ence in locoweed consumption between not consume more locoweed in the spring Range Manage. 54:265-268. cows wintered on wheat pasture or native and early summer than cattle fed grass Richards, J.B., D.M. Hallford, and G.C. range (Ralphs et al. 1997b). hay. Previous research indicated that other Duff. 1999. Serum luteinizing hormone, Many minerals and feed additives have preconditioning feeding practices or sup- testosterone, and thyroxine and growth been investigated looking for "silver bul- plements have not prevented cattle from responses of ram lambs fed locoweed (Oxytropis sericea) and treated with vitamin lets" to prevent poisoning. Mineral supple- grazing locoweed. Prevention of locoweed E/selenium. Theriogenology 52:1055-1066. poisoning at this point in time lies in not ments did not prevent poisoning, nor delay Stegelmeier, B.L., L.F. James, K.E. Panter, symptoms in sheep fed Garboncillo allowing animals to graze locoweed- and R.J. Molyneux. 1995. Serum swainso- (Astragalus wootonii Sheld) (James and infested areas when it is relatively more nine concentration and a-mannosidase activ- VanKampen 1974), or prevent cattle from palatable than associated forages. ity in cattle and sheep ingesting Oxytropis grazing white locoweed (Allison and sericea and Astragalus lentiginosus. Amer. J. Graham 1999). Clay minerals have some Vet. Res. 56:149-154. scientific theory for their use since their Literature Cited Stegelmeier, B.L., L.F. James, K.E. Panter, differing electrical charges may bind to M.H. Ralphs, D.R. Gardner, R.J. Molyneux, and J.A. Pfister. 1999. The swainsonine, but a variety of clays and Allison, C. and J.D. Graham. 1999. Reducing in pathogenesis and toxicokinetics of locoweed minerals were not effective preventing locoism with management decisions. pp. or reducing locoweed poisoning (Bachman poisoning in livestock. J. Natural Toxins 64-66 In: T.M. Sterling and D.C. Thompson 8:35-45. et al. 1992, Pulsipher et al. 1994). Neither (eds), Locoweed Research Updates and did vitamin E/selenium injections hasten Highlights, New Mexico Agr. Exp. Sta. Res. recovery from locoweed poisoning Rep. 730. Las Cruces, N.M.

JOURNAL OF RANGE MANAGEMENT 55(4) July 393 J. Range Manage. 55: 394-399 July 2002 Clipping and precipitation influences on locoweed vigor, mortality, and toxicity

M.H. RALPHS, D.R. GARDNER, J.D. GRAHAM, G. GREATHOUSE, AND A.P. KNIGHT

Authors are Range/and Scientist, USDA/ARS Poisonous Plant Lab, Logan Ut. 84341; Chemist, USDA/ARS Poisonous Plant Lab, Logan, UT 84341; Extension Agent, Union County Extension, Clayton, N.M. 88415; Ranch Manager, CSU Research Foundation Maxwell Ranch; and Chair, Clinical Sciences

Dept., Colorado State Univ. , Ft. Collins, Colo. 80523.

Abstract Resumen

White locoweed (Oxytropis sericea Nutt. in T&G) is widespread El "White locoweed" (Oxytropis sericea Nutt. in T&G) esta dis- throughout the short-grass prairies and mountain grasslands and persa a traves de las praderas de zacates cortos y los zacatales de causes chronic poisoning of cattle, sheep, and horses. The objec- montana y causa un envenenamiento cronico a los bovinos, ovi- tive of this study was to determine the effect of clipping (simulat- nos y equinos. El objetivo de este estudio fue determiner el efecto ed grazing) on vigor, mortality and toxic alkaloid concentration del corte (apacentamiento simulado) en el vigor, mortalidad y of white locoweed. One hundred locoweed plants were marked at concentracion de alcaloides toxicos del "White locoweed". Cien each of 3 locations (New Mexico, Colorado, and Utah). Plants plantas de "White locoweed" fueron marcadas en cada una de 31 were stratified into 2 age/size classes: young/small < 5 stalks; localidades (New Mexico, Colorado y Utah). Las plantas se < older/large > 7 stalks (n = 50 in each class). Pairs of plants within estratiflcaron en dos clases de tamano/edad: jovenes/pequenas each age class that were as similar as possible were selected, and 5 tallos; grandes/viejas > 7 tallos (n = 50 en cada clase). Dentro to 1 of each pair (n = 25) was clipped at ground level annually for 4 de cada clase de edad se seleccionaron pares de plantas mas years. Vigor indices included number of stalks, number of flow- similares posible, y una planta de cada par (n = 25) se corto ering heads, leaf length, and flowering head height. Mortality anualmente a nivel del suelo durante 4 anos. Los indices de vigor was recorded and the toxic alkaloid swainsonine was measured. incluyeron el numero de tallos, numero de inflorescencias, longi- Clipping did not consistently reduce vigor. Flowering heads/plant tud de la hoja y altura de la inflorescencia. Se registro la mortali- declined in most clipped plants (P < 0.05), but stalks/plant dad y se midio la concentracion del alcaloide swansonina. El declined only in large clipped plants in Utah and small clipped corte no reduce en forma consistente el vigor. Las inflorescen- plants in New Mexico (P < 0.01), and clipping did not greatly cias/planta disminuyeron en la mayoria de las plantas cortadas affect leaf length or flowering head height. Clipping did not (P < 0.05), pero el numero de tallos /planta se redujo solo en las increase mortality, and did not affect swainsonine concentration. plantas grandes cortadas en Utah y las pequenas cortadas en However, there was a natural die-off that may have been related New Mexico (P < 0.01) y el corte no afecto grandemente la longi- to precipitation. There were negative correlations between pre- tud de las hojas y la altura de la inflorescencia. El corte no incre- cipitation and locoweed mortality (r = - 0.42 to -0.84), with most mento la mortalidad ni afecto la concentracion de swansonina. of the marked plants dying during the recent drought. Grazing Sin embargo, hubo una mortalidad natural que puede haber locoweed for short periods would likely not affect its vigor or tox- estado relacionada con la precipitacion. Hubo correlaciones neg- icity, but its population dynamics were affected by drought. ativas entre la mortalidad de "locoweed"y la precipitacion (r = - 0.42 to-0.84), la mayoria de las plantas marcadas murieron durante la sequia reciente. El apacentamiento de "Locoweed" Key Words: white locoweed, Oxytropis sericea, poisonous plant, por periodos cortos de tiempo probablemente no afectaria su population cycle vigor o toxicidad, pero su dinamica de poblacion fueron afec- tadas por la sequia.

White locoweed (Oxytropis sericea Nutt. in T&G) is the most widespread locoweed on western U.S. rangelands, and causes a advantage chronic poisoning in grazing livestock and wildlife. It is relative- Weeds that are not readily grazed have competitive However, if ly palatable in the spring and fall when warm-season grasses are over palatable grasses that are frequently defoliated. dormant, and signs of poisoning appear when animals consume it animals could be forced to graze weeds, this additional stress may for 3-4 weeks. It inhabits shortgrass prairies along the foothills of reduce their competitive ability and naturally reduce their stature the Rocky Mountains and high mountain grasslands, and typical- and abundance in the plant community. Furthermore, on-off graz- ly occurs on shallow or rocky soils (Payne 1957, Ralphs and ing (short intense periods grazing locoweed then removing the Cronin 1987). White locoweed is considered an increaser species animals to allow for detoxification), has been proposed as a graz- (i.e. increasing in density as range condition declines from good ing strategy to reduce locoweed poisoning (Pfister et al. 1996). to fair condition, but decreasing as excessive grazing pressure This type of a high-intensity, short-duration grazing system may causes further deterioration from fair to poor condition, Payne force livestock to uniformly graze all forage in a pasture, thus 1957). increasing grazing pressure on locoweed. The objective of this study was to evaluate the influence of defoliation on the vigor, mortality and toxic alkaloid concentration of white locoweed. Manuscript accepted 14 Sept. 01.

394 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Weather data were available from nearby attempted to separate effects of clipping seasonal precipitation (fall, winter, spring, stations and correlations were made on young, newly established plants, vs summer) or total water year precipitation between precipitation and mortality to older plants that had been established for to determine if there was a relationship determine the influence of drought on several years. Within each size class, pairs between reduced precipitation and mortal- these parameters. of plants were selected that were as uni- ity. Annual mortality was expressed as the form as possible. One of each pair (n = 25) % of the original 25 plants in each size/ was randomly selected and clipped at clipping treatment class that died each Methods ground level each subsequent year during year. The Palmer Drought Severity Index the flower stage of growth. The unclipped (Palmer 1965) for each of the respective member of each pair served as a control. regions was graphed to illustrate the sever- Three locations were selected that had Indices of vigor included: number of ity of drought during the study period histories of cattle loss to white locoweed. stalks (equivalent to tillers in a grass), (Fig. 1). The first site was 5 km west of DesMoines, number of flowering stems, longest leaf Concentration of swainsonine, the toxic New Mexico (N 36° 45.23', W 103° length, and tallest flowering head height. alkaloid in locoweeds, was measured in 51.00'), at an elevation of 2,200 m. Soils Pinnately compound leaves of white the leaves of clipped plants. Individual were silty clay loam with interspersed out- locoweed and flowering stalks originate plant samples from each year were croppings of volcanic basalt rock and a from the crown. Measurements were taken extracted using the general procedures of gentle southern exposure. Major grasses from the crown at ground level to the tip Molyneux et al. (1991). Detection and consisted of blue grama [Bouteloua gra- of the longest leaf and flowering head. quantitation of swainsonine was accom- cilis (H.B.K.) Lag. ex Steudel], sideoats Mortality was noted each year. All 3 sites plished using atmospheric pressure chemi- grama [B. curtipendula (Michx.) Ton], lit- were fenced to prevent grazing from live- cal ionization mass spectrometry (APCI- tle bluestem [Schizachyrium scoparium stock. There was no evidence of wildlife MS) (Gardner et al. 2001). Swainsonine (Michx.) Nash], and western wheatgrass grazing within the exclosures. concentration between small and large [Elymus smithii Gould]. White (Rybd.) The study began in June 1995 at New plants was compared the first year by one- locoweed was the dominant forb. Weather Mexico and Colorado when the plants way analysis of variance (ANOVA) within data were obtained from the Capulin were selected, measured, and those desig- locations. There was no difference National Monument, about 13 km from the nated for defoliation were clipped. between size classes (P > 0.05). study site. Because of logistic problems, the study Thereafter, only large plants were extract- The second site was on the Colorado did not begin in Utah until July 1996. All ed because many of the small plants did State Univ. Research Foundation Maxwell of the large clipped plants were dead at not have enough dry plant material for Ranch, located 40 km northwest of Fort New Mexico in 1998. Most of the other accurate quantification of the alkaloid (1 Collins, Colo. (N 40° 54.73, W 105° marked plants at New Mexico and Swainsonine concentration within indi- 16.33) at an elevation of 2,000 m. Soils g). Colorado were dead in 1999. There were vidual plants over years and among loca- were a shallow gravely loam with a gently sufficient numbers of plants remaining in tions was compared by a repeated mea- sloping west exposure. Associated grasses both size/treatment classes to analyze at sures mixed model using compound sym- included needle-and-thread (Stipa comata Utah in 2000. metry covariate structure. Locations and Trin. & Pupr.), squirreltail (Elymus ely- Because of the different start times and years were main effects and plants within moides Raf. Swezey), prairie June grass differing number of years plants that sur- locations and years was the repeated fac- (Koeleria macrantha (Ledeb.) Schultes) vived at each location, vigor data were tor. In a separate analysis, swainsonine in and blue grama. Weather data were analyzed separately for each location. clipped plants was compared to 8 control obtained from the Virginia Dale weather Vigor indices were analyzed by a mixed plants (not the paired plant) that were ran- station 6 km from the study site. model in SAS in a repeated measures domly selected each year from the sur- The third site was on top of the Raft design using compound symmetry covari- rounding area to determine if clipping River mountains in northwest Utah (N 41 ° ate structure to compare the main effects affected swainsonine concentration. Data 55.61', W 113° 24.42') at an elevation of of plant size (2) and clipping treatment (2) were analyzed separately for each location 3,000 m. Soils were shallow loam (10-25 over years, and their 2- and 3-way interac- by ANOVA comparing clipped and cm to bedrock), with 35 to 65% coarse tions. Individual plants were experimental unclipped plants over years. P < 0.05 was fragments. The plant community was units. considered significant. dominated by alpine sagebrush (Artemesia Plant mortality was analyzed by failure- scopulorum Gray), muttongrass (Poa fend- time analysis (Fox 1993). Mortality over leriana (Steud.) Vasey), and Idaho fescue time was compared among locations and Elmer). Weather data Results (Festuca idahoensis between age classes using the Cox was obtained from the nearest station at Proportional Hazards model in a log-log Grouse Creek, about 25 km west of the Vigor Indices model using SAS Proc Probit. To stan- study site and 1,200 m lower. The amount There was a significant year-by-size dardize time, years were placed on a of precipitation would be greater at the interaction (P < 0.05) for most indices at numerical scale (1 to 4) rather than calen- study site, but the timing and pattern all 3 locations, therefore the statistical der years. Survival curves were used to should be similar. model was reduced and the 2 size cate- display mortality over time. At each location, 100 white locoweed gories were analyzed separately for each Below average precipitation (drought) plants were marked by placing a num- location). occurred in New Mexico in 1997 and bered tent peg in the ground next to them. 1998, in Colorado in 1998 and 2000, and Plants were selected in 2 age/size classes: in Utah in 1999 and 2000. Correlations Stalks/plant young/small < 5 stalks; older/large > 7 were made between annual mortality and The initial number of stalks on large stalks (n = 50 in each class). The design plants was 7.6 at Colorado, 16.2 at New

JOURNAL OF RANGE MANAGEMENT 55(4) July 395 New Mexico New Mexico

3 6 912 3 6 9 12 3 6 9 123 6 9 123 6 9123 6 912 1995 1996 1997 1998 1999 1995 1996 1997 1998 1999 2000 Colorado 6 Colorado

X11 4a,

-6 0 3 6 912 3 6 9 12 3 6 9 123 6 912 3 6 9123 6 912 1995 1996 1997 1998 1999 2000 1995 1996 1997 1998 1999

Utah 6 Utah

C t6 a

Cl) C

-6 36 9 12 3 6 9 12 3 6 9 12 3 6 9 12 3 6 9 12 1996 1997 1998 1999 2000 1996 1997 1998 1999 2000 2001

Fig. 1. Palmer Drought Severity Index for northeast New Mexico, north Fig. 2. The number of stalks/plant in large and small plants central Colorado, and north western Utah. Mild drought -1.0 to -2.0; that were either clipped or not clipped at New Mexico, moderate drought -2.0 to -3.0; severe drought -3.0 to -4.0; extreme Colorado, and Utah over time. P < 0.05 indicate differences drought > -4.0. between the clipped and non clipped plants in each size class.

Mexico, and 16.8 at Utah. The average between clipped and unclipped plants (P = or leaf length (Mueggler 1972). The main number of stalks on small plants was 3.9 0.91). At Colorado and Utah, the number effect of clipping tended to be significant across locations, since the selection crite- of stalks on small plants increased over (P < 0.10) in all sizes and locations except ria for small plants was less than 5 stalks. years in both clipped and unclipped plants large plants at New Mexico, verifying that The main effect of clipping and the (P < 0.0001), illustrating the growth or the clipping treatment reduced the number clip-by-year interaction was significant (P increase in number of stalks over time. of flowering heads. The clip-by-year inter- < 0.05) only in large plants at Utah. The action was significant (P < 0.05) for large number of stalks on the large clipped Flowering Heads/Plant plants at Colorado and Utah, and small plants at Utah declined 16 7 from to over The initial number of flowering plants at Utah. At Colorado, the number of the study while the unclipped plants heads on large plants was 6.3 at Colorado, 20.2 heads on large clipped plants declined remained fairly constant (Fig. 2). The at New Mexico, and 24.9 at Utah. The from 7 at the beginning of the study to 2 in main effect of year was significant (P < 1996, number of heads on small plants was 1.1, then increased to the level of the 0.05) in both plant sizes and in most loca- 0, and 5.1 at the respective locations. unclipped plants by 1998 (Fig. 3). At tions. At New Mexico, the number of Utah, the number large clipped The number of flowering heads is a of heads of stalks on large plants declined over time 5, more sensitive measure of plant vigor than plants declined from 24 to and remained (P 0.0001), but there was no difference = the number of tillers (stalks in this study) low for the remainder of the study.

396 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 New Mexico

1995 1996 1997 1998 1999

Colorado

0 1 2 4 5

w 1995 1996 1997 1998 1999

Utah

C aRS 0 1 2 3 4 N t'3 Year

1996 1997 1998 1999 2000 2001

Fig. 3. The number of flower heads/plant in large and small Fig. 4. Locoweed mortality in large and small plants, and at 3 locations: plants that were either clipped or not clipped at New Mexico, New Mexico, Colorado and Utah. Colorado and Utah over time. P < 0.05 indicate differences between the clipped and non clipped plants in each size class.

Unclipped large plants declined gradually locations, but they were slight and proba- 50% of locoweed plants died following over the length of the study. The number bly not large enough to influence the sur- the winter and spring drought in 1996, and of heads on small clipped plants at Utah vival of the plants. Leaf length of large additional mortality occurred during the declined in 1997 and 1998, and the num- plants averaged 15 cm, and was 11.4 cm moderate droughts during the spring in ber of heads in unclipped plants declined for small plants; flowering head height 1998 and 1999 (Fig. 1). At Colorado, 84% in 2000. averaged 19.6 cm over plant sizes and of small plants and 90% of large plants There was a strong year effect (P < locations. were dead after 4 years (Fig. 4), and the 0.002) for both size plants and all loca- remaining plants died during the severe tions. The number of heads in both clipped Locoweed Mortality drought in 2000 (Fig. 1). Survival rate at and unclipped large plants at New Mexico Clipping had no effect on mortality of Utah was higher (P < 0.02) with 30 % of declined from 20 in 1995 to 5 in 1996. locoweed plants at any of the locations (P the plants remaining after 4 years (Fig. 4). The number of heads of both clipped and = 0.93). However, mortality over time was Although the mortality rate was not as unclipped small plants at Colorado greater on large plants than on small plants great at Utah, the vigor measurements time. increased over (P = 0.004) (Fig. 4). Mortality rate was responded to the severe drought through- (P There were some differences < 0.05) greater at New Mexico (P < 0.01); over out 1999 and 2000 at Utah (Fig. 1). in leaf length and head height at some Mortality was negatively correlated with

JOURNAL OF RANGE MANAGEMENT 55(4) July 397 Table 1. Swainsonine concentration in the same white locoweed plants clipped annually at New Mexico, Colorado and Utah (± standard error of the mean).

Location 1995 1996 1997 1998 1999 Mean Year Diff.1 ------(% of dry weight) ------NM 0.11±0.01 0.11±0.02 0.07± 0.01 0.097 0.07 Co 0.04±0.01 0.09±0.02 0.06±0.02 0.070 0.06 UT 0.07 ±0.01 0.04±0.01 0.03±0.01 0.05 ± 0.01 0.048 0.05 IMean difference within the same plant between years. total water year precipitation at New (reviewed by Laycock 1975). Extensive drought occurs and the population dies Mexico (r = -0.84, p = 0.008) and Utah research on rangeland grasses showed that back (A. lentiginosus, Barnes 1913; A. (r = -0.75, p = 0.08). Correlation of clipping during anthesis was much more lentiginosus var. wahweapensis, Ralphs locoweed mortality with summer precipi- damaging than clipping during vegetative and Bagley 1988; A. pubentissimus, James tation at Colorado was not significant ( r = growth (reviewed by Caldwell 1984). et al. 1968). -0.42, P = 0.29). However, the correla- Defoliation reduces photosynthetic capaci- Other factors besides drought may affect tions were not totally accurate. High mor- ty and root mass and retards the plants locoweed populations. The root-boring tality in earlier years left few of the origi- ability to extract water and nutrients from weevil Cleonidius trivittatus (Coleoptera: nal marked plants in the experimental pop- the soil. Neighboring plants use these Curculionidae) prefer woolly locoweed ulation, thus the mortality rate declined in resources, further reducing their availabili- (Astragalus mollissimus Torr.) and will the last year or 2. This reduced the ty. Thus, competition is the indirect factor kill the plant when 2 or more larvae occu- strength of the correlations and underesti- causing reduction in subsequent vigor of py its root (Pomerinke et al. 1995). This mates the full impact of drought. clipped plants (Mueggler 1972). However, insect and drought are major contributors white locoweed has a deep tap root that to the crash of woolly locoweed popula- Swainsonine extends below the rooting zone of grasses, tions. The insect is found to a lesser extent Clipping had no effect on subsequent and thus may not be subjected to direct in white locoweed, but its influence on swainsonine concentration in locoweed competition for soil moisture. White vigor and mortality is not known. not to be plants (P = 0.62). This is contrary to the locoweed does appear adversely Plant Defense Theory which suggests affected by annual defoliation in the flower defoliation stimulates secondary defense stage or from competition from grasses. Conclusions compounds in some plants (Tallamy and Precipitation patterns apparently influenced white locoweed populations. Raupp 1991). There was also no differ- Clipping for 4 years did not consistently Mortality occurred during the mild to ence in swainsonine concentration reduce vigor or increase mortality of white between large and small plants (P > 0.05). severe drought periods at each of the loca- locoweed plants. Therefore, intensive tions. We found significant negative cone- There were differences among years (P < grazing pressure, that would force con- lations between locoweed mortality and 0.07), but no recognizable trends (Table sumption of white locoweed, would not precipitation at New Mexico and Utah and 1). There were differences among loca- likely reduce its population in the short weaker correlations with summer precipi- tions (P = 0.0001) with New Mexico hav- term. There was a die-off of locoweed tation at Colorado. Our methodology of ing the highest concentration of swainso- plants at all 3 locations that was associated calculating mortality on marked plants nine (0.097 % of dry wt), and Utah the with drought. The larger, and presumably without replacement may have limited the lowest (0.048%). Swainsonine concentra- older plants, had a slightly higher mortali- strength of the correlations. Furthermore, tion in a plant was not the same each year. ty rate, but a majority of small plants also mortality of individual plants is deter- Variability within the same plant over died over the study. Clipping had no effect mined by micro environmental conditions, years was greater than the difference on subsequent toxicity of white locoweed. between locations (Table 1). which are difficult to estimate by seasonal weather measurements taken some distance form the study site. Literature Cited Discussion Although white locoweed appears to be more persistent and longer-lived than many of the Astragalus locoweeds, our Barnes, W.C. 1913. Western grazing grounds Clipping white locoweed in the flower study showed that its populations still and forest ranges. The Breeder's Gazette, stage for up to 4 successive years had only cycle. Marsh (1909) also observed that Chicago. 390 pp. minor effects on vigor of the plants and white locoweed was abundant in wet Caldwell, M. M. 1984. Plant requirements for prudent grazing. 117-152, in: Develop- did not affect mortality or swainsonine years, but nearly disappeared in dry years. pp. concentration. In tall larkspur (Delphinium ing Strategies for Rangeland Management. Many of the semi-desert Astragalus Westview Press, Bolder Colo. barbeyi Huth), clipping at the bud elonga- locoweeds experience extreme population Fox, G.A. 1993. Failure-time analysis: emer- tion stage greatly reduced vigor and alka- cycles that are driven by precipitation pat- gence, flowering, survivorship, and other loid pools for 2 subsequent years (Ralphs terns (Welsh 1989). They germinate fol- waiting times. pp. 253-289, In: S.M. and Gardner 2001). Clipping other moun- lowing autumn rains, remain green over Schemer and J. Gurevitch (Eds) Design and tain forbs during the flowering stage was winter, flower in spring, and may continue Analysis of Ecological Experiments. Chapman & Hall, New York. most detrimental to subsequent vigor to grow for 1 or 2 years until the next

398 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Gardner, D.R., R.J. Molyneux, and M.H. Palmer, W.C. 1965. Meteorological Drought Ralphs, M.H. and E.H. Cronin. 1987. Ralphs. 2001. Analysis of swainsonine in Res. Pap. No 45. U.S. Weather Bureau, Locoweed seed in soil: density, longevity, locoweeds (Oxytropis spp.) Extraction meth- NOAA Library and Info., Washington D.C. germination and viability. Weed Sci. ods and detection using liquid chromatogra- 58 pp. 35:792-795. phy and tandem mass spectrometry. J. Agr. Payne, G.F. 1957. Ecology and life history of Ralphs, M.H. and D.R. Gardner. 2001. Food Chem. (in press). the poisonous plant, white locoweed Influence of defoliation on toxic alkaloid (Oxytropis Nutt.). PhD Diss. Texas James, L.F., K.L. Bennett, K.G. Parker, R.F. sericea concentration and alkaloid pools in tall lark- A&M Univ., College Station, Tex. Keeler, W. Binns, and B. Lindsay. 1968. spur. J. Chem. Ecol. (In press). Pfister, J.A., B.L. Stegelmeier, C.D. Cheney, Loco plant poisoning in sheep. J. Range Tallamy, D.W. and M.J. Raupp. 1991. Manage. 21:360-365. L.F. James, and R.J. Molyneux. 1996. Phytochemical Induction by Herbivores. Laycock, W.A. 1975. Alkaloid content of Operant analysis of chronic locoweed intoxi- John Wiley & Sons, New York. duncecap larkspur after two years of clip- cation in sheep. J. Anim. Sci. 74:2622-2632. Welsh, S.L. 1989. Astragalus L. and Oxytropis ping. J. Range Manage. 28:157-259. Pomerinke, M.A., D.C. Thompson and D.L. Marsh, C.D. 1909. The locoweed disease of Clason. 1995. Bionomics of Cleonidius DC.: definitions, distributions, and ecological the plains. USDA Anim. Ind. Bull. 112. trivittatus (Coleoptera: Curculionidae): parameters. pp. 3-13, In: L.F. James, A.D. Molyneux, R.J., L.F. James, K.E. Panter, native biological control of purple locoweed. Elbein, R.J. Molyneux, C.D. Warren (eds.), and M.H. Ralphs. 1991. Analysis and distri- Envir. Entom. 24:1697-1702. Swainsonine and Related Glycosidase bution of swainsonine and related polyhy- Ralphs, M.H. and V.L. Bagley. 1988. Inhibitors. Iowa State Univ. Press. Ames Ida. droxy indolizidine alkaloids by thin layer Population cycles of Wahweap milkvetch on chromatography. Phytochem. Analys. the Henry Mountains and seed reserve in the soil. Great Basin Natur. 48:541-547. 2:125-129. Mueggler, W.F. 1972. Influence of competition on the response of bluebunch wheatgrass to clipping. J. Range Manage. 25:88-92.

JOURNAL OF RANGE MANAGEMENT 55(4) July 399 J. Range Manage. 55: 400-405 July 2002 Long-term impacts of livestock grazing on Chihuahuan Desert rangelands

JOSEPH M. NAVARRO, DEE GALT, JERRY HOLECHEK, JIM McCORMICK, AND FRANCISCO MOLINAR

Joseph Navarro is a Rangeland Management Specialist with the BLM in the Roswell Field Office 2909 W. 2nd St. Roswell, N.M. 88201 and a Graduate Research Assistant, Department of Animal and Range Sciences, New Mexico State University; Dee Galt is a private range consultant, Las Cruces, NM 88005; Jerry Holechek is a range science professor, Dept. of Animal and Range Sciences, New Mexico State University, Las Cruces, NM 88003; Jim McCormick is assistant field manager, Las Cruces Field Office, USDI, Bureau of Land Management, Las Cruces, NM 88005; and Francisco Molinar is a range animal ecology professor, School of Veterinary Medicine and Animal Science, Autonomous University of Juarez, Chihuahua, Mexico.

Abstract Resumen

Rangeland ecological condition was monitored over a 48 year La condicion ecologica de los pastizales del desierto chi- period on 41 sites on Bureau of Land Management rangelands huahuense se determine en un estudio que abarco 48 anos y scattered across 6 counties in southwestern New Mexico. All sites que se llevo a cabo en 41 sitios dentro de 6 condados del suroeste were grazed by livestock during the study period. Sampling de Nuevo Mexico. El estudio se realize en terrenos administra- occurred in 1952, 1962, 1982, 1992, 1997, 1998, and 1999. A mod- dos por la Agencia para el manejo de Tierras de los Estados ified Parker 3 step method in conjunction with Dyksterhuis Unidos. Los sitios estuvieron bajo pastoreo durante todos los quantitative climax procedures were used to determine range- anos de duracion del trabajo. Los muestreos se realizaron en land ecological condition. At the end of the 48 year study period 1952, 1962, 1982, 1992, 1997, 1998, y 1999. Se utilize una modifi- (1952-1999), the average rangeland ecological condition score cacion del sistema de 3 pasos de Parker y los procedimientos de across study sites was the same (P > 0.05) as the beginning of the Dyksterhuis para cuantificar el climax. La finalidad fue determi- study (39% versus 41 % remaining climax vegetation, respective- nar la condicion ecologica de los pastizales. Al final de los 48 anos ly). Major changes (P > 0.05) in rangeland condition occurred del estudio (1952- 1999) se determine que el promedio de la within the study period due to annual fluctuations in precipita- condicion ecologica, al considerar todos los sitios, fue la misma tion. Ecological condition scores increased in the 1980s and early (P> 0.05) que la existente al inicio de los trabajos en 1952 (39% 1990s due to above average precipitation. However, drought in contra 41 % del remanente de la vegetacion del climax, respecti- the early to mid 1950's and again in the mid to late 1990's caused vamente). Los cambios mas notorios en la condicion ecologica rangeland condition scores to decline. At the end of the study durante los anos del estudio (P> 0.05) se debieron a las fluctua- (1997-1999), 38% of the sites were in late seral ecological condiciones de la precipitacion pluvial. Los rangos de la condicion tion, compared to an average of 25% in the 1952 to 1982 period. ecologica aumentaron en la decada de los anos ochenta y al inicio The amount of rangeland in late seral ecological condition de los noventa debido al incremento de la precipitacion. Sin increased while the amount of rangeland in mid seral and early embargo, la sequia al inicio y durante la primera mitad de los seral condition decreased in the 1990s compared to the 1952-1962 anos cincuenta y nuevamente durante la segunda mitad y hasta el period. The average percent cover of black grama (Bouteloua eri- fin de los noventa, provoco que disminuyeran los porcentajes de opoda Torr.) and tobosa (Hilaria mutica Buckley), the primary la condicion ecologica. En la parte final del trabajo (1997-1999), forage grasses in the Chihuahuan Desert, were the same (P > el 38% de los sitios se encontraron bajo condicion ecologica de 0.05) in 1952 and 1999. Over the 48 year study period, the aver- buena a excelente, en comparacion con el 25 % bajo esas condi- age cover of shrubs including honey mesquite (Prosopis glandu- ciones durante el periodo de 1952 a 1962. Al comparar los resul- (P losa Torr.) showed no change > 0.05). However major increas- tados de los anos noventa con el periodo de 1952 a 1982, la canti- es in honey mesquite basal cover occurred on 1 site and creosote- dad de pastizales en estado ecologico bueno a excelente se incre- bush (Larria tridentata [Pursh] Nutt.) increased on another. mento y la cantidad en estado pobre a regular disminuyo. Los Grazing intensity was evaluated during the last 3 years of study promedios de cobertura de los zacates mas representativos del (1997, 1998, 1999). Overall grazing use of forage across sites and desierto chihuahuense como navajita negro (Bouteloa eripoda years averaged 34% or conservative. Our research shows con- Torr.) y Toboso (Hilaria mutica Buckley) fueron los mismos (P> trolled livestock grazing is sustainable on Chihuahuan Desert 0.05) en 1952 y en 1999. Durante los 48 anos del estudio, el rangelands receiving from 26-35 cm annual precipitation. promedio de cobertura de plantas lenosas como el mezquite (Prosopis glandulosa Torr.) no mostro cambios mayores (P > 0.05). Sin embargo, se detectaron incrementos importantes en la Key Words: Rangeland condition, succession, drought, grazing cobertura basal de Prosopis en uno de los sitios y de la gober- management, plant ecology nadora (Larrea tridentata [Pursh] Nutt.) en otro. Por otra parte, se evaluo la intensidad de pastoreo durante los ultimos 3 anos (1997, 1998, 1999). This project is #14226910A970011, Monitoring Range Ecological Condition & La utilizacion promedio del forraje al con- Trend Studies in the Northern Chihuahuan Desert Grasslands located in tabilizar todos los sitios y los anos fue conservadora o del 34% Southwestern New Mexico. Funding for this research was provided by the USDI- del total. Nuestra investigacion indica que el pastoreo controlado Bureau of Land Management and the New Mexico Agricultural Experiment es viable en pastizales del desierto chihuahuense, aunque solo se Station. reciban niveles de precipitacion pluvial de 260 a 350 mm anuales. Manuscript accepted 13 Sept. 2001.

400 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Published quantitative information is and 2000 using procedures of Holechek diversity and herbage yields due to differ- lacking on the general trend in ecological and Galt (2000). ences in elevation, climate, soils, and condition for different types of public topography. Plant communities include rangelands in New Mexico and other desert grasslands, mixed desert shrub- western states. Information is available on Methods lands, desert shrublands, mountain brush, range condition and trend for many indi- and pinon-juniper/oak woodland. Our vidual public grazing allotments through- Study Area study sites were confined primarily to the out the western states. Various govern- In 1952 the Bureau of Land Manage- desert grasslands with a few desert shrub- ment reports have indicated range condi- ment initiated a study to examine long land sites. The 7 major range sites based tion on public lands is improving but these term trend in range vegetation in the on USDA-Natural Resources Conser- reports are somewhat vague and subjective Chihuahuan Desert of southwestern New vation Service guides are gravelly, regarding how range condition was mea- Mexico. This study involved use of the igneous hills, sandy, loamy, gravelly loam, sured and classified (United States malpais, and clayey. Parker 3 step method (Parker 1951). Department of Interior 1999). Studies are Parker transects on the various sites were needed that quantitatively characterize evaluated in 1952, 1962, 1982, 1992, long-term changes in rangeland vegetation 1997, 1998, and 1999. Sampling Procedures in different range biomes in response to The majority of the study area lies with- climatic conditions and livestock grazing in the northern end of the Chihuahuan Forty-one different study sites were management. This information is also Desert (106° 52' 30" W, 32°32'30N). The used in this analysis. Generally the goals needed because of increased social pres- study area is bounded by Texas and in site selection were to select accessible sure to remove livestock from public Mexico to the south, Arizona to the west, flat grassland areas with uniform soils in rangelands and emphasis on non-con- and is traversed by the Rio Grande on the medium size to large pastures 1.4 km from sumptive rangeland uses (Wuerthner 1990, eastern side. This region is administered water. Site selection was based on ease of Donahue 1999). by the USDI-Bureau of Land Management access, distance from water, soil type, veg- Recent studies in the Chihuahuan Desert Mimbres and Caballo Resource Areas in etation type, size of pasture, and terrain. of southern New Mexico have shown that the Las Cruces District. The Mimbres Ecotones where soil types and vegetation controlled livestock grazing can be biolog- Resource Area includes Hidalgo, Grant, were in transition were avoided. All study ically sustainable, economically cost Luna, and Dona Ana counties. The sites were located on uniform soils, flat and with most effective, compatible Caballo Resource Area includes Otero and terrain and in uniform, discreet plant com- wildlife species (Holechek et al. 1994, Sierra counties. There are 2.7 million ha of munities. One site had 1 transect, 26 sites Nelson et al. 1997, Winder et al. 2000). BLM administered public land intermin- had 2 transects, 10 sites had 3 transects However evaluations are needed over gled with 5.3 million ha of private, State and 4 sites had 5 transects. All sites were to the effects of broad areas determine trust, Native American, and other Federal within 1.2 km to 1.6 km of permanent livestock grazing as it has actually lands in this 6 county area. water, and were grazed during the period occurred. There is considerable concern The area is characterized by an arid to of the study. The 41 sites we studied are among many conservation groups that semiarid continental climate, with mild well scattered over 6 counties in south- ranchers may not be using sustainable winters and hot summers. Most of the western New Mexico. grazing practices (Wuerthner 1990). study area receives from 26-35 cm annual A modification of the Parker 3 step The 1950s drought had great impact on precipitation. Annual fluctuations range method was used to evaluate trends in Chihuahuan Desert rangelands (Buffing- from a low of 5 cm to a high of over 50 vegetation cover on each site (Parker 1965). However detailed ton and Herbel cm. Over half of the annual precipitation 1951). Permanent 30.5-m line transects this scientific evaluation of the impacts of arrives as rainfall during July, August, and were accurately located from existing steel on event Chihuahuan Desert vegetation September. The average frost free season rebar. Readings were taken and recorded been reported only from the USDA has exceeds 200 days and extends from April at 30.5 cm intervals using a loop 1.91-cm Jornada Experimental Range in southcen- to November. The average annual temper- in diameter along a tape stretched between tral New Mexico (Buffington and Herbel ature in the area is about 20° C. During the the 2 permanent rebar stakes. Data record- 1972, 1965, Herbel et al. Herbel and summer months, daytime temperatures ed along the transect included presence of Gibbens 1996). Drought and post drought may exceed 38° C. Through the year, a all herbaceous plants and shrubs rooted vegetation responses on the Jornada daily range of 17° C or more is common. within the loop. There were no recordings Experimental Range however may not be Both the rainfall distribution and tempera- of multiple plants within the loop. completely representative of the ture regime favor warm season perennial Evaluations were conducted during the Chihuahuan Desert in southern New plants. dormant period of the year (late fall Mexico. Typical landform includes rugged and through early spring). Transect data were The objective of our study was to evalu- steep fault-block mountain ranges, broad collected for the years of 1952-54, ate long term (48 years) trend in vegeta- basins, and volcanic uplands. Elevations 1962-64, 1982, 1992, 1997, 1998, and tion composition using 41 locatable Parker on BLM-administered public land range 1999. In all years of study, data were col- sites in Chihuahuan south- the Desert of from a low of approximately 1,067 m in lected by professionals with range science western New Mexico. Complete data is the southern Mesilla Valley to a high of training. Dr. Dee Galt, a private range available for 7 years of sampling (1952, over 2,438 m in some of the mountain consultant with 30 years experience in the 1962, 1982, 1992, 1997, 1998, 1999) on ranges. Average elevation is about 1,372 USDA-Natural Resources Conservation 22 these sites. A secondary objective of m. Our study sites are primarily in the Service, supervised all data collection in was to evaluate grazing intensity on the broad basins. the 1992 to 1999 period. study sites in late spring of 1998, 1999, The vegetation varies greatly in its Plant cover in our study is more of an

JOURNAL OF RANGE MANAGEMENT 55(4) July 401 index than an absolute value. Loop sam- 1950s were characterized by extreme In 1994 and 1995 growing season precipi- pling tends to overestimate cover for small drought receiving 73% of the 100 year tation was well below average. Annual plants (Hutchings and Holmgren 1959). average (National Oceanic and Atmos- precipitation averaged 29.74 cm for the However, as a cover index, loop sampling pheric Administration 1999). In contrast, 1992-1999 period compared to the long is reliable for detecting vegetation changes precipitation in the 1980s was 12% above term average of 30.07 cm. From a histori- over time (Smith 1962). the 100 year average. cal standpoint; the 1980s and 1990s were Range ecological condition scores were Climatic conditions in southwestern the wettest decades in New Mexico during calculated from current USDA Natural New Mexico in the 1992 to 1999 period the past 100 years while the 1950s were Resources Conservation Service New were near normal (Table 2). Four years the driest. Mexico range site guides using the were wet (1992, 1993, 1997, 1999), one Rangeland ecological condition scores Dyksterhuis (1949) procedure. Cover was year was dry (1995), and 3 years were for individual sites showed considerable used instead of biomass in calculating near average (1994, 1996, 1998) (National fluctuation in the 1997-1999 period scores. Cover and biomass indices gave Oceanic and Atmospheric Administration (Table 1). We attribute this to both sam- similar ecological condition scores across 1999). Above average growing season pling disparities and climatic variability. several sites on the Chihuahuan Desert precipitation (May through September) Because fewer plants were encountered Rangeland Research Center in southcen- occurred in 1992, 1996, 1997, and 1999. along the transects in low ecological con- tral New Mexico (Molinar 1999). During late spring (May, June) in the Table 1. Ecological condition scores (USDA-National Resources Conservation Service method) on last 3 years of study, grazing intensity was 41 sites on Bureau of Land Management rangelands in the Chihuahuan Desert of New Mexico evaluated on all study sites using proce- over a 48 year period (1952-1999). dures of Holechek and Galt (2000). This approach involves assessment of forage Rangesite County 19521 1997 1998 1999 stubble heights, remaining seed stalks, and (%) ------percentage of plants grazed. Clayey Grant 47 Comparisons among years for ecologi- Loamy Luna 24 9 cal condition scores and plant cover were Sandy Dona Ana 14 Shallow Sandy Dona Ana 26 6 analyzed by analysis of variance using a Clayey Luna 56 completely randomized design with study Clayey Grant 65 sites as replicates. The least significant Gravelly Sierra 35 difference mean separation procedure was Clayey Hidalgo 38 8 used to compare means in time if analysis Clayey Hidalgo 57 of variance indicated a significant differ- Clayey Hidalgo 44 Loamy Hidalgo 48 ence (Steel and Torrie 1980). Loamy Hidalgo 49 Sandy Hidalgo 30 Sandy Hidalgo 11 8 8 8 8 Results and Discussion Clayey Dona Ana 51 Draw Luna 63 Loamy Grant 15 Average rangeland ecological condition Sandy Luna 30 score was not different (P > 0.05) between Clayey Luna 56 1952 and 1999 (n = 22) (Table 1). For the Hills Hidalgo 32 22 sites supporting complete data sets, the Loamy Hidalgo 42 Gravelly Dona Ana 32 1950s drought caused a drop in range con- Gravelly Luna 25 dition, but steady improvement occurred Shallow Sandy Sierra --- from 1962 to 1992. Shallow Sandy Otero --- Possibly the most reliable assessment of Loamy Sierra --- change in ecological condition through Breaks Sierra ------time on our study areas is the comparison Hills Sierra Gravelly Sierra --- of the average of the first 3 years sampled Clayey Hidalgo --- in our study (1952, 1962, 1982) with the Hills Hidalgo --- last 3 years (1997, 1998, 1999) across the Gravelly Luna --- 22 sites with complete data (Table 1). On Hills Hidalgo --- this basis the condition score averaged Hills Hidalgo ------33.7% for the 1952-1982 period com- Gravelly Luna Shallow Sandy Dona Ana --- pared to 37.0% for the 1997-1999 period, Gravelly Sierra --- with no significant (P = 0.36) differences. Sandy Sierra --- Therefore, we conclude that no change has Loamy Otero --- occurred in rangeland ecological condition Shallow Sandy Hidalgo --- between early years and the final years of Sandy Dona Ana --- our study. Average x 34C 39bc (All sites) Shifting climatic conditions in south- Average - 39ab western New Mexico explain the major ii:-(First 22 sites) fluctuations in rangeland ecological condi- a,b,CMeans within rows with different superscripts differ at P < 0.05. tion from 1952 to 1999 (Table 2). The

402 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Table 2. Annual precipitation (PPT-cm) for 10 year intervals in southwestern new Mexico for the cover during the 1982-1992 period, and 1950-1999 period (National Oceanic and. Atmospheric Administration 1999).1 then sharply declined (P < 0.05) in the 1992-1999 period (n = 22) (Table 3). We Year PPT Year Year Year PPT Year PPT speculate the dry summers in 1994 and (cm) (cm) (cm) (cm) (cm) 1995 were more damaging to perennial grasses than in other decades because the 1950 14.05 1960 25.30 1970 19.41 1980 23.80 1990 37.87 long run of wet years in the 1980s and 1951 15.80 1961 28.04 1971 23.60 1981 30.86 1991 44.25 1990s may have reduced plant resistance 1952 14.99 1962 31.72 1972 37.59 1982 26.83 1992 37.72 1953 9.70 1963 23.88 1973 23.93 1983 35.53 1993 30.53 to drought. Well above average precipita- 1954 15.57 1964 21.36 1974 33.12 1984 42.95 1994 29.67 tion in the 1990-1992 period elevated 1955 25.60 1965 28.80 1975 30.10 1985 40.28 1995 21.49 perennial grass basal cover above what 1956 12.12 1966 24.66 1976 25.30 1986 44.37 1996 27.79 plant root systems could support in normal 1957 31.98 1967 28.93 1977 27.13 1987 30.76 1997 32.74 years. Plant rooting depths and root to 1958 41.43 1968 27.56 1978 36.25 1988 36.47 1998 28.22 shoot ratios may have been altered by the 1959 23.98 1969 26.64 1979 29.36 1989 22.33 1999 28.37 x 20.52 26.70 28.58 33.63 31.88 exceptionally wet conditions. This could have altered the plant's ability to cope Locations monitored include Afton, Alamogordo, Animas, Antelope Wells, Buckhorn, Cliff, Columbus, Deming, Glenwood, Hachita, Hillsboro, Jornada, Las Cruces, Lordsburg, Orogrande, Redrock, Truth or Consequences, White with the return to more normal precipita- Sands, White Signal, and Whitewater. tion conditions. The severity of consecutive summer droughts in 1994 and 1995 may have also been accentuated by the dition sites, more annual variability period was compared to the 1997-1999 spacing of rainfall events. Storms that did occurred in vegetation composition. Some period. Honey mesquite was invading 1 occur generally were well distributed sites had major fluctuations in broom site while creosotebush (Larrea tridentata through July and August, and seldom snakeweed (Gutierrizea sarothrae Pursh.) [Pursh] Nutt.) was invading another. involved more than 0.76 cm precipitation. populations during 1997-1999 period. Black grama and other perennial grasses This moisture was quickly lost to evapora- Broom snakeweed is a short lived, poiso- had a sharp increase (P < 0.05) in basal nous, perennial half shrub whose populations can fluctuate greatly from year to Table 3. Variation in mean basal cover (%) of primary plant species found on 41 different sites on year with weather (Pieper and McDaniel Bureau of Land Management rangelands in the Chihuahuan Desert of New Mexico over a 48 1989). Wet winters and springs promote year period (1952-1999). broom snakeweed establishment. Our provide only limited insight First 22 sites data Year into the various rangeland plant successional models (Holechek et al. 2001), Plant category 1952' 1962 1982 1992 1997 1998 1999 because large time gaps occurred between ------(%) ------2.3a 4.9a 10.2a 3.4a 3.3a Tobosa 4.Oa 3.Oa some of the assessments on the sites. In 1.Sa 1.la 4.7a 1.4a 1.Sa addition, grazing management changes on Black Grama 1.6" 1.2" Burrograss 2.8 1.5 2.3 3.4 0.6 0.9 0.9 the sites were not well documented across Aristida spp. 0.2 0.0 0.0 1.5 0.2 0.3 0.2 3.4a 2.3a 3.3a 1.6a 2.4a time. We do note that there was a tenden- Other perennial grasses 6.2b 1.6' cy for sites in early seral stages to show Annual grasses 0.0 0.0 0.4 0.2 Trace 1.2 1.4 11.9a 7.2a 9.7a improvement during the 1962-1999 peri- Total grasses 7.2" 12.5" 26.2b 8.4" 1.9a 2.8a od (Table 1). Nine of the 13 sites in an Total forbs 2.5" 6.6' 7.9b 2.2' 0.6" early seral stage improved to mid seral or Honey mesquite 0.2 0.1 0.3 0.6 Trace 0.0 0.2 Broom snakeweed 0.4 0.4 1.7 2.0 Trace 0.2 0.2 This provides support late seral condition. Other shrubs 0.3 0.1 0.8 1.2 0.1 Trace 0.1 for the theories of Dyksterhuis (1949) that Total shrubs 0.9 0.6 2.8 3.8 0.2 0.3 0.5 14.4a 10.9a some degraded rangelands can improve Total basal cover 15.3" 17.2" 37.9b 10.2" 10.8" through natural plant succession under All 41 sites favorable climatic conditions and con- Year trolled livestock grazing. Plant category 1962 1982 1992 1997 1998 1999 Black grama (Bouteloua eriopoda ------(%) ------Torr.), the primary decreaser forage in 3.4a 2.1a 1.9a 2.la Tobosa 1.7" 6.8b differ- 2.8a 2.1a 2.9a southern New Mexico, showed no Black Grama 2.8" 9.0b 1.9" ence (P > 0.05) in basal cover between Burrograss 1.6 2.1 2.5 0.7 0.8 0.9 1952 and 1999 (n = 22) (Table 3). Tobosa Aristida Trace 0.3 2.2 0.4 0.4 0.4 spp. 2.6a 2.0a 1.8a 2.5a (Hilaria mutica Buckley) and total peren- Other perennial grasses 2.8" 6.9b (P Annual grasses Trace 0.2 0.1 Trace 0.7 0.9 nial grass cover did not differ > 0.05) 11.4a 7.la 7.7a 9.7a Total grasses 8.9" 27.6b between 1952 and 1999 (n = 22). No 4.6a 2.5a 1.8a 0.6a (P Total forbs 5.2b 1.6" increase in shrub basal cover > 0.05) Honey mesquite Trace 0.3 0.3 Trace Trace 0.2 occurred over the 4 year period (n = 22) Broom snakeweed 0.4 1.6 1.6 Trace 0.2 0.3 (Table 3). Honey mesquite (Prosopis Other shrubs Trace 1.2 1.3 0.3 0.4 0.7 glandulosa Torn.) cover was the same (P> Total shrubs 0.4 3.1 3.2 0.3 0.4 0.7 13.9a 9.2a 9.7a 0.05) in 1952 and 1999 (n = 22). Total basal cover 17.Oa 36.0b 11.Oa abMeans However, 2 sites had major increases in within rows with different superscripts differ at P < 0.05. shrub basal cover when the 1952-1962

JOURNAL OF RANGE MANAGEMENT 55(4) July 403 provides more combined benefits to soci- Table 4. Percentages sites (41) of study in southwestern New Mexico classified as either severely ety in terms soil stability, watershed grazed, heavily grazed, moderately grazed, conservatively grazed, lightly grazed or ungrazed of by livestock in 1998, 1999, and 2000. protection, wildlife habitat, esthetic appearance, biodiversity, and forage for Grazing intensity livestock than mid seral or early seral condition rangelands. Some Chihuahuan Year Severe Heavy Moderate Desert wildlife species such as the endangered aplomado falcon require large areas 1998 0 7 38 0 7 of rangeland in near climax condition 1999 2 18 41 7 (Hector 1987). 2000 0 10 27 2 From a multiple use standpoint, it would x 1 12 35 5 be desirable to further increase the amount of Chihuahuan Desert rangeland in late tion. Generally about 2.54 cm of rain, con- None of our study sites had any type of seral or climax condition. Conservative centrated within a 1 week period, is need- shrub control applied during the 48 year stocking involving about 30-35% use of ed to initiate growth of desert grasses such study period. Additional water develop- key forage species is the surest way to as black grama and tobosa (Canfield ment had occurred within several study upgrade ecological condition and grazing 1939). This did not occur in 1994 or 1995 pastures but not near our study sites. Data capacity of arid and semi-arid rangelands in most parts of southwestern New in New Mexico Agricultural Experiment (Paulsen and Ares 1962, Valentine 1970, Mexico. Research on the Chihuahuan Station reports indicated there has been a Aldon and Garcia 1971, Martin and Cable Desert Rangeland Research Center north doubling of water developments on south- 1974, Hughes 1982, Holechek et al. 1994). of Las Cruces indicated that forage pro- ern New Mexico rangelands over the past Considerable research shows conservative duction in 1994 and 1995 was only 18% 40 years (Torell et al. 2000). Only a slight use will provide higher financial returns of the 32 year average (Holechek et a1. increase in new fencing has occurred since with lower risk to ranchers in arid and 2000). 1960. semi-arid areas than heavier use levels Based on tree ring data, extended Total livestock numbers for New (Johnson 1953, Klipple and Costello 1960, droughts lasting 5-12 years have occurred Mexico counties are reported annually by Houston and Woodward 1996, Pearson about every 50 to 60 years in the south- the New Mexico Agricultural Statistics 1973, Martin 1975, Quigley et al. 1984, western United States (Meko et al. 1995). Service (2000). In the 6 sampled counties Holechek 1992, Winder et al. 2000). The last extended drought occurred in the total beef cattle numbers steadily Shrub invasion was actively occurring on 1941 to 1956 period. Twelve of the 15 increased from 193,310 during 1960 to 2 of our 41 sites. Burning and herbicides years had below average precipitation. 251,350 head in 1980. However, cattle can be effective in controlling brush inva- Some climatologists believe the south- numbers since 1980 have declined to sions in the Chihuahuan Desert (Herbel western United States is again entering an around 220,000 head in 2000. and Gould 1995). extended dry period (Fleck 1998). Finally, we believe our study shows live- Grazing intensity across the 41 sites was stock grazing has been sustainable on most considered to be conservative (34% use of Management Implications Chihuahuan Desert rangelands in south- forage) (Table 4). Seven sites showed western New Mexico over the past 48 repeated heavy grazing during the 3 years Our study showed that 38% of Bureau years because range condition scores grazing intensity was measured. Nineteen of Land Management rangeland sites we showed no definitive change between 1952 of the 41 study sites were lightly or con- monitored in southwestern New Mexico and 1999. A higher proportion of range- servatively stocked. were in late seral or climax ecological land was in a late seral condition in 1999. Our interviews with permittees using the condition in the 1997-1999 period com- Contrary to Donahue's (1999) viewpoint, study sites indicated ranchers are better pared to 25% in the 1952-1982 period we find that conservative livestock grazing informed on the benefits of conservative (Table 5). We believe our study sites are is sustainable on arid lands receiving less stocking than they were 8 years ago fairly representative of BLM rangelands in than 26-35 cm annual precipitation. (1992). About 1 half of the ranchers inter- southwestern New Mexico. Several stud- viewed were consciously attempting to ies reviewed by Holechek et al. (2001) implement conservative stocking on their show Literature Cited grazing allotments. that late seral condition rangeland

Table 5. Percentages of Bureau of Land Management study sites (41) in southwestern New Mexico Aldon, E. F. and G. Garcia. 1971. Stocking in different ecological condition categories in 1952, 1962, 1982, 1992, and 1999. rangelands on the Rio Puerco in New Mexico. J. Range Manage. 24:344-345. Buffington, L. C. and C. H. Herbel. 1965. Excellent Good Fair Poor Vegetation changes on semidesert grassland Year (climax) (late seral) (mid seral) (early seral) range from 1858 to 1963. Ecol. Monogr., ------(%)------35:139-164. 19521 0 26 52 22 Canfield, R. H. 1939. The effect of intensity 1962 0 24 44 32 and frequency of clipping on density and 1982 0 24 50 26 yield of black grama and tobosa grass. U.S. 1992 5 32 48 15 Dept. Agr. Tech. Bull. 681. 1997 0 32 45 23 Dyksterhuis, E. 1949. Condition and man- 1998 0 30 46 24 J. agement of rangeland based on quantitative 1999 2 51 35 12 ecology. J. Range Manage. 2:104-115. 1 Only 22 sites were evaluated in 1952.

404 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Donahue, D. L. 1999. The western range revis- Hutchings, S. and R. C. Holmgren. 1959. Paulsen, H. A. and F. N. Ares.1962. Grazing ited: Removing livestock from public lands Interpretation of loop-frequency data as a values and management of black grama and to conserve biodiversity. Univ. of Oklahoma measure of plant cover. Ecol. 4:668-677. tobosa grasslands and associated shrub Press, Norman, Okla. Johnson, W. M. 1953. Effect of grazing inten- ranges of the southwest. U.S. Dept. Agr. Fleck, J. 1998. Studying droughts legacy. sity upon vegetation and cattle gains on pon- Tech. Bull. 1270. Albuquerque J. Sunday, February 1, 1998. derosa pine-bunchgrass ranges of the front Pearson, H. A. 1973. Calculating grazing Hector, D.P. 1987. The decline of the aploma- range of Colorado. U.S. Dept. Agr. Circ. intensity for maximum profit on ponderosa do falcon in the United States. American Klipple, G. E. and D. F. Costello. 1960. pine range in northern Arizona. J. Range Birds 41:381-389. Vegetation and cattle responses to different Manage. 26:277-278. Herbel, C. H. and R. P. Gibbens.1996. Post- intensities of grazing on shortgrass ranges of Pieper, R. D. and K.C. McDaniel. 1989. drought vegetation dynamics on arid range- the central Great Plains. U.S. Dept. Agr. Ecology and management of broom snake- lands in southern New Mexico. N. Mex. Agr. Circ. 929. weed. New Mexico Agr. Exp. Bull. 751. Exp. Sta. Bull. 776. Martin, S. C. 1975. Stocking strategies and net Quigley, T. M., J. M. Skovlin, and J. P. Herbel, C.H. and W.L. Gould. 1995. cattle sales on semi-desert range. U.S. Dept. Workman. 1984. An economic analysis of Management of mesquite, creosote bush, and Agr. For. Serv. Res. Pap. RM-146. two systems and three levels of grazing on tarbush with herbicides in the northern Martin, S. C. and D. R. Cable. 1974. ponderosa pine-bunchgrass range. J. Range Chihuahuan Desert. New Mexico Agr. Exp. Managing semidesert grass-shrub ranges: Manage. 37:309-312. Bull. 775. Vegetation responses to precipitation, graz- Smith, J. G. 1962. An appraisal of the loop Herbel, C. H., F. N. Ares, and R. A. Wright. ing, soil texture, and mesquite control. U.S. transect method for estimating root crown 1972. Drought effects on a semidesert grass- Dept. Agr. Tech. Bull. 1480. area changes. J. Range Manage. 15:72-78. land range. Ecol. 53:1084-1093. Meko, D., C. W. Stockton, and W. R. Steel, R. G. D. and J. H. Torrie. 1980. Holechek, J. L. 1992. Financial benefits of Boggess. 1995. The tree-ring record of Principles and procedures of statistics. 2nd range management practices in the severe sustained drought. Water Res. Bul. Ed. McGraw-Hill Book Co., New York. Chihuahuan Desert. Rangelands 14:279-282. 31:789-801. Torell, L.A., J. M. Hawkes, and S.A. Bailey. Holechek, J. L. and D. Gait. 2000. Grazing Molinar, F. 1999. Effect of honey mesquite 2000. Range livestock cost and return esti- intensity guidelines. Rangelands 22:11-14. cover and soil depth on forage production in mates from New Mexico, 1997. New Mexico Holechek, J. L., A. Tembo, A. Daniel, M. J. the Chihuahuan Desert. Ph.D. Thesis, New Agr. Exp. Sta. Res. Rep. 738. Fusco, and M. Cardenas. 1994. Long term Mexico State University, Las Cruces, N.M. United States Department of Interior grazing influences on Chihuahuan desert National Oceanic and Atmospheric (USDI). 1999. Public land statistics. U.S. rangeland. Southwestern Naturalists Administration. 1999. Climatological data Department of Interior. Bureau of Land 39:342-349. annual summary, New Mexico, 1999. Manage.. Holechek, J. L., R. D. Pieper, and C. H. National Climatic Data Center, Asheville, Valentine, K. A. 1970. Influence of grazing Herbel. 2001. Range management: princi- N.C. intensity on improvement of deteriorated 4`h ples and practices. Edition. Prentice-Hall Nelson, T., J. L. Holechek, R. Valdez, and M. black grama range. New Mexico Agr. Exp. Inc., Upper Saddle River, N. J. Cardenas. 1997. Wildlife numbers on late Stn. Bull. 553. Holechek, J. L., M. G. Thomas, D. Gait, and and mid seral Chihuahuan Desert rangelands. Winder, J. A., C. C. Bailey, M. Thomas, and F. Molinar. 2000. Conservative and moder- J. Range Manage. 50:593-599. J. Holechek. 2000. Breed and stocking rate ate stocking effects on Chihuahuan Desert New Mexico Agricultural Statistics Service. effects on Chihuahuan Desert cattle produc- forage production. Proc. West. Sect. Amer. 2000. New Mexico Agricultural Statistics, tion. J. Range Manage. 53:32-38. Soc. Anim. Sci. 51:257-260. 1960 to 2000. New Mexico Dept Agr., Las Wuerthner, G. 1990. The price is wrong. Houston, W. R. and R. R. Woodward.1966. Cruces, N.M. Sierra 25:38-48. Effects of stocking rates on range vegetation Parker, K. W. 1951. A method for measuring and beef cattle production in the northern trend in range condition in national forest Great Plains. U.S. Dept. Agr. Tech. Bull. range condition in national forest ranges. 1357. Fort Collins, Colo. U.S. Dept. of Agr., Forest Hughes, L. E. 1982. A grazing system for the Service, Rocky Mountain Forest and Range Mojave desert. Rangelands 2:17-18. Exp. Sta..

JOURNAL OF RANGE MANAGEMENT 55(4) July 405 J. Range Manage. 55: 406-411 July 2002 Population cycles of broom snakeweed in the Colorado Plateau and Snake River Plains

M.H. RALPHS AND K. D. SANDERS

Authors are Rangeland Scientist, USDA/ARS Poisonous Plant Lab., 1150 E. 1400 N., Logan Utah 84341; and Professor, Rangeland Ecology and Management, Univ. Idaho, PO Box 1827, Twin Falls, Ida 83303.

Abstract Resumen

Broom snakeweed (Gutierrezia sarothrae (Pursh) Britt. & "Broom snakeweed" (Gutierrezia sarothrae (Pursh) Britt. & Rusby) is one of the most widespread range weeds in North Rusby) es una de las malezas de pastizal mas ampliamente espar- America. The objective of this study was to monitor broom cidas en Norte America. El objetivo de este estudio fue monitore- snakeweed populations in the salt-desert shrub community of the ar las poblaciones de "Broom snakeweed" en las comunidades Colorado Plateau and in crested wheatgrass (Agropyron deserto- arbustivas del desierto salado de la meseta Colorado y en las rum (Link) Schultes) seedings typical of the Snake River Plains siembras de "Crested wheatgrass" (Agropyron desertorum (Link) and Great Basin, and determine if its population cycles are relat- Schultes) tipicas de las planicies del Rio "Snake" y la Gran ed to precipitation patterns. Foliar cover of broom snakeweed Cuenca y determinar si sus ciclos de poblacion estan relaciona- and associated plant species was measured along 7.6 or 33 m dos a los patrones de precipitacion. La cobertura foliar de transects by the line intercept technique. Density of snakeweed "Broom snakeweed" y de las plantas asociadas se midio a to age classes (seedling, juvenile, mature) was counted in 1 m2 largo de transecto de 7.6 o 33 m mediante la tecnica de intercep- quadrats at the beginning and end of each transect. Correlations cion de linea. La densidad por clases de edad de "snakeweed" were made between snakeweed cover and density, and seasonal (plantula, juvenil y madura) se conto en cuadrantes de 1 m2 al precipitation. The snakeweed population at the Colorado Plateau inicio y final de cada transecto. Se hicieron correlaciones entre la site completed 2 cycles over the 13 year study period, dying out cobertura y densidad de "snakeweed" y la precipitacion esta- in the drought of 1989-90 and again in 2001. There were positive cional. La poblacion de "snakeweed" del sitio meseta Colorado correlations between density of snakeweed classes and seasonal completo dos ciclos en los 13 anos del periodo de estudio, precipitation: seedlings with spring precipitation (r = 0.63); juve- muriendo en la sequia de 1980-1990 y de nuevo en 2001. Hubo de "snake- niles with winter precipitation (r = 0.69); and mature plants with correlaciones positivas entre la densidad de clases la precip- previous fall precipitation (r = 0.62). Only 1 cycle occurred at weed" y la precipitacion estacional: las plantulas con the Snake River Plains site. Following the snakeweed invasion itacion de primavera (r = 0.63); juveniles con la precipitacion de into crested wheatgrass seedings in the mid 1980's, the popula- invierno ( r = 0.69) y las plantas maduras con la precipitacion del tion was at the top of its population cycle when the study began otono previo ( r = 0.62), en el sitio del Rio "Snake" solo ocurrio in 1990, dropped back and fluctuated between 6-10% cover from un ciclo. Despues de la invasion del "Snakeweed" en las siembras 1992 to 1999, and died out in 2001. Although density of mature de "Crested wheatgrass", a mediados de la decada de 1980, la plants did not change much during the middle part of the study, poblacion estuvo en la cumbre de su ciclo de poblacion cuando el the change in snakeweed cover was correlated with spring (r = estudio comenzo en 1990, retrocedio y fluctuo entre 6-10% de 0.81) and total precipitation (r = 0.60), reflecting increase and cobertura de 1992 a 1999 y murio en 2001. Aunque la densidad decrease in size of plants in response to precipitation. de plantas maduras no cambio mucho durante la parte media de este estudio, el cambio de cobertura del "Snakeweed"estuvo cor- relacionado con la precipitacion de primavera (r=0.81) y la precipitacion total (r 0.60), reflejando aumentos disminuciones Key Words: broom snakeweed, Gutierrezia sarothrae, popula- = y en el tamaiio de las plantas en respuesta a la precipitacion. tion cycle, poisonous plant.

Broom snakeweed (Gutierrezia sarothrae (Pursh) Britt. & Rusby) is the most ubiquitous range weed in North America. It The early literature associated the increase of broom snake- extends from the cold-temperate climate of Canada to the sub weed with overgrazing, which reduced desirable vegetation and tropical areas of Mexico, from the sub-humid Great Plains to the allowed broom snakeweed to increase where it already existed, acid deserts of the Great Basin, and up the Sierra Nevada and and invade deeper soils and more productive sites (reviewed by Rocky Mountains. It is now a principal component of the follow- McDaniel and Torrell 1987). However, it has even increased in ing major plant communities: creosote bush, desert grassland, good condition plant communities in the absence of grazing short-grass prairie, salt-desert shrub, sagebrush steppe, (Chew 1982, Hennessy et al. 1983). Other disturbances (such as pinyon/juniper, and mountain brush. fire, drought and chaining) can also cause it to increase (Arnold et al. 1964, Parker 1939, USFS 1937). It is very competitive with desirable grasses and greatly suppresses forage production Manuscript accepted 20 Sept. 02. (McDaniel et al. 1982, Ueckert 1979). It is not palatable to most

406 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 large ungulates (Pieper 1989), and it is the beginning and end of each transect. first in the heavily grazed, spring-use toxic to livestock causing abortions Folaar cover of broom snakeweed and experimental pasture where crested wheat- (Dollahite and Anthony 1957). Platt associated species was measured by the grass had thinned out. Seed was carried by (1959) ranked it one of the most undesir- line intercept method by stretching a tape prevailing southeasterly winds from this able plants in the various regions of the between the stakes and measuring the site and established in healthy crested West. On many localized rangelands of intercept of each species along the tape. wheatgrass seedings throughout the pas- the southwest, it is the most significant Density of the following broom snake- tures during the wet years of 1982-1984, problem limiting forage and livestock pro- weed age classes: seedling (single stem < reaching its peak from 1987-1990 duction. 6 cm tall), juvenile (less than 5 stems aris- (Sanders, personal observation). Broom snakeweed is a short-lived, ing from a crown), and mature (more than At Malta, 2 macroplots were selected perennial, suffrutescent shrub whose pop- 5 stems and having a woody base), was about 100 m apart. A permanent 33 m ulations appear to be related to climatic counted in 1 m2 quadrats placed at the transect was established on each macro patterns. In southern New Mexico, broom rebar stakes at the beginning and end of plot and marked with rebar stakes. Ten, snakeweed populations died out during each transect (n = 6 both inside and out- 7.6 m lateral transects at right angles from droughts in 1970-71, 1984, and 1994, but side each exclosure). Collection of cover the main transect (5/side) were randomly rapidly reestablished from seedlings dur- data began in 1987 and both cover and age selected and marked at the ends. Foliar ing the following wet winters and springs class density data were taken from 1989 to cover of each species was measured by the (Pieper and McDaniel 1989, Beck et al. 2001 in June each year. line intercept method along each of the lat- 1996, 1999, McDaniel et al. 2000). The Snake River Plains location was on eral transects. Density of broom snake- The objective of this study was to moni- the Lee A. Sharp Experimental Range weed age classes were counted in 1 m2 tor broom snakeweed populations in the about 13 km east of Malta Ida. (N 42° quadrats at the beginning and end of each salt-desert shrub community of the 18.96', W 113° 13.05'). Elevation was of the lateral transects (n = 20 on each Colorado Plateau and in crested wheat- 1460 m, and average annual precipitation macro plot). Cover data collection began grass seedings typical of the Snake River was 27 cm. Soils were the Sublette series, in 1990 and both cover and density data Plains and Great Basin, and determine if (Coarse-loamy, mixed superactive Pache were collected from 1991 to 2001 in changes in its cover and density were Argicryoll) and originally supported a September each year. related to precipitation patterns. Wyoming big sagebrush/squirreltail/ The Fernon site was initially established Sandberg bluegrass vegetation communi- to evaluate the effects of grazing on the ty. The area was plowed and seeded in plant community. When the snakeweed Methods 1952 to control halogeton (Halogeton population trends were recognized, mea- glomeratus (Bleb.) C.A. Mey). Predomi- surements of snakeweed on this study site nant vegetation now consists of crested continued. The Malta site was later select- The Colorado Plateau site was located wheatgrass (Agropyron desertorum (Link) ed because of the increasing snakeweed 10 km south of Ferron, Ut. at 1,890 m ele- Schultes), Sandberg bluegrass (Poa secun- population. An expanded and more effi- vation (N 38° 57.74', W 111 ° 12.17'). da Presl) with varying amounts of cient experimental design was constructed Annual precipitation averaged 22 cm with Wyoming big sagebrush (Artemesia tri- to measure changes in the snakeweed pop- a bimodal pattern of winter snow and sum- dentata var. wyomingensis (Beetle & ulation. Although the experimental design mer monsoonal thunderstorms. Soils were Young) Welsh) and broom snakeweed. at the 2 locations were different, fairly uniform Ravola loam (fine-silty, necessi- Spring/fall grazing is rotated each year on tating separate mixed, active, calcareous, mesic Typic analyses, the results are the site. Broom snakeweed established similar and can be discussed together. Torrifluvent) derived from alluvium from Mancos shale and sandstone benches. The range site was a desert loam with a salt- Table 1. Correlation coefficients (r) between density of broom snakeweed age classes and cover of desert shrub plant community consisting forage classes with seasonal and total water year precipitation. Correlation of snakeweed cover of shadscale (Atriplex confertifolia (Torn. with cover of shrubs and grass. & Frem.) Wats.), Gardner saltbush (A. gardneri (Moq.) D. Dietr.), mat saltbush Snakeweed Density Cover (A. corrugata Wats), curley grass (Hilaria Location Season Mature Juvenile Seedling Snake Shrub Grass jamesii (Torn.) Benth.), Indian ricegrass ------r ------(Stipa hymenoides R. & S.), and squir- Ferron Fall 0.62 * 0.07 0.23 0.64 ** 0.15 reltail (Elymus elymoides (Raf.) Swezey). Winter 0.28 0.69 * * -0.10 -0.16 0.49 Broom snakeweed is considered an invad- Spring -0.18 0.11 0.62 * 0.26 -0.08 er species on this site (USDA/SCS 1970). Summer 0.01 -0.41 0.43 -0.02 -0.14 At Ferron, 2 sites were selected about 4 Total 0.04 0.07 0.54t 0.28 0.17 km from each other. Soils and vegetation ** ** communities were as uniform as possible. Snakeweed 1.0 -0.26 -0.22 One exclosure (50 by 110 m) was con- Malta Fall 0.26 0.02 0.36 0.09 -0.01 t structed at each site to evaluate the effect Winter 0.10 0.09 -0.07 -0.17 0.63 * of grazing on vegetation change inside and Spring -0.10 0.21 0.08 0.81** 0.15 outside the exclosure. Cattle grazed this Summer -0.04 -0.20 0.18 0.26 -0.20 t site during winter and early spring. Three Total 0.05 0.09 0.24 0.60 .031 permanent 33 m transects were systemati- Snakeweed 1.0 -0.34 ** 0.03 cally located inside and 3 outside each ** * exclosure and rebar stakes were placed at Significant correlations p < 0.01, p < 0.05, t p < 0.10.

JOURNAL OF RANGE MANAGEMENT 55(4) July 407 40 and r 0.24, 0.41 lagged 1 year) 24 = p = (Table 1), but the trends were similar for the first part of the study (Fig. 1). The 20 snakeweed population was highest when the study began in 1987, which coincided with the wet years of 1987 and 1988. Precipitation was below average in 1989 and 1990, and the snakeweed population declined and completely died out in 1990. Precipitation increased to about average in 1992 and then increased to 50% above average in 1995, with snakeweed cover continuing to increase. Both precipitation and snakeweed cover declined in 1996. Precipitation was very high in 1997 and 1999, but snakeweed cover increased only slightly, thus causing the divergence and 1986 1988 1990 1992 1994 1996 1998 2000 resulting low correlation. The snakeweed population died out again in 2001. Fig. 1. Snakeweed cover and annual precipitation at Ferron, Ut. (error bars are SE). Average Density of snakeweed age classes annual precipitation was 22 cm. increased following the 1989-90 drought (Fig. 2). Juvenile plants appeared in 1992 At Ferron, density of broom snakeweed Results following above average winter precipita- age classes, and cover of snakeweed and tion (r = 0.69, p = 0.01) and survived and associated forage classes (shrubs and Ferron, Ut. grew into about the same number of grasses) was analyzed by the Mixed pro- There was no difference between the 2 mature plants in 1994. Most of these cedure ANOVA in a repeated measures mature plants died back in 1996 in the compound symmetry Ferron sites in broom snakeweed cover or design using in previous fall covariate structure, comparing sites, density (p > 0.22). There was no differ- response to the drop the in the precipitation r 0.62, 0.02). The exclosures (grazing vs no grazing), and ence these parameters between ( = p = seedling stage did not appear in this cycle. years. At Malta, data were analyzed by a exclosures and grazed areas (p > 0.21), following similar model comparing macro plots and indicating that grazing in early spring had Perhaps germination occurred years. Correlations were made between no effect on the snakeweed population at abundant late summer rains in 1991 and broom snakeweed cover and density of this location during this 13 year study. plants were established as juveniles when each age class with seasonal precipitation There were differences among years (p < measured in June 1992. There was a large for the current water year, and also lagged 0.0001) and some correlations with sea- influx of seedlings in 1995 that corre- precipitation from the previous year. sonal precipitation (Table 1). sponded with high spring precipitation that Correlations were also made between The correlation between snakeweed year (r = 0.63, p = 0.02). Few of these cover of snakeweed and shrub and grass cover and total precipitation was not seedlings survived due to the low precipi- cover at both locations. strong (r = 0.28, p = 0.34 for current year, tation in 1996 (Fig. 2). There was a weak but significant nega- 25 tive correlation between cover of snakeweed and other shrubs (r = -0.22, p = 0.003), and between snakeweed and grasses Mature (r = -0.26, p = 0.0006), particularly curley 20 d Juvenile grass (Fig. 3). Cover of grass and shrubs Seedling increased in 1991 and 1992 when snakeweed populations were down, then declined as snakeweed increased in 1994 and 1995. Snakeweed cover declined in 1996 and 1997, and grass cover subsequently increased from 1998 to 2001, in spite of average or below average precipitation.

Malta, Ida. The snakeweed population was highest when the study began in 1990 (14% cover), declined to 6-10% cover from 1992 to1999, then nearly died out in 2001 (year effect p < 0.001) (Fig. 4). Snake- 1994 1996 weed cover was correlated with total (r Year annual precipitation = 0.60, p = 0.06), and particularly with spring precipitation Fig. 2. Density of snakeweed age classes at Ferron, Ut. (r= 0.81,p=0.004).

408 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 seedlings established at both locations and grew into uniform, even-aged stands. Most snakeweed plants were killed at Roswell by root borers (Crossidius puchellum) in 1987-88 and the site remained free of snakeweed. The Vaughn population declined in the early 1990's and has remained low. They concluded that micro environmental factors unique to each site determined germination, establishment and death of the respective populations, independent of the broad regional climatic patterns. The micro environmental factors gov- erning snakeweed germination, establishment and death have been described. Broom snakeweed is a short-lived perennial that propagates by seed in years of opti- 1986 1988 1990 1992 1994 1996 1998 2000 mal environmental conditions, and dies out from drought stress or insect damage . YEAR It is a prolific seed producer ranging from 2036-3928 seeds/plant (Wood et al. Fig. 3. Cover of snakeweed in relation to grasses and shrubs at Ferron, Ut. (error bars are 1997). Seed are dispersed over winter, SE). mainly during high winds and snowfall, and most seed fall near the parent plant. There was a progression through the and increasing in seasons of abundant pre- Seed viability remains high over winter (70 - 80%) but drops to 0 in late May and snakeweed age classes (Fig. 5 ) and some cipitation. Snakeweed population cycles in recruitment of new plants into the popula- the Southwest have also been associated June, suggesting that seed remaining on tion. Seedlings were very abundant in with climate patterns (Pieper and the soil surface degrade rapidly. 1993, corresponding to the high spring pre- McDaniel 1989, McDaniel 1989, Beck et Germination is stimulated by light, temperature and soil moisture. Seeds must cipitation that year. Another crop of al. 1996, 1999, McDaniel et al. 2000 ). seedlings grew in 1994. A few seedlings McDaniel and Ross (2001) followed 2 remain partially exposed on the soil sur- apparently survived and progressed into the snakeweed populations for 20 years. A face (Mayeux 1983) with alternating Juvenile category in 1994, and about half of severe regional drought occurred during day/night temperatures 15 and 30° C the 1994 seedlings progressed to Juveniles the summer of 1980, causing a significant (Mayeux and Leotta 1981). Germination and peaked in 1995. There was a slight die off of snakeweed plants at Vaughn, depends on saturated surface soil moisture increase in the number of mature plants in N.M., but snakeweed plants survived at for at least 4 days (Wood et al. 1997), 1996, indicating the progression of some Roswell, N.M., less than 80 km away. requiring > 25mm precipitation that keeps juveniles into the mature stage. Although The following spring of 1981 had twice the soil surface wet for several days. there appears to be relationships between the average precipitation, and snakeweed Mortality of seedlings was > 70 % the first seedlings, juveniles and precipitation, cone- lations were not significant (P > 0.05). Grass, sagebrush, and snakeweed cover 18 followed precipitation patterns during the 16 first part of the study; increasing in wet years and declining in dry years (Fig. 6). 14 Towards the end of the study, sagebrush 12 cover continued to increase as snakeweed cover declined. There was a negative cor- 10 relation between sagebrush cover and 8 snakeweed (r = -0.34, p = 0.0001) suggesting that increasing sagebrush was 6 reducing snakeweed. Grass cover appears to be more affected by total precipitation 4 (r = 0.75, p = 0.01) than competition from 2 snakeweed or sagebrush (Table 1).

0 Discussion 1990 1992 1994 1996 1998 2000 Snakeweed populations in our study cycled on the Colorado Plateau and Snake Fig. 4. Snakeweed cover and annual precipitation at Malta, Ida. (error bars are SE). Average River Plains; dying back during drought annual precipitation was 27 cm.

JOURNAL OF RANGE MANAGEMENT 55(4) July 409 tribute to snakeweed mortality. Many insects feed on snakeweed and cause considerable mortality under certain conditions (Richman and Huddleston 1981, Foster et al. 1981, Wangberg 1982, Parker 1984). Three species of insects, the red- legg grasshopper Hesperotettix viridis, the leaf-tyer Synnoma lynosyrana, and the root-boorer Crossidius pulchellus were evaluated for their potential as biological control agents for snakeweed (Thompson and Richman 1989). Perhaps the most disagreeable weedy trait of broom snakeweed is its competitive ability against desirable forage, particularly warm-season grasses. Grass production increased several fold when snakeweed was controlled (Dwyer 1958, 1992 1994 1996 1998 2000 Ueckert 1979). Greater than 75% of snakeweed must be removed to realize the Year increased forage production, else the remaining plants simply increased in size Fig. 5. Density of snakeweed age classes at Malta, Ida. and maintained the same degree of competition (McDaniel et al. 1982). McDaniel et year, but those plants that survive may growth declines (Wan et al. 1993a) and al (1993) described a negative sigmoidal live 4-7 years (Dittberner 1971). leaves are eventually shed to cope with and exponential relationship between Survival of snakeweed is dependent water stress, but stems continue photosyn- snakeweed and grass biomass, which con- upon soil moisture. Soil water content in thesis (DePuitt and Caldwell 1975). firms that almost total control is needed to the top 20-30 cm is the single most However, as drought stress increases, sto- remove snakeweed suppression of grasses. important factor determining mortality mates remain open transpiring water, However, broom snakeweed may not (Wan et al. 1993b). As soil water stress causing tissue dehydration and xylem compete with cool-season grasses. Jameson increases seasonally or during drought, embolism. Death occurs rapidly (< 10 (1966) reported no competition between stomates do not close completely, thus days) when soil water potential drops broom snakeweed and the cool-season transpiration continues (Wan et al. 1993a, below -7.5 MPa and leaf water content squirreltail (Elymus elymoides (Raf.) DePuitt and Caldwell 1975). This luxuri- declines to 50% (Wan et al. 1993b). In our Swezey), but there was a significant nega- ant use of water depletes soil moisture study, we recorded sharp declines in tive correlation with the warm-season blue when snakeweed grows in early spring, snakeweed populations during years and grama (Bouteloua gracilis (H.B.K.) Sag. ex resulting in its competitive advantage over seasons of low precipitation. Steudel). In our study, cover of the warm- warm-season grasses that grow later. Leaf Other factors besides drought may con- season curley grass was negatively correlated with snakeweed cover at the Colorado Plateau site. Curley grass increased in years 25 when snakeweed died out, and declined as snakeweed increased. Yet, there was no correlation between snakeweed and the cool-season crested wheatgrass at the 20 Snake River Plain site. Cover of crested wheatgrass was correlated with annual precipitation patterns, but was independent of a) 15 snakeweed. We speculate that since cool- season grasses grow at the same time as broom snakeweed in early spring they may compete equally for the stored winter and spring precipitation. Research is needed to determine competition between snakeweed and cool-season grasses.

0 Conclusion 1990 1992 1994 1996 1998 2000 Snakeweed populations in the Colorado Year Plateau and the Snake River Plains cycled over time and were related to precipitation Fig. 6. Cover of snakeweed in relation to grasses and sagebrush at Malta, Ida. (error bars are SE). patterns. The snakeweed population at

410 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Ferron completed 2 cycles over the 13 Foster, D.E., D.N. Ueckert and C.J. Deloach. Richman, D.B. and E.W. Huddleston. 1981. year study period, dying out in drought 1981. Insects associated with broom snakeweed Root feeding by the beatle Crossidius pul- and establishing and increasing in wet (Xanthocephalum sarothrae) and threadleaf chellus and other insects on broom snake- snakeweed (X. microcephala) in west texas weed (Gutierrezia in eastern and cen- years. Only 1 cycle occurred at the Snake spp.) and eastern New Mexico. J. Range Manage. tral New Mexico. Environ. Entomol. River Plains site. Following the snake- 34:446-454. 10:53-57. weed invasion of crested wheatgrass seed- Hennessy, J.T., R.P. Gibbens, J.M. Tramble, Thompson, D.C. and D.B. Richman. 1989. ings in the wet years of the mid 1980's, the and M. Cardenas.1983. Vegetation changes The role of native insects as snakeweed bio- population was highest at the beginning of from 1935-1980 in mesquite (Prosopis glan- logical control agents. pp 179-187 In: E.W. the study in 1990 (14% cover), dropped to dulosa) dunelands and former grasslands of Huddleston and R. D. Pieper (eds), 6% cover with the decline in precipitation southern New Mexico USA. J. Range Snakeweed: Problems and Perspectives, New in 1992, fluctuated slightly until 1999, Manage. 36:370-374. Mexico Agr. Exp. Sta. Bull. 751. then died out in response to drought in Jameson, D.A. 1966. Competition in a blue Ueckert, D.N. 1979. Broom snakeweed: effect Actinea 2000 and 2001. There was little recruit- grama - broom snakeweed - commu- on shortgrass forage production and soil nity and responses to selective herbicides. J. water depletion. J. Range Manage. ment to this population, in spite of events Range Manage. 19:121-124. 32:216-220. of seedling (1993) and juvenile establish- Mayeaux, H.S. Jr.1983. Effects of soil texture USDA FS. 1937. Range Plant Handbk. U.S. ment (1995). McDaniel and Ross (2001) and seed placement on emergence of four Gov. Printing Office. Washington D.C. summarized the management alternatives sub shrubs. Weed Sci. 31:380-384. USDA SCS. 1970. Soil survey, Carbon - to control snakeweed based on its stage in Mayeaux, H.S. Jr, and L.Leotta. 1981. Emery area, Utah. the population cycle: use prescribed burn- Germination of broom snakeweed Wan, C., R.E. Sosebee and B.L. McMichael. ing in the early stages of the population (Gutierrezia sarothrae) and threadleaf snake- 1993a. Broom snakeweed responses to cycle while there was sufficient grass weed (G. microcephala) seed. Weed Sci. drought: I. photosynthesis, conductance, and 29:530-534. water-use efficiency. J. Range Manage. cover to carry a fire; spray with herbicides McDaniel. K.C. 1989. Snakeweed populations when the population crowded out the 46:355-359. in New Mexico, 1979-1989. pp. 13-25 In: Wan, C., R.E. Sosebee and B.L. McMichael. grass; or simply wait for the population to E.W. Huddleston and R. D. Pieper (eds), 1993b Broom snakeweed responses to die out from drought or insect damage. Snakeweed: Problems and Perspectives, New drought: II. root growth, carbon allocation, Knowledge of how to prevent establish- Mexico Agr. Exp. Sta. Bull. 751. and mortality. J. Range Manage. 46:360-363 ment of snakeweed in wet years is still McDaniel, K.C. and T.T. Ross. 2001. Wangberg, J.K. 1982. Destructive and poten- lacking. Snakeweed: poisonous properties, livestock tiallly destructive insects of snakeweed in loss, and management considerations. J. western Texas and eastern New Mexico. J. Range Manage. (in review). Range Manage. 35:235-238. McDaniel, K.C. and L.A. Torell. 1987. Literature Cited Wood, B.L., K.C. McDaniel, and D. Clason. Ecology and management of broom snake- 1997. Broom snakeweed (Guterrezia weed. pp. 101-115 In. J.L. Capinera (ed.), sarothrae) dispersal, viability, and germina- Arnold, J.F., D.A. Jameson, and E.H Reid. Westview Press, Boulder, Colo. tion. Weed Sci. 45:77-84. 1964. The pinyon - juniper type of Arizona: McDaniel, K.C., D.B. Carrol, and C.R. Hart. effects of grazing fire and tree control. 2000. Broom snakeweed establishment on USDA/FS Prod. Res. Rep. 84. blue grama grasslands after fire and herbicide Beck, R.F., R.P. NcNeeley, and S.J. Muir. treatments. J. Range Manage. 53:239-245. 1996. Effects of goats and drought on snake- McDaniel, K.C., R.D. Pieper, and G.B. weed. p.7 in: Abstracts, 49th Annual Meeting, Donart. 1982. Grass response following J. Range Soc. for Range Manage, Wichita Kans. thinning of broom snakeweed. Beck, R.F., M. Nsinamwa, R. Santos, and Manage. 35:219-222. L.A. J.W. Bain. R.D. Pieper. 1999. Dynamics of Gutierrezia McDaniel, K.C., Torell, and sarothrae with drought and grazing. 1993. Overstory-understory relationships for pp. grama grasslands. J. 502-503 In: Eldridge and Freudenberger broom snakeweed-blue (Eds), Proc. VI International Rangeland Range Manage. 45:505-511. K.E. 1939. The control of snakeweed Congress, Townsville Aust. Parker, in the southwest. S.W. For. Range Exp. Sta. Chew, R.M. 1982. Changes in herbaceous and suffrutescent perennials in grazed and Res. Note 76. Parker, M.A. 1984. Local food depletion and ungrazed decertified grassland in south east- the foraging behavior of a specialist ern Arizona USA, 1958-1978. Amer. Midl. grasshopper, Hesperotettix viridis. Ecology Nat. 108:159-169. 65:824-835. DePuit, E.J. and M.M. Caldwell. 1975. Gas Pieper, R.D. 1989. Broom snakeweed content in semi-desert species in exchange of three cool herbivore diets. 203-210, in: E.W. relation to temperature and water stress. J. pp. Huddleston and R. D. Pieper (eds), Snakeweed: Ecol. 63:835-858. Problems and Perspectives, New Mexico Agr. Dittberner, D.L. 1971. A demographic study Exp. Sta. Bull. 751. of some semi-desert grassland plants. M.S. Pieper, R.D. and K.C. McDaniel. 1989. Las Cruces Thesis, New Mexico State Univ., Ecology and management of broom snake- N.M. weed. 1-2 In: E.W. Huddleston and R. 1957. pp. Dollahite, J.W. and W.V. Anthony. D. Pieper (eds), Snakeweed: Problems and Poisoning of cattle with Gutierrezia micro- Perspectives, New Mexico Agr. Exp. Sta. Amer. cephala, a perennial broomweed. J. Bull. 751. Vet. Med. Assoc. 130:525-530. Platt, K.B.1959. Plant control - some possibil- Dwyer, D.D. 1958. Competition between forbs ities and limitations. II Vital statistics for and grasses. J. Range Manage. 11:115-118. range management. J. Range Manage. 12:194-200.

JOURNAL OF RANGE MANAGEMENT 55(4) July 411 J. Range Manage. 55: 412-419 July 2002 Monitoring a half-century of change in a hardwood rangeland

KERRY L. HEISE AND ADINA M. MERENLENDER

Authors are Botanist and Staff Research Associate, Integrated Hardwood Range Management Program, Hopland Research and Extension Center, Hop/and, Calif. 95449; and Cooperative Extension Specialist, Environmental Science, Policy, and Management, University of California, Berkeley, Calif. 94720-3110.

Abstract Resumen

Changes in rangeland species composition can effect forage Cambios en la composicion de especies del pastizal pueden quality, ecosystem function, and biological diversity. afectar la calidad de forraje, la funcion del ecosistema y la diver- Unfortunately, documenting species compositional change is dif- sidad biologica. Desafortunadamente, el documentar los cam- ficult due to a lack of accurate historic records. We took advan- bios de la composicion de especies es dificil debido a la falta de tage of herbarium records dating from the early 1950's to recon- registros historicos certeros. Tomamos ventaja de los registros struct the past flora of a 2,168 ha hardwood rangeland in del herbario, que datan de inicio de la decada de 1950, para reconstruir la flora pasada de 2,168 de un con Mendocino County, California , and then compared this to the ha pastizal arboles current flora of the site. An inventory of vascular plants conduct- de madera dura en el condado de Mendocino, California, y la ed from 1996 to 2001 added 44 native and 15 non-native species comparamos con la flora actual del sitio. Un inventario de las bringing the total number of species and infraspecific taxa at the plantas vasculares conducido de 1996 al 2001 agrego 44 especies study site to 671. Of the original 612 species recorded prior to nativas y 25 especies no nativas totalizando 671e1 numero de this study, 34 native and 1 non-native species could not be relo- especies y taxas intraespecificas en el sitio de estudio. De las 612 cated. The percentage of non-native species increased from 19% especies originales registradas antes de este estudio, 34 nativas y in 1952 to 23% in 2001. Based on estimates from the early 1950's, una no nativa no pudieron ser relocalizadas. El porcentaje de mid 1980's, and 1996 to 2001, at least 13 non-native species have especies no nativas se incremento de 19% en 1952 a 23 % en increased in abundance, while some native species have 2001. Basados en las estimaciones de inicios de la decada de los decreased. Livestock grazing, competition with invasive species, 50's, mediados de la decada de los 80's y de 1996 a 2001, al conversions to different vegetation types, and transportation of menos 13 especies no nativas han incrementado en abundancia, propagules into the site by vehicles and livestock, combined with mientras algunas especies nativas han decrecido. El apacen- the difficulty of relocating rare species, are posed as the most tamiento de ganado, la competencia con especie invasoras, la likely causes for the documented changes. conversion a diferentes tipos vegetativos y el transporte de propagulos al sitio por vehiculos y ganado, combinado con la dificultad de relocalizar especies raras son propuestas como las Key Words: Northwest California, herbarium, oak woodlands, causas mas probables de los cambios documentados. livestock, flora, invasive species

Hardwood rangelands in California cover an estimated 4.5 million ha and are characterized by an overstory canopy of hard- ty, change in ecosystem function, and reduction in rangeland pro- wood trees, primarily in the genus Quercus. The overstory productivity (D'Antonio and Vitousek 1992, Campbell and Smith vides at least 10% cover; the understory is generally dominated 2000). In particular, exotic annual grasses interfere significantly by exotic annual grasses and forbs. Replacement of the native with native perennial grasses during all stages of growth understory flora with a primarily non-native one occurred rapidly (Hamilton et al. 1999), resulting in an increase in exotic annuals and on a very large scale over the past 200 years (Heady 1977, over time. Invading species can also differ functionally from the Mack 1989). The primary causes of this change are human native species that they are replacing, resulting in a great overar- induced such as livestock grazing, fire suppression, and facilitat- ching effect on ecosystems (Vitousek 1986). ed dispersal (Rejmanek et al. 1991) resulting in an increased Monitoring species compositional changes in California hard- dominance by exotics and a subsequent loss of native species wood rangelands is essential for determining the sustainability of (Mooney and Drake 1987). livestock grazing and necessary to address the lack of oak Change in understory species composition of oak woodlands, seedling survival. Because the understory of California's hard- savanna, and grasslands represents a significant component of wood rangelands are dominated by exotic grasses and forbs, global environmental change including loss of biological diversi- many lower in nutritional quality and palatability than native perennial species, the capacity of rangelands has been seriously decreased (Major 1955, Pettit et al. 1995). Annual grasses are Authors wish to thank Chuck Vaughn, Bob Keiffer, and 2 anonymous reviewers for helpful comments on the manuscript. adapted to heavy grazing and low nutrient availability (Jackson Manuscript accepted 5 Oct. 01. 1985), which has consequences for the way livestock grazing is practiced on site as well as the economic vitality of open range

412 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 grazing operations. As range conditions (Arbutus menziesii Pursh), and California areas", research pastures, and steep, wood- deteriorate, livestock seek out less palat- buckeye (Aesculus californicus (Spach) ed terrain outside of fenced pasture bound- able plant species, increasing the use on Nutt.). Grasslands and the understory of aries. Rangeland improvement practices these species, oak seedlings included. open oak woodland and savanna are domi- during the 1960's and 1970's, aimed at Additionally, the establishment of exotic nated by non-native annual grasses such as increasing agricultural production, result- annual grasses has been proposed as a lim- slender wild oat (Avena barbata Link), big ed in the conversion of approximately 300 iting factor in the regeneration of some quaking grass (Briza maxima L.), hedge- ha of oak woodland and chamise chaparral native oak species (Muick and Bartolome hog dogtail (Cynosurus echinatus L.), and to exotic grassland (Murphy 1976, 1978). 1988, Gordon and Rice 1993). barbed goatgrass (Aegilops triuncialis L.). In addition, university staff and Due to the scarcity of historic baseline Chaparral scrub consisting of chamise researchers intentionally introduced at information, few studies have examined (Adenostoma fasciculatum Hook. & Am.), least 19 exotic plant species. These were long-term change on California range- Arctostaphylos spp., Ceanothus spp., and seeded into pastures following fire or lands. Using cover data from the 1930's, Quercus spp., along with patches of closed other forms of clearing to improve range Holzman and Allen-Diaz (1991) docu- cone pine (Pinus attenuata Lemmon) are forage, and since have expanded their mented 58 years of change in 21 blue oak common above 675 m. Douglas fir ranges well beyond the initial introduction (Quercus douglasii Hook. & Am.) wood- (Pseudotsuga menziesii (Mirbel) Franco), sites. Some of the common species include land plots in California. Our study docu- California nutmeg (Torreya californica smilo grass (Piptatherum miliaceum (L.) ments change over nearly a half-century Torrey), and canyon live oak (Quercus Cosson), harding grass (Phalaris aquatica by reconstructing the past flora of a hard- chrysolepis Liebm.) form dense shaded L.), orchard grass (Dactylis glomerata L.), wood rangeland in northwest California, canopies on steep north and east facing rose clover (Trifolium hirtum All.) and comparing herbarium data to the modern slopes. Numerous seasonal creeks, sag subterranean clover (T, subterraneum L.). flora. This was feasible due to the estab- ponds, vernal pools and springs provide a lishment of an extensive reference collec- diversity of wetland habitats and plant tion of vascular plants for the site in the communities (Heise and Merenlender Methods early 1950's. A flora was later published 1999). (Murphy and Heady 1983) which included Between 1996 and 2001 we conducted data on species abundance across the site Historic Data an extensive inventory of vascular plants This comprehen- from the early 1980's. The Center was established as a range across the entire Center. First, the infor- in combina- sive set of scientific records, experiment station in 1951. The following mation from over 1,275 specimen labels tion with extensive documentation of land- year, a botanical survey was initiated to from the Center's herbarium was updated us with a means to use at the site, provided document all species of vascular plants on to reflect mis-identifications, recent scien- and identify newly established species the property. By 1954 nearly 450 species tific name changes, and reductions due to document changes in the relative abun- were recorded. During the next 30 years of synonymy (Hickman 1993), and then dance some taxa. of sporadic collecting only 137 species were incorporated into a database. From this, added to the flora, bringing the total to species lists with location, habitat, and almost 600. At this time a list of the vas- collection date were produced to facilitate Study Site cular plants for the Center was published the relocation of previously recorded taxa. (Murphy and Heady 1983). Only a few Most of the herbarium specimens (75%) Vegetation more species were added to the flora contained general location descriptions The study area encompasses the entire through the 1980's and mid 1990's, bring- such as pasture name only, while the University of California Hopland ing the total number of vascular species remaining 25% had more specific site Research and Extension Center, located in and infraspecific taxa to 612. Alfred H. locations. Because of the intensity and the interior north coast ranges of south- Murphy, the Center's first superintendent, thoroughness during the first 3 years of the eastern Mendocino County (39° 00' N, and Harold F. Heady, U.C. Berkeley original botanical survey in the early 123° 05' W). It encompasses nearly 2,168 Professor of Range Management, initiated 1950's, we assume that taxa not collected hectares of moderately steep, predomi- the inventory, and alone contributed 1,177 then were probably sparse, inaccessible, or nately southwest-facing slopes in the specimens to the Center's herbarium, absent. An extensive road system over 72 Mayacmas Mountains, with elevations about 92% of the collection. km in length facilitated access throughout ranging from 183 to 914 m. The natural the Center's property; roadless areas were vegetation reflects a mediterranean cli- Livestock and Forage Management surveyed on foot. Voucher specimens for mate of hot dry summers and cool wet Prior to the University's purchase of the previously unrecorded taxa were collected winters, with an average annual rainfall of property the area was used as a sheep and deposited into the Center's herbarium. 960 mm. Oaks and other hardwoods are range. Since then, between 1,500 and Relative terms describing abundance at the common overstory species of the lower 2,000 sheep continue to use approximately collection site ("Sparse", "Occasional", elevation woodlands and include blue oak 213 of the Center. Much of the livestock "Numerous") were assigned to 479 specimens (Quercus douglasii Hook. & Am.), black grazing pressure across this range is light (313 species) that were collected between oak (Q. kelloggii Newb.), valley oak (Q. because of the terrain and dense wooded 1952 and 1954. This site-specific information lobata Nee), Oregon oak (Q. garryana vegetation; the heaviest use occurs on provided a rough estimate of species abun- Hook.), coast live oak (Q. agrifolia Nee), approximately 300 ha and is concentrated dance across the Center in cases where Shreve oak (Q. parvula E. Greene var. in lambing and winter feeding pastures or there were similar abundance estimates for shrevei (C.H. Muller) K. Nixon), seeded grassland near the headquarters multiple collections of the same species. California bay (Umbellularia californica area. Livestock are excluded from approx- In the early 1980's Murphy and Heady (Hook. & Am.) Nutt.), Pacific madrone imately 435 ha in designated "biological (1983) estimated the abundance of 600

JOURNAL OF RANGE MANAGEMENT 55(4) July 413 vascular plant species across the entire Table 1. Floristic summary data for the University of California Hopland Research and Extension Center using similar non-absolute terms, Center, Calif. which we adopted during our field work to provide consistency. Thus, for about half of the flora we were able to obtain compa- Taxa Families Genera Taxa Taxa rable abundance estimates from the early Ferns and Fern Allies 10 16 0 1950's, early 1980's, and 1996 to 2001, Gymnosperms 3 5 4 1 and for the entire flora between the early Dicots 71 258 1980's, and 1996 to 2001. Given the sub- Monocots 11 75 jectivity of these terms, we only present Total 95 354 taxa that have undergone extreme declines or increases, thus representing notable ing pressure, 3 on moderately grazed, 9 on Changes in Species Abundance change. lightly grazed, and 10 in areas with no Many non-native species presumed grazing. A majority of the 35 species lost absent or sparse during the early 1950's occurred in grassland, chaparral, or wet- Results have since undergone dramatic increases land habitats. in abundance (Table 4). For example, New Taxa The 1996 to 2001 inventory added 59 Table 2. New species found at the University of California Hopland Research and Extension species (44 native and 15 non-native) in Center, California during 1996-2001 inventory period. Habitat: C = chaparral, G = grassland, H hardwood forest, X xeric woodland, outcrop/crevice, W wetland, R roadcut. 29 families to the flora, bringing the total = = 0 = = = Grazing: N none, L light, M moderate, H heavy number of species and infra-specific taxa = = = = at the field station to 671 (Table 1). More Species Habitat Grazing than half (34) of the new additions were native dicots, followed by non-native Dryopteridaceae Athyrium alpestre var. americanum 0 N Y dicots (14), native monocots (8), non- native monocots (1), and fern and fern Selaginellaceae Selaginella wallacei 0 N Y allies (2). Asteraceae was the most repre- Apiaceae sented family with 7 native and 4 non- Lomatium dissectum 0 N Y native species, followed by Apiaceae, Lomatium vaginatum H N Y Cyperaceae, and Scrophulariaceae (Table Oenanthe sarmentosa w N Y 2). Of the 44 native species, 22 were Tauschia kelloggii H L Y found in areas where sheep have been Torilis arvensis X,G N,L,M N excluded, such as the "Biological Areas", Asteraceae rock outcrops, or other areas inaccessible Euthamia occidentalis w M Y to livestock, 12 in areas lightly grazed, 6 Gnaphalium luteo-album w H N in moderate, and 4 in sites with heavy Hesperevax acaulis var. ambusticola C L Y Hesperevax spars iflora G L Y grazing pressure. Of the 15 new non- Layia septentrionalis X L Y native species, 3 were found in ungrazed Solidago californica w L Y areas, 3 in lightly grazed, 2 in moderately Soliva sessilis R H N grazed, 6 in heavy grazed areas, and 1 Stephanomeria elata w L Y found throughout all grazing regimes. Tolpis barbata x L N Almost half of all new species found Tracyina rostrata G H Y Tragopogon porrifolius G L N occurred in wetland habitats such as vernal pools, sag ponds, springs, and seasonal Boraginaceae Myosotis discolor w N Y creeks. Brassicaceae Carderia draba w N N Missing Taxa Erysimum capitatum 0 N Y Species loss was chiefly confined to Campanulaceae plants of low abundance and occurred Heterocodon rariflorum w L Y across a wide variety of habitat types and Caprifoliaceae grazing regimes. Of the original 612 Symphoricarpos mollis H L Y species that were documented at the Caryophyllaceae Sagina apetala W M Y Center prior to this study, 34 native and 1 non-native species could not be relocated Scleranthus annuus ssp. annuus G L N (Table 3). Over half of the missing species Crassulaceae Sedum obtusatum ssp. retusum 0 N Y (21) were native dicots, followed by native monocots (13), and non-native dicots (1). Ericaceae Arctostaphylos canescens ssp. sonomensis C L Y Although 19 families were represented, Chimaphila menziesii H N Y Poaceae, Asteraceae, and Cyperaceae Euphorbiaceae accounted for roughly half of the species Euphorbia spathulata X L Y lost. Thirteen of the missing species occurred on sites that receive heavy graz- (Table 2 continued on page 415)

414 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 (Table 2 Continued). While many introduced species have become established and have increased in Species Habitat Grazing Native abundance at the Center, some native Fabaceae herbaceous perennials, such as rancher's Lotus pinnatus W N Y fireweed (Amsinckii menziesii (Lehm.) Trifolium gracilentum var. gracilentum G M Y Nelson & J.F. Macbr.), owl's clover Vicia hirsuta W N N (Castilleja densiflora (Benth.) Chuang & Gentianaceae Heckard), Rigiopappus leptocladus A. Cicendia quadrangularis w H Y Gray, and modesty (Whipplea modesta Lamiaceae Torrey), which were considered numerous Satureja douglasii H N Y in the early fifties, are now sparse in Lamium purpureum R H N occurrence. Other species that were com- Linaceae mon in occurrence in the fifties and now Linum bienne W H N missing include Brewer's milkvetch Onagraceae (Astragalus breweri A. Grey), varied-leaf Clarkia concinna ssp. concinna H N Y Epilobium brachycarpum W H Y Collomia (Collomia heterophylla Hook.), and tufted hairgrass (Deschampsia cespi- Polemoniaceae tosa (L.) Beauv. var. cespitosa). Because Gilia sinistra ssp. pinnatisecta Y of the subjective nature of the abundance Polygonaceae smaller changes for other Rumex conglomerates N measures, species have undoubtedly occurred but Portulacaceae cannot be presented with confidence. Claytonia exigua Y

Portulacaceae Claytonia exigua 0 N Y Discussion Rubiaceae Sherardia arvensis N N The initial plant survey at the Center in Scrophulariaceae the early 1950's attempted to catalog all Collinsia parviflora X M Y the known vascular plants, establishing Cordylanthus pilosus X N Y valuable baseline floristic information for Parentucellia viscosa W,R H N a 2,168 ha rangeland. Our study represents Veronica anagallis-aquatica W H N the first step in documenting the floristic Solanaceae changes that have occurred since the early Solanum americanum W N Y 1950's. As a result of our inventory we Valerianaceae trends: 1) the Plectritis brachystemon G M Y observed the following occurrence of new native taxa previously Verbenaceae 2) new non-native species Phyla nodiflora H Y undetected, 3) Verbena lasiostachys N Y have been established at the site, some already established non-native species Cyperaceae 4) Carex barbarae W N Y have greatly increased in abundance, Carex comosa W N Y some missing taxa are clearly extirpated Carex globosa H N Y from the site, and 5) the missing taxa Carex subfusca W M Y include some species that may be too rare Scirpus californicus W N Y to relocate over the course of the study. As Orchidaceae expected, we added more non-native Corallorhiza maculata H N Y species to the Center's flora, whereas the Poaceae addition of native species gave us confi- Festuca arundinacea G L N dence in the thoroughness of our survey H L Y Melica hardfordii methods. From records at the Center there Typhaceae is evidence that some rangeland manage- eurycarpum ssp. eurycarpum W N Y Sparganium ment practices led to species loss and the spread of exotic plants. Access to pennyroyal (Mentha pulegium L.), an and abundant Japanese Hedge-Parsley management records that can help explain herbaceous perennial first collected on the (Torilis arvensis (Hudson) Link.) was first possible causes of change is an advantage of Center in 1969, now dominates the margins collected in 1996, which suggests a rapid doing this type of study on a public research of several permanent ponds and seeps in increase in abundance over the past 10 facility. We documented an increase in non- grazed and ungrazed sites alike. Barbed years. Noted as "sparse" and "occasional" native species richness which is character- goatgrass (Aegilops triuncialis L.), another respectively in 1983, big quakinggrass istic of the spread of invasive species in extremely invasive plant, was first collected (Briza maxima L.) and hedgehog dogtail western lowland California (Randall et al. here in 1961 and is now the dominant (Cynosures echinatus L.) now comprise the 1998). Non-natives at the Center are being species on approximately 125 ha of serpen- dominant understory of oak woodlands on added to the flora at a rate of 1.3 species tine grassland. The currently widespread several south and southwest-facing slopes. per year, while several of those already

JOURNAL OF RANGE MANAGEMENT 55(4) July 415 Table 3. Plants that could not be relocated at the University of California Hopland Research and areas, and are conspicuously absent in Extension Center, California following a 19902001 inventory. Habitat: R = roadcut, C = chap- dense chaparral, shaded hardwood forest, arral, G = grassland, H = hardwood forest,W = wetland, X = xeric woodland. Grazing: N = and on steep rocky outcrops. Roughly one none, L light, M moderate, H heavy = third of the Center's 155 non-native species are annual grasses, followed by Species Habitat Grazing annual forbs primarily in Asteraceae and Apiaceae Fabaceae. Today, non-native plant species Perideridia gairdneri G N Y on the Center comprise 23% of the total Asteraceae flora compared to 19% in the early 1950's, Aster radulinus H H Y Cirsium remotifolium G N Y By the time the University of California Lessingia nemaclada GL Y acquired the property in 1951, the non- Microseris laciniata ssp. laciniata G H Y native flora was already well established. Boraginaceae Herbarium records indicate that by 1954 Plagiobothrys stipitatus var. micranthus W H Y over 90 introduced species occurred on the Brassicaceae Center; of these, 22 were considered Tropidocarpum gracile G L, H Y numerous at their collection sites. Thirty Cistaceae years later there were 136 introduced Helianthemum suffrutescens C L Y species, and by 2001 the number had Crassulaceae grown to 155 (Fig. 1). This represents a Sedum radiatum C N Y 71 % increase in non-native species over a Fabaceae period of 45 years. In contrast, the entire Astragalus breweri G H Y state of California saw a 31% increase in Gentianaceae non-native plant species for a similar peri- Centaurium davyi w N Y od between 1959 and 1998 (Randall et al. Loasaceae 1998). Mentzelia laevicaulis R L Y Many of the new native species found Mentzelia micrantha R L Y during our inventory were likely over- Malvaceae looked during earlier surveys because of Abutilon theophrasti W N N their rarity or locations that were difficult Onagraceae to access. For example, spikemoss Camissonia californica W L Y (Selaginella wallacei Hieron.) was com- Clarkia unguiculata R L Y mon on several north-facing boulders Orobanchaceae camouflaged with mosses and ferns, and Orobanche uniflora G H Y an individual wallflower (Erysimum capi- Papaveraceae tatum (Douglas) E. Greene) was located Dendromecon rigida CN Y within a crevice high on a steep cliff face. Polemoniaceae Many others were found on steep wooded Collomia diversifolia H H Y Collomia heterophylla X N Y slopes outside of fenced pastures. Four in Polygonaceae plants considered rare California were Chorizanthe polygonoides C N Y also found (Skinn e r a nd Pav lik 1994) ' The Bristly is Ranunculaceae sedge (Carex comosa Boott) Aquilegia eximia C N Y considered rare across its entire range in in Cyperaceae several western states the U.S., beaked Cyperus erythrorhizos w N, M Y Tracyina (Tracyina rostrata S.F. Blake) is Cyperus squarrosus w M Y known from fewer than a dozen popula- Eleocharis radicans w M Y tions in northwest California, Sonoma Liliaceae manzanita (Arctostaphylos canescens Fritillaria recurva C L Y Eastw. ssp. sonomensis (Eastw.) P. Wells) Lilium rubescens C N Y occurs on chaparral sites in Northwest Poaceae California, and Colusa Layia (Layia Agrostis hallii X H Y septentrionalis Keck) is a serpentine Calamagrostis koelerioides C L Y endemic of California's Interior Coast Calamagrostis purpurascens C H, L Y Ranges. Documenting these locations adds Calamagrostis rubescens H H Deschampsia cespitosa W H Y to an important body of information for Festuca viridula C L Y conservation biologists and resource pro- Leersia oryzoides W H Y tection agencies monitoring rare species Pleuropogon californicus W H Y occurrences. Over half of the missing taxa were originally recorded as "rare" or "sparse" and established have undergone dramatic local sheep or vehicles. This is similar to find- may have been overlooked during our range extensions. Most of the new intro- ings on other hardwood rangelands in inventory. These species were originally duced species were found in areas California (deNevers 1985, Knops et al. collected in upland areas with little or no accessed by livestock or adjacent to roads 1995), where introduced species are pre- grazing. Furthermore, there was nothing where they could have been carried in by sent mainly in grasslands and disturbed about the species composition at the origi-

416 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Table 4. Species that have undergone notable increases or decreases in abundance at the University of California Hopland Research and Extension Center, California. Note: "absent" in 1952-55 column means plant collected after 1955 and therefore assumed not present at the Center or so rare in occurrence it was missed during first years of inventory.

Species Exotic 1952-55 1983 1996-01 Amsinckii menziesii (Fiddleneck) numerous occasional sparse Castilleja densiflora (Owl's Clover) numerous occasional sparse Rigiopappus leptocladus numerous numerous sparse Whipplea inodesta (Yerba de Selva) numerous occasional sparse Aegilops triuncialis (Barbed Goatgrass) x absent occasional Baccharis pilularis (Coyote Brush) sparse occasional numerous Briza maxima (Big Quakinggrass) x absent sparse Carduus pycnocephalus (Italian Thistle) x sparse numerous Cerastium arvense (Field Chickweed) x absent occasional Cynosures echinatus (Hedgehog Dogtail) x occasional occasional Dactylis glonierata (Orchardgrass) x sparse occasional Holcus lanatus (Velvetgrass) x sparse sparse Mentha pulegium (Pennyroyal) x absent sparse Parentucellia viscosa x absent absent Petrorhagia dubia x absent rare Soliva sessilis x absent absent Taeniatherum caput-medusae (Medusahead) x occasional numerous Torilis arvensis (Japanese Hedge-parsley) x absent absent numerous nal collection sites that would suggest grass (Holcus lanatus L.) and pennyroyal species replacement. At the same time, rare with non-native grasses and clovers in wetland habitats (Table 4). The poten- plants are more susceptible to stochastic (unpublished pasture histories). Reed grass tial impact from either of these threats may events, so it can't be ruled out entirely that (Calamagrostis koelerioides Vasey) likely play a role in the observed species loss. In some may have gone locally extinct. It is occurred on a chaparral site that has since particular, barbed goatgrass now domi- difficult to differentiate between what been cleared and seeded with non-native nates serpentine grassland sites and has might have been a stochastic extinction grasses and forbs. become a threat to rangelands across and the inability to locate a species because Thirteen of the 35 species that could not California (Kennedy 1928, Peters et al. of its rarity. Fischer and Stocklin (1997) be relocated occurred originally on sites 1996). Previously, serpentine grasslands noted a similar problem comparing old and that have been under heavy grazing pres- did not support extensive annual grass new records of plant occurrence on cal- sure. Heavy grazing can jeopardize the cover due to the low nutrient levels careous grasslands in the Swiss Jura sustainability of grassland ecosystems by (Huenneke et al.1990) and had been Mountains. They mentioned the possibility reducing soil fertility and water-holding regarded as refugia for California's native that species could have been missing in the capacity (Dormaar and Willms 1998), plants (Harrison 1999, Murphy and Ehrlich new records without actually having gone which can then lead to dominance of exot- 1989). Therefore, documenting the spread extinct, if the old and new records were not ic annual species. At the same time, these of this invasive annual grass is of particular taken at the same location. Similarly, many sites have also undergone dramatic concern for native plant conservation. of our specimens had only vague location increases in several exotic species, notably descriptions, such as "upper Orchard big quaking grass, hedgehog dogtail, and Pasture" or "Lake Biological Area", mak- barbed goatgrass in grasslands, and velvet ing it more difficult to locate the original collection site; more detailed location information was available for only 9 of the Non-natives 3 TotaI 43 missing plants. Conversions of vegetation type appear to be responsible for the extirpation of at least 5 species. Pine grass (Calamagrostis rubescens Buckley) and serpentine Collomia (Collomia diversifolia E. Greene), the latter an uncommon California endemic, were last collected in the spring of 1957. In the summer of the same year the woody vegetation of the 26 ha watershed where these plants occurred was removed and burned, and later seeded with a mixture of non-native clovers and grasses (Murphy 1976). Semaphore grass (Pleuropogon californicus (Nees) Vasey) 1950's 1960's 1970's 1980's 1990's and popcorn flower (Plagiobothrys stipitatus (E. Greene) I.M. Johnston var. micran- Decade thus (Piper) I.M. Johnston) were last col- Fig. 1. Number of vascular plant species reported for the University of California Hopland lected in 1952 and 1953 along a seasonal Research and Extension Center, California. Over 2/3 of the total number of species to date watercourse, since then this pasture has were collected between 1952 and 1954 following the establishment of the field station.

JOURNAL OF RANGE MANAGEMENT 55(4) July 417 The relatively recent establishment and new losses, as well as new additions to the Heady, H.F. 1977. Valley grassland. Pp. 491- spread of species such as big quaking flora. To better demonstrate the spread of 514 in M.G. Barbour and J. Major (eds.), grass, hedgehog dogtail, barbed goatgrass, invasive species, a monitoring program Terrestrial vegetation of California. John Wiley Sons, New York. velvet grass, and pennyroyal also has con- and should be established that includes map- Heady, H.F. and M.D. Pitt. 1979. Reactions sequences for ecosystem function. Eviner ping the distribution of non-native species of northern California grass=Woodland to and Chapin (2001) found differences in through time using the Center's existing vegetational type conversions. Hilgardia decomposition rates, net nitrogen cycling Geographic Information System. While 47:51-73. rates, and phosphorus availability between we note shifts in land-use as a likely cause Heise, K.L. and A.M. Merenlender.1999. Flora single species plots planted at the Center. for the observed changes to the flora, of a vernal pool complex in the Mayacmas In particular, barbed goatgrass plots had long-term experiments on the effects of Mountains of southeastern Mendocino County, Madrono 46: 38-45. greatly enhanced net nitrification, and sup- livestock grazing and species introduc- California. Hickman, J. 1993. The jepson manual: Higher ported the highest gopher activity, sug- tions are essential to test the direct contri- plants of California. University of California gesting that this species may provide a bution that grazing and exotics have on Press, Berkeley and Los Angeles, Calif. more stable soil surface preferred by native flora. In addition, a quantitative Holzman B.A. and B.H. Allen-Diaz. 1991. gophers. This example illustrates how study comparing vegetation composition Vegetation change in blue oak woodlands in changes in the composition and relative and abundance in similar plant communi- California. USDA For. Serv. Gen. Tech. Rep. abundance of herbaceous species can ties on comparable soil types between PSW-126. directly influence ecosystem function, grazed and ungrazed sites would be useful. Huenneke, L.F., S.P.Hamburg, R. Koide, H.A. Mooney, and P.M. Vitousek. 1990. resulting in alterations to the habitat that Studies of this sort would complement this Effects of soil resources on plant invasion and can further affect vertebrate densities. one and provide a better understanding of community structure in Californian [USA] Another consequence of the spread of species turnover and the associated causes serpentine grassland. Ecol. 71:478-491. competitive annual grasses for hardwood in north coast hardwood rangelands. Jackson, L.E. 1985. Ecological origins of rangeland conservation may be reduced California's mediterranean grasses. J. oak regeneration, due primarily to compe- Biogeog.12:349-361. tition for water (Gordon and Rice 1993). Literature Cited Kennedy, P.B. 1928. Goatgrass or wild wheat In addition to the consequences (Aegilops triuncialis). J. Amer. Soc. Agron. these 20:1292-1296. exotics have for ecosystem function and Campbell, B.D. and D.M.S. Smith. 2000. A Knops, J.M.H., J.R. Griffin, and A.C. biodiversity conservation, the quality of synthesis of recent global change research on Royalty. 1995. Introduced and native plants the livestock forage has been reduced. For pasture and rangeland production: Reduced of the Hastings Reservation, central coastal example, pennyroyal, barbed goatgrass, uncertainties and their management implica- California: A comparison. Biol. Cons. hedgehog dogtail, and big quaking grass, tions. Agr. Ecosyst. Enviroment 82: 39-55. 71:11.5-123. have all undergone substantial increases in D'Antonio, C.M. and P.M. Vitousek. 1992. Mack, R. N. 1989. Temperate grasslands vul- abundance at the Center, yet are highly Biological invasions by exotic grasses, the nerable to plant invasions: characteristics and consequences. Pp. 155-179 In J.A. Drake, unpalatable to livestock. grass/fire cycle, and global change. Ann. H.A. Mooney, F. di Castri, R.H. Groves, F.J. Early research at the Center focused on Rev. Ecol. Syst. 23:63-87. de Nevers, G. 1985. Pepperwood flora. Kruger, M. Rejmanek, and M. Williamson vegetation type conversions and reseeding Pepperwood Ranch Natural Preserve. Sonoma (eds.), Biological invasions: A global per- with non-native species in order to County, CA. California Academy of Sciences, spective. John Wiley & Sons, Chichester, enhance forage production and water San Francisco, Calif. N.Y. yield, but this was not without costs. The DiTomaso, J.M., K.L. Heise, G.B. Kyser, Major, J. 1955. Weeds on California range- conversion of a wooded 86 ha watershed A.M. Merenlender, and R.J. Keiffer. 2001. lands. Cal. Agr. 9:3-4. resulted in a dramatic increase of 2 weedy Carefully timed burning can control barb Mooney, H.A. and J.A. Drake. 1987. The ecology of biological invasions. Environ. species along with multiple goatgrass. Cal. Agr. 55(6):47-52. soil slips and 29:10-37. increases in sediment discharge (Pitt et al. Dormaar J.F. and W.D. Willms.1998. Effect of forty-four years of grazing on fescue Muick, P. C. and J. W. Bartolome. 1988. 1978, Heady and Pitt 1979). On the other grassland soils. J. Range Manage. Factors associated with oak regeneration in hand, fire has recently been used to dra- 51:122-126. California. Pp 86-91, In: T.R. Plumb and N. matically reduce barbed goatgrass infesta- Eviner, V.T. and F.S. Chapin III. 2001. Plant H. Pillsbury (Tech. Coord.). Proc. Symp. tions in 2 pastures while simultaneously species provide vital ecosystem functions for Multiple-use Management of California's increasing the abundance of meadow bar- sustainable agriculture, rangeland manage- Hardwood Resources. USFS Gen. Tech. Rep. ley (Hordeum brachyantherum Nevski), a ment and restoration. Cal. Agr. 55(6):54-59. PSW-100. native perennial bunchgrass; thereby Fischer, M. and J. Stocklin. 1997. Local Murphy, A.H. 1976. Watershed management extinctions of plants in remnants of exten- increases rangeland productivity. Cal. Agr. improving forage for sheep (DiTomaso et 30:16-21. al. 2001). The trend toward research sively used calcareous grasslands 1950-1985. Cons. Biol. 11:727-737. Murphy, A.H. 1978. Benefits of chaparral directed at oak woodland and grassland Gordon, D. R. and K. J. Rice 1993. Competitive improvement to livestock and wildlife. Proc. ecosystems and the fact that rangeland effects of grassland annuals on soil water and Ann. Calif. Weed Conf. 30:84-91. conversion is no longer done for livestock blue oak (Quercus douglasii) seedlings. Ecol. Murphy, A.H. and H.F. Heady. 1983. Vascular production will undoubtedly slow the 74: 68-82. plants of the Hopland Field Station, Mendocino spread and establishment of invasive Hamilton, J.G., C. Holzapfel, and B.E. County, California. Wasmann J. Biol. species. Mahall. 1999. Coexistence and interference 41:53-96. D.D. 1989. Hardwood rangelands are the most bio- between a native perennial grass and non- Murphy, and P.R. Ehrlich. Conservation biology of California's remnant logically rich habitat type in California native annual grasses in California. Oecologia 121:518-526. native grasslands. Pp 201-212. In: L.F. (Pavlik et al. 1991), making it essential Harrison, S.1999. Local and regional diversity Huenneke and H.F. Mooney (eds.), that we monitor changes to their flora over in a patchy landscape: Native, alien, and Grassland structure and function: California time. Continued surveys at the Center will endemic herbs on serpentine. Ecol. 80:70-80. annual grassland. Kluwer, Dordrecht. help verify local extinction and document

418 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 Pavlik, B.M., P.C. Muick, S.G. Johnson, and Randall, J.M., M. Rejmanek, and J.C. Vitousek, P.J. 1986. Biological invasions and M. Popper. 1991. Oak of California, Hunter. 1998. Characteristics of the exotic ecosystem properties: Can species make a Cachuma Press and the California Oak flora of California. Fremontia 26:3-12. difference? Pp. 163-178 In: H.A. Mooney Foundation. Los Olivos, Calif.184 pp. Rejmanek, M., C.D. Thomsen, and I.D. and J. Drake (eds.), Biological invasions of Peters, A., D.E. Johnson and M.R. George. Peters. 1991. Invasive vascular plants of North America and Hawaii. Springer-Verlag, 1996. Barb goatgrass: A threat to California California. Pp. 81-101. In: R.H. Groves, and New York, N.Y. rangelands. Rangelands 18(1): 8-10. F. di Castri (eds.). Biogeography of mediter- Pettit, N.E., R.H. Forend, and P.G. Ladd. ranean invasions. Cambridge University 1995. Grazing in remnant woodland vegeta- Press, Cambridge. tion: Changes in species composition and life form groups. J. Veg. Sci. 6:121-130. Skinner M.W. and B.M. Pavlik. 1994. Pitt, M.D., R.H. Burgy, and H.F. Heady. California Native Plant Society's Inventory 1978. Influences of brush conversion and of Rare and Endangered Vascular Plants of weather patterns on runoff from a northern California. Special Publication No.l (fifth California watershed. J. Range Manage. edition). CNPS. 31:23-27.

JOURNAL OF RANGE MANAGEMENT 55(4) July 419 J. Range Manage. 55: 420-423 July 2002 Fingerprint composition of seedling root exudates of selected grasses

JOHAN F. DORMAAR, BONNIE C. TOVELL, AND WALTER D. WILLMS

Authors are Soil Scientist (retired), Microbiologist, and Range Ecologist, Research Centre, Agriculture and Agri-Food Canada, P. 0. Box 3000, Lethbridge, Alberta, Canada TIJ 4B1

Abstract Resumen

The competitiveness of plants within a community is dictated La competitividad de las plantas dentro de una comunidad es to some extent by their association with microorganisms in the dictada en parte por su asociacion con los microorganismos del soil. That association is affected by root exudates and possibly by suelo. Esa asociacion es afectada por los exudados de la raiz y their quality. The competitiveness of species under various graz- posiblemente por su calidad. La competitividad de las especies ing regimes has been defined by their response to grazing as bajo varios regimenes de apacentamiento ha sido definida por su decreaser, increaser, or invader. To test the hypothesis that there respuesta al apacentamiento como especies decrecientes, cre- are recognisable differences in the chemical fingerprints of the cientes o invasoras. Para probar la hipotesis de que hay diferen- root exudates of decreasers, increasers and invaders, seeds of 8 cias reconocibles en la huella quimica de los exudados de la raiz grasses, representing these 3 designations, were germinated and de especies decrecientes, creciente a invasoras se tomaron semil- grown for 2 weeks in a root exudate trapping system in the labo- las de 8 zacates que representan estas 3 designaciones y se gemi- ratory. Tentative identification of the suite of compounds recov- naron y crecieron por dos semanas en un sistema de laboratorio ered from the root exudates by a solvent extraction technique que atrapaba los exudados de la raiz. La identificacion tentativa was done with the help of gas chromatography/mass spectrome- del grupo de compuestos de los exudados de la raiz recuperados try and authentic samples. Eleven identified compounds, present mediante una tecnica de extraccion con solventes se hizo con la in all exudates as major peaks, but absent in the blanks, were ayuda de cromatografia de gases y espectrofotometria de masas selected for semi-quantitatively comparing the 3 grazing y muestras autenticas. Once compuestos identificados, presentes response groups. For all 11 compounds, there was always at least en todos los exudados como picos principales, pero ausentes en 1 of the grazing response groups that had the highest percent- los blancos, se seleccionaron para comparar semicuantitativa- ages. That is to say, they were qualitatively, based on the 11 com- mente los 3 grupos de respuesta al apacentamiento. Para todos pounds selected, but not quantitatively similar. los 11 compuestos, siempre hubo al menos 1 de los grupos de repuesta al apacentamiento que tuvo los mas altos porcentajes. Es decir, ellos fueron cualitativamente, basados en los 11 com- Key Words: organic acids, native grasses, introduced grasses, puestos seleccionados, pero no cuantitativamente similares. soil quality, soil chemical properties

Rangeland plant species have been classified as decreasers, increasers, or invaders (Bedell 1998). Decreaser species often resist displacement by its associated climax dominant species. dominate the undisturbed grassland communities and normally Similarly, crested wheatgrass (Agropyron cristatum (L.) Gaertn.), have large plants. Decreaser species derive their classification an introduced species, is a strong competitor in Mixed Prairie from their response to grazing pressure. Increaser species are communities with the ability to invade and colonize undisturbed those that replace the decreasers and are more resistant to grazing native communities (Caldwell 1991, Trent et al. 1993). pressure while invader species are normally aliens that are capa- The competitiveness of plants within a community is dictated ble of competing in a grazed or ungrazed environment. to some extent by their association with microorganisms in the The consequence of replacing the dominant species in a plant soil such as the vesicular-arbuscular mycorrhizal type. That asso- community is a seral stage that can be resistant to succession ciation is affected by root exudates and possibly by their chemi- (Dormaar and Willms 1990, Dormaar et al. 1994). For example, cal quality (Bokhari 1978, Klein et al. 1988, Foster and Dormaar blue grama (Bouteloua gracilis (HBK.) Lag. Ex Steud) appears to 1991). Inhibitors of nitrifying bacteria were found in the root extracts of increases and invader species (Neal 1969). Conversely, Bremner The technical assistance of R. R. James with the initial stages of this research and McCarty (1993), based on an intensive project and of J. Elder who conducted the GC-MS analyses is gratefully literature survey, concluded that there was no satisfactory evi- acknowledged. dence to support the hypothesis that grasses inhibit nitrification in Mention of a trademark, proprietary product or vendor does not constitute a soils by exuding substances that retard oxidation of NH4+ by guarantee or warrantee of the product by Agriculture and Agri-Food Canada and nitrifying microorganisms. A later study by Neal (1973) showed does not imply its approval to the exclusion of other products or vendors that may also be suitable. that invader species had increased phosphatase activity, reflecting Lethbridge Research Centre Contribution No.387-0089 a potential for higher P uptake, in the surrounding soil as com- Manuscript accepted 17 Sept. 01.

420 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 pared to dominant, co-dominant, or Collection of Root Exudates umes of glass distilled water to eliminate increaser species. Seeds were placed in a thin bed of finely any water-soluble compounds. Next, the Organic acids play a prominent role in (particle size 0.5 to 1.0 mm) crushed lava beads were extracted with 5-bed volumes cell metabolism, affect rhizosphere pH, rock over a deeper bed of 2-cm size lava of glass distilled methanol. The methanol and microbial activity (Curl and Truelove rock. Blanks were the same but without the extract was reduced on a rotary evaporator 1986). They are good metal-chelating seeds. The lava rock was obtained from a (60° C, 30 kPa). The remaining 2 ml was compounds and aid in the absorption and garden centre. After emergence, they were adjusted to pH 2.0 with 3N HC1. Ten translocation of nutrient elements. Rice thinned down to aim for the same amount microlitres of 1-mg ml' methanol stock of (1984), after having summarised what (100 if possible; if less they were counted) suberic acid (octanedioc or 1,6-hexanedi- could be found in the literature, concluded of seedlings and allowed to grow for 2 carboxylic acid) were added to serve as that organic acids are important in weeks after germination. The seeds and internal standard. This mixture was then allelopathy. It was speculated that i) 3 plants were irrigated with 50-ml 0.1- extracted 3 times with 25-ml CH2C12, major long-chain fatty acids, i.e., palmitic, strength Hoagland solution. This was then dried over anhydrous Na2SO4, and stearic, and arachidic acids, identified circulated at the rate of 20-ml min' by air- reduced to 1-ml with a stream of N2 at from soil in rough fescue (Festuca lift. Water was added to compensate for room temperature. One half ml of the sam- campestris Rydb.) grasslands that were in losses due to aspiration and evaporation. ple was esterified by adding 1-ml 14% good condition because of low grazing The exudates were trapped with a con- boron trifluoride in methanol and placing pressure, could well contribute to the tinuous hydrophobic root exudate trapping it in a 60° C water bath for 10 min. resistance of those grasslands to the system (Tang and Young 1982). This tech- Following cooling, 1-ml saturated sodium encroachment of invading species nique was selected as a first approxima- chloride in glass distilled water was added. (Dormaar and Willms 1992) and ii) tion of either different compounds or the After shaking, 1-ml hexane was added. dioctyl adipate or di(2-ethylhexyl)phtha- same compounds at different quantities. After 10 min, 400µl of the hexane solution late could be linked to inhibition of nitrifi- To eliminate equipment related contami- was placed in a vial for analysis. cation (Dormaar et al. 1994). Neverthe- nants, a smaller, all glass and TeflonTM Initial analyses (1µl ) were carried out less, in spite of these roles and specula- system was designed. Prior to assemblage with a Hewlett Packard GC 5890 Series II tions, organic acids are rarely included in of the system, both the 2-cm size and fine- Plus, using a 30-m long capillary column lists of exudate components (Curl and ly crushed lava rock were heated to 900° (id. 0.320 mm) wall-coated with 5% Truelove 1986). C, while the all quartz glass wool was diphenyl-95% dimethyl polysiloxane (DB- Before establishing if there is a possible heated to 600° C. The quartz glass wool 5) with helium as the carrier gas (about 1 relationship between exudate composition was placed at the bottom of a 725-ml fun- ml min'). The operating conditions were: and the competitive status of plant species, nel-shaped growth reservoir. Next, the 2- 50° C for 1 min, then 25° C min-' to 100° fingerprint organic acid compositions, in cm size lava rocks were added which in C for 1 min, then 3° C min"' to 275° C for the form of gas chromatographic patterns, turn were covered with the finely crushed 1 min; the injection port temperature was of root exudates from the first 2 weeks of lava rock. The thistle of the funnel was 250° C, while the flame ionization detec- growth of a number of decreaser, increas- attached, via a taper-ground joint, to a 19 tor temperature was 275° C. The qualita- er, and invader grass species will have to x 150 mm glass tube. Root exudates were tive analyses were carried out with a be obtained. This would then allow testing collected on a 6-cm resin bed of Bio Hewlett Packard GC-MS data system. the hypothesis that there are recognisable BeadsTM 5M4 (20-50 mesh). Prior to use, Following tentative identification, the semi-quantitative differences in the chemi- the Bio BeadsTM were Soxhlet-extracted spectra of the major peaks selected were cal fingerprints of root exudates from (Tang and Young 1982) with acetone, ace- compared with authentic samples. plants that fall within these designations. tonitrile, and diethyl ether each for 24 Over 70 compounds, many in minute hours to eliminate any potential impurities quantities, were tentatively identified. Of in the product, and stored in glass-distilled these, only 11, which were present in all Materials and Methods methanol. All solvents were glass distilled exudates across the spectra but absent in OmniSolvTM grade. When in the collection the blanks, were selected for comparison. Species Selected cylinder, the Bio BeadsTM were rinsed They were then calculated as percent Seeds of 5 native and 3 introduced, with 10-bed volumes of double, glass dis- against the standard (octanediocic acid) invading grass species were collected in tilled water. The containers and collection and adjusted for 100% germination of 100 the field: Needle-and-thread (Stipa comata cylinders were wrapped in aluminum foil. seeds. Another way could have been to Trin. and Ruper.), Western wheatgrass Duplicate runs were made of a control, normalise the exudate composition by root (Agropyron smithii Rydb.), and rough fes- without plants, with each set of 4-plant weight. However, root washing, no matter cue were selected to represent decreaser systems. Runs were continued until 2 con- how carefully done, leads to losses of root species; blue grama and June grass secutive ones of a plant species were hairs and thus considered inappropriate. (Koeleria cristata (L.) Pers.) were selected obtained with good duplication. That is, The presence of seed remnants and the ash to represent increaser species; and crested considering all steps adding to errors and content of the root mass made weighing wheatgrass, Russian wildrye (Elymus the accumulative errors of the calcula- the burned lava rock before and after inap- junceus Fisch.) and timothy (Phleum tions, replicate runs with about 3% varia- propriate as well. In this study, precision pratense L.) were selected to represent tion were considered acceptable precision of the chromatographic output was consid- invader species. The seeds were collected for this study. ered of greater value in fingerprint separa- by glove-covered hands. They were then tion than accuracy. placed in paper bags and stored at room Analysis A test of the identified compounds temperature. To simulate field conditions, Following the 2-week growth period, among the grazing response groups was the seeds were not sterilised prior to use. the resin beds were rinsed with 5-bed vol- evaluated as a simple ANOVA where each

JOURNAL OF RANGE MANAGEMENT 55(4) July 421 Table 1. Tentatively identified methylated carboxylic acids (A) in root exudates collected over a 2-week period of growth following germination as determined by GL-MS.

Indigenous Decreaser Increasers Invaders Stipa Agropyron comata smithii campestris gracilis cristata cristatum junceus pratense Octanoic A 7' 15 8 Nonaoic A 27 29 1,2-Benzenedicarboxylic A2 29 25 9 Dodecanoic A 11 20 8 8 8 15 6 5 Tridecanoic A 32 29 4 6 Teradecanoic A 10 10 Pentadecanoic A 3 9 9 2-HexadecanoicA (formII) 45 18 1,2-Benzenedicarboxylic A2 39 48 7 E-9-Octadecenoic A 9 14 6 2The numbers represent the percent against the standard (Octanedioic A) and are adjusted for 100% germination of 100 seeds. Dimethyl and butyl-ethyl, respectively. species represented a replicate. Conse- selves through changes in quantity and did not elucidate why some plants are quently, the design was unbalanced with 3 quality of the chemical changes of root decreasers, while others are increasers or decreaser, 2 Increaser, and 3 invader repli- exudates (Dormaar and Willms 1995). invaders, even though they did have dif- cates. Treatment means were compared Although decreasers are plants that may ferent semi-quantitative gas chromato- with single degree of freedom contrasts. well succumb to grazing pressure, graphic fingerprint outputs. Without increasers and invaders may partially knowing what even the tentatively com- invade space vacated by the decreasers, pounds do in the rhizosphere, no conclu- Results and Discussion the chemical potential always exists that sions towards this can even be proposed. root exudates, albeit ephemeral and minor Neither may the identified compounds be in terms of contribution to total soil C, necessarily'in the soil as the same semi- All l l tentatively identified compounds may prevent decreasers from returning or quantitative fingerprint. For example, (Table 1) revealed separation (P < 0.01) plants in general from re-entering higher suc- Biondini et al. (1988) established that among the grazing response groups (Table cessional states. Callaway and Aschehoug microorganisms, in sterile and non-sterile 2) suggesting specific semi-quantitative (2000) found that diffuse knapweed's fritted clay used as growth media, signifi- fingerprint patterns. There was no consis- (Centaurea diffusa Lam.) advantage against cantly increased root exudation of A. tent ranking in the magnitude of the organ- North American species appeared to be cristatum and A. smithii, but had no effect ic compounds produced by each group. due to differences in the effects of its root on B. gracilis. Their data suggested that an Nevertheless, for all 11 compounds, there exudates and how these root exudates introduced plant species may be markedly 1 of the grazing was always at least affected competition for resources. No different from native species in the short- response groups that had the highest per- specific compounds were identified. grass steppe in terms of exudate releases. centages. In spite of all precautions, Dormaar et al. (1980) and Dormaar No specific compounds were identified. because of the polystyrene composition of (1982) demonstrated that gas chromato- This study was designed, as a first step, the Bio BeadsTM, the blank runs still had graphic patterns of solvent extractable to obtain a qualitative comparison of a It is, therefore, acknowledged that `noise.' organic acids from Chernozemic soils group of compounds identified in root by concentrating on the 11 major peaks could be related to the vegetation growing exudates collected over a 2-week period. not present in the `noise', the potential sig- on those soils. However, this present study Little, if any is known about the potential nificance of minute quantities of specific compounds, present in some but possibly absent in other root exudates, but masked Table 2. Statistical comparisons of tentatively identified methylated carboxylic acids (A) of root by the `noise', could potentially give qual- exudates among species groups (shown in Table 1). itative evidence for the decreaser/increaser/invader designation. Treatment Decreaser vs vs vs effect Increaser Invader Invader It is unfortunate that the level of 1,2-benzenedicarboxylic acid or phthalic acid ------Probability------must always be suspect when working with Octanoic A 0.008 0.003 Nonaoic A <0.001 <0.001 soil extracts. Due to the increased use of 1,2-Benzenedicarboxylic A' 0.009 0.007 plastics over the last 50 years (Dormaar Dodecanoic A 0.002 0.642 1982), phthalic acid has more or less Nonanedioic A 0.002 0.633 become a universal contaminant. However, Tridecanoic A <0.001 <0.001 0.001 under the conditions of the experiment the Teradecanoic A <0.001 <0.001 0.001 Pentadecanoic A <0.001 <0.001 is made that the phthalic assumption being 0.002 0.002 acid measured is a true product of the root 1,2-Benzenedicarboxylic A' 0.012 0.007 exudates obtained and processed, since it E-9-Octadecenoic A <0.001 0.017 was not present in the blanks. 'Dimethyl and Butyl-ethyl, respectively. Soil quality changes manifest them-

422 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 effect of the individual compounds identi- Biondini, M., D.A. Klein, and E.F. Redente. Dormaar, J.F., B.W. Adams, and W.D. fied. Although the bleeding of resins will 1988. Carbon and nitrogen losses through Willms. 1994. Effect of grazing and aban- have to be explored further, it is suggested root exudation by Agropyron cristatum, A. doned cultivation on a Stipa-Bouteloua com- that the next step ought to be to take the smithii and Bouteloua gracilis. Soil Biol. munity. J. Range Manage. 47:28-32. Biochem. 20: 477-482. Dormaar, A. 11 compounds identified and to test their J.F., Johnston, and S. Smoliak. Bokhari, U.G. 1978. Allelopathy among 1980. Organic solvent-soluble organic matter effect on some sensitive plants, such as prairie grasses and its possible ecological sig- from soils underlying native range and crest- lettuce or cucumber, as a first approxima- nificance. Ann. Bot. 42: 127-136. ed wheatgrass in southeastern Alberta, tion. However, even if this leads to further Bremner, J.M. and G.W. McCarty. 1993. Canada. J. Range Manage, 33:99-101. specific insights of the compounds identi- Inhibition of nitrification in soil by allelo- Foster, R.C. and J.F. Dormaar. 1991. fied, in the final analysis it may well be a chemicals derived from plants and plant Bacteria-grazing amoebae in situ in the rhi- combination, and most likely synergistic residues. In: J.-M. Bollag and G. Stozky zosphere. Biol. Fertil. Soils 11: 83-87. at that, of a number of factors, such as (eds.), Soil Biochemistry 8: 181-218. Marcel Klein, D.A., B.A. Frederick, M. Biondini, resource competition (moisture, nutrients), Dekker Inc., New York, N.Y. and M.J. Trlica. 1988. Rhizosphere aggressive growth behaviour, disease Caldwell, M. 1991. Underground events may microorganism effects on soluble amino be root of grass behavior. Utah Science 53 resistance, and chemical differences in acids, sugars and organic acids in the root (2): 51. zone of Agropyron cristatum, A. smithii and root exudates, that enable plants to act as Callaway, R.M., and E.T. Aschehoug. 2000. Bouteloua gracilis. Plant Soil 110:19-25. decreasers, increasers or invaders. In terms Invasive plants versus their new and old Neal, Jr., J.L. 1969. Inhibition of nitrifying of the study under discussion it can be neighbours: a mechanism for exotic invasion. bacteria by grass and forb root extracts. Can. concluded, as Dormaar and Willms (1990) Sci. 290: 521-523. J. Microbiol.15:633-635. did with NaOH-soluble organic acids Curl, E.A. and B. Truelove. 1986. The rhizos- Neal, Jr., J.L. 1973. Influence of selected between soils of needle-and-thread and phere. Springer-Verlag, Berlin, Germany. grasses and forbs on soil phosphatase activi- and Dormaar, J.F. 1982. Aliphatic carboxylic ty. Can. J. Soil Sci. 53:119-121. blue grama range, cropland, that the 2nd suite of constituents selected and identi- acids in Chernozemic soils. Can. J. Soil Sci. Rice, E.L. 1984. Allelopathy, ed. Academic 62:487-494. fied were qualitatively, but not semi-quan- Press, Inc., Orlando, Fla. Dormaar, J.F. and W.D. Willms. 1990. Effect Tang, C.-S. and C.C. Young. 1982. Collection similar. titatively of grazing and cultivation on some chemical and identification of allelopathic compounds properties of soils in the Mixed Prairie. J. from the undisturbed root system of bigalta Range Manage. 43:456-460. limpograss (Hemarthria altissima). Plant Literature Cited Dormaar, J.F., and W.D. Willms. 1992. Physiol. 69:155-160. Water-extractable organic matter from plant Trent, J.D., A.J. Svejcar, and G.J. Bedell, T.E. 1998. Glossary of terms used in litter and soil of rough fescue grassland. J. Bethlenfalvay. 1993. Growth and nutrition range management, 4th edition. Glossary Range Manage. 45:152--158. of combinations of native and introduced Update Task Group, Soc. Range Manage., Dormaar, J.F. and W.D. Willms. 1995. Root plants and mycorrhizal fungi in a semiarid Denver, Colo. mass and quality as a measure of range range. Agr., Ecosystems, Environ. 45: 13-23. health, p.121. In: Proc. 5`h Int. Rangeland Congress, Volume 1, Salt Lake City, Ut.

JOURNAL OF RANGE MANAGEMENT 55(4) July 423 Book Reviews

Restoration Ecology and Sustainable Development. Edited by be accomplished, and are being accomplished, in the field of Krystyna M. Urbanska, Nigel R. Webb, and Peter J. Edward. restoration ecology.-Tamara Boland, University of New 2000. Cambridge University Press, United Kingdom. 397 p. England, Armidale, New South Wales, Australia. US$39.95 paper. ISBN 0-521-59989. Throughout my formal education I have found, used, and been The Political Economy of Agricultural, Natural Resources, given numerous definitions of the terms ecosystem and ecosystem and Environmental Policy Analysis. By Wesley F. Peterson. processes. Many of these definitions have differed depending on 2001. Iowa State University Press, Ames, Iowa, USA. 373 p. the prevailing political environment, the country, the region, the US$69.95 hardbound. ISBN 0-8138-04329. context. Restoration Ecology and Sustainable Development This book, written by a professor of agricultural economics at resolves those inconsistencies. In this book, ecology is discussed the University of Nebraska, is intended to serve as a textbook for as a changing science, evolving as knowledge and the political upper-division courses in the economic analysis of agricultural situations surrounding our environment have developed. and natural resources policy. Part 1, comprised of 6 chapters, Part 1 introduces the meaning of restoration ecology in detail. establishes the theoretical foundations of policy analysis. Chapter Particular emphasis is placed on the phrase "restoration ecology" 1 categorizes situations requiring collective action and public pol- and the scope of its meaning. icy. Collective action then is analyzed in an economic `market' Part 2 follows this holistic approach to restoration ecology. Its setting (Chapter 2), and a political setting through an economic development over time is emphasized. Ecology has come from lens (i.e., public choice theory, Chapter 5). Chapters 3 and 4 being a barely recognized discipline to a multi-disciplinary, high- cover standard resource-economics topics including the funda- ly researched, recognized science. The authors and editors do not mental theorems of welfare economics, public goods, externali- try to offer the "right" solution per se, but give numerous ties, common pool resources, and perfect and imperfect competi- accounts of successes and failures in the restoration process. tion. Chapter 6 addresses philosophical issues providing the foun- Overall, they provide the reader with a better understanding of all dation for ethical considerations of economic justice and equality the encompassing factors involved in restoration ecology and sus- in formulating public policies. tainable development. Part II, comprised of 9 chapters, translates the theoretical foun- The contributing authors provide a range of examples from dation into practical tools for policy analysis and applies them to North America, Iceland, Australia, and Europe. Each of these case a number of case studies. Chapter 7 covers benefit-cost analysis, studies presents issues beyond that of restoration ecology, looking which is applied to 2 case studies in Chapter 8 (revitalization of at impacts on various temporal and spatial scales. Bakker et al. the banana industry in West Africa) and Chapter 9 (groundwater contribute an account of coastal salt marshes, outlining impacts contamination). Chapter 10 presents partial equilibrium analysis, from 700-500 BC to the present day. Throughout this discussion which is applied to case studies of the U.S. sugar program in are restoration successes and failures. This open-minded approach Chapter 11 and international trade conflicts over genetically mod- to restoration is a continual theme throughout the book. ified organisms in Chapter 12. Chapter 13 discusses multisector, Part 3 illustrates the need for recognizing restoration ecology systems, and economy-wide models. These tools are applied to research worldwide. Jonathon Major, the author of the Australian study the North American Free Trade Agreement in Chapter 14. case study, raises an interesting issue in Figure 11.2. This figure Chapter 15 concludes the book with a discussion of the place of is a graphical representation of the origin of references cited in policy analysis in the policy process. issues of Restoration in Ecology in America, Britain, Australia, In my opinion, this is an excellent book for use in advanced and elsewhere. This "concerning observation" shows an over- undergraduate and master's level courses in natural resource and whelming tendency to cite from only the country where the issue environmental economics. The book is well organized and the is authored. writing is very clear. Introductions to the 2 parts of the book and Jonathon Major, and the other contributing authors in Part 3 the constituent chapters provide students with a useful road map highlight the importance of looking beyond national borders. for the information to follow. The theoretical section in most text- Every continent is different, as is every state, county, and region. books can be rather dry, but the book avoids this problem by Each situation is unique. As a result of this uniqueness, situations interlacing the discussion with several interesting real-world poli- need to be treated individually. To learn from mistakes and bene- cy examples. The examples are presented in an evenhanded man- fit from others' knowledge, we must expand our resources. ner that allows students to evaluate both sides of complex issues. This concept of expansion is developed further in the final part The book's progression from theory to practical policy assess- of the book. Expansion or restoration is illustrated, and rightly so, ment tools applied to real-world case studies is a very effective as encompassing much more than just the ecology of an area. way to teach policy analysis. Many of the contributing authors give examples which illustrate The book's use of mathematics is also very effective. The the importance of incorporating ecological, economic, and social mathematics of the subject matter is nicely presented in shaded aspects. boxes and in appendices. This allows the material to be taught Without overloading the reader with extensive technical lan- equally well with or without the mathematics, depending on the guage and equations, the book balances between scientific analy- mathematical proficiency of the class. ses and management options. The management options are abun- In sum, the book effectively combines economics, philosophy, dant and described effectively. Restoration ecology is acknowl- political science, and mathematics to give students a very well- the economic analysis of agricultural and edged as a vast field encompassing a broad array of changing van - rounded presentation of ables. The book suggests and illustrates enormous potential for natural resource policy. The extensive use of case studies and in learn- recovery in a vast range of habitat and land types. In Restoration real-world examples gives students excellent experience Ecology and Sustainable Development, each author contributes an ing how theory and practical analytical tools can work together to important aspect of restoration and sustainable development. The produce meaningful analysis.-Ray Huffaker, Department of State University, Pullman, reader is left with a feeling of satisfaction that so many things can Agricultural Economics, Washington Washington.

424 JOURNAL OF RANGE MANAGEMENT 55(4) July 2002 INSTRUCTIONS FOR AUTHORS (from revised Handbook and Style Manual)

Eligibility received an unsatisfactory review may file an appeal with the Editor. The Editor then may select another Associate Editor to review the appeal. The Associate Editor reviewing the appeal will be provided with copies of an The Journal of Range Management is a publication for reporting and correspondence relating to the original review of the manuscript. If the documenting results of original research and selected invitational papers. appeal is sustained, a new review of the manuscript may be implemented Previously published papers are unacceptable and will not be considered at the discretion of the Editor. for publication. Exceptions to this criterion are research results that were Authors should feel free to contact the Associate Editor assigned to originally published as department research summaries, field station their manuscript at any stage of the review process: to find out where the reports, abstracts of presentations, and other obscure and nontechnical paper is in the process; to ask questions about reviewer comments; to ask handout publications. Manuscripts submitted to the JRM are the property for clarification or options if a paper has been rejected. of the Journal until published or released back to the author(s). Manuscripts may not be submitted elsewhere while they are being considered for this journal. Papers not accepted for publication are automat- Page Proofs ically released to the authors. Page proofs are provided to give the author a final opportunity to make Kinds of Manuscripts corrections of errors caused by editing and production. Authors will be charged when extensive revision is required because of author changes, Journal Articles report original findings in Plant Physiology, Animal even if page charges are not assessed for the article. One author per paper Nutrition, Ecology, Economics, Hydrology, Wildlife Habitat, will receive page proofs. These are to be returned to the Production Methodology, Taxonomy, Grazing Management, Soils, Land Recla- Editor, 3059A Hwy 92, Hotchkiss, Colorado 81419-9548, within 48 mation (reseeding), and Range Improvement (fire, mechanical chemical). hours after being received. If a problem arises that makes this impossible, Technical Notes are short articles (usually less than 2 printed pages) authors or their designates are asked to contact the Production Editor reporting unique apparatus and experimental techniques. By invitation of immediately, or production and publication may proceed without the the Editorial Board, a Review or Synthesis Paper may be printed in the author's approval of his edited manuscript. journal. Viewpoint articles or Research Observations discussing opinion or philosophical concepts regarding topical material or observational data Page Charges and Reprint Orders are acceptable. Such articles are identified by the word viewpoint or observations in the title. Authors are expected to pay current page charges. Since most research is funded for publication, it will be assumed that the authors are able to Manuscript Submission pay page charges unless they indicate otherwise in writing, when submitting a manuscript. When funds are unavailable to an author, no page Contributions are addressed to the Editor, Gary Frasier, Journal of charges will be assessed. Only the Editor will have knowledge of fund Range Management, 7820 Stag Hollow Road, Loveland, Colorado status of page charges; the Associate Editors and reviewers will accept or 80538. Manuscripts are to be prepared according to the instructions in the reject a manuscript on content only. Journal's Handbook and Style Manual. If the manuscript is to be one of An order form for reprints is sent to one author with the page proofs. a series, the Editor must be notified. Four copies of the complete manu- Information as to price and procedure are provided at that time. The min- script, typed on paper with numbered line spaces are required. Authors imum order is 100; no reprints are provided free of charge. We are also may retain original tables and figures until the paper is accepted, and send providing electronic reprints for a cost of $50.00. good quality photocopies for the review process. Receipt of all manuscripts is acknowledged at once, and authors are informed about subse- Basic Writing Style quent steps of review, approval or release, and publication. It Manuscripts that do not follow the directives and style in Journal hand- Every paper should be written accurately, clearly, and concisely. through materi- book will be returned to the authors before being reviewed. A manuscript should lead the reader from a clear statement of purpose should be reported number and submission date will be assigned when the paper is received als and methods, results, and to discussion. The data tables, drawings, and in the appropriate format. in coherent sequence, with a sufficient number of photographs to clarify the text and to reduce the amount of discussion. Review Tables, graphs, and narrative should not duplicate each other. Manuscript Authors should have manuscripts thoroughly reviewed by colleagues Manuscripts are forwarded to an Associate Editor, who usually obtains in their own institution and elsewhere before submitting them. Peer 2 or more additional reviews. Reviewers remain anonymous. These review before submission insures that publications will present signifi- reviewers have the major responsibility for critical evaluation to deter- cant new information or interpretation of previous data and will speed mine whether or not a manuscript meets scientific and literary standards. JRM review process. Where reviewers disagree, the Associate Editor, at his discretion, may Particular attention should be given to literature cited: names of obtain additional reviews before accepting or rejecting a manuscript. The authors, date of publication, abbreviations or names of journals, titles, Associate Editor may also elect to return to the author those manuscripts volumes, and page numbers. that require revision to meet format criteria before the Journal review. It is not the task of Associate Editors or Journal reviewers to edit poor- The Associate Editor sends approved manuscripts, with recommenda- ly prepared papers or to correct readily detectable errors. Papers not prop- tions for publication, to the Editor, who notifies the author of a projected erly prepared will be returned to the author. publication date. Manuscripts found inappropriate for the JRM are released to the author by the Associate Editor. Manuscripts returned to an author for revision are returned to the Associate Editor for final acceptability of the revision. Revisions not returned within 6 months, are considered terminated. Authors who consider that their manuscript has