R Ange Vhnag"Nt

Journal of Range Management, Volume 56, Number 2 (March 2003)

Item Type Journal; text

DOI 10.2458/azu_jrm_v56i2_board

Publisher Society for Range Management

Journal Journal of Range Management

Download date 04/10/2021 00:18:21

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

Version Final published version

Link to Item http://hdl.handle.net/10150/650597 Joum'af of Range Vhnag"nt

TABLE OF CONTENTS: VOL. 56, NO. 2, March 2003

FEATURE ARTICLE 106 State and transition modeling: An ecological process approach by Tamzen K. Stringham, William C. Krueger, and Patrick L. Shaver 114 Development and use of state-and-transition models for rangelands by Brandon T. Bestelmeyer, Joel R. Brown, Kris M. Havstad, Robert Alexander, George Chavez, and Jeffrey E. Herrick Grazing Management 127 Hay-meadows production and weed dynamics as influenced by management by Daniele Magda, Jean-Pierre Theau, Michel Duru, and Francois Coleno 133 Moderate and light cattle grazing effects on Chihuahuan Desert rangelands by Jerry Holechek, Dee Galt, Jamus Joseph, Joseph Navarro, Godfrey Kumalo, Francisco Molinar, and Milt Thomas Measurement/Sampling 140 A digital photographic technique for assessing forage utilization by P.W. Hyder, E.L. Fredrickson, M.D. Remmenga, R.E. Estell, R.D. Pieper, and D.M. Anderson 146 An index for description of landscape use by cattle by Haitao Zuo and Mary S. Miller-Goodman Hydrology 152 Hydrologic and sediment responses to vegetation and soil disturbances by J.H. Giordanengo, G.W. Frasier, and M.J. Trlica Improvement 159 Woody vegetation response to various burning regimes in South Texas by Donald C. Ruthven III, Anthony W. Braden, Haley J. Knutson, James F. Gallagher, and David R. Synatzske

Plant/Animal 167 Stocking rate effects on goats: A research observation by Miguel Mellado, Raul Valdez, Laura M. Lara, and Ramiro Lopez Published bimonthly-January, March, May, July, 174 Mineral concentration dynamics among 7 northern Great Basin September, November grasses by Dave Ganskopp and Dave Bohnert Copyright 2003 by the Society for Range Management Plant Ecology

INDIVIDUAL SUBSCRIPTION is by membership in 185 Vegetation dynamics from annually burning tallgrass prairie in the Society for Range Management. different seasons by E. Gene Towne and Ken E. Kemp LIBRARY or other INSTITUTIONAL SUBSCRIP- TIONS on a calendar year basis are $140.00 for Reclamation the United States postpaid and $165.00 for other countries, postpaid. Payment from outside the 193 Prescribed fire effects on dalmation toadflax by James S. Jacobs and. United States should be remitted in US dollars by international money order or draft on a New York Roger L. Sheley bank. 198 Overcoming dormancy in New Mexico mountain mahogany seed BUSINESS CORRESPONDENCE, concerning collections by Lee S. Rosner, John T. Harrington, David R. Dreesen, and subscriptions, advertising, reprints, back issues, and related matters, should be addressed to the Leigh Murray Managing Editor, 445 Union Blvd., Suite 230, Lakewood, Colorado 80228. Book Reviews EDITORIAL CORRESPONDENCE, concerning manuscripts or other editorial matters, should be 203 The Politics of Precaution: Genetically Modified Crops in Developing addressed to the Editor, Gary Frasier, 7820 Stag Countries By Robert L. Paarlberg; Invasive Exotic Species in the Sonoran Hollow Road, Loveland, Colorado 80538. Page proofs should be returned to the Production Editor, Region Edited by Barbara Tellman 3059A Hwy 92, Hotchkiss, CO 81419-9548..

INSTRUCTIONS FOR AUTHORS appear on the inside back cover of most issues. THEJOURNAL 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 HAFERKAMP H. RALPHS SAM ALBRECHT GERALD L. ANDERSON USDA-ARS USDA-ARS 445 Union Blvd., Ste 230 USDA-ARS Fort Keogh-LARRL Poisonous Plant Lab Lakewood, Colorado 80228 1500 N. Central Ave. 243 Fort Keogh Rd. E 1400 N (303) 986-3309 Sidney, Montana 59270 Miles City, Montana 59301 Logan, Utah 84341-2881 Fax: (303) 986-3892 e-mail address: R. JAMES ANSLEY ROBERT LYONS LARRY REDMON sam_albrecht@ rangelands.org Texas Agricultural Texas A&M University TAREC Editor/Copy Editor Experiment. Station Agricultural Research & P.O. Drawer E GARY FRASIER/JO FRASIER P.O. Box 1658 Extension Ctr. Overton, Texas 74684 7820 Stag Hollow Road Vernon , Texas 76385 P.O. Box 1849 Loveland, Colorado 80538 Uvalde, Texas 78802 NEIL RIMBEY University of Idaho e-mail address: KLAAS BROERSMA 16952 S. 10th Ave. [email protected] Agriculture Agri-Food ROBERT A. MASTERS and Canada Caldwell, Idaho 83607 Production Editor 3015 Ord Rd. Dow AgroSciences LLC 3618 South 75th Street PATTY RICH Kamloops, British Columbia CAROLYN HULL SIEG Lincoln, Nebraska 68506 Society for Range Management V2B 8A9 CANADA S.W. Science Complex 3059A Hwy 92 2500 South Pine Knoll Hotchkiss, Colorado 81419-9548 DAVID ELDRIDGE MITCHEL McCLARAN Flagstaff, Arizona 86001 e-mail address: Dept. Land and Water Conserv. University of Arizona [email protected] School of Geography 301 Biological Science East STEVE WARREN Book Review Editor University of New South Wales Tucson, Arizona 85721-0001 Colorado State University DAVID L. SCARNECCHIA Sydney, New South Wales Center for Environmental M. ANNE NAETH Dept of Natural Res. Sci. 2052 AUSTRALIA Management of Military Land University of Alberta Washington State University Dept. of Forestry Sciences Pullman, Washington 99164-6410 ROBERT GILLEN Dept. Renewable Resources Fort Collins, Colorado 80523 e-mail address: USDA-ARS 751 General Services Bldg. scarneda @ mail.wsu.edu Southern Plains Range Res. Sta. Edmonton, Alberta T6G 2H1 CANADA Electronic JRM Editor 200018th Street M. KEITH OWENS Woodward, Oklahoma 73801 Texas A&M University ROBERT PEARCE Research Center ELAINE E. GRINGS 5028 Highway 6 1619 Garner Field Road USDA-ARS Bishop, California 93514 Ulvade, Texas Fort Keogh-LARRL e-mail address: 243 Fort Keogh Rd. [email protected] Miles City, Montana 59301 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 President Society was changed in 1971 by amendment of the Articles of Incorporation. BOB BUDD The objectives for which the corporation is established are: Red Canyon Ranch 350 Red Canyon Rd -to properly take care of the basic rangeland resources of soil, plants, and water; Lander, Wyoming 82520-9417 develop an understanding of range ecosystems and of the principles applicable to the 1st Vice-President -to MORT KOTHMANN management of range resources; Texas A&M University assist all who work with range resources to keep abreast of new findings and Dept. Rangeland Ecology & Mgt. -to College Station, Texas 77843-0001 techniques in the science and art of range management; 2nd Vice-President -to improve the effectiveness of range management to obtain from range resources the ANGELA WILLIAMS products and values necessary for man's welfare; USDA-NRCS RR1 Box 108 -to create a public appreciation of the economic and social benefits to be obtained from Paolia, Oklahoma 73074-9772 the range environment; Executive Vice-President -to promote professional development of its members. SAM ALBRECHT 445 Union Blvd. Suite 230 Membership in the Society for Range Management is open to anyone engaged in or interested in Lakewood, Colorado 80228-1259 any aspect of the study, management, or use of rangelands. Please contact the Executive Vice- (303) 986-3309 President for details. Fax: (303) 986-3892 e-mail address: samalbrecht_rangelands.org Directors 2001-2003 is a publication of the Society for Range JOHN TANAKA The Journal of Range Management Eastern Oregon Agr. Res. Center-Union Management. It serves as a forum for the presentation and discussion of facts, ideas, and philosophies P.O. Box E pertaining to the study, management, and use of rangelands and their several resources. Accordingly, all Union, Oregon 97883 material published herein is signed and reflects the individual views of the authors and is not necessari- GREG TEGART ly an official position of the Society. Manuscripts from anyone-nonmembers as well as members-are BCMAFF welcome and will be given every consideration by the editors. Editorial comments by an individual are 1690 Powick Rd, Suite 2000 also welcome and, subject to acceptance by the editor, will be published as a "Viewpoint." Kelowna, BC V 1 X 7G5 CANADA In Cooperation With: Some of the articles appearing in The Journal of Range Management (JRM) 2002-2004 are presented in cooperation with The American Forage and Grassland Council (AFGC). JOHN MALECHEK A This cooperation consists of JRM acceptance of professional papers in forage grazing man- FG Utah State University agement and related subject areas from AFGC members and the appointment of 2 AFGC Dept. of Rangeland Resources UMC 5230 affiliated associate editors to JRM's Editorial Staff. The American Forage and Grassland Logan, Utah 84322-0001 Council Offices: P.O. Box 94, Georgetown, Texas 78627; Larry Jeffries, President; Dana Tucker, MARTIN VAVRA Executive Secretary. EOARC HC 71 Box 451, Hwy 205 Burns, Oregon 97720-9807 Society for Range Management may accept donations of real 2003-2005 Contribution Policy: The JEFF BURWELL and/or personal property subject to limitations set forth by State and Federal law. All donations shall be 655 Parfet Street subject to management by the Executive Vice President as directed by the Board of Directors and their Room E2000 discretion in establishing and maintaining trusts, memorials, scholarships, or other types of funds. Lakewood, Colorado 80215 Individual endowments for designated purposes can be established according to Society policies. Gifts, bequests, legacies, devises, or donations not intended for establishing designated endowments will be ALLEN RASMUSSEN College of Agr. & Human Sciences deposited into the SRM Endowment Fund. Donations or requests for further information on Society MSC 156 policies can be directed to the Society for Range Management, Executive Vice-President, 445 Union Kingsville, Texas 78363 Blvd., Suite 230, Lakewood, Colo. 80228-1259. We recommend that donors consult Tax Advisors in The term of office of all elected officers regard to any tax consideration that may result from any donation. and directors begins in February of each year during the Society's Annual Meeting.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 105 J. Range Manage. 56:106 -113 March 2003 State and transition modeling: An ecological process approach

TAMZEN K. STRINGHAM, WILLIAM C. KRUEGER, AND PATRICK L. SHAVER

Authors are assistant professor and professor, Department of Rangeland Resources, Oregon State University, Corvallis, Ore. 97331; and rangeland management specialist, USDA, Natural Resources Conservation Service, Grazing Land Technology Institute, Corvallis, Ore. 97331.

Abstract Resumen

State-and-transition models hold great potential to aid in Los modelos de estados-y- transicion presentan un gran understanding rangeland ecosystems' response to natural and/or potencial para ayudar a entender la respuesta de los ecosis- management-induced disturbances by providing a framework temas de pastizal a los disturbios naturales y/o inducidos por el for organizing current understanding of potential ecosystem manejo al proveer una estructura para organzzar el dynamics. Many conceptual state-and-transition models have conocimiento presente de las dinamicas del potential del ecosis- been developed, however, the ecological interpretation of the tema. Muchos modelos conceptuales de estados-y-transicion model's primary components, states, transitions, and thresholds, han sido desarrollados, sin embargo, la interpretacion ecologi- has varied due to a lack of universally accepted definitions. The ca de los componentes principales del modelo: estados, transi- lack of consistency in definitions has led to confusion and criti- ciones y umbrales han variado debido a la carencia de defini- cism indicating the need for further development and refinement ciones universalmente aceptadas. La falta de consistencia en las of the theory and associated models. We present an extensive definiciones ha conducido a confusion y critica indicando la review of current literature and conceptual models and point out necesidad de un mayor desarrollo y refinamiento de la teoria y the inconsistencies in the application of nonequilibrium ecology los modelos asociados. Nosotros presentamos una revision concepts. The importance of ecosystem stability as defined by the extensiva de la literatura actual y modelos conceptuales y pun- resistance and resilience of plant communities to disturbance is tualizamos las inconsistencias en la aplicacion de los conceptos discussed as an important concept relative to state-and-transition de la ecologia de no equilibrio. La importancia de la estabilidad modeling. Finally, we propose a set of concise definitions for del ecosistema, detinida como la resistencia y resilencia de las state-and-transition model components and we present a concep- comunidades vegetales a los disturbios, se discute como un con- tual model of state/transition/threshold relationships that are cepto importante relativo al modelaje de estados-y- transicion. determined by the resilience and resistance of the ecosystems' Finalmente, proponemos un grupo de definiciones concisas primary ecological processes. This model provides a framework para los componentes del modelo de estados-y-transicion y pre- for development of process-based state-and-transition models for sentamos un modelo conceptual de las relaciones de management and research. estados/transiciones/umbrales que estan determinadas por la resilensia y resistencia de los principales procesos ecologicos del ecosistema. Este modelo provee un marco para el desarrollo de Key Words: state, transition, threshold, modeling, ecological, modelos de estados-y-transicion basados en procesos para process manejo a investigacion.

Applied ecology disciplines, such as range management, are necessarily organized around a response model based on theoretical supposition. Thus, the litmus test for an ecological or mecha- 1986, Tausch et al. 1993). After 50 years of applying the quanti- nistic model is its ability to predict the consequences of natural tative climax model of Dyksterhuis (1949) to rangeland manage- disturbances and/or management activities with acceptable preci- ment its predictive capabilities have come under scrutiny. The sion over timescales relevant to management. Traditional theories inability of the model to incorporate multiple pathways of change of plant succession leading to a single climax community have has led some ecologists to abandon the model completely been found to be inadequate for understanding the complex suc- (Wilson 1984, Smith 1988). The recognition of this inadequacy cessional pathways of semi-arid and arid rangeland ecosystems has generated a search for an alternative theory that more correct- considering timescales important for making management adjust- ly reflects the observed dynamics of rangeland ecosystems. As ments (West 1979, Westoby 1980, Anderson 1986, Foran et al. many scientists were questioning the validity of the climax model, Westoby et al. (1989) developed a foundational discus- Authors wish to thank L.E. Eddleman, N.E. West, M. Westoby, W.A. Laycock, sion and conceptual model based on non-equilibrium ecology. and G.L. Peacock for insightful review. Numerous scientists have utilized these concepts as a basis for This article is submitted as Technical Paper No. 11874 Oregon Agricultural Experiment Station, Corvallis, Ore. the development of conceptual models of vegetation dynamics Manuscript accepted 8 Oct. 2002. which incorporate multiple successional pathways, multiple

106 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 steady states, thresholds of change, and discontinuous and irreversible transitions (Archer 1989, Friedel 1991, Laycock 1991, Fuhlendorf et al. 1996, Stringham 1996, Rietkerk and Van de Koppel 1997, Davenport et al. 1998, Oliva et al. 1998, Petraitis and Latham 1999, Plant et al. 1999, West 1999, West and Young 2000, Stringham et al. 2001). However, the ecological interpretation of Westoby's model has varied due to a lack of universally accepted definitions of the key concepts. The lack of consistency in definitions has led to confusion and criticism indicating the need for further development and refinement of the theory and associated models (Iglesias and Kothmann 1997). The USDA Natural Resources Conser- vation Service (NRCS) adopted the use of state-and-transition vegetation dynamics in describing rangeland ecological sites. The attempt to use this concept illustrated the in the and con- inconsistency definitions Fig. 1. Broad applications of the state-and transition concepts. Derived from the Society for cepts. The NRCS recognizes the need for Range Management, Task Group on Unity in Concepts and Terminology (1995). The consistency in the application of the con- plane labeled SCT (site conservation threshold) represents a change from 1 ecological site cepts (USDA 1997). For management to to another and may also be considered a threshold between 2 states. The individual boxes utilize the non-equilibrium ecological or ovals represent plant communities or seral stages that exist within 1 site. model the definitions of model objects must be succinctly stated and validated. change the system does not stabilize until As defined, Friedel's thresholds mirror the transition is complete. Westoby et al.'s (1989) definition of per- Quantitative approaches to ecological sistent or irreversible transitions. However, Background thresholds have been presented by May the use of thresholds in current state-and- (1977), Wissel (1984) and Rietkerk and transition models has not been consistent van de Koppel (1997). Archer (1989) nor clear on whether thresholds exist Westoby et al. (1989) was the first to introduced the qualitative concept of a between all states or only a subset of states. apply the use of state-and-transition termi- transitional threshold. He modeled the Conceptual models, based on these nology to non-equilibrium theory for the expansion of a woodland community into ideas, have incorporated states and transi- of producing a management purpose a grassland domain using a transitional tions but not always thresholds. As a focused model that describes vegetation threshold as the boundary between the result, there have been both a broad inter- dynamics in a non-linear framework as an respective grassland and shrub domains. pretation of states, more or less separated alternative to the linear continuum process Whisenant (1999) proposed a model of by thresholds, and a narrow interpretation incorporated in the quantitative climax degradation based on the stepwise degra- of states that approximate seral stages or model. The authors defined a "state" as an dation concept of Milton et al. (1994). phases of vegetation development. alternative, persistent vegetation commu- Whisenants model is similar to Archer's, Broadly applied, states are climate/soil/- nity that is not simply reversible in the lin- which incorporates 2 transition thresholds, vegetation domains that encompass a large ear successional framework. We interpret the first being controlled by biotic interac- amount of variation in species composi- Westoby's transitions as trajectories tions and the second by abiotic limitations. tion. Specifically a grassland state would between states with the characteristic of The concept of a transitional threshold as include many seral stages of the overall the transition being either transient or per- used by both Archer and Whisenant is sim- grassland community. These seral stages sisting. Transitions between states are ilar to the persistent transition as the suc- are within the amplitude of natural vari- often triggered by multiple disturbances cessional processes shift from grass con- ability characteristic of the state and repre- including natural events (e.g., climatic trolled to shrub controlled, however, in sent responses to disturbances that do not events or fire) and/or management actions Whisenant's (1999) model the focus is on force a threshold breach. Westoby et al. (grazing, farming, burning, etc.). ecological processes not vegetative groups. (1989), Archer (1989), and Archer and Transitions may occur quickly, as in the Friedel (1991) focused on the concept of Smeins (1991) provided examples of this case of catastrophic events like fire or thresholds of environmental change broad definition of state where domination flood or slowly over an extended period of between domains of relative stability. She of successional processes determine the time as in the case of a gradual shift in defined a threshold as a boundary in space boundary of the state (e.g. grass controlled weather patterns or repeated stresses like and time between 2 domains or states, succession versus shrub controlled succes- frequent fire. Regardless of the rate of which is not reversible on a practical time sion). The Society for Range Manage- scale without substantial inputs of energy. ment, Task Group on Unity in Concepts

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 107 and Terminology (1995) developed a graphical depiction of the broad applica- Late Seral tion of states with multiple vegetative Sagebrush Steppe stages diagrammed within one state (Fig. 1). Milton et al. (1994) and Whisenant (1999) de-emphasized the species component of the ecosystem within their models, Depauperate Late Native Perennial focusing instead on the functional integrity Seral Sagebrush and self-repair thresholds of the site for Grasses dominate Steppe determining state boundaries. In the broad definition of state the natural variability ---4- -t --- _ _ .Threshold characteristic of plant communities within ------a site is the result of, and contributes to, Introduced the current functional integrity of the site's Brush and grass pastures primary ecological processes (hydrology, introduced annuals nutrient cycling, and energy capture). The narrower interpretation of state allows for far less variation in plant com- 1 munity composition. States are typically Introduced grass depicted as seral stages or phases of vege- Cheatgrass and/or pastures with tation development. In the narrow applica- medusahead shrub reinvasion tion of the model a state change does not necessarily represent a movement across a threshold as envisioned by Friedel (1991). Figure 2 represents the narrow interpretation of states as adapted from West (1999). Boxes represent states and arrows indicate the transitions between states. Note that many of the transitions are reversible, however, the threshold indicates a persis- Fig. 2. Specific, or narrow, application of states with each state (box) representing 1 phase or seral stage of vegetation development. Transitions between states are indicated by tent transition. Other examples of specific arrows and the dashed line represents a threshold. The dashed transitional line signifies or narrow applications of states are pre- the requirement of substantial energy input to move the state back across the threshold. sented by Weixelman et al. (1997), Oliva Modified from West (1999) and West and Young (2000). et al. (1998), Allen-Diaz and Bartolome (1998), West (1999), and West and Young recover after it has been disturbed. Thus, dynamics. A state change, on the other (2000). The specific approach to state-and- fully functioning ecosystems are both hand, requires a shift across a boundary or transition modeling may be the reason for resistant to change and resilient or able to threshold, defined by a change in the statements that such models are structural- recover without external energy inputs integrity of the site's primary ecological ly similar to traditional linear climax-seral thereby maintaining stability while allow- processes, resulting in a different potential stage models. The significant difference ing for fluctuating combinations of plant set of plant communities. being the description of communities as species over time. States, by definition are discrete entities as opposed to the continu- relatively stable (Westoby et al. 1989), um concept of the quantitative climax therefore it follows that a state change is Rangeland Ecological Processes model (Iglesias and Kothmann 1997). only possible when a threshold is crossed. Accepting this concept points out the con- Ecological processes functioning within fusion that is apparent in the current a normal range of variation will support a Ecological Resistance and attempts to produce state-and-transition suite of specific plant communities. The Resilience models. The specific or narrow approach important primary processes are (1) has produced models, which depict state hydrology (the capture, storage, and redis- The concept of stability as defined by changes occurring without having crossed tribution of precipitation); (2) energy cap- the resistance and resilience of plant com- a threshold. Often such changes are dia- ture (conversion of sunlight to plant and munities have been discussed in the litera- grammed as reversible and perhaps occur animal matter); and (3) nutrient cycling ture for sometime and offer important without the input of management (the cycle of nutrients through the physical insights for state-and-transition models resources (Fig. 2). Rather than consider and biotic components of the environment (Margalef 1969, Verhoff and Smith 1971, these vegetation dynamics as state changes (Pellant et al. 2000, Whisenant 1999). Holling 1973, May 1977, Noy-Meir and it is more appropriate to consider them as Pellant et al. (2000) defines the function- Walker 1986). Resistance is defined as the phase shifts or plant community dynamics ing of an ecosystem by "the degree to ability of the system to remain the same within a state. Therefore, within a state which the integrity of the soil, vegetation, while external conditions change whereas there exists the potential for a large varia- water, and air, as well as the ecological resilience is the ability of the system to tion in species composition, which is processes of the rangeland ecosystem, are merely a reflection of plant community

108 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 balanced and sustained". Integrity is Temporal Scale Vegetation Structure: a component defined as the "maintenance of the func- The definition of threshold as presented resulting from above ground communi- tional attributes characteristic of a locale, by Friedel (1991) indicates that once a ties of living organisms, whose vital including normal variability" (Pellant et threshold has been breached return to the attributes (Noble and Slatyer 1980) al. 2000). Degradation of an ecosystem previous state is precluded within a time competitively capture and utilize the occurs when the integrity of the system is frame relevant to management, without system's available energy, water, nutri- damaged or lost. Maintenance of a func- substantial inputs of energy. Ecological ents, and space. tional site or repair of a damaged site management models should focus on the The interaction between the structural requires management focused on soil sta- time required to repair damaged ecological attributes of soil and the vegetative commu- bility, nutrient cycling, and the capture, processes not on a time scale predicated nities, through the processes of energy cap- storage and safe release of precipitation. by management. Careful consideration of ture, hydrology and nutrient cycling defines Vegetation goals should be based on the the threshold concept negates the need for the resilience and resistance of the state. concept of vegetation as a tool for main- including management timescales in the taining or repairing damaged ecological definition of ecological thresholds as these Resilience and Resistance processes rather than predefined species thresholds represent a permanent change The stability of a state is defined above groups. Monitoring of species groups may in the function of the state. Thus, restating in terms of resilience and resistance. be a mechanism for evaluating or detecting the threshold definition, independent of Resilience and resistance are inherent change in the site's ecological processes. management timescales, results in the con- properties of an ecosystem that are deter- clusion that once a threshold has been vio- mined by the physical components of the lated return to the prior state is precluded system and the functional capacity of the Clarification of the Concepts without substantial inputs of energy. and Definitions associated ecological processes. Resilience Therefore, under the current climatic con- focuses on how far a system can be dis- ditions and without substantial inputs of placed from equilibrium before return to Spatial Scale energy, state changes are permanent. The equilibrium is precluded. The emphasis is Ecosystems are difficult to define or temporal scale is defined by the perma- placed on the persistence of relationships delimit in space and time. Hierarchy theo- nence of the current climate regime. as they affect the systems ability to adapt ry, as applied to ecological systems, sug- to change (Walker et al. 1981), therefore, gests several levels of organization exist, State resilience relates to the functioning of the i.e., organisms, populations, communities, A state is a recognizable, resistant and system's ecological processes. Resistance ecosystems, landscapes (Archer and resilient complex of 2 components, the indicates the ability of a system to remain Smeins 1991). Each level of organization soil base and the vegetation structure. The at or near its equilibrium condition by encompasses one or more of the primary vegetation and soil components are neces- maintaining control of its ecological ecological processes that are operating at sarily connected through integrated eco- processes. Thus, the strength of this con- specific spatial and temporal scales. logical processes that interact to produce a trol determines a system's inherent resis- Although landscape scale management sustained equilibrium that is expressed by tance to change. Consequently, under an may be the goal, our current understanding a specific suite of vegetative communities. existing climate, stability of a state is a of organization function declines with function of the combination of its inherent increasing spatial and temporal scale. resilience and resistance. The ecological site concept has long Soil Base and Vegetation Structure been utilized as an organization level that The base of any rangeland ecosystem is provides an appropriate spatial scale for the soil resource that has developed Thresholds and Transitions inventory, evaluation, and management of through time from a specific parent materi- Thresholds are points in space and time rangelands (USDA 1997). Organisms, al, climate, landscape position, and interac- at which one or more of the primary eco- populations, and communities exist within tion with soil and terrestrial biota. These logical processes responsible for maintain- this spatial scale and interact with one factors are the primary determinants of the ing the sustained equilibrium of the state another through the flow of water and ecological site's capability. The integrity degrades beyond the point of self-repair. energy, and the cycling of nutrients. An of the soil resource, as reflected by site These processes must be actively restored before the return ecological site has evolved a kind of char- hydrology and nutrient cycling, is directly to the previous state is possible. In acteristic plant community such as cool connected to the composition and energy the absence of active restora- season shrub-grass or warm season grass- capture process of the above-ground vege- tion a new state, which supports a differ- land. Within an ecological site numerous tative component. The interaction between ent suite of plant communities and a new expressions of the various developmental the soil resource and the associated vegeta- threshold, is formed stages of the characteristic plant communi- tive community determines the functional Thresholds: boundary in space and time ty can occur. The concept and definition status of the state's ecological processes. between any and all states, or along irre- of an ecological site fits the large-scale Soil Base: a component that results from versible transitions, such that one or interpretation of the state-and-transition the interaction of climate, abiotic soil more of the primary ecological processes model. We define the ecological site as the characteristics, soil biota and topogra- has been irreversibly changed and minimum scale for definition of a state. phy that determines the hydrologic char- must be actively restored before return acteristic and biotic potential of the sys- to a previous state is possible. tem. Transitions are trajectories of change that are precipitated by natural events

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 109 and/or management actions which degrade the narrow application of the non-equilib- Milton et al. 1994, and Whisenant 1999). the integrity of 1 or more of the states pri- rium approach to states and transitions Plant community phase changes within mary ecological processes. Transitions are (Fig. 5). States are diagrammed as the states, in addition to transitions of change, often composed of 2 separate properties large boxes and are bordered by thresh- thresholds and multiple stable states are that are defined by the state threshold. The olds. Thresholds are the boundaries of any illustrated in Figure 5. The management first property is reversibility and it occurs and all states, but may also occur during and natural mechanisms responsible for within the state. The second property is the transition between states. For a state community phase shifts and transition ini- irreversibility and it occurs once a thresh- change to occur a threshold must be tiation must be defined in terms of ecolog- old has been breached. Transitions are breached. The small boxes within the state ical processes and included in the model vectors of system change that will lead to are referred to as plant community phases description. For example, prolonged a new state without removal of the stres- or seral stages and are joined by communi- drought or overgrazing leads to a reduc- sor(s). The primary difference between the ty pathways that flow in both directions. tion in the perennial herbaceous understo- reversible and irreversible property of a Transitions are reserved for a trajectory of ry. The decrease in perennial understory transition is defined by the systems' abili- change with the dashed line inside the leads to a decrease in total energy capture ty or inability to repair itself. state indicating the portion of the transi- and nutrient cycling. In addition, the plant Transition: a trajectory of system change tion that is reversible with minimal input community's ability to protect the soil away from the current stable state that is from management. Figure 4 illustrates the from raindrop impact and potential soil triggered by natural events, management process of a state change. Once the thresh- erosion declines. The mechanism (or actions, or both. old is crossed the state has lost control of mechanisms) of disturbance have led to a - Reversible Property of the Transition: its primary ecological processes, is no change in the 3 primary ecological trajectory of change that occurs within a longer able to self-repair and will transi- processes and a phase shift as diagrammed state and indicates the system is moving tion to a new equilibrium with a different by community phase pathway P1 (Fig. 5). toward a threshold. Reversal requires ecological capability. The entire trajectory In the case of prolonged drought return to elimination of the stress or stresses from a vegetation phase in State 1, across the late seral sagebrush steppe phase responsible for triggering the transition. the threshold to the formation of State 2 is would gradually occur with a return to a - Irreversible Property of the Transition: considered a transition and represents a normal or above normal precipitation peri- trajectory of change that occurs after a degradation of ecological capability. The od (P2). Increased available moisture threshold has been breached. The sys- portion of the transition contained within leads to an increase in biomass of the tem can no longer self-repair even with the boundary of State 1 is reversible with herbaceous understory that translates into removal of the stressor(s). The system removal of the stressor(s), however, once an increase in energy capture, nutrient will not come to rest until a new equilib- the trajectory crosses the threshold it is not cycling and an improvement in soil pro- rium (i.e., new state) is established that reversible without active restoration tection and site hydrology. The degrada- supports a different suite of plant com- including substantial energy input. tion mechanism of overgrazing would munities. Additional thresholds may occur while the need to be addressed through grazing system is in transition, changing the direc- management with the goal of improving tion of the trajectory away from State 2 the function level of the primary ecologi- Model Structure towards State 3 (Fig. 4). State-and-transi- cal processes. Continued overgrazing tion modeling efforts indicate the first would further decrease the vigor of the The conceptual model, illustrating the threshold is forced by a change in the biot- native herbaceous understory and further above definitions, is represented in ic component of the system whereas addi- impact the community's ability to main- Figures 3 and 4. The model accommodates tional thresholds would involve changes in tain control of the primary ecological both the quantitative climax approach and the soil resource (Westoby et al. 1989, processes. As the vigor of the native

Objects within a State

Reversible transition b

Threshold of the state Plant community phases or seral stages within a state

Community pathways i i

Fig. 3. Conceptual model depicting the objects of 1 state. Note the linear response, retrogression-succession model may be modeled within the state (i.e., a to b to c and vice-versa).

110 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Threshold

State 1'

State 2 Threshold

Reversible transition Community Pathway

Irreversible transition

Community Phases or a seral stages within a state

Fig. 4. Conceptual state-transition model incorporating the concepts of community pathways between plant community phases within states, reversible transitions, multiple thresholds, irreversible transitions, multiple pathways of change, and multiple steady states.

herbaceous community declines, the site is reduction in the amount of organic materi- Conclusions opened up for invasion by annual species. al being added to the soil and an increase The transition from State 1 towards State 2 in the potential for damage to soil surface Definitions and model concepts as dis- has begun and will continue without the structure from raindrop impact. Return to cussed in this paper are being adopted by removal of the stress from improper graz- State 1 would be impossible without the the USDA Natural Resources Conserva- ing (Tla). At the point in time where use of intensive management inputs. The tion Service as the standard for describing annuals dominate the herbaceous under- practicality of this level of management vegetation dynamics in rangeland ecologi- story and fire frequency intensifies, the would preclude its use. State 3 may be the cal site descriptions. State-and-transition state has crossed a threshold and is transi- practical state of choice. models hold great potential to aid in tioning to a new state (Tlb). During this Although many scientists have recog- understanding rangeland ecosystems' transition phase the plant community may nized the short-comings of the quantitative response to natural and/or management- still retain a minor component of sage- climax model developed by Dyksterhuis induced disturbances by providing a brush; however, this is not representative (1949) there are ecosystems, generally of framework for organizing understanding of a stable state and with increased fire more mesic climates, where the linear of potential ecosystem dynamics. Many frequency the brush will be eliminated and model is appropriate. It is important to real- state-and-transition model applications are the new equilibrium state formed. The ize that any modeling approach is a best-fit available in the literature, although the new state is defined as a Bromus tectorum solution, not a perfect-fit solution. scale of interpretation of the concepts has (cheatgrass) and/or Taeniatherum asperum Therefore, the retrogression-succession varied. We have attempted to review and (medusahead) dominated community with continuum can be modeled within the states clarify a large amount of information into a fire frequency interval of 2 to 3 years. to depict the situation where plant commu- a proposed conceptual model of state/tran- Energy capture has declined and the time nity phases do respond linearly. However, sition/threshold relationships that are period for energy capture has been it is also possible for linear response mech- determined by the resilience and resistance reduced. Nutrient cycling in both the verti- anisms to be pushed past an ecological of the systems' primary ecological cal and horizontal plane has decreased threshold, resulting in a state change. processes. Most of the components pre- with the shift to a shallow rooted, primari- sented are not new; however, the proposed ly monoculture community. The hydrolo- model attempts to clarify the definitions gy of the site will be impacted through a

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 111 Anderson, J.E. 1986. Development and structure of sagebrush steppe plant communities. Late Seral In.. P.J. Joss, P.W. Lynch, and O.B. Williams (eds.) Rangelands: A resource under siege - 2nd Sagebrush Steppe State 1 Proc. of Internat. Rangeland Congress. Aust. Acad. of Sci., Canberra, Aust. Archer, S. 1989. Have southern Texas savan- Pl P2. nas been converted to woodlands in recent I history? The Amer. Natur.134:545-561. Native Perennial Archer, S. and F.E. Smeins. 1991. Grasses dominate Ecosystem-level processes, p. 109-139. In: Depauperate Late K. Rodney and Jeffery W. Stuth (eds.) Seral Sagebrush Grazing Management an Ecological Steppe I, Perspective. Timber Press, Portland Ore. Davenport, D.W., D.D. Breshears, B.P. Wilcox, and C.D. Allen. 1998. Viewpoint: Sustainability of pinon-juniper ecosystems - a unifying perspective of soil erosion thresholds. J. Range Manage. 51:231-240. Tlb Dyksterhuis, E.J. 1949. Condition and management of rangeland based on quantitative Introduced ecology. J. Range Manage. 2:104-115. grass pastures Foran, B.D., G. Bastin, and K.A. Shaw. 1986. Range assessment and monitoring in arid lands: the use of classification and ordination Cheatgrass and/or in range survey. J. Environ. Manage. 22: medusahead 67-84. l r Friedel, M.H. 1991. Range condition assessment and the concept of thresholds: A view- I Introduced State 2 grass pastures with point. J. Range Manage. 44(5):422-426. Fuhlendorf, S. D., F. E. Smeins, and W. E. shrub reinvasion Grant. 1996. Simulation of a fire-sensitive ecological threshold: a case study of Ashe State 3 juniper on the Edwards Plateau of Texas, USA. Ecol. Modelling 90:245-255. Holling, C. S. 1973. Resilience and stability of 9 ecological systems. Ann. Review Ecol Systematics 4:1-23. Iglesias, R.M.R. and M.K. Kothmann.1997. Structure and causes of vegetation change in Fig. 5. Modification of the West (1999) and West and Young (2000) specific sagebrush steppe state and transition model applications. J. model (see Fig. 2) to illustrate the broad concept of state with plant community phases and Range Manage. 50(4):399-408. community pathways (i.e., P1 and P2) within states. Tla and Tlb signify the reversible and Laycock, W.A. 1991. Stable States and thresh- irreversible properties of the transition between State 1 and State 2. For additional discus- olds of range condition on North American sion of the mechanisms leading to community phase shifts see West (1999) and West and rangelands: A viewpoint. J. Range Manage. Young (2000). 44(5):427-433. Margalef, R. 1969. On certain unifying principles in ecology. Amer. Natur. 97:357-374. and concepts and to link them together that the ecological site is the minimum May, R. M. 1977. Thresholds and breakpoints into a process-based model for manage- scale associated with a state, understand- in ecosystems with a multiplicity of stable ment and research. The management and ing ecological processes at the landscape states. Nature 269:471-477. Milton, S. W. R. Dean, M. A. duPlessis, This model con- J., J. natural mechanisms responsible for com- scale should be the target. and W. R. Siegfried. 1994. A conceptual munity phase shifts and transition initia- tains the flexibility to accommodate land- model of arid rangeland degradation. Biosci. tion must be included in the model scape level dynamics; however, further 44:70-76. description. The description of these research is needed to clarify the ecological Noble, I. R. and R. 0. Slatyer.1980. The use mechanisms should contain information relationships occurring at that scale. This of vital attributes to predict successional changes in plant communities subject to on their impact on the primary ecological effort is not viewed as completed, but recurrent disturbances. Vegetatio 43:5-21. processes and the resulting change in the rather as another step in the process to fur- Noy-Meir, I. and B.H. Walker. 1986. biotic community and system function. ther develop understanding of rangeland Stability and resilience in rangelands. p. Further research is needed to identify indi- ecosystems. 21-25. In: P.J. Joss, P.W. Lynch, and O.B. Williams. (eds.) Rangelands: a resource cators of change for ecological processes 2nd under siege - Proc. of the Internat. that will allow management to intervene Rangeland Congress. Australian Acad. of prior to a threshold change. Once a thresh- Literature Cited Sci.. Canberra, Aust. old has been crossed, the focus of manage- Oliva, G., A. Cibils, P. Borrelli, and G. ment should be on restoration of the dam- Allen-Diaz B. and J.W. Bartolome. 1998. Humano. 1998. Stable states in relation to aged ecological processes, not on reestab- Sagebrush-grass vegetation dynamics: com- grazing in Patagonia: a 10-year experimental trial. J. Arid Environ. 40:113-131. lishing a specific plant community. paring classical and state-transition models. Although this conceptual model suggests Ecological Applications 8(3):795-804.

112 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Pellant, M. P. Shaver, D. A. Pyke, and J. E. Stringham, T.K., W.C. Krueger, and P.L. West, N. E. 1999. Managing for biodiversity of Herrick. 2000. Interpreting indicators of Shaver. 2001. States, transitions, and thresh- rangelands. p. 101-119. In: W. Collins, C. rangeland health, ver. 3, Tech. Reference olds: Further refinement for rangeland appli- and Qualset (eds.) Biodiversity in 1734-6. USDI, BLM, Nat. Sci. and Tech. cations. Agr. Exp. Station Special Rep. 1024. Agroecosys-tems. CRC Press Washington, Center, Denver, Colo Oregon State Univ., Corvallis, Ore. D.C. Petraitis, P.S. and R. E. Latham. 1999. The Tausch, R.J., P.E. Wigand, and J.W. West, N. E. and J. A. Young. 2000. Inter- importance of scale in testing the origins of Burkhardt.1993. Viewpoint: plant commu- mountain valleys and lower mountain slopes alternative community states. Ecol. 80(2): nity thresholds, multiple steady states, and p. 256-284 In: Barbour, M. G. and W. D. 429-442. multiple successional pathways: legacy of Billings (eds.) North American Terrestrial Plant R.E., M.P. Vayssieres, S.E. Greco, the Quaternary? J. Range Manage. Vegetation, Second Edition. Cambridge M.R. George, and T.E. Adams. 1999. A 46:439-47. Univ. Press, New York, N.Y. qualitative spacial model of hardwood range- USDA, Natural Resources Conservation Westoby, M. 1980. Elements of a theory of land state-and-transition dynamics. J. Range Service. 1997. National Range and Pasture vegetation dynamics in rangelands. Israel J. Manage. 52:51-59. Handb.: USDA. Washington D.C. Bot. 28:169-194. Rietkerk, M. and J. van de Koppel. 1997. Verhoff, F. H. and F. E. Smith. 1971. Westoby, M., B. Walker, and I. Noy-Meir. Alternate stable states and threshold effects Theoretical analysis of a conserved nutrient 1989. Opportunistic management for range- in semi-arid grazing systems. Oikos ecosystem. J. Theor. Biol. 33:131-147. lands not at equilibrium. J. Range Manage. D. Ludwig, C.S. Holling, and 79:69-76. Walker, B.H., 42(4):266-274. of semi-arid Smith E.L. 1988. Successional concepts in R.M. Peterman.1981. Stability S. G. 1999. Repairing damaged , grazing systems. Ecol. 69:473-498. relation to range condition assessment. p. savannah A process-oriented, landscape- Weixelman, D.A., D.C. Zamudio, K.A. Univ. Press. 312 . Vegetation approach. Cambridge 113-133 In: P.T. Tueller (ed.) and R.J. Tausch. 1997. science applications for rangeland analysis Zamudio , pp. Classifying ecological types and evaluating A.D. 1984. What is range condition: and management . Kluwer Academic degradation. J. Range Manage. 50(3): of concepts in Australia. Bull. Publishers , Boston , Mass. Management Task Group Soc. Amer. 65:171. Society for Range , E., 1979. Basic synecological rela- Terminology. West, N. C. 1984. A universal law of the charac- on Unity in Concepts and tionships of sagebrush-dominated lands in 1995 . New concepts for assessment of range- return time near thresholds. Great Basin and the Colorado Plateau. In: 65:101-107. land condition . J . Range Manage. Anon, The sagebrush ecosystem: A sympo- 48:271-282. sium. Utah State University, College of Stringham, T.K. 1996. Application of non- Natural Resources, Logan, Utah. equilibrium ecology to managed riparian ecosystems. Ph.D. Diss.. Dept. of Rangeland Resour., Oregon State Univ., Corvallis, Ore. pp.156.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 113 J. Range Manage. 56:114-126 March 2003 Development and use of state-and-transition models for rangelands

BRANDON T. BESTELMEYERI, JOEL R. BROWN2, KRIS M. HAVSTAD3, ROBERT ALEXANDER4, GEORGE CHAVEZ5, AND JEFFREY E. HERRICK6

'Ecologist, USDA-ARS, Jornada Experimental Range, New Mexico State University, Las Cruces, N.M. 88003, ZCooperating Research Scientist USDA- NRCS, Jornada Experimental Range, New Mexico State University, Las Cruces, N.M. 88003, 3Supervisory Scientist, USDA-ARS, Jornada Experimental Range, New Mexico State University, Las Cruces, N.M. 88003, 4State Range Specialist, Bureau of Land Management, 1474 Rodeo Rd., Santa Fe, N.M. 87502, 5State Rangeland Management Specialist, USDA Natural Resources Conservation Service, 6200 Jefferson, Albuquerque, N.M. 87109. 6Soil Scientist, USDA-ARS, Jornada Experimental Range, New Mexico State University, Las Cruces, N.M. 88003

Abstract ships can be readily understood. A knowledge of mechanisms is closely related to the use of ecological indicators to anticipate transitions. We conclude that models should include 1) reference State-and-transition models have received a deal great of atten- values for quantitative indicators, 2) lists of key indicators and tion since the introduction of the concept to range management descriptions of changes in them that suggest an approach to a in 1989. Nonetheless, only recently have sets of state-and-transi- transition, and 3) a rigorous documentation of the tion models been produced theory and that can be used by agency personnel assumptions (and their alternatives) underlying and private citizens, the structure of and there is little guidance available for each model. developing and interpreting models. Based upon our experiences developing models for the state of New Mexico, we address the following questions: 1) how is information assembled to create site-specific models for entire regions, 2) what ecological issues Key Words: community stability, ecological sites, ecosystem should be considered in model development and classification, health, indicators, New Mexico, vegetation dynamics and 3) how should models be used? We review the general structure of state-and-transition models, emphasizing the distinction between changes among communities within states (pathways) that are reversible with changes in climate and "facilitating prac- Resumen tices" (e.g. grazing management), and changes among states (transitions) that are reversible only with "accelerating prac- Los modelos tipo tices" such as seeding, shrub control, or the recovery of soil sta- del de estado y transicion han recibido mucha bility and historical hydrologic function. Both pathways and atencion desde su introduccion en 1989. Sin embargo hasta hace transitions occur, so these models are complementary. Ecological poco tiempo se cuento con diferentes modelos de `estado y transi- sites and the climatically-defined regions within which they occur cion' que pueden ser usados por agencias gubernamentales y (land resource units) serve as a framework for developing and particulares. Existen pocos lineamientos disponibles para su selecting models. We illustrate the importance of clearly delineat- desarrollo a interpretacion. Basados en nuestra experiencea en el desarrrollo de estos modelos ing ecological sites to produce models and describe how we have para el estado de Nuevo Mexico, dealt with poorly-delineated sites. Producing specific models hacemos las siguentes preguntas: 1) como debe ensamblarse la requires an understanding of the multiple ecological mechanisms informacion para crear modelos para sitios especificos en grandes 2) underlying transitions. We show how models can represent and regiones? cuales aspectos ecologicos deben ser con- distinguish alternative and complementary hypotheses for transi- siderados en el desarollo del modelo y su clasificacion? y 3) tions. Although there may be several mechanisms underlying como deben usarse estos modelos ? Revisamos la estructura gen- transitions, they tend to fall within discrete categories based eral de los modelos de `estado y transicion' enfatizando los cambios upon a few, fundamental ecological processes and their relation- entre comunidades vegetales dentro de estados que son reversibles con los cambios en el clima y practices facilitadoras (i.e. manejo de pastizales). Asi como cambios entre estados (tran- We thank several agencies, institutions and individuals who contributed data, siciones) que son reversibles solamente mediante practicas acel- knowledge, and criticisms that are represented in this manuscript. From the BLM, eradoras como la resiembra, control de arbustos, y recuperacion Phil Smith, Lane Hauser, Rusty Stovall, Mike Ramirez, Susan Britt, and John de la estabilidad y funcion hidrologica del suelo. las regiones Spain; from the NRCS, David Trujillo, Gene Adkins, Garth Grizzle, Darrel climaticamente definidas (unidades de recursos terrestres) los Reasner, Tim Henry, Frank Corn, Bill Schwebke, Arlene Tugel, y and Pat Shaver; sitios ecologicos dentro de ellos sirven como estructura el from the U. S. Forest Service, Wayne Robbie and Ralph Pope, from USDA-ARS, para Bob Gibbens, from New Mexico State University, Gary Donart and Barbara desarollo y seleccion de los modelos. Tambien, ilustramos la Nolen; and private citizens, Jim Powell and Lee Gile. Alfonso Serna-Perez and importancia de delinear claramente los sitios ecologicos para Isaac Reyes-Vera translated the abstract. We thank Sam Fuhlendorf, George producir los modelos y describir como hemos resuelto el proble- Peacock, Nathan Sayre, Mark Westoby, 2 anonymous reviewers, and especially ma de los sitios pobremente delineados. Para producir modelos Mitch McClaran for many valuable suggestions. That should not imply that these especificos se requiere la comprension de los mecanismos ecologi- individuals agree with all that we have written, however. cos que Manuscript accepted 21 May 02. determinan las transiciones. Adicionalmente, mostramos como los modelos pueden representar y distinguir hipotesis alter-

114 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 nativas ylo complementarias para explicar tion in determining community patterns ecosystem change and the roles that man- las transiciones. Aun cuando puede haber (Kingsland 1985), paved the way for agement can play in directing these varios mecanismos para explicar transi- Westoby et al. (1989, Westoby 1980) to processes. The details of these models can ciones, estos se consideran dentro de cate- propose a revised framework for range draw upon a wealth of recent conceptual gorias discretas basadas en unos pocos management. The state-and-transition and technological advances in community procesos ecologicos fundamentales y sus model formally acknowledged the multi- and landscape ecology, including the rela- relaciones pueden ser facilmente com- equilibrial nature of many rangeland tionships between positive feedback prendidas. El conocimiento de los mecan- ecosystems and the rapid and unanticipat- mechanisms and threshold changes in ismos esta estrechamente ligado al use de ed shifts among these equilibria. processes such as erosion (Davenport et al. indicadores ecologicos para anticipar las Furthermore, Westoby et al. (1989) 1998), the dependency of threshold modelos transiciones. Concluimos que los focused attention on the multiple mecha- changes on processes operating on differ- de incluir 1) valores de referencia deben nisms underlying alternative equilibria and ent scales of space and time (Scheffer et para indicadores cuantitativos, 2) lista de an "opportunistic" style of al. 2001), and understanding the linkages indicadores cave descripcion de sus emphasized y in which vary among processes using ground-based and cambios que sugieran una aproximacion a management strategies (Rango et al. in una transicion, 3) una documentacion depending upon which mechanisms are remotely-sensed patterns y important. The state-and-transition con- press). Implementing these advances to rigurosa de las teorias y asunciones (y sus alternativas) que dan base a la estructura cept provides a means for anticipating improve on-site management of range- de cada modelo. departures from the monoclimax model lands, however, presents substantial chal- and incorporating this understanding into lenges, among them: 1) how do we draw management plans. Consequently, this together the detailed information required Ecological theory provides a basis for concept is being widely espoused within to create site-specific models that are 2) land management. An understanding of the range science community of the applicable across entire regions, what the processes inferred to cause population United States (Society for Range ecological issues need to be considered in or community patterns determines how Management 1995, USDA NRCS 1997). developing and classifying sets of models, managers should respond to patterns. A For agencies such as the Natural and 3) given the models that can be pro- prime example is the succession-retrogres- Resources Conservation Service (NRCS) duced, what do we want to use them for sion (or range condition) model of and Bureau of Land Management (BLM), and how should they be used? In address- Dyksterhuis (1949) that is based on the state-and-transition models promise to ing these general questions about state- successional theory of Clements (1916) improve assessment, monitoring, and and-transition model development, we and the edaphic polyclimax concept of management in many semiarid rangelands. draw on examples from state-and-transi- Tansley (1935). This model emphasized Twelve years after the seminal publications models developed in a range of land- the return of disturbed communities to a tion of the state-and-transition concept, scapes in New Mexico. competitively-determined climax state and however, few applications of the concept has been a guiding principle in range man- exist that can be used by land managers. What is a state-and-transition agement (Westoby 1980). Upon recogniz- This is not to say that there has not been a model? ing an undesirable trend in plant commu- great deal of work on the concept. The idea that rangeland vegetation nity composition, managers could respond Researchers have provided refinements to exhibits multiple states, and transitions (Laycock 1991, Freidel by reducing or redistributing grazing pres- underlying ideas among them, has been referred to general- and Kothmann sure and effect a return to desirable condi- 1991, Rodriguez-Iglesias ly as the state-and-transition model. These Koppel 1997), tions. An important reason for the success 1997, Reitkerk and van de concepts have been adequately reviewed in the production of this model is that it provided a method technological advances elsewhere (Laycock 1991, Brown 1994, (Wiegand to measure and compare land condition and quantification of models Rodriguez-Iglesias and Kothmann 1997, Allen-Diaz and against the expectations of the model (i.e., and Milton 1996, Stringham et al. 2001). To be operational, the similarity index), thus providing a con- Bartolome 1998, Plant et al. 1999), and however, specific state-and-transition models (e.g., crete link between theoretical expectations some site-specific conceptual models must be created that describe the et al. 1994, and management response. Archer et al. 1988, Ash details of vegetation dynamics for particu- There have been few Rangeland managers have long recog- George et al. 1992). lar land areas. The graphical and concep- sets of site- nized that semiarid grasslands can trans- attempts, however, to develop tual format provided by Westoby et al. based upon existing infor- form into shrub-dominated states that can- specific models (1989), with modifications most recently by land man- not be returned to grassland through graz- mation that can be applied summarized by Stringham et al. (2001; way over broad ing management (Laycock 1991), contrary agers in a systematic Fig. illustrate several key elements that of the insights 1), to applications of the succession-retrogres- areas. Here, we relay some are communicated in these conceptual the production of state-and- sion model. Assuming that a single, com- gained during models. The model format presented here for rangelands in the petition-defined equilibrium plant commu- transition models is based upon that currently used and pre- Mexico, USA. nity should exist for each site, alternative state of New sented by the NRCS. states, New Mexico states, and the rangelands in which they Like other western As in vegetation mapping (Grossman et dominated by semiarid rangelands and occur, have been referred to as "non-equi- is al. 1998), the most basic unit is the plant the monoclimax model librial" (following Wiens 1984). In fact, the limitations of community. This is the relatively homoge- in many of these ecosys- these alternative states may be highly are very apparent neous assemblage of plants that occurs at a models have the equilibrial (e.g., Muller 1940) after the tems. State-and-transition particular point in space and time, and can potential to provide a framework for orga- transition, so these systems are better be defined at a scale relevant to a land ideas about the termed multi-equilibrial. The increasing nizing complex sets of manager (e.g., 0.1-10 ha). In the models processes driving emphasis on processes other than competi- multiple, interactive for New Mexico, plant communities are

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 115 1997) are often expensive to apply. Generally, transitions among ecosystem Transition 1 states are thought to be caused by a combination of external and internal, positive feedback mechanisms that alter constraints on the presence or abundance of particular plant species (e.g., Schlesinger et al. 1990; Fig. 2). Three general classes of constraints can be recognized: 1) the dispersal of propagules to a site and subsequent reproduction, 2) "neighbor" constraints, including the effects of competitors, State A State B predators, or parasites, as well as the ten- Transition 2 dency of certain life-forms to facilitate fire disturbance, and 3) "site" constraints, including soil properties, hydrology, and climate. Transitions occur when 1 or more constraints are altered by external factors State C and this change catalyzes changes in positive feedbacks that produce relatively Fig. 1. The general structure of state-and-transition models after Stringham et al. (2001). important shifts in vegetation structure and The small boxes represent individual plant communities and the dashed arrows between soil properties. Multiple external factors them represent community pathways along which shifts among communities occur. These can be affected by singular processes, such shifts are reversible through facilitating practices and fluctuations in climate. The large as livestock grazing boxes containing communities are states that are distinguished by differences in structure (e.g., by introducing and the rates of ecological processes (such as erosion). The transitions among states (solid shrub propagules and decreasing competi- arrows) are reversible only through accelerating practices (e.g. seeding, shrub control, or tion with them). Furthermore, changes in addition of soil) that can be applied at relatively great financial expense. one class of constraints may reinforce changes in other constraints, such that sev- often defined by dominant species, or plant functional groups and ecosystem eral positive feedback mechanisms operate species that indicate the operation of par- processes and, consequently, in vegetation together (c.f. Archer 1989). ticular processes (e.g., an encroaching structure, biodiversity, and management For example, heavy, continuous live- shrub species, whether or not it is now requirements. For example, a grassland, stock grazing may initiate changes to dominant). Descriptions of plant commu- whether it is dominated by one grass shrub colonization ability by providing a nities may also contain information on soil species or another, may provide sufficient dispersal pathway for seeds and reducing conditions that indicate processes (e.g., the cover to prevent rain-drop compaction of competition and fire disturbance by cover of physical, chemical or cryptogam- the soil surface and to intercept water and removing grasses. Reduced competition ic crusts). It is important to identify the nutrients before they are lost from the sys- and fire disturbance may permit shrub range of plant communities in a land area tem. In doing so, dominance by either establishment and growth. The presence of because these are the observable and mea- grass species can sustain the soil condi- adult shrubs 1) increases shrub seed avail- surable links to the processes embodied in tions required by the other, and replace- ability through reproduction and facilitates the remaining components of the model. ments along community pathways may the dispersal of additional seeds to the site Plant community identity at a location occur within the state. Once grass cover by attracting birds, 2) increases competi- may vary in time and the arrows between has been reduced below a critical amount tion with grasses and limits grass reestab- communities indicate changes (or "com- (Davenport et al. 1998), or a shrub species lishment, and 3) increases erosion rates munity pathways") among them. These invades that leads to grass loss (Brown and nutrient loss from shrub-interspaces by shifts in plant composition, as opposed to and Archer 1999), infiltration is reduced reducing grass basal cover. Alternatively, those referred to as "transitions" (see and erosion accelerates, a change in soil prolonged and severe drought or mechani- below), may be caused by climate or land conditions occurs, and the system crosses cal disturbance (e.g., off-road vehicles) use but are reversible by simply altering into a new state. This new state is charac- might catalyze a similar sequence of events the intensity or direction of the factors that terized by a distinct set of plant communi- if shrub seeds were already present. caused the change in composition (e.g., ties and a distinct range of values for Thus, different states can be viewed as practices that reduce grazing pressure or ecosystem attributes. separating positive feedbacks between dif- increases in rainfall after a drought). The The shift between states is referred to as ferent kinds of plants and different ecosys- "facilitating practices" as defined by the a "transition". Unlike community path- tem processes. For example, by retaining NRCS (USDA NRCS 1997) would pro- ways, transitions are not reversible by sim- soil nutrients and high infiltration capacity duce responses along community path- ply altering the intensity or direction of the over relatively homogeneous areas ways. Thus, the succession-retrogression factors that produced the change (c.f. the (Ludwig et al. 1994), or by promoting fire model operates along community path- "amplitude" of Westman 1978) and (McPherson 1995), grasslands sustain ways and is embedded within the state- instead require the application of distinct themselves. By promoting intershrub ero- and-transition model. factors such as the addition of seeds, the sion and heterogeneous nutrient distribu- Communities are aggregated into states removal of shrubs, or the addition of top- tions, or by outcompeting grasses for (Fig. 1) that are distinguished from other soil. These "accelerating practices" as water, shrublands promote shrublands states by relatively large differences in defined by the NRCS (USDA NRCS (Schlesinger et al. 1990). Once competi-

116 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Climate change or ferences in important environmental fac- changed hydrology tors, including soil properties, slope, and landscape position (e.g., in an upland or swale). These differences correspond to differences in the structure of plant com- Altered site munities, and with respect to state-and- transition models, plant community (e.g.reduced grass constraints (e.g. decrease in dynamics in the face of natural and increases erosion, site suitability decreases infiltration) human-caused disturbance (Society for for grass) Range Management 1995). Ecological sites are mapped by grouping soil map- (e.g. shrub introduction ping units on which plant communities are permits competition) assumed to behave similarly. Ecological Altered neighbor Altered demographic sites occur together in a landscape as a constraints /dispersal constraints mosaic determined by patterns of geomor- (e.g. reduction of phology (Fig. 3c). Like other vegetation grass populations classification systems, ecological sites are promotes shrub nested within a hierarchy of climatically- establishment) defined regions (Fig 3a, 3b). The extent of Grazing or fire Species introduction or a particular ecological site is bounded (life-form specific large, general disturbance within an area of similar geology and cli- disturbance) mate, the land resource unit of the NRCS or ecoregion of the U.S. Forest Service, produce an Fig. 2. The relationships among the classes of constraints that are altered to beyond which analogous ecological sites ecosystem transition, and the internal and external processes that affect these constraints. may a External processes (dashed arrows) act as environmental triggers that set in motion posi- exist. Thus, particular state-and- tive feedbacks represented by the internal processes (solid arrows). Processes affecting transition model is intended to apply to particular constraints interact with one another such that multiple mechanisms produce one ecological site that can be found with- and maintain transitions. See text. in only one land resource unit. The definition of ecological sites tive dominants are introduced, species are Historically, ecological sites (previously (Creque et al. 1999) and land resource lost, or soil properties are significantly called range sites) were based solely on units is often arbitrary. Given the impor- altered, transitions can be difficult and similarities in the composition and produc- tance of soils and climate for the nature of expensive to reverse. An understanding of tivity of dominant, climax vegetation vegetation dynamics, the validity of any the constraints to ecosystem change, and (Shiflet 1975). Ideally, ecological sites are state-and-transition model will depend the relationships between the external and a classification of land types based on dif- upon the amount of variation in important internal mechanisms affecting them, suggest strategies for predicting and avoiding transitions and devising restoration strategies (Whisenant 1999). Although the causes of individual transitions in models are varied, the mechanisms involved fall into readily-understood classes (Fig. 2) that are common to all models. State-and-transition models, then, represent postulates about the causes of both ephemeral and persistent changes in vegetation at a site and should offer testable predictions. Moreover, the models provide a logical framework in which assumptions and concepts about how rangelands work must be specified. This can add clarity to our ideas about rangelands, and often reveals how little we really know about them.

Model classification using ecological Fig. 3. The land-unit classification framework within which state-and-transition models are sites being developed by the Natural Resources Conservation Service. a) the Land Resource Before a model can be created and test- Region scale, highlighting the Western Range and Irrigated Region within which land ed, it is critical to define the extent over resource units are embedded, b) the Major Land Resource Area scale: Southern Desertic which a single model will apply. For and Basins, Plains, and Mountains within New Mexico. Land resource units (SD-I, SD-2, SD-3) with A unique set of ecological sites and state-and-transition models are rangelands in the United States, this extent are noted arrows. common to each subresource area. c) a map of ecological sites for part of the Jornada the is currently defined by ecological sites Experimental Range based on groups of related soil series (draft soil map courtesy of Dr. of the NRCS (USDA NRCS 1997). Lee Gile and Barbara Nolen, the Desert Soils Project).

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 117 soil properties and climate within ecologi- lighted in the ecological site description. creosotebush had encroached and dis- cal sites and land resource units, respec- Furthermore, this community is often placed climax native grasses. Defining tively. If land assigned to ecological sites believed to be the one in which soil appropriate management standards via the does not exhibit consistent properties, it resources and native biodiversity is best state-and-transition (or the succession-ret- may not be clear whether the variation in conserved (but see Belsky 1996). Thus, rogression) approach requires that we plant communities observed between 2 the state bearing this community (along acknowledge and accept data limitations, areas within the same ecological site is with related successional stages) is the take into account the multi-equilibrium due to a vegetation transition or to static global management standard of the BLM nature of plant communities, and develop differences in environmental conditions and the standard is implicit in the activities ecological sites based on a detailed under- (Friedel et al. 1993). For example, burro- of other agencies and non-governmental standing of plant-soil relationships. grass (Scleropogon brevifolius Phil.) is a organizations (c.f. USFS 2000, Bureau of Defining the range of alternative com- mat-forming native perennial grass of the Land Management 2001, Strittholt and munities occurring in several ecological Chihuahuan Desert, is often dominant on Boerner 1995). sites can be accomplished using monitor- loamy or clayey ecological sites, and is There are, however, several limitations ing data, such as those gathered by the relatively unpalatable to livestock. Thus, to the historic climax plant community Long-Term Ecological Research program, dominance by burrograss is often assumed concept. First, in much of the western the BLM, the NRCS, private individuals, to be related to management practices and United States, the historic climax plant as well as repeated aerial or terrestrial drought. While this may be the case on community does not now exist and cannot photography from a variety of sources Stellar clay loam soils that were formerly be reliably estimated from historical (e.g., Callaway and Davis 1993, Miller dominated by tobosa (Pleuraphis mutica records. Thus, historic climax plant com- 1999, McClaran et al.). In many cases Buckl.), burrograss may have been histori- munities are sometimes estimated with the however, the number of ecological sites cally dominant on Reagan clay loam soils hidden assumption that the plants that monitored or the duration of monitoring is (Leland Gile, personal communication). were most palatable to livestock were the limited. Interviews of rangeland profes- Reagan soils are highly calcareous at shal- competitive dominants in the historic cli- sionals, researchers, and ranchers low depths relative to Stellar (see Gile and max plant community (Westoby 1980). (Bellamy and Brown 1994) in conjunction Grossman 1997) and are less pervious to Second, the notion of the competitively- with structured, rapid vegetation surveys water than are Stellar soils in a similar determined climax state is explicitly based on soil maps and associated with landscape position (Herbel and Gibbens acknowledged in the historic climax plant soil series determinations can add signifi- 1989). Although Stellar, Reagan, and community, and this is at odds with the cantly to the number of communities other soils have been grouped within the multi-equilibrium concept now embraced encountered and provide rigorous associa- same ecological site, these soils clearly by agencies (Svejcar and Brown 1991). tions of communities with soil properties. exhibit distinct properties and have proba- That is, climaxes may shift, even without A practical limitation of using profes- bly always harbored distinct communities. human influence, such that a climax is a sional testimony and casual field observa- To distinguish management-produced "moving target" over broad time scales tions to define communities is that it is from natural patterns (especially in the due to climate change (Brown et al. 1997). often unclear how communities are related absence of historical data), it will be nec- For example, it is possible that the domi- in time. The number of communities essary to rigorously delineate ecological nance of black grama (Bouteloua eriopoda included within an ecological site may be sites (Creque et al. 1999). Doing so will (Torr.) Torr.) on sandy soils of southern an artifact of the persistent environmental require that we distinguish ecological sites New Mexico was a consequence of cli- heterogeneity in space included within an based upon values of soil and climatic matic conditions peculiar to the late nine- ecological site definition, rather than tem- variables that correspond to differences in teenth century (Neilson 1986). A regional poral variability at points in space. When the nature of state-and-transition models. change in climate may now preclude suffi- it is suspected that 2 communities do not In turn, this necessitates a detailed docu- cient sexual reproduction to reestablish occur at the same points in space, the cre- mentation of the relationships between black grama as a dominant in many areas, ation of a new ecological site may be plant communities and their dynamics to soil and it is possible that overgrazing and called for. Alternatively, (or while new series. Such efforts will in many cases lead drought have only hastened the demise of ecological sites are developed), we can to reassignments of soils to ecological sites this species. Given this case, it is question- denote the absence of temporal relation- and the creation of new ecological sites. able whether the restoration of a produc- ships among communities by having them tive black grama-dominated community "float" within the state and not be connect- Building state-and-transition models would be a suitable management goal. ed to other communities by community Defining communities and community Third, aggregating soils with distinct pathways (Fig. 4). In other cases, areas pathways inherent properties into ecological sites with differing aspect or slope are circum- Once an ecological site has been select- leads to the development of uniform scribed in ecological site definitions, and ed, the first task in creating a state-and- expectations where they may not be war- northern and southern exposures may transition model is to define the communi- ranted. For example, creosotebush (Larrea exhibit distinct species composition and ties that can occur within that ecological tridentata (DC) Coy.) has likely been dynamics. Nonetheless, these areas func- site. A key benchmark is represented by dominant on erosional fan remnants at the tion similarly enough (or are so predictably the "historic climax plant community" of base of Mount Summerford in southern associated) that they are considered within the NRCS (USDA NRCS 1997). This New Mexico since before European settle- the same ecological site and state. community represents the composition of ment (Wondzell et al. 1996). The ecologi- plants that is known or is presumed to cal site in which these soils are now Defining states and transitions grouped (Bullock and Neher 1980), how- have dominated an ecological site prior to States have been defined based on shifts ever, would lead one to the conclusion that the settlement of Europeans, and is high- in plant community structure using multi-

118 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 and conditions that lead to dominance by other species (Neilson 1986). Changed cli- Black grama -4- Bush muhPy Bush muh' Black rams mate or further soil degradation may lead a to black grama extinction and dominance Tobosa by bunchgrasses that are able to reproduce Bush muhl Creosotebush/tarbush by seed (transition 2). The transition to a Grassland date Tobosa shrub-invaded state is catalyzed by either Creosotebush the introduction of mesquite propagules Burrograss into grassland (Brown and Archer 1987) Tarbush Tobosa or by the competitive release of existing mesquite seedlings through the reduction Tobosa Eroded shrubland state Creosotebushrtarbush I of grass cover (Van Auken and Bush 1997), fire frequency (Wright et al. 1976), or shrub seedling herbivores (Weltzin et Bush muhly Creosotebush al. 1997; transition 3a). Alternatively, Creosotebush 1+ Bu_rrogrrass propagule introduction might occur after with a change in climate or Bush muhly or concurrent Tobosa soil degradation (Hennessy et al. 1983; Tobosa-bush muhlylshrub state transitions 4a, 5a). Initiated by these external triggers, the loss of black grama may be caused by a shift in positive feedbacks Fig. 4. A draft state-and-transition model for the gravelly loam ecological site within the SD- involving competition, erosion, and physi- 2 land resource unit of southern New Mexico. Transition 1 is caused by grass loss and subsequent shrub invasion, whereas transition 2 is caused by soil degradation. Transition 3 cal and chemical changes to soils requires shrub removal, grass seeding, and restoration of soil fertility and permeability. (Schlesinger et al. 1990, Herrick et al. The communities within the eroded shrubland state are not connected by arrows, indicat- 2002) due to the presence of shrubs (tran- ing that there is no evidence for replacement among these community types. Instead, these sition 6). Shrubs would need to be distinct community types seem to reflect variation among soils included within the ecologi- removed to return to the black grama- cal site, although there is not yet enough information to reliably split soils into separate eco- dominated (transition 3b) or limited (4b) logical sites. Black grama, bush muhly (Muhlenbergia ported Scribn.), tobosa, and burro- state. Shrub expansion may need to be grass are perennial grasses; burrograss is usually least palatable to cattle. Creosotebush and controlled in order to remain in the shrub- tarbush (Flourensia cernua D.C.) are shrubs. invaded state, although coexistence variate analyses of long-term data sets (Allen-Diaz and Bartolome 1998). While this approach is objective and potentially repeatable, it does not consider the Black grams Mack grams Black or mechanisms involved in I grams processes Dropseeds j_ e_su te community changes. Furthermore, most Iv b 1J applications do not demonstrate that plant r N Shrub-invaded 1 communities have entered a new domain Black grams Drop seeds grasslands of variability (e.g., Friedel 1991) that indi- Threeawns Black grams Bunchgrssses cates a fundamental change in the func- Mesquite Snakeweed 4 tioning of the ecosystem (Whisenant Black grams 1999). Thus, it is unclear whether changes in plant composition can be reversed Black gra ma-dominated grassland i through facilitating or accelerating practices (Stringham et al. 2001). 5a Mesquite Bunch grasses Given these uncertainties and the pauci- Threeawns Dropseeds I ty of long-term data supporting the exis- Black grams Black rams r tence of multiple domains of variability, Black grams-limited grassland Mesquite states and transitions can be constructed (snakeweed) based upon postulates of vegetation Mesquite shrubland change in combination with empirical Bunchgrssses observations of community structure and Snakeweed) I environmental conditions. For example, a Bunchgrass grassland number of explanations for the well-documented loss of black grama and increase Fig. 5. A state-and-transition model for the sandy ecological site within the SD-2 land in honey mesquite (Prosopis glandulosa resource unit of southern New Mexico. Black grama, dropseeds (Sporobolus R. Br. spp.) Torr.) may pertain to Sandy ecological threeawns (Adstida L. spp. ) are perennial grasses. Black grama is palatable to cattle for a sites in south-central New Mexico (Fig. longer duration than the other species and comparatively sensitive to grazing pressure. 5). The transition to a black grama-limited Snakeweed (Xanthocephalum Willd. spp.) is a subshrub that tends to invade with reduc- state (transition la) represents the shift tions in grasses or with adequate winter-spring precipitation. Mesquite is a large shrub that between climatic or soil fertility condi- tends to invade intact or degraded grasslands, promote the loss of grasses in intershrub tions conducive to black grama dominance spaces, and concentrate resources beneath its canopy.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 119 between bunchgrasses and mesquite with- lands to blue grama (Bouteloua gracilis may be sufficient to describe persistent out shrub control may persist for long (H.B.K.) Lag. ex Steud. Hitchcock) grass- changes due to various causes and creating periods if soil degradation is not aggravat- lands characteristic of upland sites. new ecological sites would be an unneces- ed by environmental stress. If grazing, Subsequently, agradation of soil into chan- sary semantic complication. drought, or shrub encroachment continue nels will lead to a return of flooding Unexpected transitions among states (or to reduce the remaining grass cover, ero- cycles, and bottomland grasslands may sites) may depend upon the transitions sion and/or increases in rodent and rabbit reestablish with improved grazing man- occurring in adjacent sites. On the loamy densities (Campbell 1929) eventually lead agement. Ecological sites and regions may soils of the Jornada Experimental Range, to the formation of stable coppice dunes differ in the degree to which hydrology, for example, wind erosion of degraded with the loss of most grasses (transition 7). climate, or other physical features create sandy soils to the west of loamy soils has At this point, natural reestablishment may stability over particular time scales, blur- resulted in the accumulation of sand on the not be possible for most grass species. In ring the distinction between pathways and loamy soils (C. Monger, personal commu- principle, the mechanical restoration and transitions among "stable" states nication). These patches support black stabilization of topsoil and addition of soil (Fuhlendorf et al. 2001). Nonetheless, grama grass that is absent on the unaltered nutrients and microorganisms following when intensive accelerating practices soils. Herbel et al. (1972) speculate that shrub removal could be used to initiate (such as gully stabilization) can be used to the increased abundance of tobosa relative grassland recovery (transitions 8, 9) hasten relatively slow natural processes or to black grama and other grasses on lower In this example, the definition of each of recovery, the recognition of distinct states piedmont clayey sites is due to the the states depend critically on our notions is useful. Stability must be defined relative increased water run-on from degraded of the processes driving vegetation change to a time scale, so it is important to con- gravelly sites occurring upslope. It is and our responses to them. For example, sider how the scale over which natural important to recognize that some transi- the shrub-invaded state may be defined by recovery is observed matches management tions may have extrinsic causes that either the presence of shrub seedlings, or timeframes when defining states. depend upon landscape context rather than the presence of shrubs seedlings in the local management. context of particular stresses or changes to The relationship between states, tran- disturbance regimes. If the presence of sitions, and ecological sites Patterns in sets of state-and-transi- shrubs with grasses for long periods accel- When a transition involves changes in tion models erates grass loss, a shrub-invaded state soil structure, the resulting vegetation/soil Given the issues and approaches dis- may be supported only by intensive accel- change may be so permanent that it is con- cussed in the preceding sections, what erating practices. Alternatively, this situa- sidered a new ecological site rather than a generalizations can be drawn from the tion might simply be considered as an state. The NRCS National Range and models we have created? To date, we have early stage of a shrub-dominated state if Pasture Handbook states "Severe physical produced about 60 draft or completed shrub control was not implemented. In deterioration can permanently alter the models spanning 4 land resource units in addition, we proposed that some commu- potential of an ecological site to support southwestern-southcentral New Mexico. nities (e.g. the dropseed/black grama com- the original community" (USDA NRCS These land resource units intergrade with munity) occur in 2 distinct states. This 1997). This may apply when soils are one another, and differ in subtle ways denotes that communities with similar truncated to the point where an eroded based on moisture and temperature (Table vegetation structure may have very differ- phase is recognized in soil classification. 1). Some of the patterns that we have ent responses to management due to dif- For example, the loss of sandy surface observed can be compared among land ferences in climate or soil properties horizons on some sandy loam soils may resource units using models from a set of between the states. States are human con- expose clay-rich strata that no longer sup- common ecological site types representing structs that represent our understanding of port the germination or survival of former- a gradient of landscape position and soil and relationships with rangelands. ly dominant species (Gile and Grossman properties. In general, we see that more By definition, the recognition of states 1997). The point at which a new ecologi- states per ecological site were generated and transitions also depends on temporal cal site should be recognized, however, for thermic, aridic soils of SD-2 (i.e., the scale (Friede11997) and, thus, our ideas of depends on the fuzzy distinction between southern desert unit 2 land resource unit) stability and equilibrium. In many cases, changes that are "persistent" (Stringham et than for the soils experiencing more ustic such as when grasslands are converted to al. 2001) without the use of accelerating and/or mesic regimes in higher elevation mesquite dunes, the difference between practices (i.e., a transition) and "perma- community pathways and transitions or more northerly land resource units. This nent" change. We should also be aware of may be due to 1 or 2 non-exclusive caus- between states is clear. In other cases, it is the consequences of losing track of the es: 1) more ecological sites within rela- not. For example, bottomland/draw system occurrence of historic communities at a tively warm and arid regions are subject to degradation in the New Mexico Plateaus site-do we want to maintain documenta- a variety of processes that lead to several and Mesas land resource unit may exhibit tion that a particular area used to be grass- states (e.g. erosion plus shrub invasion or a cyclic sequence of "states" over a period land but is now shrubland? The amount of expansion; Table 1) and 2) land-use pro- of 50-100 years without the use of accel- financial resources required to apply accel- fessionals recognize more states in SD-2 erating practices (G. Adkins, NRCS, per- erating practices to reverse a transition may because of the relatively extensive sonal communication). Erosion and chan- be a suitable criterion for deciding when to research conducted there. Another feature nelization due to reduced grass cover leads create a new ecological site. Constraints apparent in the groups of models is the to decreased soil moisture availability and due to biotic interactions, for example, are wide range in the number of states. Some a transition from giant sacaton often less expensive to overcome than abi- models (e.g. SD-1 Hills) have only one (Sporobolus wrightii Munro) or alkali otic limitations (Whisenant 1999). We sug- state, implying that the site is resilient sacaton (S. airoides (Torn.) Torr.) grass- gest, however, that the concept of "state" with facilitating practices alone and that

120 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 1. The number of states and key constraints defining the states for 7 common classes of ecological sites within 4 Land Resource Units (e.g., SD-2) in southern New Mexico. The Land Resource Units chosen differ in mean annual precipitation (MAP) and soil moisture regime (MR) and/or mean annual temperature (MAT) and soil temperature regime (TR) but all occur within south-central to southwestern New Mexico. The number of states and key constraints are based on published literature and interviews with rangeland professionals. These models may be viewed at the NRCS New Mexico website: http:llwww.nm.nres.usda.gov/techserv/fotg/Section2/esd.htm

MLRA 42, SD-2* 42, SD-1* 42, SD-4* 36, WP-3* ° Latitude: 31° 19' - 34° 24 Latitude: 33° 27' - 35° 21' Latitude: 31 40' - 32° 30' Latitude: 32° 22' - 34° 35' Longitude: 105° 50' -109° 02' Longitude: 106° 25' - 107° 07' Longitude: 105° 39' -105° 55' Longitude: 106° 58' -109° 02' 8-10" MAP;15.5°C MAT 8-11" MAP; 13.3°C MAT 12-14" MAP; 15°C MAT 12-16" MAP;13.3°C MAT MR: Aridic MR: Ustic-Aridic MR: Ustic-Aridic MR: Ustic-Aridic TR: Thermic TR: Thermic-mesic TR: Thermic TR: Mesic Ecological Site # of Key constraints of constraints of constraints constraints states states states states Bottomland 5 Gullying 3 2 3 Blocked run-on' Blocked run-on Blocked run-on Blocked run-on Mesquite invasion

Swale/Draw 5 Gullying 2 Soil sealing Blocked run-on Blocked run-on Mesquite invasion

Clayey/Clay upland 6 Soil sealing 2 sealing 2 run on 2 Blocked run-on Erosion/soil loss Soil sealing Shrub invasion Erosion/soil loss

Loamy (Draw) 4 Soil fertility loss 4 fertility loss 3 invasion 2 fertility loss? Shrub invasion Erosion/soil loss Erosion/soil loss Gullying Erosion/soil loss

Sandy/Loamy sand 6 Soil fertility loss 2 fertility loss? 2 sage3 expansion 3 invasion Mesquite invasion Sand sage expansion Soil-surface instability Erosion

Gravelly 3 Creosotebush 3 fertility loss 2 expansion 4 loss expansion2 Creosotebush Juniper4 invasion Erosion/soil loss invasion

Hills 2 Erosion/soil loss 3 invasion Mean number of states** 4.8 2.8

MLRA refers to the Major Land Resource Area (USDA NRCS 1997), a climatically-defined unit within which Land Resource Units are nested. Land Resource Unit designations include SD 1-4 (Southern Desert units 1, 2, and 4) and WP 3 (Western Plains and Mesas unit 3). See also Fig. 3. ** SwalelDraw and Hills Ecological Sites were excluded from the mean because they do not occur in all Land Resource Unit's (Hills) or are functionally different (Swale/Draw). zBlocked run-on refers to the alteration of surface hydrology by features such as dams, ditches, or roads that inhibit natural water run-on patterns Invasion refers to the presence of a plant species that was not present in historic communities, whereas expansion refers to increased density of a plant relative to its density in historic communities. 3Artemisia filifolia Ton. 4Juniperus monosperma (Engelm.) Sarg. the succession-retrogression model ade- subject to erosion and loss of soil fertility. rounding past vegetation changes on eco- quately describes its dynamics. At a conti- The importance of shrub or tree invasion, logical sites and to use this information to nental level, this may be true of many on the other hand, seems to depend on the anticipate and interpret changes in the rangeland communities that are resistant to ecological site/land resource unit combi- future. Beyond this general function, iden- soil degradation, not dependent on surface nation in question, although in some land tifying the operation of processes underly- hydrologic inputs, or not subject to inva- resource units (e.g. SD-2) it is more ing transitions will be key elements in sion by competitive species. important than in others. Overall, we see using models to improve land manage- Can we generalize about the importance that a subset of common processes in vari- ment. In many cases this is difficult to do, of particular processes in different ecolog- ous combinations explain vegetation however, because we often do not fully ical site types or land resource units? In dynamics within different ecological sites. understand the causes of observed vegeta- some cases, we find that the same types of This allows us to make some generalization changes, particularly their interac- processes are invoked to explain transitions, while requiring that these general- tions. tions in similar ecological sites: lowland izations be carefully evaluated for each A conceptual approach to deal with this sites such as bottomlands, draws and ecological site and land resource unit. is to postulate that there is usually a domi- clayey sites are subject to changes in sur- nant proximate variable or a characteristic face hydrology and surface soil structure Using models: proximate variables, pattern of correlation among several vari- and chemistry with respect to infiltration. indicators, and predicting transitions ables that regulates particular transitions Changes among states in upland sites such on ecological sites. In the southwestern The primary use of state-and-transition as sandy, gravelly, and hills sites are often U.S., the duration and timing of available models is to depict the circumstances sur-

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 121 soil water may be a key variable involved in many kinds of transitions (Whitford et Tobosa grassland Mixed-shrub grassland al. 1995, Devine et al. 1998, Breshears and 12 Barnes 1999; Table 1) and also relates closely to nutrient availability (Stafford Smith and Morton 1990, Reynolds et al. 4 1999, Dodd et al. 2000). Transitions may

be due to threshold (non-linear) changes in 2030 2030 30- 50-100 100200 200+ the availability of soil water over time due to shifts in runoff patterns (e.g., percola- tion thresholds; Davenport et al. 1998) or to threshold responses of species performance to gradual changes in the soil water 12 Burrograss grassland Tarbush shrubland 10 (e.g., species-specific soil moisture limita- sa

tions). Thus, nonlinear responses may be 6 caused by both positive feedbacks that 4 create threshold changes in variables as 2 1 well as through species tolerance limits 0 11 20.30 30810 90100 10000 200+ 2030 30.60 90100 100200 200+ (e.g., Austin 1985). Iii To anticipate transitions, we need to Bare ground patch size class (cm) know something about the changes in proximate variables underlying threshold responses in vegetation and soils. Despite Fig. 6. Histograms for gap-intercept data (see Herrick et at. 2001) representing different a wealth of information, these issues have states within the loamy ecological site in the SD-2 land resource unit of southern New model. Gap (bare-patch) seldom been directly addressed in New Mexico. Graphs are arranged to parallel the state-and-transition sizes were measured along a 50-m tape. Gaps were the distance between perennial plant For example, is the Mexico, or elsewhere. bases (where stems emerge from the ground) that were intercepted by the edge of the tape. persistent reduction of black grama abun- Only gaps greater than 20 cm were tallied. Data were gathered on representative sites on dance due to a reduction in soil moisture the Jornada Experimental Range. availability, reductions in the abundance of mycorrhizal fungi, a change in rodent ground patches or shrubs, reduction in soil the plant composition data gathered by herbivory levels, some other unconsidered stability, or the formation of rills, litter management personnel to identify states. factor, or a combination of factors? If soil dams, and terracettes indicate erosion that Together, the suite of qualitative and moisture changes are important for black reduces soil fertility and microbial popula- quantitative measurements can do for grama germination, survival, or reproduc- tions, water infiltration, and may eventually state-and-transition models what the simi- tion, then, across what values of moisture truncate entire soil horizons (Unpublished larity index does for succession-retrogres- availability do threshold responses occur? data, Herrick et al.). Thus, a trend towards a sion models: they provide a means to con- Are threshold or continuous changes in transition involving one or several interact- nect field observations with theoretical these variables observed in nature and ing mechanisms may be revealed by indica- expectations and management responses. what are the causes of this variation? Are tors, even when the precise mechanisms It is important to recognize that transi- threshold responses of particular species (and critical values) are not well known. tions usually do not happen simultaneous- determined by the same environmental By linking specific indicators to state- ly across entire landscapes, grazing allot- variables in all instances? Addressing and-transition models, model developers ments, or even pastures. In many cases, questions for key grass and shrub these can provide tools to help land managers transitions occur 1 patch at a time, occur- species will greatly improve our capacity determine the state that land is in and to ring first on areas within ecological sites to provide flexible, predictive models of evaluate the probability of a transition. that are most sensitive to change due to transitions with management utility that Different qualitative indicators (Pellant et slight variations in soils or landscape posi- link retrospective data and observations al. 2000) relate to different processes, and tion (Fig. 7). Over time, or across space, with comparative studies and experiments. models can point to specific indicators that these patches may coalesce to produce But what is a manager to do with state- signal an approach to a particular transi- landscape-scale phenomena (Gosz 1993). and-transition models while we gather and tion. Models can include ranges in values Furthermore, changes in ecosystem func- synthesize these critical data? Even when for quantitative indicators such as perenni- tioning in 1 patch may affect adjacent mechanisms are fully understood, proxi- al plant cover, shrub density, bare ground patches. In many arid systems, nutrients mate variables such as soil moisture or patch size, frequency, and spatial arrange- lost from patches that have undergone mycorrhizal populations are going to be ment, soil compaction, and soil surface transitions may be redistributed to adja- difficult to monitor directly or may change stability to define the range of structure cent patches, increasing local production only after a transition. A promising and function characterized by states and and grass cover in those patches (Ludwig approach is to measure factors related to that provide benchmarks for measure- and Tongway 1995). These patch-scale key proximate variables that indicate the ments of the processes leading to transi- dynamics indicate that 1) monitoring operation of processes that can be altered tions. In particular, indicators such as changes in the frequency, size, or spatial to prevent (or facilitate) a transition basal and canopy gap size (Unpublished arrangement of patch-scale transitions (Herrick 2000, Ludwig et al. 2000, Kuehl data, Herrick et al.; see Fig. 6) are cone- may be used to indicate the consequences et al. 2001). For example, the presence, lated with other indicators of soil quality of management activities, 2) monitoring expansion, and spatial arrangement of bare and erosion and can easily complement transects located in "nutrient sinks" or

122 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 can convey what has happened on different soils, and what will happen given the operation of the processes embodied in a model. Managers can use case-specific information, supplemented with indicators in many instances, to evaluate the likelihood that particular processes are operating, evaluate management options with respect to those processes, and make a determination of the costs and benefits of those options. The refinement of transition-specific indicators and reporting of reference values with specific models will aid this process. In using reference indicator values, it is ill-advised to consider maximizing grazing returns by pushing a rangeland as close as possible to a point of transition without "crossing" into another state. It is clear that temporally unpredictable and uncon- trollable factors (e.g., climate) and a high degree of spatial variability in site properties (e.g., amounts of run-on, variation in soil gravel content) interact with factors Fig. 7. An area within the clayey ecological site on the Sevilleta National Wildlife Refuge under management control to cause rapid north of Socorro, N.M. in the SD-1 land resource unit. Grasses in the foreground include changes in key, proximate variables in galleta (Pleuraphis jamesii Torr.), alkali sacaton, and burrograss. This refuge has been space and time. Thus, it is doubtful that excluded from livestock grazing since 1973. Despite rest, some large bare patches remain. indicator values for specific transitions These patches tend to have low soil aggregate stability indicating reduced capacity for will 1999, infiltration. This suggests that a transition has occurred, albeit at a small scale. be precise (Hobbs and Morton Muradian 2001). Management with respect to state-and-transition models calls insensitive locations will be incapable of agement responses. For example, by for conservative strategies aimed at detect- providing early warning of a broad-scale assigning land within the gravelly ecologi- ing and reversing trajectories toward vari- 3) transition, and it is necessary to inter- cal site to a creosotebush-invaded state, it able environmental thresholds. This is the pret fine-scale indicators within patches is believed that the recovery of grass pro- essence of "adaptive management" in the (e.g. Pellant et al. 2000) by considering duction may occur through use of a shrub- state-and-transition model context. patterns across entire landscapes. specific herbicide. In this case, competition for water and nutrients and propagule limitation define the properties of the Conclusions and recommendations Implications of state-and-transition State-and-transition models organize the models for management state. On the other hand, if there had been a transition from the creosotebush-invaded combined understanding of scientists and As agencies and individuals adopt the state to the creosotebush-dominated state, managers to explain ecosystem dynamics state-and-transition model framework, and the application of herbicide and seeding across variable rangeland landscapes. It is specific models are produced, it is impor- would be ineffective for regenerating a important to recognize that this framework tant to consider how these concepts can grassland state. State-and-transition mod- is not a blanket replacement of an outdated change on-the-ground management and els provide a framework for recognizing succession-retrogression model, but a how the models will be used. First, the distinctions among the causes of vegeta- complement to it that accounts for the state-and-transition model implies that, for tion change and distinguishing among existence of multiple equilibria, as well as all practical purposes, some areas are inca- these alternatives in the field. the return to equilibrium following pertur- pable of being restored to a historic state The grassland-shrubland transition on bations. Furthermore, the contrast between by intensive accelerating practices. What gravelly soils also illustrates a potential communities and states can be used to dis- constitutes "for all practical purposes" hazard of the state-and-transition tinguish the need for facilitating and accel- depends on the processes maintaining an approach. Degradation-caused shrublands erating management practices. The results undesirable state and the costs of reversing on gravelly soils may be presumed to be of a broad range of studies and personal them. If a transition is determined by the practically unrecoverable when they may, experiences can be summarized within this lack of propagules of a dominant species in fact, be recoverable through manage- framework and the resulting views can be at a site, it may be relatively inexpensive ment practices (J. Powell, personal com- continually updated as new observations to reverse it. On the other hand, if the tran- munication). If states are misidentified, and ideas emerge (Bradshaw and Borchers sition is caused by soil degradation, then sites may be "written off" prematurely, 2000). the cost of reversing it would be far leading to missed opportunities and con- In order for the understanding represent- greater. tinued degradation. For this reason, local ed in state-and-transition models to be By assigning land to a state, we assert knowledge and open debate are necessary communicated, refined, and compared the existence of particular processes and ingredients for applying state-and-transi- against the field observations of managers, constraints in that land that indicate man- tion models. State-and-transition models we suggest the inclusion of several com-

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 123 ponents in individual models. First, the Third, both academic ecologists and Bradshaw, G. A. and J. G. Borchers. 2000. catalog of states proposed by Westoby et resource managers should attempt to link Uncertainty as information: narrowing the al. (1989) can contain information that can information on other components of science-policy gap. Conserv. Ecol. 4: 7. be used to place land in ecosystem states rangelands valued by land owners and [online] URL: http://www.consecol.org/vol4/issl/art7, 15 to and, consequently, provide benchmarks society (e.g., biodiversity ) state-and- May 2002 for assessing the risk of a transition. transition models so that their responses Breshears, D. D. and F. J. Barnes. 1999. Ranges for quantitative indicator values, can be interpreted alongside those of Interrelationships between plant functional including line-intercept data of cover and plants and soils. (e.g., Van der Haegan et types and soil moisture heterogeneity for species composition, belt transect data for al. 2000, Bestelmeyer and Wiens 2001). semiarid landscapes within the grassland/for- invasive or encroaching species, gap inter- Our efforts in New Mexico lead us to est continuum: a unified conceptual model. cept data (Fig. 6), and soil aggregate sta- conclude that a great deal of information Landscape Eco1.14:465-478. bility index values (Herrick et al. 2001) has yet to be gathered, but that a great deal Brown, J. H., T. J. Valone, and C. G. Curtin. relate closely to many of the processes of understanding already exists that had 1997. Reorganization of an arid ecosystem in response to recent climate change. Proc. Nat. states in the New Mexico models not yet been captured in a useful form. defining Acad. Sci. USA 94:9729-9733. have been well- that we have produced and form the core The models we produced Brown, J. R.1994. State and transition models of the standard set of measurements we received by many field managers, for both for rangelands. 2. Ecology as a basis for are currently gathering. The reference complementing and organizing their rangeland management: performance criteria areas in which these data are gathered, (or knowledge as well as highlighting uncer- for testing models. Trop. Grassl. 28:206-213. estimates of reference values where such tainties. Models can improve the capacity Brown, J. R. and S. Archer. 1987. Woody areas no longer exist) are selected by of managers to evaluate the costs and con- plant seed dispersal and gap formation in a range professionals and scientists that consequences of management decisions and North American subtropical savanna wood- tribute to model development. Second, the rally researchers around unanswered ques- land: the role of domestic herbivores. Vegetatio 73:73-80. catalog of transitions can include a list of tions. For this reason alone, the production Brown, R. and S. Archer. 1999. Shrub is a worth- J. key qualitative and quantitative indicators, of state-and-transition models invasion of grassland: recruitment is continu- and descriptions of changes that occur in while endeavor. As models become firmly ous and not regulated by herbaceous biomass them, that suggest progress towards a tran- linked to a mechanistic understanding of or density. Ecol. 80:2385-2396. sition. Finally, the hypotheses, assertions, plant pattern, dynamics and indicators of Bulloch, H. E. and R. E. Neher. 1980. Soil literature and observations justifying the the processes underlying them, the models Survey of Dona Ana County Area, New structure of each model should be docu- will become invaluable. Mexico. USDA Soil Conservation Service, mented. This allows the mechanistic foun- Washington, D.C. dations of models to be evaluated and Bureau of Land Management. 2001. H-4180- challenged by the science and manage- Literature cited 1- Rangeland Health Standards, BLM Manual, Release 4-107, Bureau of Land ment communities so that model structures Management, Washington, D.C. can evolve. Allen-Diaz, B. and J. W. Bartolome 1998. Campbell, R. S. 1929. Vegetative succession Collaborations between the management Sagebrush-grass vegetation dynamics: com- in the Prosopis sand dunes of southern New agencies and the research community will paring classical and state-transition models. Mexico. Eco1.10:392-398. continue to play a large role in improving Ecol. App!. 8:795-804. Callaway, R. M. and F. W. Davis. 1993. the utility of the state-and-transition model Archer, S. 1989. Have southern Texas savan- Vegetation dynamics, fire, and the physical approach and several specific directions nas been converted to woodlands in recent environment in coastal central California. are indicated in this review. First, ecologi- history? Amer. Natur. 134:545-561. Ecol. 741567-1578. cal sites and land resource unit delin- Archer, S., C. Scifres, C.R. Bassham, and R. Clements, F. E. 1916. Plant succession: an in a of vegetation. eations should be refined to better reflect Maggio. 1988. Autogenic succession analysis of the development subtropical savanna: conversion of grassland Carnegie Inst., Washington Pub. 242:1-512. in and key differences soil, topographic, to thorn woodland. Ecol. Monogr. Comrie, A. C. and E. C. Glenn. 1998. climatic properties that create differences 52:111-127. Principal components-based regionalization in vegetation dynamics. Multivariate Ash, A. J., J. A. Bellamy, and T. G. H. of precipitation regimes across the southwest analyses of broad-scale soils, vegetation, Stockwel1.1994. State and transition models United States and northern Mexico, with an geomorphological (Gile and Grossman for rangelands. 4. Application of state and application to monsoon precipitation vari- 1997, McAuliffe 1994), and climatic transition models in northern Australia. Trop. ability. Climate Res. 10:201-215. (Comrie and Glenn 1998) datasets can be Grassl. 28:223-228. Creque, J. A., S. D. Bassett, and N. E. West. extremely useful in the effort. Second, Austin, M. P. 1985. Continuum concept, ordi- 1999. Viewpoint: Delineating ecological efforts must be continued to gather long- nation methods, and niche theory. Annu. sites. J. Range Manage. 52:546-549. Rev. Ecol. Syst.16:39-61. Davenport, D. W., D. D. Breshears, B. P. term data (in an experimental context Bellamy, J. A. and J. R. Brown. 1994. State Wilcox, and C. D. Allen. 1998. Viewpoint: where possible), stratified by soil series, and transition models for rangelands. 7. Sustainability of pinon-juniper ecosystems- and at appropriate spatial scales. Few Building a state and transition model for a unifying perspective of soil erosion thresh- experimental studies have directly management and research on rangelands. olds. J. Range Manage. 51:231-240. addressed the constraints to plant dynam- Tropical Grass!. 28:247-255. Devine, D. D., M. K. Wood, and G. B. ics on different soils (e.g., Northrup et al. Belsky, A. J. 1996. Viewpoint: Western Donart. 1998. Runoff and erosion from a 1999, Vandekerckhove et al. 2000), juniper expansion: Is it a threat to and north- mosaic tobosagrass and burrograss communi- despite clear evidence that these con- western ecosystems? J. Range Manage. ty in the northern Chihuahuan Desert grass- straints vary among sites. Studies of the 49:53-59. land. J. Arid Environ. 39:11-19. mechanisms producing transitions are Bestelmeyer, B. T. and J. A. Wiens. 2001. Dodd M. B., W. K. Lauenroth, and I. C. Ant biodiversity in semiarid landscape Burke 2000. Nitrogen availability through a needed to interpret long-term vegetation mosaics: the consequences of grazing vs. nat- coarse-textured soil profile in the shortgrass data and develop suitable indicators. ural heterogeneity. Ecol. App!. 11:1123-1140. steppe. Soil Sci. Soc. Amer. J. 64:391-398.

124 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Dyksterhuis, E. J. 1949. Condition and man- Hobbs R. J. and S. R. Morton. 1999. Moving Plant, R. E., M. P. Vayssieres, S. E. Greco, agement of rangeland based on quantitative from descriptive to predictive ecology. M. R. George, and T. E. Adams. 1999. A ecology. J. Range Manage. 2:104-115. Agrof. Systems 45:43-55. qualitative spatial model or hardwood range- Friedel, M. H. 1991. Range condition assess- Kingsland, S. E. 1985. Modeling Nature: land state-and-transition dynamics. J. Range ment and the concept of thresholds: a view- Episodes in the History of Population Manage. 52:51-59. point. J. Range Manage. 44:422-426. Ecology. Univ. of Chicago Press, Chicago, Rango, A., S. Goslee, J. Herrick, M. Freedel, M. H. 1997. Discontinuous change in Ill. Chopping, K. Havstad, L. Huenneke, R. arid woodland and grassland vegetation Kuehl, R. 0., M. P. McClaran, and J. Van Gibbens, R. Beck, and R. McNeely. (In along gradients of cattle grazing in central Zee. 2001. Detecting fragmentation of cover press). Remote sensing documentation of his- Australia. J. Arid Environ. 37:145-164. with line intercept measures in simulations of toric rangeland remediation treatments in Friedel, M. H., G. Pickup, and D. J. Nelson. Desert Grassland conditions. J. Range southern New Mexico. J. Arid Environ. 1993. The interpretation of vegetation change Manage. 54:61-66. Reitkerk, M. and J. van de Koppel. 1997. in a spatially and temporally diverse and Laycock, W. A.1991. Stable states and thresh- Alternate stable states and threshold effects Australian landscape. J. Arid Environ. olds of range condition on North American in semiarid grazing systems. Oikos 24:241-260. rangelands: a viewpoint. J. Range Manage. 79:69-76. Reynolds, F., R. A. Virginia, P. R. Kemp, Fuhlendorf, S. D., D. D. Briske, and F. E. 44:427-433. J. A. G. De Soyza, and D. C. Tremmel.1999. Smeins. 2001. Herbaceous vegetation change Ludwig, J. A. and D. J. Tongway. 1995. of landscapes and its Impact of drought on desert shrubs: effects of in variable rangeland environments: The rel- Spatial organization function in semiarid woodlands, Australia. seasonality and degree of resource islands ative contributions of grazing and climatic Landscape Ecol.10:51-63. development. Ecol. Monogr. 69:69-106. variability. Appl. Veg. Sci. 4:177-188. Ludwig, A., D. Tongway, and S. G. Rodriguez Iglesias, R. M. and M. M. George, M. R., R. Brown, and J. Clawson. J. J. J. Marsden.1994. A flow-filter model for sim- Kothman. 1997. Structure and causes of 1992. Application of nonequilibrium ecology ulating the conservation of limited resources vegetation change in state and transition to management of Mediterranean grasslands. in spatially heterogeneous, semi-arid land- model applications. J. Range Manage. 50: J. Range Manage. 45:436-440. scapes. Pacific Conservation Biol. 399-408. R. B. Grossman. 1997. The Gile, L. H. and 1:209-213. Scheffer, M., S. Carpenter, J. A. Foley, C. Desert Project Soil Monograph. Natural Ludwig J. A., G. N. Bastin, R. W. Eager, R. Folke, and B. Walker. 2001. Catastrophic Resources Conservation Service, USDA Nat. Karfs, P. Ketner, and G. Pearce. 2000. shifts in ecosystems. Nature 413:591-596. Soil Surv. Center, Lincoln, Neb. Monitoring Australian rangeland sites using Schlesinger, W. H., J. F. Reynolds, G. L. Gosz, J. R. 1993. Ecotone hierarchies. Ecol. landscape function indicators and ground- Cunningham, L. F. Huenneke, W. M. Appl. 3:369-376. and remote-based techniques. Environ. Mon. Jarrell, R. A. Virginia, and W. G. Grossman, D. H., D. Faber-Langendoen, and Assess. 64:167-178. Whitford. 1990. Biological feedbacks in A.S. Weakley, M. Anderson, P. McAuliffe, J. R. 1994. Landscape evolution, global desertification. Sci. 247:1043-1048. Bourgeron, R. Crawford, K. Goodin, S. soil formation, and ecological patterns and Shiflet, T. N.1975. Range sites and soils in the Landaal, K. Metzler, K. Patterson, M. processes in Sonoran desert bajadas. Ecol. United States. p. 26-33 In: D. N. Hyder (ed.) Pyne, M. Reid, and L. Sneddon. 1998. Monogr. 64:111-148. Arid Shrublands: Proceedings of the Third Terrestrial Vegetation of the United States. McClaran, M. P, D. L. Angell, and C. Workshop of the U.S./Australia Rangeland Volume 1: The National Vegetation Wissler. In press. Santa Rita Experimental Panel. Soc. for Range Manage., Denver, Classification System:Development, Status, Range Digital Database: User's Guide. Colo. and Applications. The Nature Conservancy, USDA Forest Serv. Rocky Mountain Res. Society for Range Management, Task Group Arlington, Virg. Sta., Gen. Tech. Rept., RMRS-GTR-XXX. on Unity in Concepts and Terminology. Hennessy, J. T, R. P. Gibbens, J. M. McPherson, G. R. 1995. The role of fire in 1995. New concepts for assessment of range- Tromble, and M. Cardenas. 1983. desert grasslands. p 130-151 In: M. land condition. J. Range Manage. Vegetation changes from 1935-1980 in McClaran and T. R. Van Devender (eds) The 48:271-282. mesquite dunelands and former grasslands of desert grassland. Univ. of Arizona Press, Stafford Smith, D. M. and S. R. Morton. southern New Mexico. J. Range Manage. Tucson, Ariz. 1990. A framework for the ecology of arid 36:370-374. Miller, M. E. 1999. Use of historic aerial pho- Australia. J. Arid Environ.18:255-278. Herbel, C. H. and R. P. Gibbens. 1989. tography to study vegetation change in the Stringham, T. K., W. C. Krueger, and P. L. Matric potential of clay loam soils on arid Negrito Creek watershed, southwestern Mew Shaver. 2001. States, transitions, and thresh- rangelands in southern New Mexico. J. Mexico. The Southwestern Natur. olds: Further refinements for rangeland Range Manage. 42:386-392. 44:121-137. applications. Oregon State Univ. Agr. Exp. Herbel, C. H., F. N. Ares, and R. H. Wright. Muller, C. H. 1940. Plant succession in the Sta., Special Rep. 1024. 1972. Drought effects on a semidesert grass- Larrea-Flourensia climax. Ecol. 21206-212. Strittholt, J. R. and R. E. J. Boerner. 1995. land range. Ecol. 53:1084-1093. Muradian, R. 2001. Ecological thresholds: a Applying biodiversity GAP analysis in a Herrick, J. E. 2000. Soil quality: an indicator survey. Ecol. Econ. 38:7-24. regional nature-reserve design for the edge of of sustainable land management? Appl. Soil Neilson, R. P. 1986. High resolution climatic Appalachia, Ohio (USA). Conserv. Biol. 9: Ecol.15:75-83. analysis and southwest biogeography. Sci. 1.492-1505. Herrick, J. E., J. R. Brown, A. Tugel, P. L. 232:27-34. Svejcar, T. and J. R. Brown. 1991. Failures in Shaver, and K. M. Havstad. 2002. Northrup, B. K., J. R. Brown, and J. A. Holt. the assumptions of the condition and trend Application of soil quality to monitoring and 1999. Grazing impacts on the spatial distrib- model for managing natural ecosystems. management: Paradigms from rangeland ution of soil microbial biomass around tus- Rangelands 13:1.65-167. A. and abuse veg- ecology. Agro. J. 94:3-11. sock grasses in a tropical grassland. Appl. Tansley, J.1935. The use of Ecol. Herrick J. E., W. G. Whitford, A. G. de Soil Ecol. 399:1-12. etational concepts and terms. M., P. D. Pyke, and 16:284-307. Soyza, J. W. Van Zee, K. M. Havstad, C. Pellant, Shaver, J. indicators of USDA Natural Resources Conservation A. Seybold, and M. Walton. 2001. Field Herrick. 2000. Interpreting rangeland health, version 3. Tech. Reference Service. 1997. National Range and Pasture soil aggregate stability kit for soil quality and rangeland health evaluations. Catena 1734-6. U. S. Department of the Interior, Handbook. U.S. Dep. Agr., Washington, DC. Land Management, National USFS. 2000. National Forest System Land 44:27-35. Bureau of Science and Technology Center Information Resource Management Planning; Final Rule. and Communications Group, Denver, Colo. Federal Register 65 (219.20): 67574-67575.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 125 Van Auken, 0. W. and J. K. Bush. 1997. Westman, W. E. 1978. Measuring the inertia Whisenant, S. G. 1999. Repairing damaged Growth of Prosopis glandulosa in response and resilience of ecosystems. Biosci. wildlands: a process-orientated, landscape- to changes in aboveground and belowground 28:705-710. scale approach. Cambridge Univ. Press, interference. Ecol. 78:1222-1229. Westoby, M. 1980. Elements of a theory of Cambridge, U.K. Van der Haegen W. M, F. C. Dobler, and D. vegetation dynamics in arid rangelands. Whitford, W. G., G. Martinez-Turanzas, Israel J. Bot. 28:169-194. J. Pierce. 2000. Shrubsteppe bird response to and E. Martinez-Meza.1995. Persistence of Westoby, M, B. Walker, and I. Noy-Meir. habitat and landscape variables in eastern desertified ecosystems: explanations and 1989. Opportunistic management for range- Washington, USA. Conserv. Biol. implications. Environ, Monitoring and lands not at equilibrium. J. Range Manage. Assess, 37:319-332. 14:1145-1160 42:266-274. Wondzell, S. M., G. L. Cunningham, and D. Vandekerckhove, L., J. Poesen , D. 0. Wiegand, T. and S. G. Milton. 1996. Bachelet.1996. Relationships between land- Wijdenes, J. Nachtergaele, C. Kosmas, M. Vegetation change in semiarid communities: forms, geomorphic processes, and plant com- J. Roxo, and T. deFigueiredo. 2000. Simulating probabilities and time scales. on a in the Vegetatio 125:169-183. munities watershed northern Thresholds for gully initiation and sedimenta- Chihuahuan desert. Landscape Ecol. tion in Mediterranean Europe. Earth Surface Wiens, J. A. 1984. On understanding a non- 11:351-362. Processes and Landforms 25:1201-1220. equilibrium world: Myth and reality in community patterns and processes, p. 439-457 In. Wright, H. A., S. C. Bunting, and L. F. Weltzin J. F., S. Archer , and R. K. Neuenschwander. 1976. Effect of fire on Heitschmidt. 1997. Small-mammal regula- D. R. Strong, D. Simberloff, L. G. Abele, and A. B. Thistle (eds). Ecological Communities: honey mesquite. J. Range Manage. tion of vegetation structure in a temperate Conceptual Issues and Evidence Princeton 29:467-471. Ecol. Savanna. 78:751-763. Univ. Press, Princeton, N.J.

WE CATER TO SOME PRETTY TOUGH CUSTOMERS

... with the West's largest selection of pasture, range and reclamation seed. Over 600 native and adapted varieties for custom seed blends, plus a full line of erosion control and hydroseeding products. ErrAronmental consulting services and training available.

SEED

Call or fax for our catalog. (801) 768-4422, fax (801) 768-3967 Granite Seed Co., I 697 W. 2100 North, Lehi, UT 84043 www. graniteseed. corn

126 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 J. Range Manage. 56: 127-132 March 2003 Hay-meadows production and weed dynamics as influenced by management

DANIELE MAGDA, JEAN-PIERRE THEAU, MICHEL DURU, AND FRANOIS COLENO

Authors are researchers, Institut National de la Recherche Agronomique, Departement Systemes Agraires et le Developpement, Unite d'Agronomie, B.P. 27, 31326 Castanet-tolosan, France.

Abstract Resumen

Managers of extensive livestock systems generally have 2 goals Los manejadores de sistemas extensivos de ganado general- for permanent grassland management: to obtain sufficient dry mente tienen dos metas para el manejo de zacatales perennes: matter to feed animals and to avoid the establishment and domi- Obtener suficiente materia seca para alimentar a los animales y nance of unpalatable species. Hay production in French evitar el establecimiento y dominancia de especies no gustadas Pyrenean meadows is dependant on the need to balance grazing por el ganado. La produccion de heno en praderas "French and cutting dates to produce maximum biomass for hay stock Pyrenean" depende de la necesidad de balancear las fechas de and to prevent seed recruitment of Chaerophyllum aureum L., apacentamiento y corte para producir la maxima biomasa de one of the major invasive unpalatable species. Experiments and heno para almacenar y prevenir la acumulacion de semilla de observations on a set of meadows within farms show that optimal Chaerophyllum aureum L., una de las principales especies inva- dates calculated from degree-days for cutting or spring grazing soras sin gustocidad. Experimentos y observaciones hechas en un of C. aureum fitted to seed production and apex development grupo de praderas dentro de granjas muestran que las fechas respectively, decreases hay yield. This decrease is related to the optimas calculadas a partir de grados-dia para corte o apacen- earliness of the cut in regard to sward growth or to the biomass tamiento en primavera del C. aureum se ajustaron a la produc- loss by senescence due to the vegetative regrowth of the sward cion de semilla y desarrollo del spice respectivamente y dismin- after spring grazing. Compromises and choices have to be made uyen el rendimiento de heno. Esta disminucion es relacionada a for each meadow by the farmer according to its potential produc- to temprano del corte con respecto al crecimiento de la pradera o tion, the risk of invasion by C. aureum, and its role in the forage a la perdida de biomasa por senescencia debida al rebrote vege- system. tativo de la pradera despues del apacentamiento en primavera. Arreglos y elecciones tienen que ser hechos por el granjero para cada pradera de acuerdo a su potential de produccion, el riesgo de invasion de C. aureum y su papel en el sistema forrajero. Key Words: management sequences, permanent grasslands, herbage growth, weed population demography

species are often driven out by competition. For these reasons, In European livestock systems in mountain areas, grazed mead- total dry matter production and the proportion eaten by animals is ows often represent around 80% of the permanent grassland area reduced significantly. at the farm level. The remaining 20% is grass used exclusively to Generally, the impact of management practices on biomass pro- yield hay or silage for animal wintering. duction and species population dynamics have been studied sepa- A critical point of this management system is to ensure, each rately and very often only with regard to a simple technical oper- year, sufficient dry matter yield of hay or silage to feed animals ation (fertilization or grazing and cutting regimes). The purpose during the entire winter period (Theau et al. 1998). Any shortfall of this study was to define efficient grassland management sys- in stored feed requires hay purchases which threaten the econom- tems that maximize dry matter yield and limit the population den- ic sustainability of such systems. Topography and socio-econom- sities of undesirable species. Weed population control involves ic trends favor valley bottom meadows, which are used both for finding a better match between management patterns and sensi- grazing and hay production. Relatively recent changes in productive phenological stages of the target species to maintain a con- tion organisation and in grassland management schedules stant or decreasing population growth rate. In this study, we used (chronological order of farming operations and characteristics of observations, surveys and experiments on a set of permanent the operations), have reinforced competition between grazing and meadows belonging to French Pyrenean livestock owners to cutting. Moreover, populations of undesirable species in perma- assess the effect of cutting date, presence and date of spring graz- nent grasslands have markedly increased during the last few ing and manuring on dry matter yield at haymaking time and on decades as a result of less intensive management practices in the population density of Chaerophyllum aureum L. representing farming systems, leading to a reduction of grazing or cutting a major invasive perennial weed of meadows. We focused on the intensity (de Hullu et al. 1985, Bobbink and Willems 1993). Due impact of these technical operations on recruitment of C. aureum to the unpalatable nature of these species, desirable forage seed as the most sensitive stage in the life cycle of this species (Magda and Jarry 2000). Manuscript accepted 11 Jul. 02.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 127 Materials and Methods The consequences of defoliation by were chosen to avoid underestimating grazing of the C. aureum adult apex on the populations as a result of missing juve- Study site: recording of manage- development of reproductive structures niles in spring or intensively grazed adults. were tested experimentally. The dynamics Chosen at random over the whole mead- ment operations of apex development of C. aureum adults ow, the sample of 8 different plots allows Surveys of management regimes, from spring regeneration were followed to the population to be estimated at the field herbage production and C. aureum popu- determine the height of the apex within the level, taking into account thef spatial varia- lation density were carried out on 52 sward at different times. Twenty-two plots tion in density. Abundance was visually meadows belonging to 4 farms in the Erce (2.50 x 1.20 m) were located in an assessed for each plot, noting 1 if C. valley in the Central French Pyrenees (0° ungrazed meadow colonized by C. aureum aureum was present, otherwise 0. Total 30' E, 42° 48' N). These permanent mead- with 8 replicate plots at each of the 4 scores for the whole meadow could thus ows are natural grassland communities observation dates. Ten main stems per plot vary from 0 to 8. with high species diversity. About 19 dif- were sampled every 12 days between the Correlations between management ferent species represent 90% of the grazed beginning of spring regrowth (23 March, regime and abundance of C. aureum plants meadow biomass: about 60% of them are i.e. at 394 degree days) and the end of were analyzed by a non-parameterc test grasses (data not published). vegetative growth (4 May, 782 degree adapted for small and independant sam- These 19 or so species, along with high- days). The height above ground and length ples, the so-called Wilcoxon-Mann- er altitude communally owned pasture, of each stem apex were measured at each Whitney test (Scherrer 1984). form the basis of traditional Pyrenean date. After each sampling date, the 8 plots meat production systems. Each meadow were cut to record the effect of removing management regime was identified from Herbage Mass Measurement the main stem apex on the reproductive Each meadow's production was estimat- direct observations and surveys of farm- capacity of the individual. Plots were cut ers. The different technical operations ed by measurement of standing herbage with a mower, leaving a residual herbage mass from the same plots which were used (e.g., cutting, grazing, manuring, etc.) and height of 5 cm. This height was measured their timing were recorded. Grazing takes to assess C. aureum abundance and taking with a sward stick in most of the meadows the place in spring and fall, before and after into account the spatial variation of after grazing. sward over the whole meadow area. These haymaking. Sward mean height after Germination tests were done on C. spring grazing was estimated from 50 measurements were made before spring aureum seeds after a period in manure to and sum- sward stick measurements along a transect grazing (20 April-19 May 1998) see whether manuring might affect long- (4 within each meadow (Duru and Ducrocq mer cutting June-13 July 1998) for the distance seed dispersion for this species. grazed meadows. For the ungrazed mead- 1998). This height assesses the grazing Four cotton bags, each containing 100 intensity. ows, a first measurement of herbage mass seeds harvested from C. aureum popula- was also made at the time of spring graz- tions established in the study site, were ing. Each sward plot was cut at 2 cm C. aureum life cycle, apex develop- placed in January in 2 different manure above ground level with a small hand ment and seed viability heaps made the month before. Bags were mower. Each plot biomass was weighed Chaerophyllum aureum is one of about placed at 2 different depths within each after drying at 80°C for 48 hours. 300 species present naturally in permanent heap at 10 and 40 cm under the surface. Temperatures were recorded regularly grasslands between 500 and 1,800 m in indices determination Europe (Gonnet 1989). Being very com- with a Campbell thermocouple probe. Nutrient The herbage sampled for mass measure- petitive, it is responsible for major infesta- Seeds were excavated the following spring ment was milled through a 0.8 mm screen tions of hay meadows in less intensive (after 4 months) or a year later (after 16 and analyzed for nitrogen (Kjeldhal 1883), livestock systems. Grazed only at a very months) to represent the times generally phosphorus (Murphy and Riley 1962) and early stage, it very soon becomes unpalat- observed from farmers' practices between potassium. To assess nitrogen status, we able to sheep and cows because of the manure collection and its spreading. Seeds used the "dilution curve" method (Lemaire concentration of lignified tissues in its were buried in boxes filled with soil taken and Salette 1984). During herbage shoots. It is a long-lived perennial (at least from the study site but from meadows not regrowth (after cutting or grazing), nitrogen 7 years) that spreads solely by sexual colonised by Chaerophyllum aureum and (N) decreases as the above- reproduction, producing achenes which placed directly in the environmental con- concentration ground herbage mass (HM) increases ripen from late July to late August and are ditions of the site. Seed germination rate according to the relation: N% a * HM a, distributed by dehiscence up to 20 m from was calculated from seedling surveys = nitrogen being the nitrogen concentration in the mother plant (Gonnet 1989). The seed- made in spring, which is the only time of per 100 of biomass, herbage mass the bank is transient (Grime et al. 1988). year this species germinates. This rate was g g above-ground herbage mass in tonnes of Previous experimental studies on demog- compared with that of control seeds which dry matter ha', a the nitrogen concentra- raphy of C. aureum populations have elu- had never been stocked in manure but har- tion when herbage mass 1 t ha', the cidated its population dynamics strategy vested at the same time and from the same = 1 coefficient of nitrogen dilution. With opti- and identified sensitive stages in its life populations. mum nitrogen nutrition, the a and (3 para- cycle in relation to management practices meters are constant for all species, even (Magda 1998, Magda and Jarry 2000). Chaerophyllum aureum abundance legumes: 4.8 and (3 0.32 Duru et al. Restricting recruitment of new individuals a = = ( surveys 1997, Lemaire and Gastal 1997). We used is the only way to maintain or decrease Chaerophyllum aureum abundance was the parameters of this control curve to cal- population density by management of recorded within each meadow at 2 differ- culate a nitrogen index (Ni) - the ratio adult fecundity, seed dispersion and ent dates in spring and summer on 4 x between the measured nitrogen concentra- seedling survival. 0.25 m2 plots. These 2 sampling dates tion of the above-ground herbage mass

128 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 (HM) and the optimum nitrogen concen- allowed the plants to reach 70% of their 1999). This result confirms experimental tration as previously defined (Lemaire and growth potential (Lemaire and Gastal data for a range of nutrient supply treat- Gastal 1997): 1997). The nutrients phosphorus ( Fig. lb) ments over 2 growing seasons (Calviere and potassium (not shown) indices were 1994) where it was observed that the dry Ni=l00xN14.8HM -0.32 (1) more variable, the lowest values being matter accumulation increased up to 1500 below 50 and the highest greater than 100. degree days. potassi- Critical herbage phosphorus and They were significantly mutually correlat- Following spring grazing, hay cutting um concentrations were established, in ed (Test Wilcoxon-Mann-Whitney, time, averaged between 750 and 1000 relation with nitrogen concentration, and P<0.001). Standing herbage mass at the degree days (Fig le). It was significantly and Ki indices Pi (Phosphorus) hay stage varied from 1500 to 8000 kg/ha lower (P < 0.05) than that of ungrazed as the (Potassium)were defined ratio (Fig le). meadows (54% on average) and positively between the actual phosphorus and potas- At hay cutting time, the standing correlated to the phosphorus status and the ones sium concentration and the optimal herbage mass (SHM) on ungrazed mead- residual sward height (which reflects the (Duru and Thellier 1997): ows averaged between 1000 and 1500 intensity of spring grazing) and negatively to spring grazing date: Pi=100xP%/(0.15+0.065(N%)) (2) degree days (Fig le). It depended on the correlated interaction between phosphorus and potas- Ki=l00xK%/(1.6+0.525(N%)) (3) sium nutrient status and time of harvest: GT SHM =1931+32.6 Pi +132 RSH- O(3.1); The indices could vary from 30 for very HT Febl deficient herbage to 100 when the avail- (SHM=-463 + 9[0.035Pi + 0.022 Ki]; n = 27, r2 = 0.74, se = 400). ability of nutrients allowed maximal Febl This means that the later the grazing growth. n=22, r2 = 0.42, se = 742, P < 0.001) (4) time ( 6) and the lower the residual sward Time scale definition There was no effect of nitrogen status height, the smaller was the standing Two difficulties occur when comparing on late hay yield but only one of phospho- herbage mass at the time of the hay cut. meadows for herbage mass production. rus and potassium through stem growth The difference in accumulated tempera- First, meadows ranged from 600 to 1,000 (Duru and Calviere 1996). As there was a tures between spring grazing and hay yield m in altitude; secondly, biomass measure- positive significant effect of accumulated times were not significant, meaning that ments were not made on all plots on the temperature over the growth period, it the maximum yield was reached in the same date. For this reason, time is means that for most of the plots, the maxi- interval between the shortest (444 degree expressed in degree days (degree days) i.e. mum yield was not reached at the time days) and the longest (956 degree days) the accumulated daily mean temperatures where the herbage yield was measured, regrowth times after grazing. This obser- period for 21 of the 22 vation was in agreement with leaf lifespan from 1 February or from the previous whereas the harvest defoliation. This date corresponds to a plots occurred after 1100 degree days, for vegetative swards of cocksfoot change in the response of leaf extension to which corresponds to the flowering period (Dactylis glomerata L.) and tall fescue the senescence temperature as a consequence of the tran- of most of the species (Calviere and Duru (Festuca arundinacea L.), sition from vegetative to reproductive development (Parsons and Robson 1980). A correction (-0.6°C per 100 m) was made to take into account the effect of altitude on temperature and so on growth processes.

Results

Management schedules 50 55 60 65 70 75 80 85 90 95100 50 55 60 65 70 75 80 85 90 95 100 0 200 400 600 800 1000 1200 Herbage N index Herbage P index Accumulated temperatures Spring grazing occurred on 32 of the 52 plots between about 500 and 1100 degree days after 1 February (Fig lc), and the residual sward height varied from 3 to 18 cm (Fig id). Manure was spread on 40% of spring grazed meadows and on 82% of ungrazed meadows. The hay was cut after about 1500 degree days for ungrazed meadows and between 1200 and 1500 degree days for grazed meadows. . _. 3 6 9 12 15 18 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8

Residual sward height (cm) Herbage mass ( t per ha) Abundance of C. aureum (score) Herbage nutrient status and stand- Fig. 1. Distribution of meadows (ungrazed in white; grazed in black) according to different ing herbage mass variables. a) Herbage N Index, b) Herbage phosphorus index, c) Defoliation dates (degree from 55 to 85 (Fig. Herbage Ni varied days from 1 Feb. ), d) Residual sward height after defoliation (cm), e) Herbage mass la). This means that on average, the nutri- (kg/ah), 0 Abundance classes of C. aureum. ent supply from fertilizers and soil

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 129 process beginning at about 800 degree days (Duru et al. 1993), and the maximum 0 yield being reached between 800 and 1000 degree days. For both forms of spring management, maximum yield occurred when senescence I rate became equal to growth rate (Parsons r

1988, Scarnecchia 1988). When there is r no spring grazing, it is reached at a high cr2 standing herbage mass and a later date V because of the presence of stem material. a Indeed, over the stem elongation period, 0 200 400 600 800 1000 1200 1400 1600 1800 2000 photosynthesis rate increased up to 65% Daily mean of accumulated temperatures from first February (dd) (Woledge 1978) and was related to the height of the apex (Leafe 1978, Parsons Fig. 2. C. aureum apex growth dynamics. Height of apex above ground (Y, cm) plotted 1988). This increase in photosynthesis rate against thermal time (X, Degree-Days) (dotted line) with a confidence interval calculated acts as a sink (Varlet-Grancher et al. from a total of 280 measurements. The beginning of seed maturation period is represented 1982), but the proportion of stem depends by the vertical line. Horizontal line indicates the actual hay yield period. on nutrient status (Duru et al. 2000) and on grazing management (Briske 1996). cant and negative effect of grazing on C. tion according to date of grazing or cutting. Intense grazing in the spring removes the aureum population density (Test For example, for a spring grazing date at apex of grasses and forbs, causing the Wilcoxon-Mann-Whitney, z = 5.6, P = 600 degree days, the maximum threshold plants to remain vegetative. Senescence 0.05), (Fig, if). About 56 % and 21 % of of residual herbage height was 11.8 cm. rate depends on leaf lifespan and phenolo- whole fields sampled fell within the abun- Moreover, at this date, C. aureum is not gy of the species. Leaf lifespan, which dance classes 0 and 1 respectively. lignified and is entirely palatable for ani- determines the beginning of senescence, is Seventy six percent of the sampled fields mals. Combining the data of Fig lc, id and relatively constant for a given species which were not colonized by C. aureum Fig 2, we deduce that for 24 of the 27 when expressed in degree-days (Duru et were grazed. The 2 highest abundance val- plots, the C. aureum apex must have been al. 1993). For vegetative swards, the ues of C. aureum (abundance classes 7 removed and that in most of the swards senescence process involved the whole and 8) were found only in ungrazed fields. this species remained vegetative. leaf, whereas over the reproductive phase, These results clearly indicate the role of For the ungrazed fields, manure applica- the stem component of the sward did not spring grazing in limiting C. aureum pop- tion was correlated significantly and posi- senesce, at least to begin with. ulation density. Because adult resistance tively to C. aureum population densities Spring grazing management had the to defoliation has been already demon- (Test Wilcoxon-Mann-Whitney, z = 2.26 greatest effect on standing herbage mass at strated (Magda 1998), these results con- P = 0.05). This confirms the hypothesis the hay stage, and on the pattern of firm that grazing affects the recruitment that seed can be imported with manure. herbage mass accumulation. The effect of process. It is likely that spring grazing first Experiments on seed viability after storage herbage nutrient status was less signifi- increases the seedling mortality rate due to in manure show that C. aureum seeds are cant. We expected that, due to apex trampling and biomass defoliation. We capable of germination even after several removal, standing herbage mass with suspect that seedlings from emergence to months storage in manure and with greater spring grazing would be about 50% lower the juvenile stage differ in sensitivity to success for seeds distributed near the heap and would increase only up to 1000 degree biomass removal, but the most sensitive surface when it is not turned over (Table days after grazing time, while it increased stage to grazing has not been determined. 1). These results confirm the potential role at least up to 1500 degree days after the Secondly, spring grazing decreases the of manuring in dispersing C. aureum seeds end of winter when there is no spring overall fecundity of the population by and in increasing the risk of seed immigra- grazing. removing the shoot apex. tion into fertilised meadows. We show that apex removal during The fact that the effect of manure supply spring regrowth leads to purely vegetative was not significant for grazed fields sup- Impact of the technical operations stem development for C. aureum. To ports the view that grazing is a major fac- schedule on C. aureum population remove the main apex, date and height of tor limiting seedling establishment and/or density defoliation through cutting or grazing has population fecundity. Chaerophyllum aureum was observed to be correlated (Fig. 2). This figure shows The presence of C. aureum in 13% of throughout the range of nitrogen index apex development schematically, so as to grazed fields can be explained by the spa- (Ni) measured within the whole sampled define the most efficient height for defolia- tial heterogeneity of grazing, which main- meadows. However, it was less abundant at nitrogen indices < 56 (Fig la, 1f). Table 1. Percentage of germination on C. aureum seeds after a short or long period of storage in Previous experimental studies have shown manure at 2 different depts,10 cm and 40 cm under surface. that a similar minimum fertility threshold (55 nitrogen index) is necessary for the Short stocking Long stocking establishment of C. aureum (Magda Control (4 months) (16 (months) -10 cm -40 cm -10 cm 40 cm 1998). In this study site, the herbage nitrogen index varied from 55 to 85, indicating (%) ------that all areas are liable to be invaded by 69.6 72 0 61 0 71 0 54 0 this species (Fig. 1a). There was a signifi-

130 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 tains favorable microsites and consequent- the residual herbage height, which deter- yield. This risk is also low if the nutrient ly allows seedlings to escape from distur- mines the probability that the apex escapes index is sufficiently low to prevent any bance and mortality. Some adults of C. grazing. The lower the grazing intensity, natural colonization. aureum could be also leniently grazed, the greater is the likelihood that C. aureum Manuring greatly increases the probabil- particularly with late grazing. This phe- adults will not be grazed. Nevertheless, ity of seeds reaching non-invaded mead- nomenon, in addition to heavy seed shed- the efficiency with which grazing removes ows, particularly when the meadows are ding, can maintain populations at a rela- the C. aureum apex will be related to the not close to other seed sources. In this lat- tively high density within grazed fields. It ingestion behavior of animals, grazing ter case, farmers can adopt a prevention can also correspond to transitory changes intensity (number of animals per unit of strategy and maintain spring grazing to of management, creating an opportunity land area) and number of days of grazing. control potential seedling populations or for C. aureum seedlings to be established. Moreover, as time passes and C. aureum take the risk and alter the management To define the timing of the sensitive plants mature, diet selection by animals schedule only when adults appear in the period for efficient defoliation for seed will increase the probability for apex to meadow. production control, phenological surveys escape to defoliation. Therefore, a lower The demographic strategy of C. aureum have been conducted within C. aureum limit of grazing date at 600 degree days is populations is clearly characterized by a populations already established in hay recommended, with a residual herbage great ability to reach high densities and to meadows. Results show that cutting for height of 11.8 cm (Fig. 2). This date cone- maintain these densities for a long time. hay before 1200 degree days prevents the sponds to the lower limit of apex height. Prevention of establishment therefore seed maturation process. Before this date, grazing does not remove appears to be the best management strate- the apex of many species and especially C. gy for this species. aureum. Delaying grazing, as well as Discussion increasing the grazing intensity, increases the probability of removing the stem apex Conclusions for many grass species (Gillet 1980). From results on herbage yield and C. If the farmer wants to minimize any risk aureum population demography, it is pos- This study shows that the schedule of of seed production, hay cutting should sible to propose a grazing and cutting management operations to control weed occur before 1200 degree days. But in this schedule to optimize fodder production on species can be proposed from a knowledge case, hay yield is considerably depressed. hay meadows colonized by C. aureum of the developmental dynamics of the Cutting should be done not later than 1000 (Table 2). For ungrazed meadows, 1 cut species and the demographic strategy of degree days during the regrowth period to before 1200 degree days would prevent C. its population. Differentiation of popula- avoid losses through senescence. Finally, a aureum seed production. Also, sward bio- tion dynamics traits among grassland choice has to be made by farmers: whether mass losses associated with the develop- weed species is possible and allows differ- to ensure maximum yield production and ment of reproductive structures would be ent appropriate control strategies to be accept a risk of recruitment of new indi- limited because of the earliness of the cut. proposed. We have shown that spring viduals or to accept a compromise for hay Spring grazing could limit recruitment grazing and cutting for hay can efficiently production and ensure long term C. of new C. aureum individuals through 2 control C. aureum infestation when they aureum population control. processes: enhancement of seedling mor- are applied, relatively constantly, at appro- In the case of hay meadows not already tality (Magda and Jarry 2000) and preven- priate dates for the morphological and colonized by C. aureum, we believe that tion of development of reproductive struc- reproductive patterns of this species. the risk of invasion is largely dependent through apex removal for C. aureum However, C. aureum population dynamics tures on manure spreading. If no manure is and also for 3 grasses associated with C. are not independent of forage species applied, the risk of immigration of C. aureum (not shown). Grazing efficiency is dynamics within the grassland community. aureum seeds is relatively low and man- dependent on its intensity as defined by The sequence of management operations agement can be based on maximizing hay

Table 2. Sequences and timing of technical operations (cutting and grazing) for control of C. aureum and maximazation of hay yield (degree days, standing herbage mass).

Sequence of technical Limitation of C. aureum abundance Maximization of hay yield operations Hay cutting <1200 degree days <1500 degree days, standing herbage mass = f(herbage nutrient index)

Grazing - Cutting Grazing Cutting If grazing begins at < 600 degree days, the regrowth is reproductive - hay yield up to 1500 degree days should minimize herbage losses. If grazing begins at >600 degree days, the regrowth is vegetative - yield time around 1000 degree days after beginning of grazing (about 1500 degree days after 011 02 if the swards were grazed at 600 degree days) should minimize herbage losses. <600 degree days <1200 degree days >600 degree days <1200 degree days or not f(risk level accepted)

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 131 needed to control C. aureum depresses Duru, M. and I. Calviere. 1996. Effets of Murphy, J. and J.B. Riley 1962. A modified herbage yield and could adversely affect phosphorus and N nutrition status and of single solution method for the determination forage production. Spring grazing limits botanical composition of permanent pastures of phosphate in natural waters. Analytica the establishment of new individuals of on their growth in spring. Agronomie. Chimica Acta. 27:31-36. 16:217-229. A.J. 1988. The effect of season and this species but also reduces hay yield to Parsons, Duru, M. and H. Ducrocq. 1998. La hauteur management on the growth of grass swards, that of the regrowth biomass. du couvert prairial : un moyen d'estimation p.129-178. In: M. Jones, and M. B. Lazenby In a given year, estimation of herbage de la quantite d'herbe disponible. Fourrages. (ed.)., The Grass Crop. Chapman and Hall. losses, with or without spring grazing, has 154:173-190. Parsons, A.J. and M.J. Robson. 1980. to be made according to the size of the C. Duru, M. and L. Thelier. 1997. N and P-K Seasonal changes in the physiology of 524 aureum population and the corresponding status of herbage : use for diagnosis of grass- perennial ryegrass (Lolium perenne L.). 1. herbage mass (mainly C. aureum stems) lands. In Diagnostic procedures for crop N Response of leaf extension to temperature which was rejected when animals were fed management, Colloque de l' INRA. during the transition from vegetative to indoors. Without control, C. aureum will Poitiers.p.125-138. INRA. reproductive growth. Ann. Bot. 46: 435-444. always become rapidly dominant, sup- Duru, M., G. Lemaire, and P. Cruz. 1997. Scarnecchia, D.L. 1988. Grazing, stocking and The N requirements of grasslands, p. 59-72. production efficiencies in grazing research. J. pressing the remaining species in the com- In: G. Lemaire (ed.), Diagnosis of the N Range. Manage. 41:279-281. munity. In this case, biomass is productive Status in Crops. Springer Verlag Press, Scherrer, B. 1984. Biostatistique. Gaetan only if C. aureum is maintained at a leafy Berlin. Morin (ed). stage, preventing lignified stem develop- Duru, M., V. Delprat, C. Fabre, and E. Theau, J.P., F.C. Coleno, M. Duru, and Y. ment and this is possible only if defolia- Feuillerac. 2000. Effect of nitrogen fertiliser Rauzy.1998. L'utilisation de 1'herbe paturee tion occurs early in spring before stem supply and winter cutting on morphological et fauchees en reference au potentiel de pro- apex development. composition and herbage digestibility of a duction des prairies. Fourrages. 156:589-601. The conflict between forage production Dactylis glomerata L. sward in spring. J. Sci. Varlet-Grancher, C., R. Bonhomme, M. and weed control can be resolved at farm Food and Agr. 80:33-42. Chartier, and L. Artis 1982. Efficience de Duru, M., E. Justes, A. Langlet, and V. la conversion de 1'energie solaire par un con- level. Limiting C. aureum recruitment and Tirilly. 1993. Comparaison des dynamiques vert vegetal. Acta Oecol. Plant. 3 :3-26. improving hay yields call for the revision d'apparition et de mortalite des organes de Woledge, J.1978. The effect of shading during of grazing and cutting operations, select- fetuque elevee, dactyle et luzerne (feuilles, vegetative and reproductive growth on the ing productive areas for hay, reserving taller et tiges). Agronomie.13:237-252. photosynthetic capacity of leaves in a grass invaded areas for grazing, and generally Gillet, M. 1980. Les graminees fourrageres. sward. Ann. Bot. 42:1085-1089. preventing any seed immigration. Silage Gauthier-Villars (ed.). production could allow earlier cutting, i.e. Gonnet, J.F. 1989. Apport de la Biologic before weed seeds mature. Reducing C. Micromoleculaire (flavonoides) a la com- aureum seed viability by stamping manure prehension de la structure et du fonction- could be another way of reducing the risk nement de l'espece allogame Centaurea montana (Composees) et de 1'espece autogame of seed propagation, especially on mead- Chaerophyllum aureum (Ombelliferes) au ows on which there is a lot of biomass sein de deux groupements vegetaux sub- production. alpins: prairie et megaphorbiaie. PhD Thesis, Univ. Claude Bernard. Lyon I. Grime, J.P., J.G. Hodgson, and R. Hunt. Literature Cited 1988. Comparative plant ecology. Unwin Hyman, London. Kjeldahl, J. 1883. Neue Methode zur Bobbink, R. and J.H. Willems.1993. Restoration Bestimmung des Stickstoffs in organischen management of abandoned chalk grassland in Korpern. Z. Anal. Chem. 22:366-382. the Netherlands. Biod. and Conserv. 2:616-626. Leafe, E.L.1978. Physiological, environmental Briske, D.D. 1996. Strategies of plant survival and management factors of importance to in grazed systems: a functionnal interpreta- maximum yield of the grass crop, p. 37-79. tion, p. 37-68. In: Hodgson, J. and Illius In: ADAS/ARC (ed.), Maximising Yields of (eds) The ecology and management of graz- crops, ADAS/ARC Symposium. Harrogate, ing systems. A.W.CAB International. G.B. Calviere, I. 1994. Effets de la nutrition Lemaire, G. and F. Gastal. 1997. N uptake minerale et de la composition botanique sur and distribution in plant canopies, p.3-43. In: la croissance de prairies permanentes au G. Lemaire (ed.), Diagnosis of the N Status printemps : consequences pour la prevision et in Crops. Springer Verlag Press, Berlin. le diagnostic de la biomasse recoltable et de Lemaire, G. and J. Salette. 1984. Relation sa qualite. Ph. D. Thesis, Univ. Toulouse III. entre dynamique de croissance et dynamique Toulouse. de prelevement d'azote pour un peuplement Calviere, I. and M. Duru.1999. The effect of de graminees fourrageres. I-Etude de 1'effet N and phosphorus fertilizer application and du milieu. Agronomie. 4:423-430. botanical composition on the leaf/stem ratio Magda, D. 1998. Effects of extensification on patterns in spring in pyrenean meadows. invasive populations of Chaerophyllum Grass and Forage Sci. 54:255-266. aureum in grasslands. J. Veg. Sci. 9 de Hullu, E., T. Brouwer, and S.J. Ter Borg. :409-416. 1985. Analysis of the demography of Magda, D. and M. Jarry 2000. Prediction of Rhinanthus angustifolius populations. Acta cutting effects on a population of Botanica Neerlandica. 34:5-22. Chaerophyllum aureum - a demographic approach. J. Veg. Sci. l l: 485-492.

132 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 J. Range Manage. 56:133-139 March 2003 Moderate and light cattle grazing effects on Chihuahuan Desert rangelands

JERRY HOLECHEK, DEE GALT, JAMUS JOSEPH, JOSEPH NAVARRO, GODFREY KUMALO, FRANCISCO MOLINAR, AND MILT THOMAS

Authors are Professor, Department ofAnimal and Range Sciences, New Mexico State University, Las Cruces, N.M. 88003; private Range Consultant, 3000 Devandale Drive, Las Cruces, N.M. 88005; Graduate Research Assistants, Department of Animal and Range Sciences, New Mexico State University, Las Cruces, N.M. 88003; Professor, Center for Biological Studies, University of Juarez, Chihuahua, Mexico; and Assistant Professor, Department of Animal and Range Sciences, New Mexico State University, Las Cruces, N.M 88003.

Abstract Resumen

Vegetation changes were evaluated over a 13 year period Durante un periodo de 13 anos (1988-2000) se evaluaron los Chihuahuense (1988-2000) on moderately grazed and lightly grazed rangelands cambios de vegetacion en pastizales del Desierto in the Chihuahuan Desert of south central New Mexico. During de la region sur central de New Mexico y que fueron apacenta- el de estu- the study period, grazing use of primary forage species averaged dos ligeramente y moderadamente. Durante periodo promedio 49 and 26% on moderately and lightly grazed rangelands, dio la utilizacion de las principales especies forrajeras con moderado respectively. Autumn total grass and black grama (Bouteloua eri- de 49 y 26% para los pastizales apacentamiento y A traves del estudio la biomasa total en opoda Torr.) standing crop were consistently higher on the light- ligero respectivamente. pie de los zacates en otono la del "Black grams" (Bouteloua eri- ly than moderately grazed rangeland throughout the study. Total y en el grass standing crop declined on the moderately grazed rangeland opoda Torr.) fueron consistentemente mayores pastizal que en el apacentado moderadamente. when the last 3 years of study were compared to the first 3 years apacentado ligeramente los 3 anos del estudio con los primeros 3 (10 versus 124 kg ha"1), but showed no change on the lightly Al comparar ultimos se detecto que la biomasa total en pie disminuyo en el pasti- grazed rangeland (320 versus 357 kg ha"1). Black grama, the pri- anos mary perennial grass in the Chihuahuan Desert, increased in zal apacentado moderadamente (10 versus 124 kg ha 1), pero no autumn standing crop on the lightly grazed rangeland, but mostro cambios en el pastizal apacentado ligeramente (320 ver- principal zacate perenne decreased on the moderately grazed rangeland. Dropseed sus 357 kg ha"1). El "Black grama", el biomasa en pie en otono (Sporobolus autumn standing crop decreased on both del desierto Chihuahuense, aumento su spp.) ligeramente, pero disminuyo en el pas- rangelands during the study. However, this decrease was greater en el pastizal apacentado moderadamente. Durante el estudio la biomasa on the moderately grazed rangeland (97% decline) than on the tizal apacentado (Sporobolus producida en otono dis- lightly grazed rangeland (67% decline). Perennial grass survival en pie del "Dropseed" spp.) Sin embargo, esta disminucion fue following a 3-year period of below average precipitation was minuyo en ambos pastizales. moderadamente (97% de dis- higher on the lightly grazed (51%) than the moderately grazed mayor en el pastizal apacentado ligeramente (67% de disminu- rangeland (11%). Severe grazing intensities on the moderately minucion) que en el apacentado los zacates perennes despues de un grazed rangeland during the dry period (1994-1996) appear to cion) La sobrevivencia de de la precipitacion promedio fue explain differences in grass survival between these 2 rangelands. periodo de 3 anos por debajo ligero que en el apacen- Our study and several others show that light to conservative mayor en el apacentamiento (51%) de apacen- grazing intensities involving about 25-35% use of key forage tamiento moderado (11%). Intensidades severas el periodo seco (1994-1996) en el species can promote improvement in rangeland ecological condi- tamiento ocurridas durante explicar las difer- tion in the Chihuahuan Desert, even when accompanied by pastizal apacentado moderadamente parecen los zacates entre estos dos pastiza- drought. encias en la sobrevivencia de les. Nuestro estudio y algunos otros muestran que intensidades de apacentamiento de ligeras a conservadoras con un 25 - 35% de use de las especies forrajeras cave pueden promover el mejo- Key words: Stocking rate, arid lands, livestock, range manage- ramiento de la condicion ecologica del Desierto Chihuahuense, ment aun cuando se presenten sequias.

Chihuahuan Desert rangelands in the southern half of New Mexico and north central Mexico support nearly a half million needed on plant successional changes on Chihuahuan Desert animal units of livestock and numerous species of wildlife. rangelands under different stocking strategies. About 70% of the Because of aridity and fragile soils, these rangelands are easily Chihuahuan Desert in southern New Mexico is controlled by the damaged by poorly controlled livestock grazing (Paulsen and Bureau of Land Management (BLM). Generally, BLM range- Ares 1962, Buffington and Herbel 1965). Better information is lands are managed for multiple uses with a goal of about 50% use of perennial grasses over long time periods. In some years, graz- on This research was supported by the New Mexico Agr. Exp. Sta., Las Cruces, ing use may be lighter and other years heavier depending N.M. 88003 and was part of project 0172944. annual precipitation. Various studies reviewed by Holechek et al. Manuscript accepted 8 Jun. 02. (1999) have indicated conservative grazing (about 35% use of

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 133 perennial grasses) through time promotes increased forage productivity and may give higher financial returns than moderate grazing (about 45% use of perennial grasses). During the 13-year period from 1988 to 2000, we evaluated long-term vegetation changes on 2 adjoining Chihuahuan Desert rangelands in southwestern New Mexico with similar soils, climate, and terrain (flat) that were assigned different stocking strategies. One rangeland unit was grazed lightly with the goal of 30% use of perennial grasses while the other was grazed moderately for 50% use. Variable stocking rates were applied to both rangelands taking into account changing forage conditions among years. Our objectives were to evaluate trend in forage production, herbaceous cover, brush cover, species composition, and rangeland condition on the 2 rangelands. This research is a continuation of a long-term study initially reported by Holechek et al. (1994). During the 1982-1990 period, they found forage production increased on the 2 rangelands in our study in response to above average precipitation and conservative grazing use. Our study provides a detailed evaluation of vegetation changes on the 2 rangelands in the 1988-2000 period.

Materials and Methods

Study Areas Our 2 study areas were a lightly grazed site on the New Mexico State University Chihuahuan Desert Rangeland Research Center and an adjoining Bureau of Land Management (BLM) moderately grazed site (32° 32' 30" N 106° 52' 30" W) (Fig. 1). Both study areas are approximately Fig. 1. The fenceline between the moderately grazed range (left) and lightly grazed range 1,300 ha in size. The area is bound by the (right) in February 1991 (top) and February (2001) bottom. San Andres Mountains on the east and several isolated mountains on the west. annually at 10 Elevation varies from 1,190 to 1,380 m locations on the ton (Croton pottsii Lam.), nightshades Chihuahuan Desert with level or gently rolling hills. Soils are Rangeland Research (Solanum spp.), globemallows (Sphaeralcea primarily shallow, fine sandy loams Center rangeland (Table 1). spp.), and Russian thistle (Salsola iberica Vegetation is classified as (Aridisols) of the Simona-Cruces associa- Chihuahuan L.). The presence of these forbs is depen- Desert grassland and shrubland tion (Tembo 1990). Topography is rela- (Paulsen dent on seasonal precipitation. and Ares 1962). tively flat with all slopes under 5%. Most of the grassland A detailed grazing history of the areas have been invaded Long-term (1930-2000) average annual by woody species Chihuahuan Desert Rangeland Research during the precipitation on the Chihuahuan Desert last 100 years (Brown 1950, Center and BLM study areas is provided Dick-Peddle 1966). The principal vegeta- by Rangeland Research Center is 23.6 cm yr'. Holechek et al. (1994). Both areas were tion communities Seasonal patterns of precipitation are char- are black grama predominantly black grama grassland with (Bouleoua eriopoda Torr.) a acterized by small amounts in spring and a grassland, minor woody component when the honey mesquite peak in late summer (August) with gradu- (Prosopis glandulosa Chihuahuan Desert Rangeland Research Torr.) shrubland, creosotebush ally reduced amounts during fall. A small- (Larrea tri- Center was established in 1927. Zones of er peak occurs in early dentata Lar.) shrubland, and tarbush degradation were minimized because winter (January) (Flourensia (Pieper and Herbel 1982). Total and grow- cernua D.C.) shrubland watering points were widely spaced. (Paulsen ing season precipitation were collected and Ares 1962, Pieper and Herbel Although information is vague, stocking 1982). Annual forbs include leatherleaf cro- rates on the Chihuahuan Desert Rangeland

134 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 1. Total and growing season (July, August, September) precipitation (cm) on the Chihuahuan Desert Rangeland Research Center from 1988-2000.

13 Long-term Year Average 1988 1989 1990 1994 1998 1999 2000 Average (1930-2000)

Total precipitation 30.0 19.3 21.2 38.4 39.1 25.1 17.8 17.0 20.1 29.5 20.8 28.0 27.4 25.6 23.6 Growing season (July, August, September) Precipitation 17.8 10.2 19.1 18.3 12.4 13.5 5.1 10.2 12.7 14.0 9.6 15.8 3.4 12.5 12.4

Research Center range averaged 40 ha per BLM) to evaluate vegetation canopy cover Late June of each year was used to eval- animal unit (AU), forage production aver- and standing crop. During the autumn of uate grazing intensity because it is the end aged near 360 kg ha', and forage use aver- 1988, 1989, 1990, 1998, 1999, and 2000, of the forage cycle prior to new growth of aged about 35% during the 1930's and incremental measurements of herbaceous perennial grasses which usually occurs in 1940's. On the adjacent BLM study site, foliar cover were taken seasonally along July (Paulsen and Ares 1962). Grazing stocking rates averaged 24 ha AU' and each transect using a modification intensity was evaluated on each study area grazing use averaged between 60-70%. A (Holechek and Stephenson 1983) of the using procedures of Anderson and Currier continuous (year-long) grazing system has line intercept procedure outlined by (1973) as modified by Holechek and Gait been used on both study areas from the Canfield (1941). At 100-m intervals, a 1- (2000). These procedures involved assess- past to the present. Black grama cover was m rod incremented at 1-mm intervals was ing grazing intensity on each study area greatly reduced on both study areas during placed perpendicular to the transect and through a combination of perennial grass extended drought in the 1953-1956 peri- percent foliar cover was recorded by stubble heights and ungrazed residual bio- od. Herbicide treatments to control brush species. Each transect consisted of 64 mass of forage plants. Four permanent key were applied to approximately 90% of the sampling points. Three detailed measure- areas were systematically established Chihuahuan Desert Rangeland Research ments of shrub foliar cover were made at within each study area for these assess- Center study site in the 1957 to 1964 peri- 2-km intervals along each transect using ments. These key areas were selected by od (Holechek et al. 1994). Mesquite kill 40 x 2-m belt transects (Canfield 1941). dividing each pasture into 4 equal parts varied from 64 to 93%. The BLM study These belt transects covered an area of and then locating the key area near the area received herbicide treatment in 1955, 240-m2 per transect. center of each part. All key areas were 1.3 but no evaluations were made of percent Above ground standing crop (kg ha-') to 1.8 km from water. mesquite kill or herbage response. was sampled at the end of the summer In October 1999, the percentages of live Grazing on the Chihuahuan Desert growing season (October) from 1988 to and dead perennial grasses were evaluated Rangeland Research Center study area has 2000. Vegetation was clipped at ground on all transects. The procedure involved been carefully controlled since 1967 when level from 10 systematically located recording the nearest plant at 200-m inter- the stocking rate was initially reduced to quadrats (0.5 x 1.0-m) placed at approxi- vals along transects as live or dead based 67 ha AU' (Beck 1978, Holechek 1991, mately 600-m intervals along each of the on the presence or absence of live above Holechek et al. 1994). During the 24-year 6.2-km transects. Vegetation was hand ground biomass. Dead plants were charac- period from 1967 to 1991, forage utiliza- separated by species in the field, oven terized by all blackish above ground bio- tion averaged about 30%. Both rangeland dried at 60°C for 72 hours, and weighed. mass while presence of green or yellow ecological condition and forage produc- Only current-year's growth was measured. above ground biomass characterized living tion steadily increased. This allowed a Livestock grazing was suspended on the plants. gradual stocking rate increase from 67 to BLM study area from 1 May 1998 to 1 Rangeland ecological condition scores 45 ha AU-' with no increase in forage use November 1999 and during the growing were calculated from current USDA or sacrifice in cattle performance season (1 July through 30 September) in Natural Resources Conservation Service (Holechek 1992). 2000. This allowed us to evaluate herbage site guides for New Mexico (shallow The general goal on the BLM rangeland standing crop in the autumn on the BLM sandy site) using the Dyksterhuis (1949) since 1967 has been to remove about 50% rangeland during the last 3 years of study procedure. Relative percent composition of the perennial grass production. The without herbage removal by livestock dur- of autumn current-year standing herbage stocking rate from the late 1960's to 1981 ing the growing season. Because the on the Chihuahuan Desert Rangeland averaged 42 ha AU'. In the 1981 to 1990 Chihuahuan Desert Rangeland Research Research Center and BLM rangelands was period, the rancher destocked the range to Center rangeland was continuously (year- used to calculate rangeland ecological an average rate of 72 ha AU'. This man- long) grazed, autumn current-year stand- condition scores for each year of study agement change and above average pre- ing crop of forage underestimated (1988-2000). cipitation resulted in a major increase in ungrazed forage production. Forage use A repeated measures analysis of vari- condi- forage production and ecological levels were light to conservative on both ance using the mixed model procedure of tion (Holechek et al. 1994). rangelands in the 1988-1990 period. SAS (Littell et al. 1996) was used to com- Therefore, we consider autumn current- pare autumn current-year total standing Procedures year growth to be a reasonable estimate of herbage, total grass standing herbage, Eight permanent transects (each 6.2 km baseline herbage production. We made no black grama standing herbage and ecologi- in length) spaced 500-m apart were locat- adjustments for utilization. cal scores across stocking strategies (2) ed across each study site (CDRRC and and years (13). Transects were used as

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 135 replications (8 per stocking strategy). Results and Discussion drought. Some minor adjustments were Autumn herbaceous standing crop, stand- made in cattle numbers in the late autumn ing crop relative composition, vegetation of each year. If at any time average stub- A comparative summary of stocking foliar cover, and foliar cover relative com- ble height of black grama dropped below rate, forage utilization, vegetation standing position were compared on the lightly 5.5 cm, the pasture would be completely crop, and rangeland ecological scores on grazed and moderately grazed rangelands destocked for at least 1 growing season. the lightly and moderately grazed range- using data pooled across the first 3 years This occurred in late July 1994 and the lands during the 13 year study period is and the last 3 years of study. A random- pasture was immediately destocked. given in Table 2. Stocking rate data in ized factorial analysis of variance was Restocking was not initiated until January Table 2 are based on the forage cycle year used with the 2 grazing treatments (light 1997 because of poor forage growth in starting 1 July (beginning of forage and moderate) and the 2 time periods 1996 (Table 2). growth) rather than on the calendar year. (1988-1990, 1998-2000) as factors and On the moderately grazed rangeland, the This coordinates vegetation standing crop transects (8 per stocking strategy) as repli- approach was to stock at grazing capacity data collected in October with forage uti- cations. In an analysis of grazing experi- (230 animal unit years) unless excessive lization data collected the following June. ments, Holechek et al. (1999) found data grazing (over 55% use of forage) was Over the 13 year study period, total pooled across the first and last 3 years of observed at the end of the forage cycle annual precipitation averaged 109% of the study gave the most meaningful compar- (June) for 2 consecutive years. If this con- long-term average (23.6 cm yr') (Table isons of long-term vegetation changes. dition was met, then the stocking rate 1). During the first 6 years of study, annu- This was because ecological condition and would be reduced for at least 1 growing al precipitation averaged 122% of the forage production were often not equiva- season. Recent precipitation conditions long-term average. However, during the lent across grazing treatments at study ini- and forage availability in the pasture were last 7 years, total annual precipitation was tiation as in our case. The least significant also used in stocking rate decisions. If for- 97% of the long-term average. Drought difference mean separation procedure was age became severely depleted, complete occurred in 1994. Major stocking rate used to compare means if analysis of vari- destocking was an option. Major stocking adjustments were made on both range- ance indicated a significant (P < 0.05) dif- rate reductions were made in 1994, 1995, lands in response to changing precipitation ference (Steel and Torrie 1980). 1998, 1999, and 2000 on the moderately and forage conditions (Table 2). The gen- Comparisons of percentage of live plants grazed rangeland (Table 2). eral approach on the lightly grazed range- between the Chihuahuan Desert Rangeland Total vegetation, total grass, and black land was to stock at 70% of grazing capac- Research Center and BLM rangelands in grama standing crop differed (P < 0.05) ity in years of above average to near aver- 1999 were made using the standard t-test among stocking strategies and years age growing season precipitation and to 8 2). Interactions between stocking with the transects in each treatment as destock under conditions of severe (Table replicates. strategies and years were significant (P <

Table 2. Stocking rate, forage use, forage production, and rangeland ecological condition scores for the lightly grazed and moderately grazed rangelands from 1988-2000.

88/89 89/90 90/91 91/92 92/93 93/94 94/95 95/96 96/97 97/98 98/99 99/00 00/01 Lightly Grazed Rangeland f)f Stocking rate (ha AUY 49 49 49 44 40 50 320 0 210 48 Forage utilization (%)2 33 28 33 22 18 33 50 0 10 33 Autumn vegetation 1)3 527ed 335e of 579e standing crop (kg ha 41.4' 716b 1006' 281 llh 150g 255f Autumn grass standing I 270ed 349e 7e 57e 92e Crop (kg ha )3 452b 554b 842' 225' 301cd Autumn black grama 1)3 95ef 169ed standing crop (kg ha 68g 59g 119' 380' 86fg 5' 48h 66g Rangeland ecological 57e 57e 59e 53e 56e condition score 62b 48`d 64b 81 72b Moderately Grazed Rangeland i)1 Stocking rate (ha AUY 83 61 46 47 47 47 71 98 47 47 Forage utilization (%)2 21 31 38 33 30 48 60 75 80 75 Autumn vegetation t)3 45de 22e 160e standing crop (kg ha 229b 227b 420' 405' 500' 210b 78' Autumn grass 104e 108e 201ab standing crop (kg hal)3 159" 251b 140bc 13d 3d 18' 12d Autumn black grama 1)3 2e 34ab 2e 49a

136 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 3. Average autumn herbaceous standing crop (kg ha'1) and relative herbaceous composition Gibbens 1996). Total forb standing crop (%) on lightly grazed (LG) and moderately grazed (MG) Chihuahuan Desert rangelands in increased (P < 0.05) on both rangelands in south central New Mexico for the 1988-1990 and 1998-2000 periods. 1998-2000 compared to 1988-1990. Forbs in the Chihuahuan Desert are Autumn Standing Crop Standing Crop Relative Composition responsive to winter-spring precipitation. Species or group LG MG LG Winter-spring precipitation was higher in 88-90 98-00 88-90 98-00 98-00 98-00 1998-2000 than 1988-1990. ------(kgha 1 )------Broom snakeweed (Gutierrizea 60a Aristida spp. 45b 13c 2d sarothrae Pursh), the primary poisonous Bouteloua eriopoda 74b 152 13c ld plant found on both rangelands, showed Erioneuron pulchellum 4b tb 21" 3b no change (P > 0.05) in autumn standing 5a to 2a Muhlenbergia porteri 1 1" crop on either rangeland when the last and 169a 3 Sporobolus spp. 56b 62b 3c first years of study were compared 44a 66a b Other grasses 13b tc (Table 3). Broom snakeweed is a short- 357a 320a Total grasses 124b 10c lived, cyclic half-shrub that can cause 73a abortion in livestock and has severely forbs 14c 12c 40b Total depressed productivity of perennial grass- 157a Gutierrezia sarothrae 54b 77b 143' es on New Mexico rangelands (McDaniel 425a Total vegetation 471' 292b 193 et al. 1993). Populations of broom snake- Data were pooled across the first 3 years and last 3 years of study for trend comparisons as suggested by Holechek et weed are closely related to climatic condi- alb (1999). tions (Pieper McDaniel a' and 1990). Above Means within rows with different letters differ (P < 0.05). normal autumn through spring precipita- t trace = tion favors broom snakeweed establishment. Broom snakeweed standing crop 0.05). We will focus our discussion on crop increased on the lightly grazed range- levels averaged nearly twice as high (P < standing crop trends over time for each land, but decreased on the moderately 0.05) on the moderately grazed as on the stocking strategy concentrating on com- grazed rangeland during the study period. lightly grazed rangeland. parisons between the first and last 3 years Black grama is the primary decreaser Total grass, total forb, and black grama of the study (Table 3). perennial grass in the Chihuahuan Desert foliar cover averaged higher (P < 0.05) on Standing crop of total vegetation and (Canfield 1939, Paulsen and Ares 1962). the lightly grazed than the moderately change (P > 0.05) Dropseed (Sporobolus standing crop total grasses showed no spp.) grazed rangeland (Table 4). However, on the lightly grazed rangeland, but decreased (P < 0.05) on both rangelands. total shrub cover was higher (P < 0.05) on (P < moderately We attribute this decrease to below aver- declined 0.05) on the the moderately grazed rangeland. Most the last 3 years age precipitation in the 1994-1996 period. grazed rangeland when canopy cover components had significant were compared to the first 3 years of the Dropseeds are less drought tolerant than interactions (P < 0.05) between stocking black grama (Campbell 1929, Herbel and study (Table 3). Black grama standing strategy and time period. Generally, vegetation foliar cover Table 4. Average autumn vegetation foliar cover (%) and relative composition on lightly grazed (Table 4) showed the same trends as (LG) and moderately grazed (MG) Chihuahuan Desert rangelands in south central New Mexico autumn standing crop for primary herba- for the 1988-1990 and 1998-2000 periods. ceous components with a few exceptions. We cannot explain why total autumn Vegetation Foliar Cover Relative Foliar Vegetation Composition standing crop of forbs increased (P < 0.05) group LG MG LG MG Species or between 1988-1990 and 1998-2000, but 88-90 98-00 88-90 98-00 88-90 98-00 88-90 98-00 foliar cover showed no change (P > 0.05). ------(%)------4a 2a We believe this is probably a sampling Aristida spp. 0.4" 0.3" 0.2" 0.2" 1" 1" 5a 0.05) dur- 0.9ab Sporobolus spp. 1.4" 0.4c tc 13" ing the period of study on either range- 2.9a Total grasses 2.2" 1.6b 0.4' 26' land. Honey mesquite canopy cover on the to 2a Croton pottsii 0.3" 0.2" to lightly grazed range was only half (P < Total forbs 0.4a 0.4a 0.2c 0.5a 4a 0.05) that on the moderately grazed range- 6.8a 8.4a Gutierrezia sarothrae 1.3b 2.0b 12b land. This is explained in part by more 0.3a intensive with Ephedra spp. 0.2" 0.2" 0.3" 2" control of mesquite herbi- Yucca elata 1.2" 1.3" 0.7" 0.7" 11" cides on the lightly grazed rangeland prior Acacia constricta 0.6b 0.6b 1.3" 1.5" 5" to the initiation of our study. Both crop and Prosopis glandulosa 4.5b 6.2b 11.2" 12.6" 41" herbaceous standing 59a foliar cover showed major changes (P < Total shrub cover 6.5b 8.3b 13.5" 15.1" 0.05) in relative vegetation composition b 24.4a Total vegetation cover 11.1 12.9b 22.0" 100 100 100 100 when the last and first 3 years of the study Data were pooled across the first 3 years and last 3 years of study for trend comparisons as suggested by Holechek et were compared (Tables 3 and 4). They al. (1999). were consistent in showing a shift to lower a,bMeans within rows with different letters differ (P < 0.05). (P < 0.05) total grass and dropseed com- t = trace

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 137 ponents on both rangelands and a higher rangeland remained black. Very few stud- grazing during drought can quickly forb component on the moderately grazed ies have evaluated perennial grass survival degrade these rangelands. Our research rangeland. under different grazing intensities. In the shows the commonly used guideline of Rangeland ecological scores, based on Edwards Plateau of Texas, perennial grass take half and leave half does not work the USDA-NRCS method and current survival during the 1950's drought was well on arid and semi-arid rangelands. The New Mexico range site guides, averaged closely related to stocking rate (Young problem with this guideline is that major higher (P < 0.05) on the lightly than mod- 1956). Lightly stocked pastures had 66% destocking will be required in 50% of the erately grazed rangeland (Table 1). They higher grass survival than those heavily years to avoid excessive use of primary showed an upward trend (P < 0.05) on the stocked and 33% higher than those moder- grasses (Hutchings and Stewart 1953, lightly grazed rangeland, but a downward ately stocked. Better photosynthetic capa- Paulsen and Ares 1962, Martin 1975, Galt trend (P < 0.05) on the moderately grazed bility and more extensive root systems of et al. 2000). In contrast, major destocking rangeland (Table 2). Rangeland ecological lightly grazed plants under stress are the is only needed in about 2 years out of 10 condition scores on both rangelands primary explanations for this relationship. when rangelands are stocked lightly. The showed considerable fluctuation among An additional explanation is better soil difficulties of estimating forage produc- years (Table 2). moisture conditions under light grazing tion and buying and selling livestock make We believe our study provides support due to more mulch and soil organic matter management for 50% grazing use a risky for the model of Dyksterhuis (1949) (Molinar et al. 2001). and unsound proposition. Improvements regarding rangeland vegetation responses Dropseed mortality was over 70% on in forage production and rangeland health to grazing and climate. However, we can- both rangelands (Table 4). Other studies are unlikely and the probabilities of range- not reject the state-and-transition model of have shown that black grama is more land degradation and financial ruin are Westoby et al. (1989). They are not com- drought resistant than dropseeds (Paulsen high unless the rancher is extremely savvy peting models. Both high mesquite foliar and Ares 1962, Wright and Van Dyne and capable of reacting quickly. cover (13%) and lack of perennial grasses 1976, Herbel and Gibbens 1996). This make it doubtful that meaningful improve- may be explained by black grama having a ment can occur in ecological condition on more extensive root system and more Summary and Conclusions the moderately grazed rangeland through capability to extract moisture from dry grazing management alone (Herbel et al. soils than dropseeds. Our 13 year study (1988-2000) showed to during the 13 year 1983). Mesquite control will be needed Forage utilization an upward vegetation trend on a lightly a study period averaged 26% on the lightly restore this rangeland to healthy condi- grazed rangeland while a downward trend rangeland and 49% on the moder- tion as defined by the United States grazed occurred on an adjacent moderately grazed Interior (2000). ately grazed rangeland. Generally forage Department of rangeland (Fig. 1). Various measures of Perennial grass plant survival percent- use levels in excess of 50% are considered vegetation change were used to monitor (P 3 a by Holechek et ages were higher < 0.05) for all grass- heavy based on review trend in our study. es [black grama, dropseeds, and threeawns al. (1999). Heavy grazing is defined as a Even though forage use was estimated that does not permit (Aristida spp.)] evaluated on the lightly degree of herbage use to average 49% on the moderately grazed compared to the moderately grazed range- desirable forage species to maintain them- rangeland during the 13 year study period, land (Table 5). Most perennial grass mor- (Klipple and Bement 1961). selves a sharp downward trend occurred. The Grazing reached the heavy level only in rancher attempted to keep livestock num- on the lightly grazed Table 5. Percentages of live perennial grass the summer of 1995 bers in balance with forage supplies, but plants on long-term lightly grazed (LG) and rangeland, but was heavy to severe for 5 was unsuccessful in 5 of the 13 years stud- long-term moderately grazed (MG) years (spring 1995 to spring consecutive ied. Although sharp reductions were made Chihuahuan Desert rangelands in south cen- moderately grazed rangeland. 1999) on the in cattle numbers during drought years, tral New Mexico in Autumn 1999. Black grama stubble height averaged 11.3 heavy to severe grazing use still occurred. cm on the lightly grazed rangeland, but 5.6 LG MG By the last 3 years of study, palatable cm on the moderately grazed rangeland perennial grasses had been nearly elimi- Grass species during the 13 year study period. A mini- nated (Fig. 1). Both brush control and - - - (% live plants) - - - mum stubble height of 7.6 cm is recom- 80a seeding may be needed to obtain meaning- Bouteloua eriopoda lOb mended for maintenance of black grama 23a ful increases in forage production on this Sporobolus spp. 5b (Valentine 1970). Dropseed stubble 67a rangeland. Aristida 12b lightly spp. heights averaged 22.5 cm on the In contrast to the moderately grazed Other grasses' 35a 15b but 11.2 cm on the mod- grazed rangeland, rangeland, the lightly grazed rangeland had 51a rangeland. A minimum Average 11b erately grazed a strong upward trend in ecological condi- stubble height of 15.0 cm is recommended 1Primarily Setaria leucopila and Muhlenbergia porteri. tion. This was primarily due to an increase for dropseeds (Holechek and Gait 2000). in production of black grama during the Our study is consistent with earlier stud- last 3 years of study. Major adjustments in ies by Canfield (1939), Paulsen and Ares cattle numbers on the lightly grazed range- on likely occurred (1962), and Valentine (1970) that light to tality both rangelands land were only needed in 1994 and 1995 below average precip- grazing intensities can be during the period of conservative when it was completely destocked due to (Table 2). the effective in maintaining and increasing itation in 1994-1996 During drought and lack of forage. production on Chihuahuan Desert summer of 1997 when above average rain- forage Our research indicates that severe graz- fall occurred, most black grama plants on rangelands dominated by black grama. ing during drought greatly increases the lightly grazed rangeland showed green Our study is also consistent with these perennial grass mortality compared to tops, but those on the moderately grazed studies in showing that heavy to severe light grazing. Rangeland retrogression in

138 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 the Chihuahuan Desert can occur within 2 Campbell, R. S. 1929. Vegetation succession Littell, R. C., G. A. Milliken, W. W. Stroup, years when drought and severe grazing are in the Prosopis sand dunes of southern New and R. B. Woulfinger. 1996. SAS system coupled together. On the other hand, Mexico. Ecol.10:392-398. for mixed models. SAS Inst., Cary, N.C. recovery from short-term drought can Canfield, R. H. 1939. The effect of intensity Martin, S.C. 1975. Stocking strategies and net and frequency of clipping on density and cattle sales on semi-desert range. U.S. Dept. occur within 3 years if rangelands are yield of black U.S. Agr. destocked before excessive use occurs. grama and tobosa grass. For. Serv. Res. Pap. RM-146. Dept. Agr. Tech. Bull. 681. McDaniel, K. C., L.A. Torell, and J.W. Bain. Light to conservative grazing is important Canfield, R. H. 1941. Application of the line 1993. Overstory - understory relationships in non-drought years because it allows intercept method in sampling range vegeta- for broom snakeweed - blue grama range- black grama to increase crown area and tion. J. Forest. 39:388-394. lands. J. Range Manage. 46:506-511. root growth (Canfield 1939, Paulsen and Dick-Peddie, W. A. 1966. Changing vegeta- Molinar, F., D. Gait, and J. Holechek. 2001. Ares 1962,.Valentine 1970, Young 1980). tion patterns in southern New Mexico, p. 234 Managing for mulch. Rangelands 23(4):3-7. Our study supports the recommendation -235. In: 16`" Field Conf. New Mexico Geol. Paulsen, H. A. and F. N. Ares.1962. Grazing by various researchers that grazing intensi- Soc. Albuquerque, N.M. values and management of black grama and ties on Chihuahuan Desert rangelands be Donahue, D. L. 1999. The western range revis- tobosa grasslands and associated shrub kept at around 30 to 35% use of perennial ited: Removing livestock from public lands ranges of the southwest. U.S. Dep. Agr. to conserve biodiversity. Univ. Oklahoma Tech. Bull. 1270. grasses (Canfield 1939, Paulsen and Ares Press, Norman, Okla. Pieper, R. D. and C. H. Herbel. 1982. 1970, 1980, 1962, Valentine Young Dyksterhuis, E. J. 1949. Condition and man- Herbage dynamics and primary productivity Holechek et al. 1994). agement of rangeland based on quantitative of a desert grassland ecosystem. New Our data show that a combination of ecology. J. Range Manage. 2:104-115. Mexico Agr. Exp. Stat. Bull. 695. information on precipitation, forage pro- Gait, D., F. Molinar, J. Navarro, J. Joseph, Pieper, R. D. and K. C. McDaniel. 1990. duction, grazing intensity, livestock num- and J. L. Holechek. 2000. Grazing capacity Ecology and management of broom snake- bers, range condition, and range trend are and stocking rate. Rangelands 22(6):7-11. weed. New Mexico State Univ. Agr. Exp. needed for sound management decisions. Herbel, C. H. and R. P. Gibbens.1996. Post- Sta. Bull. 751. drought vegetation dynamics on arid range- Steel, R. G. and J. H. Torrie.1980. Principles We recognize that it is seldom possible to 2nd collect all this information due to limita- lands in southern New Mexico. New Mexico and procedures of statistics, Edition. tions of money, labor, time, and technolo- Agr. Exp. Sta. Bull. 726. McGraw-Hill Book Co., New York. Herbel, C. H., W.L. Gould, W.F. Liefiste and Tembo, A. 1990. Influence of watering points Light to grazing is most gy. conservative R.P. Gibbens. 1983. Herbicide treatment and range condition on vegetation of the needed where intensive monitoring is not and vegetation response to treatment of Chihuahuan Desert. Ph.D. Diss., New possible or practical. This minimizes the mesquite in southern New Mexico. J. Range Mexico State Univ., Las Cruces, N.M. risk of destructive grazing in drought Manage. 36:149-151. United States Department of Interior - years from failure to adequately destock. Holechek, J. L. 1991. Chihuahuan desert Bureau of Land Management. 2000. Recently, several range professionals have rangeland, livestock grazing, and sustainabil- Interpreting indicators of rangeland health. advocated the use of a 25% harvest coeffi- ity. Rangelands 13(3):115-120. Tech. Ref. 1733-6. cient in arid and semi-arid areas when Holechek, J. L. 1992. Financial benefits of Valentine, K. A. 1970. Influence of grazing stocking rates are set to reduce risk and range management practices in the Chihuahuan intensity on improvement of deteriorated 14(5):279-284. facilitate range improvement (Lacey et al. desert. Rangelands black grama range. New Mexico Agr. Exp. Holechek, J. L. and D. Gait. 2000. Grazing Sta. Bull. 553. 1994, Johnston et al. 1996, White and intensity guidelines. Rangelands 22 Ward, N. 1999. Ranchers need support for sus- McGinty 1997, Ward 1999, Galt et al. (3):11-14. tainable ranching: What government can do: 2000). Our research on Chihuahuan Desert Holechek, J. L. and T. Stephenson. 1983. A rancher's perspective. Rangelands rangelands supports this recommendation. Comparison of big sagebrush in northcentral 21(3):13-17. As a final point, we believe our study New Mexico under moderately grazing and Westoby, M. B., B. Walker, and I. Noy-Meir. contradicts Donahue's (1999) viewpoint grazing excluded conditions. J. Range 1989. Opportunistic management for range- that livestock grazing is not sustainable on Manage. 36:455-456. lands not at equilibrium. J. Range Manage. arid lands receiving less than 30 cm of Holechek, J. L., H. Gomez, F. Molinar, and 42:266-274. D. Gait. 1999. Grazing studies: What we've annual precipitation. During the 13 year White, L. D. and A. McGinty. 1997. Stocking (2):12-16. study period, the lightly grazed pasture learned. Rangelands 20 rate decisions. Texas A&M Univ. Agr. Ext. Holechek, J. L., A. Tembo, A. Daniel, M. improved from late seral to climax ecologi- Serv. Publ.13-5036. Fusco, and M. Cardenas. 1994. Long term Wright, R. G. and G. M. Van Dyne. 1976. cal condition. grazing influences on Chihuahuan Desert Environmental factors influencing semi- Rangeland. Southw. Nat. 39:342-349. desert grassland perennial grass demography. Hutchings, S. S. and G. Stewart. 1953. Southw. Nat. 21:259-274. Literature Cited Increasing forage yields and sheep produc- Young, S. A. 1980. Phenological development tion on intermountain winter ranges, U.S. and impact of season and intensity of defotia- Dept. Agr. Circ. 925. tion of Sporobolus flexuosus and Bouteloua Anderson, E. W. and W. F. Currier. 1973. Johnston, P. W., G. M. McKean, and K. S. eriopoda. Ph.D. Diss., New Mexico State Evaluating zones of utilization. J. Range Daily. 1996. Objective `safe' grazing capaci- Univ., Las Cruces, N.M. Manage. 26:87-91. ties for southwest Queensland Australia. V. A. 1956. The effect of the Beck, R. F. 1978. A grazing system for semi- Young, Aust. Rangeland J.18:244-258. 1949-1954 drought on the ranges of Texas. arid lands. Proc. Internat. Rangeland Congr. Klipple, G. E. and R. E. Bement.1961. Light J. Range Manage. 9:139-142. 1:569-672. grazing - Is it economically feasible as a Brown, A. L. 1950. Shrub invasion of southern range improvement practice? J. Range Arizona desert grasslands. J. Range Manage. Manage. 14:57-62. 3:172-177. Lacey, J., E. Williams, J. Rolleri, and M. Buffington, L. C. and C. H. Herbel. 1965. Marlow.1994. A guide for planning, analyz- Vegetation changes on semi-desert grassland ing, and balancing forage supplies with live- range from 1858 to 1963. Ecol. Monogr. stock demand. Montana State Univ. Ext. 35:139-164. Publ. EB-101.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 139 J. Range Manage. 56:140-145 March 2003 A digital photographic technique for assessing forage utilization

P. W. HYDER, E. L. FREDRICKSON*, M. D. REMMENGA, R. E. ESTELL, R. D. PIEPER, AND D.M. ANDERSON

Authors are Senior Research Assistant and Research Scientist, USDA, Agricultural Research Service, Jornada Experimental Range, Las Cruces, N.M. 88003, associate professor, University Statistics Center, New Mexico State University, Research Animal Scientist, USDA, Agricultural Research Service, Jornada Experimental Range, Las Cruces, N.M. 88003, Professor Emeritus, Department of Animal and Range Sciences, New Mexico State University, and Research Animal Scientist, USDA, Agricultural Research Service, Jornada Experimental Range. *Corresponding author.

Abstract Resumen

Changes in forage utilization have been difficult to measure Los cambios en la utilizacion de forraje han sido dificiles de non-destructively without some level of subjectivity. This subjec- medir en forma no destructiva sin un grado de subjetividad. Esta tivity, combined with a lack of reproducibility of visual estimates, subjetividad, combinada con la carencia reproduction de las has made forage utilization measurement techniques a topic of estimaciones visuales, ha hecho que las tecnicas de medicion de considerable discussion. The objective of this study was to devel- utilizacion de forraje sea un topico de considerable discusion. El op and test the accuracy and repeatability of an objective, com- objetivo de este estudio fue desarrollar y probar la certeza y puter-based technique for measuring changes in plant biomass. repetibilidad de una tecnica computational objettva para medir Digital photographs of target plants acquired before and after la biomasa vegetal. Fotografias digitales de plantas de interes partial defoliation were analyzed using readily available image tomadas antes y despues de ser sujetas a una defoliation partial analysis software. Resulting data were used to develop a simple fueron analizadas utilizando programas de computation ya linear random coefficient model (RC) for estimation of plant bio- disponibles para el analisis de imagenes. Los datos resultantes mass removed based on the area of the plant in the photo. fueron utilizados para desarrollar un modelo lineal simple de Sample collection took approximately 20 minutes/plant for alfal- coeficiente aleatorio (RC) para estimar la biomasa vegetal fa (Medicago sativa L.). Analysis of images took another 60 to 90 removida basandose en el area de la planta en la fotografia. La minutes. Regression analysis gave an R2 of 0.969 for predicted vs. coleccion de la muestra tomo aproximadamente 20 observed plant weights. Testing this model using 10 alfalfa plants minutos/planta de "Alfalfa" (Medicago sativa L.), el analisis de yielded weight estimates of defoliated plants accurate to within las imagenes tomo otros 60 a 90 minutos. El analisis de regresion +/- 8.5%. The advantage of the RC model is its ability to use eas- produjo una R2 de 0.969 para los pesos de planta predichos con- ily obtained coefficients from simple linear regression models tra los observados. La prueba de este modelo usando 10 plantas developed from each plant in a way that accounts for the lack of de "Alfalfa" produjo estimaciones de peso de las plantas defoli- independence between samples within an individual plant. The adas con una certeza dentro de +/- 8.5%. La ventaja del modelo technique described here offers an objective and accurate RC es su capacidad para usar coeficientes facilmente obtenidos a method for measuring changes in plant biomass with possible partir de modelos de regresion lineal desarrollados de cada plan- applications in ecology, botany, and range science. In particular, ta en una manera que toma en cuenta la falta de independencia application of this technique for estimating forage utilization entre muestras dentro de una planta individual. La tecnica may improve accuracy of estimates and, thereby, improve range descrita aqui ofrece un metodo objetivo y certero para medir management practices. cambios en la biomasa de las plantas con posibles aplicaciones en ecologia, botanica y manejo de pastizales. En particular, la aplicacion de esta tecnica para estimar la utilizacion de forraje Key Words: image analysis, alfalfa, Medicago sativa, random puede mejorar la certeza de las estimaciones, por to tanto, mejo- coefficient model. rar las practicas de manejo de pastizales.

Measurement of current-year's forage production that is either consumed or destroyed by grazing animals (Society for Range allow greater accuracy and can detect small differences in animal Management 1974) is critical for successful rangeland manage- use are extremely tedious and labor intensive. These latter prob- ment. Despite the importance of accurately measuring range uti- lems severely limit quantitative research because large sample lization, methods currently available have various constraints that sizes are required to obtain accurate forage use estimates (Estell in this limit their usefulness (Holechek et al. 1989). Techniques that et al. 1998). The objective of the work described paper was to develop a rapid, accurate, and repeatable quantitative tech- The authors would like to thank 2 anonymous reviewers for their constructive com- nique for objective assessment of forage utilization. We also ments. Field assistance from David Hu and Andrine Morrison is greatly appreciated. desired a technique that was readily available to most researchers Trade names used in this publication are solely for the purpose of providing spe- at reasonable costs. To accomplish this task, we focused on cific information. Mention of a trade name does not constitute a guarantee, recent technological advancements in digital photography and endorsement, or warranty of the product by the U. S. Department of Agriculture or The photographic aspect this method adds a New Mexico State University. image analysis. of Manuscript accepted 25 May 02. level of flexibility and reliability not found in other methods.

140 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Photographs provide a permanent record remaining above ground tissue was Table 1. Example data used to demonstrate that allows an objective estimate of the removed. Each sample was dried at 60° C statistical methods. area, and through modeling, the mass of a for 24 hours prior to being weighed to the plant to be calculated. nearest 0.5 g. The weight of each defolia- Plant Sample Area remaining Weight of and then summed to in photograph remaining tion was recorded plant determine total weight of the plant. Materials and Methods (%) (%) 1 1 87 Image Analysis Field Technique Jandel's SigmaScan Pro® 3.0 was used 3 The field component of this study was to analyze the digital images. Any image 1 conducted at the USDA, Agricultural analysis software could be used provided 1 Research Service, Sheep Research Facility it has something approximating the fol- 1 located at the La Tuna Federal lowing capabilities: image cropping, cali- 2 2 bration of scale, color/black and 3 Correctional Institute in Anthony, Tex. 2 Alfalfa (Medicago sativa L.) was used white/contrast transforms, intensity selec- 2 because of its importance as a forage tion procedures, and some form of data program. These are used to pre- 3 1 plant. Fifty alfalfa plants were randomly reporting 2 selected from a 10 x 2-m plot. Each plant pare the image of the plant for areal mea- 3 was transplanted to a 5-gallon bucket. surement. Image cropping helps isolate the 4 Transplanting each plant allowed us to target image from incidental objects. This 3 5 deal with developing the technique on a will become important as the technique 4 1 plant-by-plant basis with no interference evolves for use in the field. Once the 4 2 from neighboring plants or other objects. image is adequately prepared, the program 4 To separate the plant from the back- counts pixels of various values that cone- 4 4 4 5 13 ground, a sheet of white, dry-erase board spond to the image. These pixels are then was placed behind, but not touching, the transformed to area based on the measure- image, in plant (Fig. 1A). Plant identification, date, ment of a known distance on the variety of alfalfa (Malone) at this particu- sample number, and photo angle were this case the 10-cm ruler, to the number of lar stage of growth. The random coeffi- recorded on the dry-erase board. A ruler pixels contained in the length of the image cient (RC) model was used and can be and white/contrast attached to a wire was placed to the side of of the ruler. Color/black written as + + + + intensity selection proce- [yij = (30 si (1 di)xij the plant on a plane bisecting the midpoint transforms and i=1..... n and j=1, k] where is the dures enhance isolation and measurement yij of the plant. Each ruler was held in place percentage of the ith plant remaining after 1B, C, and D). In by inserting the attached wire into the soil. of the target image (Fig. the jth sample is removed, is the remain- was A second ruler was placed perpendicular our method, the color image trans- ing area covered in the photograph of the due to the first, also on a plane bisecting the formed to black and white (grayscale) ith plant after the jth sample is removed, plant's midpoint. These rulers served to to the program we used being better able to (30 is the y-intercept, 1i is the slope, and calibrate the image and to establish 0 and delineate the image in black and white. Si, di, and are random effects in the 90° reference angles. To improve calibra- Once the measurements are made, such model associated with the random devia- that a calculation of area for each image is tion accuracy, the rulers were trimmed to a tion of the ith plant's intercept from the length of 10 cm, eliminating the need available, the data can be copied into a total intercept (3p, the random deviation of the to see the marks on the ruler. spreadsheet for further analysis. ith plant's slope from the slope f 31, and ran- A digital camera (Kodak® DC260) was Sample weights and photographic areas dom error, respectively (Graybill 1976). used to photograph the plants. Two pho- were converted to percent weight remain- The first step in building the RC model tographs were taken at each sampling ing and percent area remaining, respec- is to estimate a simple linear regression from stage (sample 0-k) with the first 2 taken tively. We used a mean calculated (SLR) equation for each plant. Each SLR prior to defoliation (sample 0). The first the percent remaining areas for angles 0° equation describes the straight-line rela- photograph of each pair was taken at the and 90° to address the 3 dimensional tionship between the percentage of Forty midpoint of the plant height and width aspect in a 2 dimensional form. remaining plant (y) and the percentage of from a direction minimizing shadows. The plants were used to develop a statistical remaining area covered in the photograph 10 to test the model. To second was taken at 90°, on a horizontal model and plants (x) for a single plant. In general notation, plane, from the first. The white backdrop test for repeatability of the image analysis, b let bpi and 1 i denote the estimated y- we calculated the area for 1 plant image was also rotated at each step. intercept and slope for the ith plant (i = Plants were hand-defoliated in 5 steps. 10 times. To test sampling precision, we 1,..., n). Computations were performed Approximately 20% of the total plant calculated the mean deviation from the using fifteen significant digits and rounded size and weight was removed and collected as a desired 20% defoliation sample to one for illustration purposes. The esti- standard deviation of this mean for the sample during each step. As each sample mated SLR equation for the ith plant is samples taken from the 40 plants used to was collected, the sample number was = + build the model. b0i blixi. recorded on the backdrop and the plant re- A small subset of the data set (n = 4, k = photographed. Succeeding photos were 5) is given in Table 1 to use in illustrations taken from as close to the same position as Model Building of the statistical computations. The esti- possible. Samples collected at each step A statistical model was constructed to mated SLR equations and associated sums were bagged and labeled separately. provide estimates of individual plant uti- of squared residuals for each of the 4 all During the last defoliation step, lization, specifically for this particular plants in Table 1 are reported in Table 2.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 141 Table 2. Estimated Simple Linear Regression Photographing plants from perpendicular film speed, of slides vs. prints may argue (SLR) equations and Sum of Squared directions and taking the mean area of the against the use of print film. This differ- Residuals relating the percentage of remain- 2 images helped to reduce variation incor- ence in resolution is due to the inherent ing plant (y) to the percentage of remaining area covered in the photograph (x) for each porated into the model by the 3-dimen- loss of information that occurs when an plant in Table 1. sional shape being reduced to 2 dimen- image is transferred to a different medium, sions during analysis. Estimates of the i.e., film negative to paper. It is possible Plant SLR Equation Sum of Squared remaining percentages of plant weights that this loss of resolution would be too Residuals were accurate to +/-21% using the 0° small to have an effect on this technique.

1 y =13.25 + 0.9337x 6.9594 angle photographs and to +1-17% using A fast film (ISO of 200 or more) used with 2 y =13.77 + 0.8829x 4.7088 the 90° angle photographs, but were accu- flash would enable working during periods 3 y=12.15+0.9241x 7.5185 rate to +/- 8.5% using the mean of the 0° of light to moderate winds. The use of a 4 y =13.24 + 0.9283x 4.9426 and 90° angle photographs. The mean area flash should help reduce problems created for the repeated images was 676 cm2 (SD by shadows. Depth of field should be = 21.61, CV = 3.19, n = 10). The basic maximized so that the entire image of the The y-intercept and slope of the RC premise, that a pair of photographs taken plant is as sharp as possible. equation are estimated by computing the at perpendicular angles can adequately Time investments in this technique are sample averages of the estimated individ- model the 3-dimensional plant, seems to somewhat longer than with previous meth- ual plant y-intercepts and slopes, respec- hold up well. ods listed in Table 3, but the improved tively. The estimators for the y-intercept Problems encountered were the inabili- and slope can be written as ty to distinguish plant tissue from back- Table 3. Comparison of methods for estimating ground clutter, poor image quality, and utilization. l n _ -b0i (1) increasing area with decreasing biomass. n t=1 The backdrop solved the background clut- Method Time Accuracy and ter problem. Using a digital camera and (days)a (%) - (2) image manipulating software allows the Digital Photographic 4 ± 8.5 1 operator to manipulate image factors and Ocular by Plotb 2 + 19 does much to solve image quality prob- Ocular by Plantb + 2 respectively. For our example, lems. The area-biomass problem is typi- Leaf Lengthb + 16 Countb + 4 cally the result of allowing the background Plant 26 Height/Weight Ratio + 10-25 fl0 = -- b01=(13.2+13.8+12.1+ 13.2)/4 =13.1 screen to contact the plant or photograph- aTime 41=1 ing the plants under windy conditions includes model building and training. thereby changing the apparent area bPechanec and Pickford (1937) (these values are means). Clark (1945). l between photographs. These 2 factors are and /1 = - b11= (0.93+0.88+0.92+0.93)14 = 0.92 the main contributors to the plane of visi- 4=1 bility problem mentioned above. To results offset the increased time invested. Thus, the estimated e nation for the RC ensure the repeat positioning of the cam- Collecting samples and photographing era 2 model is y ij = (30 + 1 x = 13.1+0.92x for sequential shots, one could use alfalfa plants took an average of 20 min- for this example. tripods placed at right angles to the plant utes/plant. A large part of the post-sam- The estimated mean of y at a given and simply move the camera between the pling time spent depends on the camera value of x (say x*) is tripods. The use of quick release heads on used. Digital cameras provide usable the tripods would facilitate this approach. images immediately. Turn-around time for (Y1XX*)fl0+fllx* (3) With SigmaScan Pro® 3.0, the set thresh- development of film is dependent on old routine is a potential source of subjec- whether the film is sent to a lab or devel- tivity. Carefully setting the threshold val- oped in house. Once slides, or digitized For the example, an estimate of the per- ues in the first image so that plant tissue is images, are available, the process takes 3 centage of a plant remaining that was maximized and extraneous material is to 5 minutes per image to calculate the observed to have, say, 80% of the area minimized and then holding this value variables of interest. This data is then tran- remaining in a photograph of the plant constant throughout the series of images scribed to spreadsheet form and appropri- after it was browsed by an animal, is y will minimize the subjectivity inherent in ate statistical analyses completed. Training given x = 80 or (91 x = 80) = 130 + (31(80) this step. Figure 1D shows an example of of inexperienced individuals could proba- =13.1 + 0.92x = 86.5. a threshold setting where plant tissue is bly be done in 2 sessions of roughly 4 highlighted, but shadows and grass stems hours each, an hour each for photography, are not. digitizing, image analysis, and data collec- Results and Discussion There are additional considerations if tion plus some time to practice the tech- film is used. In a previous trial, we used nique. Once simple linear regression equa- Kodachrome, 64 slide film. Upon return tions are computed for each plant, con- Theoretically, the limit for detection of of the processed slides, they were digitized struction of the random coefficient model biomass removal should correspond to the with a Nikon Coolscan® slide digitizer. is relatively easy. area of the photo covered by a pixel given This gave us good control over the image Statistical analysis of this data set using that the tissue removed is on a plane visi- quality with respect to setting up the simple linear regression across all plants ble to the camera, i.e., not hidden by inter- image for further analysis. Prints could would seem to be an intuitive approach. vening tissue. This plane of visibility is possibly be used with a flatbed scanner, An important assumption associated with probably the most important source of but the superior resolution, for a given simple linear regression (SLR) is that the variation in the individual plant models.

142 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 set in black and white. C. Cleaned image. D. Image ready for measuring. Fig. 1. A. Apparatus for separating plant from background. B. Image 9.06 + 0.976x]. Figure 2 shows the random variable (y) at each value of x is each plant by simple linear regression as [Yip = From a statistical estimated and observed percentages of independent of the random variable y at the attribute of interest. preferred plant remaining for sample points from 10 any other value of x. In the present work, perspective, this approach is precise and plants not used in constructing the model. sequential samples were taken from each because the estimates are more can be There is an increased sensitivity at lower plant resulting in measures of biomass (y) because the sources of variability to improve levels of defoliation. The greater sensitivi- at various values of area (x) that are not partitioned and examined ty is advantageous due to desired utiliza- independent of each other. The RC model future models. 40 of tion levels, which usually requires leaving addresses the problem of a lack of inde- The RC model fit to the data from more than 50% of the plant. pendence by using the line produced for the alfalfa plants resulted in the equation

143 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 grasses or woody plants. Any situation where individual plants can be delineated in the field of view will probably work. Situations where plants are in close prox- imity, such as turf, may be problematic. As software of this type develops, the use of variation in color will likely improve the ability to differentiate specific targets and thereby improve the utility of digital- photographic methods such as the one presented here.

Conclusions

The digital-photographic method described herein accurately estimated the weight of individual alfalfa plants at sequential defoliation episodes to within ± 0 20 40 60 60 100 8.5%. The weight values were calculated from paired perpendicular photographs Observed Ptent Weight (g) taken of each plant at each level (0, 20, 40, Fig. 2. Estimated vs. observed plant weight remaining with 95% confidence intervals for 10 60, 80, and 100%) of defoliation. Time alfalfa plants with 4 samples from each plant. needed to complete the model building varied from 22 to 32 hours total time The optimal sample allocation between plants are used as the basic units and a depending on familiarity with techniques. number of plants and number of samples weight/area relationship is developed Once an appropriate model is developed per plant is dependent on plant morpholo- (Lommasson and Jenson 1938, Cook and for a plant species, objective estimates of gy. However, because of statistical Stubbendieck 1986). The digital-photo- plant material present can be readily cal- assumptions of normality, we recommend graphic method minimizes subjectivity culated based on photographs of the plant. no fewer than 30 plants. The optimal num- through the use of photographs. Repeated Before and after photographs will give ber of measures per plant is dependent on measurements of an image gave a coeffi- estimates of plant biomass removed, or the researcher's experience with the sam- cient of variation of 3.19%. Due to varia- remaining, if photographs are obtained in ple/photographic methodology. However, tion in plant morphology across species, short enough temporal intervals that plant for purposes of fitting the RC model, a seasons, or sites, a model may need to be growth or leaf abscission does not become minimum of 4 measurements per plant developed for each application (Caird a confounding factor. Within this context, including the whole plant measurement is 1945, Clark 1945). Once a model for a plant phenology should not affect esti- required. The researcher must conscien- given plant is developed, it should be use- mates. The use of a digital-photographic tiously attempt to select samples at the ful indefinitely. We do not know at the method for estimating plant biomass lost same x-values for each plant to meet sta- present time what the limits on usefulness to herbivores has applications in agricul- tistical assumptions of the model. For our are for a given model, say alfalfa, when ture, ecology, and other fields. As image samples, the mean deviation from the ideal applied to a plant with a different mor- analysis technology improves, the applica- 20% intervals between areas of consecu- phology. bility and accuracy of techniques such as tive samples within a plant for the 40 Collection of plant samples and their this will improve. One aspect in particular plants used in building the model was related photographs can be done fairly need of improvement is the ability to sepa- 12.5% (SD = 6.2, n = 160). rapidly in the field and subsequent model- rate target plants or groups of plants from One of the biggest drawbacks with cur- ing and analysis completed in the labora- background objects. For most shrubs and rent methods of forage and browse utiliza- tory. Additionally, photographs from sites forbs in arid and semiarid environments tion is the level of subjectivity required. to be analyzed for forage utilization can be that have a random or regular distribution, Ocular estimates rely on the experience of taken and analyzed at a time convenient to the problem associated with plant separa- the observer and may be subject to indi- the individual doing the assessment. A tion should be minimal. In all cases, the vidual bias, particularly if the observer is series of forage conditions could be pho- use of blocking objects, such as the back- not well trained or fails to periodically tographed at intervals throughout the sea- ing board used here, will be needed to sep- check his observations against a standard son and an analysis of season-long arate the plant from its background. The (Bement and Klipple 1959). Height/weight changes could be carried out. This tech- use of readily available computer technol- methods generally perform well, but are nique will be difficult to use on large ogy and photographic equipment will dependent on plants used to develop the shrubs or trees due to the limitations of decrease the subjectivity and increase the model being similar in morphology to the harvesting the entire plant for modeling. accuracy of field measurements of plant plants being estimated (Clark 1945). The Estimates of weight/unit area may allow geometry and associated losses due to her- technique described in this paper incorpo- the use of this method for large shrubs or bivory and/or other factors. rates aspects of several methods. It is simi- trees although indirect methods are proba- lar to ocular-estimate-by-plant and bly of more value (Bonham 1989). Also, height/weight methods in that individual this method has not yet been tested on

144 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Literature Cited Cook, C.W. and J. Stubbendieck. 1986. Lommasson, T. and C. Jensen. 1938. Grass Range research: basic problems and tech- volume tables for determining range utiliza- niques. Soc. Range Manage., Denver, Colo. tion. Sci. 87:444. Bement, R.E. and G.E. Klipple.1959. A pas- Estell, R.E., E.L. Fredrickson, D.M. Pechanec, J.F. and G.D. Pickford. 1937. A ture comparison method of estimating uti- Anderson, K.M. Havstad, and M.D. comparison of some methods used in deter- lization of range herbage on the central Great Remmenga. 1998. Relationship of tarbush mining percentage utilization of range grass- Plains. J. Range Manage. 12:296-298. leaf surface terpene profile with livestock es. J. Agr. Res. 54:753-765. Bonham, C.D.1989. Measurements for terres- herbivory. J. Chem. Ecol. 24:1-12. Society for Range Management. 1974. A trial vegetation. John Wiley & Sons, New Graybill, Franklin A. 1976. Theory and appli- glossary of terms used in range management, York, N.Y. cation of the linear model. Duxbury Press, 2nd Edition. Soc. Range Manage., Denver, Caird, R.W. 1945. Influence of site and graz- North Scituate, Mass. Colo. ing intensity on yields of grass forage in the Holechek, J.L., R.D. Pieper, and C.H. Texas Panhandle. J. Forest. 43:45-49. Herbel. 1989. Range management: princi- Clark, I. 1945. Variability in height of forage ples and practice. Prentice Hall, Engelwood grasses in central Utah. J. Forest. Cliffs, N.J. 43:273-283.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 145 J. Range Manage. 56:146-151 March 2003 An index for description of landscape use by cattle

HAITAO ZUO AND MARY S. MILLER-GOODMAN

Authors are former Graduate Research Assistant and Associate Professor, Department of Agronomy and Soils, Auburn University, Auburn, Ala., 36849- 5412. Research supported by an Alabama Department of Environmental Management Section 319 Nonpoint Source Program Grant and an Auburn University Dissertation Research Award.

Abstract Resumen

Understanding the role of landscape diversity in livestock dis- Entender el papel de la diversidad del paisaje en los patrones tribution patterns is an important consideration for design of de distribucion del ganado es de considerable importancia para effective grazing systems. The objective of this study was to el diseno de sistemas de apacentamiento efectivos. El objetivo de develop and evaluate a Distribution Evenness Index (DEI) based este estudio fue desarrollar y evaluar in Indice de Uniformidad on the Shannon-Wiener index to characterize cattle distribution de Distribucion (DEI) basado en el indice de Shannon-Weener patterns for a heterogenous landscape within a given period of para caracterizar los patrones de distribucion del ganado en un time. Observations of diurnal behavior of beef cattle (Bos taurus) paisaje heterogeneo dentro de un periodo de tiempo dado. Se were made in grassland, wooded, and riparian habitats within a hicieron observaciones del comportamiento diurno del ganado fenced landscape from March to October 2000 at a farm in para carne (Bos taurus) en habitats de zacatal, boscoso y north-central Alabama. The DEI was calculated based on obser- ribereiio, las observaciones se realizaron de marzo a abril del vation records at different time intervals (15-, 30-, and 60-min) 2000 dentro de un paisaje cercado en una granja de la region and different levels of grassland habitat subdivision (18-, 9-, and norte-centro de Alabama. El DEI fue calculado en base a los reg- 6-zones). Comparisons of calculated DEI values were made istros de observacion a diferentes intervalos de tiempo (15, 30 y among different habitat types, observation intervals, landscape 60 min) y diferentes niveles de subdivision del habitat de zacatal subdivision levels, and daytime periods. Annual DEI means indi- (18, 9 y 6 zonas). Se realizaron comparaciones de los valores cal- cated low evenness of cattle distribution in riparian (0.517) and culados del DEI entre diferentes tipos de habitat, intervalos de wooded habitats (0.606), and consistently high evenness in the observacion, niveles de subdivision del paisaje y periodos del grassland habitat (0.860). Although grazing activity in the grass- dia. Las medias anuales del DEI indicaron una baja uniformidad land habitat was uneven between different daytime periods de la distribucion del ganado en los habitats ribereno (0.517) y (0.565 to 0.679), when combined for the total daytime period, boscoso (0.606) y una uniformidad consistentemente alta en el grazing activity in the grassland habitat had a high evenness habitat de zacatal (0.860). Aunque la actividad de apacentamien- value (0.855). Relative stability of the DEI calculated between to en el habitat de zacatal fue desuniforme entre los diferentes selected spatial and temporal scales in this study indicated that periodos del dia (0.565 a 0.679) cuando se combinaron para el the index may be useful for comparison of evenness of livestock total del periodo del dia, la actividad de apacentamiento en el habitat use and grazing patterns between different studies at habitat de zacatal tuvo un valor de uniformidad alto (0.855). La similar spatial and temporal scales. estabilidad relativa del DEI calculada entre las escalas espaciales y temporales seleccionadas en este estudio indico que el indice puede ser util para la comparacion de la uniformidad del use del habitat por el ganado los patrones de apacentamiento entre Key Words: y beef cattle, diurnal grazing behaviors, location diferentes estudios en escalas espaciales temporales similares. choice, distribution evenness, riparian areas y

Grazing distribution patterns affect optimal forage utilization, nutrient recycling, and ultimately, pasture persistence and grazing Multiple regression (Senft et al. 1983), probability distributions capacity. Thus, an important principle of grazing management is (Arnold and Maller 1985), and inverse Gaussian distribution to maintain an even distribution of grazing animals within a graz- function (Pickup 1994) have been used to predict or measure ing unit or area (Vallentine 2001). In addition, animal agricultural livestock grazing distribution patterns. A drawback to these production practices are being increasingly scrutinized for their approaches is that these models cannot be transferred from 1 site impact on water quality throughout the USA (Martin 1997). to another since relationships between distribution patterns and Therefore, livestock distribution is a fundamental concern in environmental characteristics vary from location to location grazing system design. This is especially true for grazing units (Bailey et al. 1996). The objective of this study was to develop an that include wooded riparian areas, since cattle have been report- index based on modification of the Shannon-Wiener index ed to spend more time near shade and water sources (Blackshaw (Shannon and Weaver 1949) by which evenness of the distribu- and Blackshaw 1994). tion patterns of cattle location choice and behaviors could be easily quantified within a heterogeneous landscape over a given period of time. Sensitivity and stability of the distribution evenness The authors wish to thank Mr. James Glenn, Glendale Farms, Moulton, Ala. for index was tested at different temporal and spatial scales. generously allowing access to his property. We also thank Mr. Rick Zellmer, GIS Specialist, USDA-MRCS, Auburn, Ala. for assistance with map development. Manuscript accepted: 15 Jun. 02.

146 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Materials and Methods averaged 5 AU ha (20 total head) during the cool season (October to April) and 4 Description of study site AU ha` (17 total head) during the warm The study was conducted between season (May to September). March and October 2000 at Glendale Three habitat types were defined within Farms, close to the town of Moulton the studied landscape as riparian (stream, (196.6 m, 34°29'N, 87°18'W) in the Flint streambanks, and streamside woods), Creek Watershed of north-central grassland (open pasture area), and wooded Alabama, USA. The average annual tem- (wooded areas along fence line and perature is 13 to 16°C. Average annual drainage way). The area ratio of different precipitation is 925 to 1,400 mm with habitat types was 1 (wooded) : 1.6 (ripari- maximum in midwinter and midsummer, an): 6 (grassland). Ground cover composi- and minimum in autumn. Stream dis- tion was quantified in both the grassland charge is generally greatest in late winter and riparian habitats using a point sam- and spring in response to precipitation. pling technique (Buckner 1985). Precipitation is generally adequate for for- Endophyte-infected tall fescue (Fescuta age growth, however, dry periods early in arundinacea L.) was the predominant veg- summer and in autumn can reduce bio- etation cover (84%) in the grassland habi- mass production. Rainfall during the study tat of the studied landscape; common period ranged from a high of 220 mm in bermudagrass (Cynodon dactylon L.) con- April to a low of 0 mm in October (Fig. tributed up to 11 % of the grassland vege- indicated 1). Total rainfall for the study period was tation cover. Measurements uni- 257.3 mm lower than the previous 30-year form fescue production and utilization average. throughout the grassland habitat (Zuo The studied landscape was fenced to 2001). Sycamore (Platanus occidentalis about 3.3 ha in rectangle and was part of a L., 54.3%) and oak (Quercus sp., 19.8%) larger grazing system that used rotational were the dominant overstory species in the stocking. A second-order stream (Sheats 5- to 10-m wide wooded portions of the Branch) flowed through the studied land- riparian habitat; riparian overstory cover scape (Fig. 2). The producer had an 80- averaged 48% in winter and 84% in summer. Ground cover of the riparian habitat Fig. 2. Map of the studied landscape (3.3 head beef cow-calf (Bos taurus; Hereford) ha), Glendale Farms, north-central Ala. herd that was allowed yearlong access to understory was dominated by the combi- the stream. During observation periods, nation of litter (41%) and bare ground stocking density of the studied landscape (28%). The total vegetative portion of the understory ground cover averaged 21% during the study period; tall fescue comprised approximately 40% of the total understory vegetation cover.

Observation of cattle behavior Two-day diurnal observations of cattle behavior were made in March 2000;1-day diurnal observations were made in May, July, August, and October 2000. From daybreak to dark, observations of weather conditions and cattle behavior were conducted at 15-min intervals with the assistance of binoculars from a convenient point that avoided disturbance of cattle. Data recorded during observations included: temperature, wind direction (Table 1), total number of head at each location, and activities of individual animals, such as grazing, lying, and loafing. Grazing activity represented times when cattle were harvesting and masticating forages; lying activity represented times when cattle were lying down at a given location; loafing activity represented activities other than grazing and lying, such as moving, standing, itching, and playing. The loca- Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec tion, numbers, and behavior categories of on a map Fig. 1. Monthly rainfall during 2000 compared to the previous 30-year average, north-central cattle were recorded landscape Ala. for each time interval during the observa-

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 147 tion periods; those maps were referred to z the 2 probabilities of cattle choice of each as distribution plots. Total time was calcu- H'= p; In p. (1) zone are equal, the DEI is largest and - to 0 when 1 freedom of lated as the sum of time cattle spent in the reduces zone's riparian habitat (including the stream), the Where pi = the proportion of total number choice is gone. If there are many, rather grassland habitat, and the 2 areas of wood- or behavioral categories of beef cattle in than 2 zones, then the DEI is largest when ed habitat. Grazing time, lying time, and the ith zone; the probabilities of the choices of each loafing time combined represented the zone are as near to equal as possible during a given period of time, namely, cattle total time period of the observation. The Z = total number of zones studied. percentage of cattle participation in each spend nearly identical time in each zone. For zones without a record of any category the other hand, the choice of 1 zone activity within each habitat type was cal- On if of behavioral activity, 0.001 was assigned has a probability near 1 so that all the culated. Based on the record of 15-min to make the calculation mathematically other choices have probability near 0, the 30- and interval observations, 60-min- feasible. The statistic H' was then stan- indication is that the cattle's choices are interval records were generated by dele- dardized for the number of zones involved heavily influenced toward one particular 1 3, tion of or 15-min intervals in the cal- (Z) to achieve the Distribution Evenness zone or cattle have little freedom of culation. Index (DEI): choice. In that case, the DEI does calculate to have a very small value, i.e., the distrib- Calculation of the Distribution z ution evenness is low. DEI =H'/1nZ=(-p;lnp;)/InZ (2) Evenness Index i=1 For each distribution plot recorded at a Comparisons at different temporal given time interval, the total landscape The Shannon-Wiener index was initially and spatial levels area was subdivided into zones by over- developed for human communication the- The Distribution Evenness Index (DEI) laying a transparent zone-subdivision of ory (Shannon and Weaver 1949) and has was calculated at different time intervals the landscape on each distribution plot. been widely applied in ecology as a mea- (15-, 30-, and 60-min) and different subdi- The transparent overlay had 5 zones even- sure of species diversity. Two assumptions vision levels (18-, 9-, and 6-zone) in the ly delineated in the riparian habitat, 18 must be satisfied to use the index for grassland habitat for both total activities zones in the grassland habitat, and 3 zones species diversity measurement: (1) indi- and categorized activities. When total in the 2 wooded habitats within the studied viduals are randomly sampled from an activities were taken into account, the DEI landscape. For the grassland habitat only, `infinitely large' population and (2) all represented the evenness patterns of loca- transparent overlays with 9- and 6-zone species from a community are included in tion choice or distribution within the subdivisions were also applied separately the sample (Kent and Coker 1992). grassland habitat. The DEI was also calcu- to each distribution plot. Based on the Although the original purpose of the lated based on the 15-min interval data for summary record of cattle numbers and Shannon-Wiener index was to describe cattle location choice in different habitat behavior categories for each zone at a many types of human behavior, the DEI types. To describe the distribution pattern given period of time, an index of cattle modification should be suitable to describe of grazing behavior in the grassland habi- distribution evenness was calculated by a the distribution of cattle location choice tat during daylight, the DEI was also cal- modification of the Shannon-Wiener index since cattle behavior is a continuous culated based on the 15-min interval data process and the (H'). The information parameter H' was number of subdivision for morning (before 1100 hours), midday zones is fixed for a given area studied. defined as: (1100 to 1300 hours), afternoon (1300 to To simplify the mathematical explana- 1700 hours), and evening (after 1700 tion of the DEI, suppose there are only 2 hour) periods. zones within a grazed landscape. When

Table 1. Predominant weather conditions for behavior observation periods, north-central Ala, March to October 2000.

Date Sunrise Sunset temp. temp. conditions (h) 9 March 0606 1749 13.3 3.3 skies, strong wind from 0600 to 1400 hours with one and one-half rain beginning at 0830 hour; remaining periods were sunny 10 March 0604 1750 Sunny before 0800 hour, cloudy and windy from 0800 to 1300 hours, then rain occurred until a thunderstorm at 1700 hours that lasted until dark 11 March 0603 1751 Light rain occurred before 1100 hours, then cloudy skies until dark 12 March 0602 1751 and calm 22 May 0542 1950 2-hour rain began at 1230 hours, remaining periods were sunny

23 May 0541 1951 18 July 0546 1958 19 July 0547 1958 and clear 15 August 0602 1933 16 August 0603 1932 with occasional partly-cloudy conditions 17 October 0653 1810 during morning periods; remaining periods were sunny 18 October 0654 1809 during morning periods; remaining periods were sunny

148 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 ness of cattle distribution was low in ripar- Habitat type ian and wooded habitats, and high in the grassland habitat. About 90% of total Riparian Grassland Wooded activities that occurred in the grassland habitat were grazing and about 75% of the total activities that occurred in wooded or riparian habitats were lying and loafing (Zuo 2001). Thus grazing activities caused consistently higher distribution evenness in the grassland habitat while unstable evenness patterns in riparian or wooded habitats were most closely related to seasonal effects on loafing and lying activities.

Effects of temporal and spatial levels Higher evenness of cattle distribution in the grassland habitat was indicated by the DEI calculated using 15-min interval observation records for both location choice and grazing activities during the cool-season versus the warm-season (Table 2). The most uneven distribution occurred in August, the warmest period during this study, when cattle spent the majority of Mar May Jul Aug Oct Annual diurnal time lying or loafing in wooded habitats and their foraging activities Month occurred mainly in shaded areas of the grassland habitat close to wooded or shad- Fig. 3. Monthly and annual comparison of cattle distribution eveness indicated by the DEI ed riparian habitats. The similar evenness (Distribution Evenness Index) for different habitat types in the studied landscape, pattern of total activities and grazing activ- Glendale Farms, north-central Ala. Annual means with the same letter indicate no signifi- ities demonstrated the major influence of cant difference (P < 0.05). grazing activities on cattle location choice in the grassland habitat (Table 2). On the Because the indices themselves will be riparian habitat during March compared to other hand, this similarity further indicated normally distributed if the Shannon- May, July, and October (Fig. 3). However, that cattle spent the majority of time in the Wiener index is calculated for a number of extremely uneven use of the riparian habi- grassland habitat for grazing activities. samples (Taylor 1978), it is possible to tat by cattle was noted in August since No significant difference was detected use parametric statistics to compare sets of evenness equaled 0 at that time. The between the DEI values calculated using DEI values (Magurran 1988). To test the wooded habitats had very high DEI values 30-min interval records and those based sensitivity of the DEI to specific cattle in May, July, and October, and very low on 15-min interval observation records behaviors, the paired t-test (P < 0.05) was Distribution Evenness Index (DEI) values (Table 2). However, the DEI values calcu- performed (PROC MEANS, SAS® version in March and August. Annual means of lated using 60-min interval records were 6.12) to test differences between the DEl diurnal observations indicated that even- values for total activities and grazing activities at different time intervals in the grass- Table 2. Comparison of DEI (Distribution Evenness Index) values for cattle location choice for all land habitat and grazing activities in the activities and grazing activity in the grassland habitat at different time-interval levels, Glendale Alabama 2000. grassland habitat at different zone-subdivi- Farms, north-central sion levels. Analysis of variance (PROC All activities Grazing activity ANOVA, SAS® version 6.12) and the least 15-min 30-min 60-min significant difference (P < 0.05) were used to detect overall differences for total activi- Marcht 0.883 0.890 ties in different landscape habitat types, May 0.896 0.820 and for grazing activities in the grassland July 0.822 0.820 habitat at different daytime periods. August 0.756 0.780 October 0.944 0.930

Results and Discussion Paired t-test 15-min 15-min 30-min vs. vs vs. Location choice distribution 30-min 60-min 60-min Consistently high cattle distribution Difference 0.012 0.087 0.075 evenness was indicated for the grassland (probability) (0.5153) (0.0225)* (0.0050)* habitat throughout this study. Higher dis- October through April = cool season; May through September = warm season. tribution evenness was detected in the *Indicates significant difference at P < 0.05.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 149 Temporal distribution of grazing activities Consistently high evenness of diurnal grazing patterns in the grassland habitat was indicated by the DEI based on total daytime observations, although uneven grazing patterns were detected for different daytime periods from March to W October (Fig. 4). This result suggested that D cattle could choose different zones at different times with small overlap between b b b ab a them, or their patch preference at different periods of daytime was complementary across the total daytime period. This temporal distribution pattern of cattle grazing activities supports the suggestion made by Senft et al. (1987) that observed distribution patterns are the cumulative effects of Morning Midday Afternoon Evening Total diet selection and feeding station behav- Daytime iors. It also supports the conclusion made Diurnal period by Bailey (1995) that no patch preferences will be measured in a homogeneous area if Fig. 4. Grazing distribution evenness in the grassland habitat indicated by the mean DEI data are pooled within a day, and time (Distribution Evenness Index) for different diurnal periods, March to October 2000, spent in patches is not consistent through- Glendale Farms, north-central Ala. Means with different letters indicate a significant dif- out the day. ference (P < 0.05). Advantages and drawbacks consistently and significantly lower than inflated the unevenness of cattle distribu- An important criticism of grazing the DEI values based on 15-min or 30-min tion when the true value of DEI was actu- behavior studies has been that there is no interval records. This decrease indicated ally relatively low. On the other hand, the standardized technique of observation that information about actual cattle distrib- subdivision of a given habitat area itself (Bailey et al. 1996), thus making compari- ution patterns in this study could have should not be so small as to impact animal son of animal behavioral patterns difficult been lost if the observation interval had aggregation behavior, namely, each zone between different studies. The use of the been greater than 30 min. should be able to hold all animals together DEI in this study demonstrated that the Relative stability of the DEI was detect- with enough inter-animal distance, about index has relative stability for observation ed at various levels of grassland habitat 78.5 m2 head' (Phillips 1993). For exam- intervals less than 30-min and different subdivision for both 15- and 30-min inter- ple, the 18-zone grassland habitat subdivi- grassland habitat subdivision levels. We val records, especially during the cool-sea- sion in this study allowed approximately hypothesize that the relevant sensitivity or son (Table 3). Higher DEI values were 80 m2 head-` when all cattle were in 1 stability range of DEI could be acquired obtained during the warm season when the zone, even during the cool-season when a for larger landscape areas based on live- 6- or 9-zone subdivision was used for the higher stocking density was employed. stock activities pooled across several grassland habitat; an exception was the 6- weeks. This information could then be zone subdivision in August. It appeared used to describe overall grazing patterns in that subdivision with a larger zone size the landscape as well as evenness of forage utilization. In these circumstances, Global Table 3. Comparison of DEI (Distribution Evenness Index) values for 15- and 30-min interval records at different levels of grassland habitat subdivision, Glendale Farms, north-central Ala., Positioning System (GPS) technology 2000. could be used to obtain animal landscape positions at 5-min intervals (Turner et al. 15-min interval records 30-min interval records 2000). Thus, additional information about 18-zone 9-zone 6-zone livestock distribution could be obtained for Marcht 0.868 0.865 larger landscapes through combination of May 0.903 0.968 the DEI with GPS technology. July 0.823 0.880 The DEI appears to be a relatively sim- August 0.749 0.769 ple and direct method of obtaining infor- October 0.934 0.965 mation about uniformity of cattle distribu- Paired t-test tion in heterogeneous landscapes. The advantages of temporal and spatial stabili- 18-zone 18-zone 9-zone 18-zone 18-zone 9-zone ty should facilitate use of the DEI vs. vs vs. vs. vs. vs. for 9-zone 6-zone 6-zone 9-zone 6-zone 6-zone rapid evaluation of the effects of certain management practices on cattle distribu- Difference -0.034 -0.023 -0.011 -0.038 -0.032 0.006 tion patterns for the same pasture, or com- (probability) (0.0515) (0.5365) (0.7250) (0.0144)* (0.3317) (0.8283) t parison between different studies with October through April = cool season; May through September = warm season. similar spatial and temporal scales.

150 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Summary and Conclusions Literature Cited Martin, J.H., Jr. 1997. The Clean Water Act and animal agriculture. J. Environ. Qual. 26: 1198-1203. Uniformity of cattle distribution patterns Arnold, G.W. and R.A. Maller. 1985. An Pickup, G. 1994. Modeling patterns of defolia- in a grazed landscape was described by a analysis of factors influencing spatial distrib- tion by grazing animals in rangelands. J. Distribution Evenness Index (DEI) based ution in flocks of grazing sheep. Appl. Anim. Appl. Ecol. 31:231-246. Behav. Sci.14:173-189. on modification of the Shannon-Wiener Phillips, C.J.C. 1993. Cattle behaviour. Bailey, D. W. 1995. Daily selection of feeding Farming Press, Ipswich, U.K. index. The DEI also provided general areas by cattle in homogeneous and heteroge- Senft, R.L., L.R. Rittenhouse, and R.G. information about evenness of cattle habi- neous environments. Appl. Anim. Behav. Woodmansee. 1983. The use of regression tat use for various behaviors within the Sci. 45:183-200. equations to predict spatial patterns of cattle heterogeneous landscape studied. When Bailey, D.W., J.E. Gross, E.A. Laca, L.R. behavior. J. Range Manage. 36:553-557. the DEI was calculated based on observa- Rittenhouse, M.B. Coughenour, D.M. Senft, R.L., M.B. Coughenour, D.W. Bailey, Swift, and P.L. Sims. 1996. Mechanisms tions at a given time interval, cattle distri- L.R. Rittenhouse, O.E. Sala, and D.M. that result in large herbivore grazing distribu- bution evenness could be described for Swift. 1987. Large herbivore foraging and tion patterns. J. Range. Manage. 49:386-400. ecological hierarchies. BioSci. 37:789-799. any behavior type or temporal period. This Blackshaw, J.K. and A.W. Blackshaw.1994. Shannon, C.E. and W. Weaver. 1949. The study, conducted at a small spatial scale Heat stress in cattle and the effect of shade mathematical theory of communication. (3-4 ha), indicated that observations on production and behavior: A review. Aust. Univ. of Ill, Urbana, Ill. should be made at time intervals of 30-min J. Exp. Agr. 34:285-295. Taylor, L. R. 1978. Bates, Williams, Hutchinson or less since behavioral information was Buckner, D.L. 1985. Point-intercept sampling - a variety of diversities. p. 1-18. In: Mound, in revegetation studies: Maximizing objectiv- L. A. and N. Warloff (ed.) Diversity of insect lost when longer intervals were used. 2nd ity and repeatability. p.110-113. In: Proc. faunas: 9th Symposium of the Royal However, relative stability of the DEI cal- Annu. Meeting. Bridging the gap between Entomological Society. Blackwell, Oxford. culated between selected spatial and tem- science, regulation, and the surface mining Turner, L.W., M.C. Udal, B.T. Larson, and poral scales in this study indicated that operation. Amer. Soc. Surface Mining and S.A. Shearer. 2000. Monitoring cattle comparison of the evenness of livestock Reclamation, Princeton, W.Virg. behavior and pasture use with GPS and GIS. habitat use and grazing patterns between Kent, M. and P. Coker. 1992. Vegetation Can. J. Anim. Sci. 80:405-413. and A practical different studies at similar spatial and tem- description analysis: Vallentine, J.F. 2001. Grazing Management approach. CRC Press, Inc., Boca Raton, Fla. poral scales is possible and should be (second edition). Academic Press, San Magurran, A.E. 1988. Ecological diversity Diego, Calif. explored through further research. and its measurement. Princeton University Zuo, H. 2001. Cattle behavior and impacts on Press, Princeton, N.J. streamwater quality. Ph.D. Diss. Auburn Univ., Auburn, Ala.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 151 J. Range Manage. 56:152-158 March 2003 Hydrologic and sediment responses to vegetation and soil disturbances

J.H. GIORDANENGO, G.W. FRASIER, AND M.J. TRLICA

Authors are former Graduate Research Assistant, Colorado State University, Rangeland Ecosystem Science Dept., Fort Collins, Colo.; Research Hydraulic Engineer (Retired), USDA, Agricultural Research Service, Fort Collins, Colo.; Professor, Colorado State University, Rangeland Ecosystem Science Dept., Fort Collins, Colo. 80523.

Abstract Resumen

Soil erosion has been linked to stream sedimentation, ecosys- La erosion del suelo ha sido vinculada a la sedimentacion de tem degradation, and loss of rangeland productivity. However, las corrientes de agua, la degradacion de ecosistemas y la perdi- knowledge of soil loss, as it affects rangeland productivity or da de productividad de pastizales. Sin embargo, se carece de ecosystem sustainability is lacking. We evaluated the effects of 3 conocimiento de como la perdida de suelo afecta la productivi- levels of vegetation cover reduction (0, 27%, and 43%) and soil dad de los pastizales o la sostenibilidad del ecosistema. removal (0, 12, and 24 tonnes ha') on soil surface runoff and sed- Evaluamos el efecto de 3 niveles de reduccion de cobertura vege- iment yield in a sagebrush [Artemisia tridentata var. vasseyana tal (0, 27 y 43%) y remocion de suelo (0,12 y 24 ton ha') en el (Rydb.) Beetle] steppe under simulated rainfall. Time to runoff escurrimiento superficial y la produccion de sedimento en una initiation was affected by the vegetation cover reduction treat- estepa de "Sagebrush" [Artemisia tridentata var. vasseyana ments, but not by the soil removal treatments. The 43% vegeta- (Rydb.) Beetle] bajo lluvia simulada. El tiempo de inicio del tion canopy reduction treatment resulted in a shorter time to escurrimiento superficial fue afectado por los tratamientos de runoff initiation than did the 27% and 0% canopy reduction reduccion de cobertura vegetal, pero no por los de remocion de treatments (p = 0.002). Results from analysis of covariance indi- suelo. El tratamiento de 43% de reduccion de cobertura vegetal cated that vegetation reduction and soil removal did not signifi- resulto en un menor tiempo de iniciacion del escurrimiento que cantly affect sediment yield or runoff quantities in the first year el obtenido por los tratamientos de reduccion de cobertura de 0 y following treatments. Multiple regression analysis revealed total 27% (P = 0.002). Los resultados del analisis de covarianza indi- sediment yield was related to forb cover, sand in the upper soil caron que la reduccion de la vegetacion y la remocion del suelo profile (0-5 cm), and the amount of bare ground. Time to runoff no afectaron significativamente el rendimiento de sedimento o initiation was positively correlated with slope. Despite the lack of las cantidades escurrimiento en el primer ano despues de aplica- significant treatment differences, we do not conclude that these dos los tratamientos. El analisis de regresion multiple revelo que soil removal and vegetation reduction treatments had no affect la produccion total de sedimento estuvo relacionada a la cobertu- on soil surface hydrology and sediment yield. There are numer- ra de hierbas, contenido de arena en el perfil superior del suelo ous studies that show a strong relationship between vegetation (0--5 cm) y la cantidad de suelo desnudo. El tiempo de inicio del reduction and soil erosion. Future research at this site may escurrimiento estuvo positivamente correlacionado con la pendi- reveal long-term treatment effects that were not apparent in first ente. A pesar de la falta de diferencea significativa entre year results. tratamientos nosotros no concluimos que estos tratamientos de remocion de suelo y reduccion de vegetacion no tuvieron efecto en la hidrologia superficial del suelo y la produccion de sedimen- Key Words: rainfall simulation, erosion, runoff, hydrology, sage- to. Hay numerosos estudios que muestran una fuerte relacion brush steppe entre la reduccion de vegetacion y la erosion del suelo. Futuras investigaciones en este sitio pueden revelar efectos a largo plazo de los tratamientos que no fueron aparentes en los resultados del Soil erosion is a major problem throughout the world (Meyers primer ano. 1984, Pimentel et al. 1995) and has been recognized as a problem in the United States since the early 1900's (Sampson 1918). In the United States, an estimated 4.4 billion tonnes of soil are erod- (Buckhouse and Gaither 1982), and impair fish habitat (Binkley ed by wind and water every year (Bills and Heimlich 1984). In and Brown 1993). In addition to ecological impacts, soil erosion many cases, these soil losses lead to increased sediment loads in can lead to decreased rangeland productivity through the loss of streams and rivers which can reduce productivity of aquatic organic matter and plant nutrients. A principle challenge for ecosystems, shorten the life span of ponds and reservoirs rangeland managers is to optimize forage production for herbivores without reducing the ecological integrity of rangelands or This research was funded by the United States Department of Agriculture, decreasing their societal benefits. Agricultural Research Service and the Colorado State University Agricultural Not all erosion is a result of land mismanagement. Soil erosion Experiment Station. The authors wish to thank the Arapaho National Wildlife is a natural process of dislodgment of soil particles from the sur- Refuge and all the support personnel for their assistance on the study. Manuscript accepted 22 May 02. face and subsequent transport by water and wind (Brooks et al.

152 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 1997). Erosion at a rate greater than 11.2 intermountain glacial basin known as developed by the University of Wyoming tonnes ha' yr' exceeds the estimated rate North Park and encompasses 10,037 (Linse 1992) was used to apply simulated at which most parent material weathers hectares. The study site (40° 36' 48.5"N, rainfall to plots at an intensity of 100 mm (McCormack and Young 1981). 106° 16' 29.6W) was grazed by livestock hour' for 30 minutes (dry run) at Accelerated erosion and changes in soil prior to being designated as a wildlife antecedent soil moisture. After 30 min- surface hydrology have been reported refuge in 1967 (Zanier 1999) and is repre- utes, the third of 3 spray nozzles was acti- under conditions of reduced vegetation sentative of a sagebrush steppe ecosystem vated and water pressure was increased to cover and altered soil structure (Sampson classified in good rangeland health. apply rainfall at an intensity of 150 mm 1918, McCalla et al. 1984, Linse et al. Current management of the research site is hour' for an additional 30 minutes (wet 2001). Preventing accelerated soil erosion exclusion of livestock grazing, but run). has been regarded as the key to maintain- includes winter grazing by 200-500 free- The rainfall intensity of 100 mm hour' ing rangeland ecosystem sustainability roaming Rocky Mountain Elk (Cervus ela- was chosen to simulate a high intensity- (Buckhouse and Gaither 1982). Land man- phus canadensis, Erxleben). low frequency storm that would exceed agers, however, do not have scientifically The site is at an elevation of approxi- the infiltration capacity and produce mea- based evidence for the amount of vegeta- mately 2,500 m, receives an average annu- surable runoff (Linse et al. 2001). The tion cover necessary to maintain rangeland al precipitation of 240 mm, and experi- increased application rate, in the wet run, sustainability. The influence of soil sur- ences an average of 30 consecutive frost- was used to determine whether sediment face degradation (i.e., removal of the soil free days per year. Soils are listed as a yield from this ecosystem was energy lim- "A" horizon, changes in surface rough- cabin sandy loam in the subgroup Argic ited. Six wedge rain gauges were spaced at ness, decreasing micro-channel sinuosity, Cryoboroll. These soils are deep, well equal distances around the perimeter of soil compaction and previous soil erosion) drained soils that formed in gravelly allu- each plot to measure the total amount of on rangeland sustainability is even less vium, and are underlain by gravelly sand simulated rainfall that reached the plot understood. at a depth of 50 to 100 cm (USDA 1981). during the dry and wet runs. Runoff sam- To prevent soil erosion by water, it is The average soil texture was 61 % sand, ples were manually collected every 2 min- necessary to maintain the soil surface in a 22% silt, and 17% clay, but ranged from utes, for a duration of 6 seconds, from condition that readily accepts water 55-64% sand, 19-27% silt, and 14-25% each plot. (Brooks et al. 1997). The soil "A" horizon clay. often contains significant amounts of Treatments organic matter which improves infiltration Experimental Design Soil and vegetation treatments were and increases soil water holding capacity. The experimental design was a random- applied in July 1999. Soil was vacuumed Therefore, removal of the "A" horizon can ized complete block experiment with a 3 x from plot surfaces using a gas-powered lead to increased erosion of underlying 3 factorial arrangement of treatments. The blo-vac. Soil was vacuumed from bare soil horizons (Pimentel et al. 1995). The treatments consisted of 3 levels of soil ground, coppice dunes, and through the effects of soil removal on soil surface removal (0, 12, and 24 tonnes ha') and 3 crowns of all forb and grass plants to hydrology and sediment yield from range- levels of plant canopy reduction (0, 27, achieve a uniform soil removal. lands are not well understood and must be and 43%) with 3 replications of each treat- Vacuumed soil was field weighed and clarified before accurate soil loss thresh- ment (27 plots). Individual treatments taken back to the lab to dry, obtain a more olds are developed. were randomly assigned to pairs of plots. accurate weight, and determine particle The purpose of this research project was Each plot was paired with an adjacent plot size. Non-woody canopy cover was killed to determine the effects of vegetation of the same size according to similar vege- by spraying Roundup herbicide through cover reduction and soil removal on soil tation characteristics and amount of bare one of 2 perforated board templates onto surface hydrology and sediment yield in a ground. One of the plot pairs was used for circular patch areas within the plots. The big sagebrush [Artemisia tridentata var. destructive sampling (above ground bio- exposed area of template 1 represented vasseyana (Rydb.) Beetle] steppe ecosys- mass and soil bulk density) and the other 30% of the plot area. The exposed area of tem. The main hypotheses of this project plot was used for collecting sediment yield template 2 represented 60% of the plot were that (1) a 27% reduction in vegeta- and runoff variables. The plots were locat- area. However, vegetation cover reduction tion cover would not result in altered soil ed on a 7% slope within an exclosure area from these treatments averaged 27% and hydrology, (2) a 43% reduction in cover of approximately 5,000 m2. 43% for the 30% and 60% templates, would result in increased runoff and sedi- respectively. Where shrub cover was ment production, (3) soil removal of 12 Plot Installation and Rainfall exposed beneath the perforations, the tonnes ha' would not affect soil surface shrub canopy was spray painted and hydrology, and (4) soil removal of 24 Simulator removed with pruning shears. The result- tonnes ha' would result in increased Plots measuring 2 x 0.6 m were delin- ing soil-vegetation characteristics were by inserting 15 cm wide metal sheets runoff and sediment yield. eated designed to represent variations in degrad- to a depth cm on the into the soil of 3-6 ed rangeland sites. up-slope end and sides of each plot. Plot Methods and Materials pairs, separated by one meter, were oriented lengthwise with the slope. Runoff col- Plot Characterization lection troughs were placed at the down- Vegetation cover and soil surface rough- Site Description slope end of each plot and sealed against ness were measured before and after treat- on an Field experiments were conducted the soil surface with an expanding foam. ments were applied. Vegetation cover was upland area of the Arapaho National Rainfall simulations took place during the estimated by individual species using the Wildlife Refuge (ANWR), 15 km south of period from 27 July 1999 to 5 August point frame method (Bonham 1989, Linse is located in an Walden, Colo. The ANWR 1999. A rotating boom rainfall simulator et al. 2001). A 2 x 0.6 m 100 point hori-

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 153 zontal plane pin table was placed onto as runoff during the equilibrium runoff ment yield residual plots indicated the each plot and pins were lowered until period and was calculated for the dry and need for a natural logarithm (ln) transfor- either vegetation, standing litter, rock, sur- wet runs. mation of these data. Therefore, all data in face litter or bare ground was intercepted. this study were analyzed using a In trans- The initial pin hit was used to determine Sedigraph formation. This transformation was also canopy cover by species. As many as 2 pin Every runoff sample was transferred to used by Sharpley (1985) to analyze sedi- hits were recorded to determine canopy an individual sediment bottle until peak ment yield. cover of sagebrush and under story herba- runoff was reached (determined as the ceous species cover. Each pin was then point at which runoff leveled off or lowered to the soil surface to characterize showed an initial decline). After the initial Results and Discussion soil surface cover. Absolute canopy cover peak runoff rate was attained, every other by species and absolute soil surface cover runoff sample was transferred to a sedi- Vegetation and Soil Parameters was recorded as the percentage of total pin ment bottle for subsequent sediment Average pre-treatment vegetation strikes for that cover class per total pin analysis. Prior to sediment filtration, paper canopy cover across all plots was 45%, strikes for the plot. Cover, by plant life filters with 1 micrometer pores were dried but ranged from 30 to 65%. Average soil form, and soil surface characteristics were at 40° C for 4 hours and weighed. bulk density across plots was 1.22 cm3 categorized into 8 classes: grasses g (by Sediment samples were gravity filtered in the upper 0-5 cm and 1.25 cm3 in the species), forbs species), shrubs g (by (by and then dried for 6 hours at 80° C and lower 5-10 cm of the soil profile. Soil bulk species), standing litter (current year litter weighed (Gutierrez-Castillo 1994). A sedi- density was not significantly different which had not fallen to the ground and graph was generated by plotting sediment among treatments. standing woody debris), cryptograms, bare yield vs. time. The area under the curve of Water erosion typically removes fines ground, rock, and litter. Surface roughness this graph was integrated to determine (clay- and silt-sized soil particles) and was measured using a digital caliper rest- total sediment yield. Sediment yield is leaves coarse particles behind (Ellison ing on top of the lowered pins. Surface reported in kg ha' mm runoff'. These 1944, 1948, Pearce et al. 1998). Although roughness was calculated as the standard units adjust sediment yield for the amount particles larger than clay and silt, and deviation of the elevation of 100 pins of runoff carrying the sediment to the large soil aggregates, may be detached by (Kuipers 1957, Linse et al. 2001). Finally, trough. In addition to these parameters, raindrop splash, they settle out of the over- the hill slope (%) was calculated by deter- sediment yield which occurred between land flow much sooner than finer particles mining the regression equation for the the time to runoff initiation and the end of (Ellison 1944, Pearce et al. 1998). The soil average slope of a plane of the plot as the dry run was subdivided into 3 periods texture of the vacuumed soil was 65% determined by pin height measurements. (early, middle, and late). sand, 22% silt and 13% clay, with ranges In each destructively sampled plot, two, of 64-69%,19-24%, and 12-15%, respec- 192 cm3 soil samples were collected to a tively. This texture was not statistically depth of 10 cm. These samples were Data Analysis different from the in situ soil texture. This divided into 2 subsamples at 0-5 cm and Data were analyzed using standard sta- suggests that vacuuming of the surface 5-10 cm to determine soil bulk density tistical programs (SAS Institute Inc. 1998) soil did not selectively remove silt- and using the core method (Blake and Hartge available for analysis of covariance (PROC GLM), multiple regression analy- clay-sized particles, as would be expected 1986) and soil texture according to the under natural water erosion events. hydrometer method (Bouyoucos 1962). sis (PROC REG), and repeated measures Although soil particle size was not affect- Immediately prior to each rainfall simula- analysis of variance (PROC MIXED). ed by the vacuuming, the soil removal tion, three, 51 cm3 soil Differences were considered significant at cores were treatment did affect surface roughness. removed adjacent to the plot pair. These P < 0.10. Analysis of covariance was used The average surface roughness of vacu- samples were divided into subsamples at to determine whether soil removal or veg- umed plots (27 S.D.) was significantly 0-5 cm and 5-10 cm to determine etation reduction significantly affected greater (P < 0.001) than the average pre- antecedent soil moisture using the gravi- runoff or sediment yield. Covariates treatment surface roughness (22 S.D.). metric method (Gardner 1986). included slope, antecedent soil moisture, soil bulk density, total live vegetation cover, sagebrush cover, litter cover, grass Runoff Hydrograph Runoff cover, forb cover, bare ground, soil tex- The amount of runoff is the difference Hydrographs were generated by plotting ture, and surface roughness. Multiple lin- between the amount of water applied runoff values over time for dry and wet ear regression, with stepwise selection, minus the water retained on the soil and runs. The shape of the hydrograph was was used to determine whether significant plant surfaces, and the amount that infil- used to evaluate the time to runoff initia- relationships existed between independent trated into the soil (Frasier et al. 1998a). tion, equilibrium runoff rate, and equilibri- variables (bare ground, grass cover, sage- Three runoff parameters were analyzed: um runoff ratio (Frasier et al. 1998a). brush cover, litter cover, total vegetation time to runoff initiation, equilibrium Time to runoff initiation was defined as cover, sand, slope, soil bulk density, runoff ratio in the first 30 minutes (dry the time when the runoff rate exceeded 5% antecedent soil moisture, surface rough- run) of the rainfall simulation, and equilib- of the rainfall rate. The equilibrium runoff ness, and rainfall intensity), and sediment rium runoff ratio in the last 30 minutes period was defined as the time when the yield or runoff values (SAS Institute Inc. (wet run) (Table 1). rising limb of the hydrograph leveled off 1998). The average sediment yield in each In many rangeland situations, the runoff until the end of the appropriate time period period was used in repeated measures characteristics at the beginning of the (dry or wet run). The equilibrium runoff analysis of variance to determine whether storm, such as time to runoff initiation, are ratio was defined as the average percent- there was a significant decrease in sedi- most important because storm durations age of applied rainfall that was collected ment yield over time. Analysis of sedi- are too short to develop equilibrium runoff

154 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 1. Means (standard errors) for runoff parameters by treatment from sagebrush steppe rain- ution of the vegetation over the plot. The fall simulation plots. Treatment differences were not significant at P < 0.10. spatial distribution of vegetation influences the microchannel network which, in Soil Vegetation Time to Runoff Ratio Runoff Ratio turn, may have dominated the runoff Removal Reduction Runoff Initiation Dry Run Wet Run processes in this study. (tonnes had) (%) (minutes) ( %) The equilibrium runoff ratio, averaging 0 0 8.3 (7.5) all treatments, was significantly greater (P 0 27 2.3 < 0.0001) in the wet run (67%) than in the 0 43 4.3 (9.8) dry run (43%). In addition to the increase 12 0 5.7 in rainfall intensity, surface sealing of 12 27 3.7 (6.9) these soils may account for the increased 12 43 4.3 (8.7) runoff ratio (Farres 1978). Also, saturation 24 0 6.3 (9.8) of surface litter, plant surfaces and soil 24 27 3.0 (9.8) macropores would result in an increased 24 43 5.7 (2.3) proportion of rainfall reaching the soil surface and a decrease in infiltration.

(Frasier et al. 1998a). Our results showed time, that regrowth of vegetation in the Sediment Yield that time to runoff initiation was affected herbicide-treated areas may help to main- There was an interaction between soil by the vegetation canopy reduction treat- tain the degree of channel formation pre- and vegetation treatments (P = 0.096) for ments, but not by the soil removal treat- sent at the time of rainfall simulations. total sediment yield. A 27% vegetation ments. The 43% vegetation canopy reduc- Unless annual treatments are applied, the reduction resulted in greater sediment tion treatment resulted in a shorter time to desired degraded rangeland conditions yield, at the soil removal level of 0 tonnes runoff initiation than did the 27% and 0% may not be maintained, and it is unlikely had, than did a 0 or 43% vegetation reduc- canopy reduction treatments (P = 0.002). that long term effects from these vegetation. Analyses of sediment data were done Nyhan et al. (1984) and Giordanengo tion treatments will be noticeable. by soil treatment, averaging over vegeta- (2001) found that antecedent soil moisture Past research has shown that amount of tion treatment, and visa versa. Four sedi- had a pronounced effect on time to runoff slope (Wischmeier and Smith 1978, ment yield parameters were analyzed: total initiation. However, their results were Sharpley 1985) and bare ground (Branson sediment yield over the 60 min rainfall from consecutive rainfall simulations that and Owen 1970, Wischmeier and Smith simulation, sediment yield over the first 10 occurred over successively wet conditions 1978) were positively correlated with min, sediment yield for the dry run, and on the same plots. In this study, antecedent runoff. However, our results indicated that sediment yield for the wet run (Table 2). soil moisture did not vary great enough slope and bare ground were not significant Analysis of covariance indicated that from plot to plot to affect runoff variables. covariates and did little to explain treat- total sediment yield was not significantly Equilibrium runoff rates have been used ment differences in runoff data. Other affected (P = 0.441) by soil or vegetation as indicators of treatment differences in covariates that were used, but did not sig- treatments. Likewise, sediment yield in the previous rainfall simulation studies nificantly affect equilibrium runoff ratios, first 10 minutes (P = 0.469), first 30 min- (Simanton et al. 1991, Frasier et al. included total live vegetation cover, sage- utes (P = 0.624), and last 30 minutes (P = 1998b). Equilibrium runoff occurs when brush cover, grass cover, litter cover, sur- 0.229) after rainfall simulations began was soil surface layers are saturated and is rep- face roughness, antecedent soil moisture, not significantly affected by soil or vege- resentative of long duration precipitation and soil bulk density. Soil bulk density tation treatments. Given the slow decom- events that exceed the infiltration rate only represented bare interspace areas and position rates in this arid environment, the (Frasier et al. 1998b). Because equilibrium did not vary significantly enough among roots and shoots of herbicide-treated runoff rates are influenced by rainfall treatments to influence runoff. The plants should have been intact when the intensity, we used equilibrium runoff absolute cover of vegetation may not be as rainfall simulations occurred. Therefore, ratios [(runoff rate divided by rainfall rate) important a covariate as the spatial distrib- the indirect effects of vegetation on sedi- x 100] to compare treatment effects. Equilibrium runoff ratios in the dry and Table 2. Means (standard errors) for sediment yield parameters by treatment from sagebrush wet runs were not significantly affected by steppe rainfall simulation plots. Treatment differences were not significant at P < 0.10. treatments (P = 0.732 and 0.872, respectively). These results support findings by Soil Vegetation Sediment Yield Busby and Gifford (1981), who reported Removal Reduction Total sr 10 Minutes Dry Run Run that removal of vegetation did not have an (tonnes ha') (%) ------(kg ha 1 mm runoffs) ------immediate effect on infiltration. 0 0 761 (429) 218 (111) 461 (234) 300 (217) Because the herbicide-treated dead veg- 0 27 4506 (2823) 1124 (572) 2439 (1420) 2067 (1404) etation cover was not removed prior to 0 43 685 (332) 256 (135) 434 (218) 244 (122) rainfall simulations, the standing litter that 12 0 836 (77) 222 (48) 443 (37) 393 (114) remained was likely affecting rainfall 12 27 999 (167) 235 (80) 601 (176) 397 (44) interception in the same manner as live 12 43 1502 (616) 323 (165) 632 (285) 871 (422) vegetation. In addition, the vegetation 24 0 2178 (1776) 1108 treatments most likely did not have an 250 (206) (974) 1070 (804) affect on channel formation. Although not 24 27 504 (97) 150 (44) 253 (83) 250 (57) measured, it can be expected that over 24 43 1164 (456) 335 (113) 665 (263) 499 (232)

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 155 ment yield (i.e., increased litter cover, yield over time (P < 0.001). Sediment to a greater degree than the independent reduced bulk density and increased infil- yield in the late period was significantly variables used in this analysis. trability), reported by Wilcox et al. (1988), greater than in the early period (P < 0.001) (P was probably operating during rainfall and the middle period = 0.024). This Sediment Yield simulations. In fact, our results indicated may be an indication that soil erosion in Many researchers (Meeuwig 1969, that an instantaneous reduction of live this sagebrush steppe is not a source-limit- McCalla et al. 1984 and Linse et al. 2001) vegetation cover did not significantly ed process. As such, removal of 24 tonnes have documented changes in sediment affect runoff or sediment yield. Research ha' may not be drastic enough to affect yield with changes in litter cover, vegeta- by Busby and Gifford (1981) and Wilcox subsequent sediment yield. This may help tion cover, bare ground, and surface et al. (1988) has also shown that vegeta- to explain the absence of significant dif- roughness. In our study, total sediment tion removal does not immediately affect ferences in sediment yield among the soil yield was negatively correlated with forb or runoff. Similarly, Johnson removal treatments. infiltration cover (P = 0.018) and the amount of sand and Gordon (1988) reported that sage- in the upper 5 cm of the soil profile (P = brush canopy removal did not significantly Predicting Runoff and Sediment 0.016), and positively correlated with bare affect soil loss. If rainfall interception, Yield ground (P = 0.055) for a model R2 of 0.60 infiltration, runoff and channel sinuosity Runoff (Table 3). Sediment yield in the first 10 not being affected by vegetation treat- are Three runoff parameters were analyzed: minutes also showed a negative correla- ments, then sediment yield will most like- time to runoff initiation, equilibrium tion with forb cover (P = 0.011) and sand ly not be altered. A more drastic change in runoff ratio in the dry run, and equilibrium in the upper 5 cm of the soil profile (P = not just an aboveground vegetation cover, runoff ratio in the wet run. Independent 0.014), for a model R2 of 0.56. The nega- is likely necessary to influence sedi- kill, variables included total vegetation cover tive correlation between sand in the upper ment yield. (%), forb cover (%), grass cover (%), 5 cm of the soil profile and sediment yield The effect of soil removal on subsequent sagebrush cover (%), litter cover (%), bare may be attributed to the fact that sandier sediment yield is not well documented. ground (%), rainfall intensity (mm/hr), soils experience higher infiltration rates Results of this study did not support the slope (%), surface roughness (SD), sand in which results in a lower volume of over- hypothesis that soil removal of 24 tonnes the upper 5 cm of the soil profile (%), and land flow. This lower overland flow pro- would result in increased sediment ha' soil bulk density (g/cm3). Results of the vides a lower capacity for carrying soil yield. possible explanation is that the One multiple regression analysis indicated that particles and lower energy for dislodging significant increase in surface roughness time to runoff initiation was positively soil particles as it moves down slope. In created by the soil vacuuming reduced correlated with slope (R2 0.39, P addition, sand particles are more difficult down slope. = = sediment movement 0.005, Table 3). Other researchers to transport down slope compared to finer However, Linse et al. (2001) reported a (Flenniken 1999) have reported a correla- grained particles such as silt or clay. weak correlation between surface rough- tion between time to runoff initiation and Bare ground was not strongly correlated yield. The spatial vari- ness and sediment rainfall intensity. In our study, rainfall with any of the sediment yield parameters ability of surface roughness, as opposed to intensity was not correlated with time to analyzed (e.g., partial R2 values were 0.13 an absolute value, may be a more accurate runoff initiation, but did show a positive and 0.14 for total sediment yield and sedi- predictor of sediment yield. Linse (1992) correlation with the dry run equilibrium ment yield in the wet run, respectively). explained how the spatial variability of runoff ratio (R2 0.34, P 0.002). Branson and Owen (1970) also reported a roughness can affect sediment = = surface Equilibrium runoff ratio in the wet run low correlation between bare ground and yield. If surface roughness is greater at the was not correlated with any of the inde- sediment yield (R2 of 0.03 to 0.24). Some it may act to trap sedi- bottom of the plot, pendent variables used in the multiple researchers, however, have reported high- ment from the top of the plot, whereas sur- regression analysis. Variables such as the ly positive correlations between bare roughness concentrated at the up- face spatial distribution of microtopography, ground and sediment yield (Hofmann et al. slope end of the plot will not trap as much channel sinuosity, and average infiltrabili- 1983). The lack of correlation in this study sediment before it reaches the collection ty for each plot may be controlling runoff may have resulted from differences in the trough. Also, if surface roughness is inter- connected along one side of the plot, the depressions may join to form a micro- Table 3. Results of stepwise multiple regression analysis for sediment and runoff variables from channel or rill. This may allow for consid- sagebrush steppe rainfall simulation plots. erable erosion from the plot. Using the point-frame method to measure surface Partial R2 Values for Independent Variables roughness, it is possible to describe the Slope Forb Sand Rainfall spatial variability of surface roughness, Dependent Variables Cover (0-5 cm) Intensity but without quantifying the spatial (mm hf') arrangement of vegetation within the Sediment (kg ha' mm runoff') topography (e.g., degree of channel forma- Total - 0.15 tion). Such an analysis is incomplete. First 10 min - 0.34 0.22 - Sediment yield in runoff water generally First 30 min 0.29 - - Second 30 min 0.11 0.31 0.11 decreases over time during a rainfall event (Ellison 1944, Gutierrez-Castillo 1994). Runoff Contrary to their findings, results from the Time to runoff (min) 0.39 - Equil. Runoff Ratio (dry run) - 0.34 0.34 repeated measures analysis of variance in Equil. Runoff Ratio (wet run) this study revealed an increase in sediment -

156 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 spatial distribution of bare ground from Because rainfall events in this sagebrush Binkley, D. and T.C. Brown. 1993. plot to plot. If bare ground dominates the steppe are typically of short duration, mea- Management impacts on water quality of lower portion of the plot, significant chan- surable parameters such as time to runoff forests and rangelands. USDA For. Ser. Gen. neling may form throughout the rainfall initiation and sediment yield within 10 Tech. Rep. RM-239. Ft. Collins, Colo. Blake, G.R. and K.H. 1986. Bulk simulation, allowing high sediment yields minutes after rainfall initiation have prac- Hartge. density, 363-375. In: A. Klute (ed.) to occur. However, if bare ground is pp. ran- tical value for land managers. Time to Methods of Soil Analysis, Part 1, Physical 2nd domly distributed in small patches across runoff initiation was significantly affected and Mineralogical Methods. ed. No. 9 the plot, then significant sediment yield by the 43% vegetation reduction treat- Agron. Amer. Soc. Agron., Madison, Wis. will be unlikely given the lack of connec- ment, supporting findings of previous Bonham, C.D. 1989. Measurements for tivity between areas of higher erosion research and strengthening the argument Terrestrial Vegetation. John Wiley and Sons, potential and low channel formation. Even for grazing intensities which do not reduce Inc., New York, N.Y. if bare ground is dominating the up-slope vegetation cover beyond this point. Bouyoucos, G.J. 1962. Hydrometer method portions of the plot, the remaining vegeta- However, determining a vegetation reduc- improved for making particle size analysis of soil. Agron. J. 54:464-465. tion at the down-slope end of the plot can tion at threshold would, the very least, Branson, F.A. and J.B. Owen. 1970. Plant be effective at creating pools and allowing require a finer gradation of vegetation cover, runoff, and sediment yield relation- sediment to settle out of the runoff before reduction treatments. ships on Mancos Shale in Western Colorado. water leaves the plot. In addition, the up- Almost half of the variability in sedi- Water Resources Res. 6:783-790. slope end of the plot should experience ment yield is unexplained by the indepen- Brooks, K.N., P.F. Ffolliott, H.M. Gregersen, lower runoff volumes than the down-slope dent variables measured in this study, and L.F. DeBano.1997. Hydrology and the 2nd portions of plots. making a meaningful prediction model Management of Watersheds, ed., Iowa difficult to obtain. Despite some statisti- State Univ. Press, Ames, Iowa. cally significant multiple regression Buckhouse, J.C. and J.L. Gaither. 1982. Variability of Runoff and Sediment Potential sediment production within vegeta- results, we can not conclude Yield Parameters that factors tive communities in Oregon's Blue affecting sediment yield in the first 10 Spatial variability in infiltration rates Mountains. J. Soil and Water Cons. minutes (ie., forb cover and sand in the and erosion have been reported by Ellison 37:120-122. upper soil profile) would be reliable field Busby, F.E. and G.F. Gifford. 1981. Effects (1945), Gard and Van Doren (1949), and indicators of erosion potential in this of livestock grazing on infiltration and ero- Nyhan et al. (1984). The spatial variability ecosystem. While the negative correlation sion rates measured on chained and in soil characteristics, flow paths, depth to between sand in the upper soil profile and unchained pinyon-juniper sites in parent material and infiltration rates over sediment yield was an intuitive result, Southeastern Utah. J. Range Manage. this seemingly uniform range site probably interpreting the negative correlation 34:400-405. masked any real treatment effects on the Ellison, W.D. 1944. Studies of raindrop ero- between forb cover and sediment yield small plots. During rainfall simulation sion. Agr. Eng. April:131-136. was not attempted. Though this correlation events, qualitative observations revealed Ellison, W.D. 1945. Some effects of raindrops was statistically significant, further and surface-flow significant ponding where the borders of on soil erosion and infiltra- research is necessary to confirm that such tion. Amer. Geophy. Union Trans. the plots intersected sagebrush dunes. Plot a correlation is ecologically valid. 26:415-429. borders may create artificial ponds and Soil surface and vegetation parameters Ellison, W.D. 1948. Erosion by raindrop. Sci. interrupt flow paths to a greater degree in are inherently variable in sagebrush steppe Amer. 179:40-45. small narrower plots than in large wider Farres, P. 1978. The role of time and aggre- and other plant communities. Many of plots. Therefore, the edge effect may be gate size in the crusting process. Earth Surf. these parameters, such as connectivity of lower in large plots than in small plots. In Processes 3:243-254. flow paths, are difficult to quantify and addition, larger plots should integrate Flenniken, M. 1999. Flow characteristics and may greatly influence results. Without more spatial heterogeneity of soil and sediment movement in a montane riparian quantifying the connectivity of flow paths, M.S. plant characteristics, which influence ecosystem. Thesis, Colorado State researchers employing rainfall simulators Univ., Fort Collins, Colo. runoff and sediment variability. will continue to be challenged to find Frasier, G.W., M. Weltz, and L. Weltz. treatment differences as influenced by 1998a. Technical Note: Rainfall simulator vegetation cover and soil parameters. runoff hydrograph analysis. J. Range Summary and Conclusions Manage. 51:531-535. Future studies should consider quantifying Frasier, G.W., M.J. Trlica, W.C. Leininger, flow path connectivity, increasing replica- R.A. Pearce, and A. Fernald. 1998b. Rainfall simulations were conducted tions and improving blocking structure to Runoff from simulated rainfall in 2 montane within 1 month after vegetation reduction more accurately determine the factors that riparian communities. J. Range Manage. and soil removal treatments were applied. control runoff and erosion processes in 51:315-322. Results within this short period indicated sagebrush steppe rangelands. Gard, L.E. and C.A. Van Doren. 1949. Soil that soil removal and vegetation reduction losses as affected by cover, rainfall, and treatments did not significantly affect total slope. Soil Sci. Soc. Amer. Proc. runoff, equilibrium runoff ratios, or any of Literature Cited 14:374-378. Gardner, W.H. 1986. Water content, pp. the sediment yield parameters analyzed. 493-544. In. A. Klute (ed.) Methods of Soil However, from these short-term results, Bills, N.L. and R.E. Heimlich. 1984. Analysis, Part 1, Physical and Mineralogical 2nd we do not conclude that vegetation reduc- Assessing erosion on U.S. cropland: land Methods. Ed. No. 9 Agron. Amer. Soc. tion and soil removal will not eventually management and physical features. USDA, Agron. J., Madison, Wis. influence runoff and sediment yield. Econ. Res. Serv., Natur. Resource Econ. Div. Giordanengo, J.H. 2001. Hydrologic and Soil Future research at this site may reveal Agr. Econ. Rep. No. 513. Erosion Responses to Soil Removal and long-term treatment effects that were not Vegetation Reduction. M.S. Thesis, apparent in first year results. Colorado State Univ., Fort Collins, Colo.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 157 Gutierrez-Castillo, J. 1994. Infiltration, sedi- McCormack, D.E. and K.K. Young. 1981. SAS Institute Inc. 1998. SAS Procedures ment, and erosion under grass and shrub Technical and societal implications of soil Guide, Version 6, 3rd ed., SAS Institute Inc., cover in the southern high plains. Ph.D. loss tolerance, pp. 365-376. In: Soil Cary, N.C. Diss., Texas Tech Univ., Lubbock, Tex. Conservation-Problems and Prospects. Sampson, A.W. 1918. Range preservation and Hofmann, L., R.E. Ries, and J.E. Gilley. John Wiley & Sons, New York, N.Y. its relation to erosion control on western 1983. Relationship of runoff and soil loss to Meeuwig, R.O. 1969. Infiltration and soil ero- grazing lands. USDA Bull. 675. USDA, ground cover of native and reclaimed grazing sion as influenced by vegetation and soil in Wash., D.C. land. Agron J. 75:599-602. northern Utah. J. Range Manage. Sharpley, A.N. 1985. Depth of surface soil- Johnson, C.W. and N.D. Gordon. 1988. 23:185-189. runoff interaction as affected by rainfall, soil Runoff and erosion from rainfall simulator Meyers, N. 1984. Gaia: An Atlas of Planet slope, and management. Soil Sci. Soc. Amer. plots on sagebrush rangeland. ASAE Trans. Management. Anchor Press/Doubleday and J. 49:1010-1015. 31:421-427. Company Inc., Garden City, N.Y. Simanton, J.R., M.A. Weltz, and H.D. Kuipers, H.1957. A reliefineter for soil culti- Nyhan, J.W., G.L. DePoorter, B.J. Drennon, Larsen. 1991. Rangeland experiments to vation studies. Neth. J. Agr. Sci. 5:255-262. J.R. Simanton, and G.R. Foster. 1984. parameterize the water erosion prediction Lanier, M. 1999. Facsimile correspondence. Erosion of earth covers used in shallow land project model: vegetation canopy cover United States Dept. of the Interior, Fish and burial at Los Alamos, New Mexico. J. effects. J. Range Manage. 44:276-282. Wildl. Serv., Arapaho Nat. Wildl. Refuge, Environ. Qua1.13:361-366. USDA - Soil Conservation Service (SCS). Walden, Colo. Pearce, R.A., M.J. Trlica, W.C. Leininger, 1981. Soil Survey of Jackson County area, Linse, S.J. 1992. The influence of ground D.E. Mergen, and G.W. Frasier. 1998. Colo., U.S. Gov. Print. Off., Wash., D.C. cover on upland range erosion. M.S. Thesis, Sediment movement through riparian vegeta- Wilcox, B.P., M.K. Wood, and J.M. Univ. Wyoming, Laramie, Wyo. tion under simulated rainfall and overland Tromble. 1988. Factors influencing infiltra- Linse, S.J., D.E. Mergen, J.L. Smith, and flow. J. Range Manage. 51:301-308. bility of semiarid mountain slopes. J. Range. M.J. Trlica. 2001. Upland erosion under a Pimentel, D.C., H.P. Resosudarmo, K. Manage. 41:197-206. simulated most damaging storm. J. Range Sinclair, D. Kurz, M. McNair, S. Crist, L. Wischmeier, W.H. and D.D. Smith. 1978. Manage. 54:356-361. Shpritz, L. Fitton, R. Saffouri, and R. Predicting rainfall erosion losses - a guide McCalla II, G.R., W.H. Blackburn, and L.B. Blair. 1995. Environmental and economic to conservation planning. USDA Agr. Merrill. 1984. Effects of livestock grazing costs of soil erosion and conservation bene- Handb.. No. 537. USDA, Sci. and Ed. on sediment production, Edwards Plateau of fits. Sci. 267:1117-1123. Admin., U.S. Gov. Print. Off., Wash. D.C. Texas. J. Range Manage. 37:291-294.

158 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 J. Range Manage. 56:159-166 March 2003 Woody vegetation response to various burning regimes in South Texas

DONALD C. RUTHVEN III, ANTHONY W. BRADEN, HALEY J. KNUTSON, JAMES F. GALLAGHER, AND DAVID R. SYNATZSKE

Authors are Natural Resource Specialist (DCR, JFG), Research Assistants (AWB, HJK), and Area Manager (DRS), Chaparral Wildlife Management Area, Texas Parks and Wildlife Department, Artesia Wells, Tex. 78001.

Abstract Resumen

Responses of woody plant communities on native rangelands in La respuesta de las comunidades arbustivo-lenosas de los ter- the western South Texas Plains to fire are not clearly understood. renos de pastizal de las planicies del suroeste de Texas al fuego Our objective was to compare woody plant cover, density, and prescrito no ha podido ser entendida claramente. El objettvo de diversity on burned and nontreated rangelands. Five rangeland este estudio fue comparar la cobertura, densidad, y diversidad sites that received 2 dormant-season burns, 5 rangeland sites that de plantas arbustivo-lenosas en areas tratadas con fuego pre- received a combination of 1 dormant-season and 1 growing-sea- scrito y en areas no tratadas. Quince sitios fueron seleccionados son burn, and 5 sites of nontreated rangeland were selected on en el Area de Manejo de Fauna Silvestre Chaparral ubicada en the Chaparral Wildlife Management Area, Dimmit and La Salle los condados de Dimmit y La Salle en Tex.: 5 sitios recibieron Counties, Tex. Woody plant cover was estimated using the line fuego prescrito en 2 ocaciones, ambas durante la estacion de dor- intercept method, and stem density was estimated in 25-x 1.5-m mancia (invierno); otros 5 sitios recibieron fuego prescrito en 2 plots. Species richness did not differ among treatments. Percent ocasiones, una vez durante la estacion de dormancia y otra woody plant cover was reduced by 50 and 41 % on winter and durante la estacion de crecimiento activo (verano); y 5 sitios no winter-summer combination burned sites, respectively. Honey recibieron tratamiento. La cobertura de plantas arbustivo- mesquite (Prosopis glandulosa Torr.), twisted acacia (Acacia lenosas se estimo por medio del metodo de intercepcion de linea, schaffneri S. Wats.), Texas persimmon (Diospyros texana y la densidad de area basal se estimo usando parcelas de 25 x 1.5- Scheele), lotebush [Ziziphus obtusifolia (Hook.) T. & G.], wolfber- m. El parametro de riqueza de especies fue similar entre ry (Lycium berlandieri Dunal), and tasajillo (Opuntia leptocaulis tratamientos. El porcentaje de cobertura de plantas arbustivo- Cand.) canopy cover was greatest on nontreated sites. Woody lenosas se redujo en 50% en los sitios tratados en el invierno y en plant density declined by 29 and 23% on winter and winter-sum- 41 % y en los sitios tratados en invierno y verano. La cobertura mer combination burned sites, respectively. Density of guayacan del dosel de especies como Prosopis glandulosa Torr., Acacia (Guajacum angustifolium Engelm.), wolfberry, and tasajillo was schaffneri S. Wats., Diospyros texana Scheele, Ziziphus obtusifo- less on all burning treatments. Percent cover of spiny hackberry lia (Hook.) T. & G., Lycium berlandieri Dunal, y Opuntia lepto- (Celtis pallida Torr.) and density of Texas pricklypear (Opuntia caulis Cand. fue mayor en sitios no tratados. La densidad de engelmannii Salm-Reif.-Dyck) declined on winter burned sites. plantas arbustivas se redujo en 29% en los sitios tratados en el Inclusion of summer fire into the burning regime did not invierno y en 23% en los sitios tratados en invierno y verano. La increase declines in woody plants. Fire created a post-fire envi- densidad de Guajacum angustifolium Engelm., Lycium ronment which resulted in the decline of many woody plant berlandieri Dunal, y Opuntia leptocaulis Cand. fue menor en species. It is unclear to what degree other environmental factors todos los sitios que recibieron tratamiento. El porcentaje de such as herbivory and competition between woody plants and cobertura de Celtis pallida Torr. y la densidad de Opuntia engel- among woody and herbaceous vegetation may have interacted mannii Salm-Reif.-Dyck se redujo en los sitios tratados durante with fire in producing woody plant declines. Fire may be a useful el invierno. La aplicacion de fuego prescrito durante el verano tool in managing woody vegetation on native south Texas range- no contribuyo a la reduccion del nunero de especies arbustivas. lands, while maintaining woody plant diversity. El ambiente creado por el fuego despues de su aplicacion, resulto en la reduccion de muchas especies arbustivas. El grado de interaccion entre el fuego prescrito y otros factores ambientales, como los herbivoros, la competencia entre plantas arbustivas entre Key Words: fire, diversity, range improvement, wildlife habitat. y plantas arbustivas y vegetacion herbacea, en la reduccion de plantas arbustivas, no es del todo claro. El fuego prescrito The Rio Grande Plains of South Texas is the southern most pudiera ser una herramienta muy titil en el manejo de la veg- extension of the Great Plains Grasslands. Fire, along with other etacion arbustivo-lenosa de los pastizales nativos del sur de climatic variables such as drought presumably maintained the Texas, y en el mantenimiento de su diversidad. mesquite (Prosopis glandulosa Ton) savannas and interspersed grasslands of pre-European settlement South Texas (Scifres and 1993). Frequency of fire appeared to be highly variable Hamilton and ranged from 5-30 years (Wright and Bailey 1982). Following European settlement, suppression of fire combined with heavy Manuscript accepted 14 Jun. 02. livestock grazing lead to the current thorn woodlands common

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 159 throughout South Texas (Archer et al. wildfire can severely damage woody ing, application. All study plots received 1988, Archer 1994), plants (Cable 1965), suggesting that con- 100% coverage by burns. All burned plots Beginning in the mid-twentieth century, trolled growing-season fire may provide were subjected to fire during winter South Texas landowners began to convert an effective means of managing woody (December-March) 1997-1998. Winter thorn woodlands back to grasslands to vegetation. Data on the effects of summer burn plots were again burned in winter enhance rangelands for livestock produc- fire on vegetation and wildlife in South (November-January) 1999-2000. A sum- tion. Mechanical treatments such as root Texas are lacking. mer fire was applied to winter-summer plowing were commonly used to achieve Our objective was to determine the burn plots in August 1999. Relative this goal. However, once treated range- effects of combinations of dormant and humidity and air temperature, using a sling lands are revegetated by woody species, growing-season fire on woody plant diver- psychrometer, and surface wind speed, woody plant diversity can be greatly sity, canopy cover, and density on native using the Bufort Scale, were estimated reduced (Fulbright and Beasom 1987, rangelands in the western South Texas before ignition and at the completion of Ruthven et al. 1993). In South Texas, Plains. We hypothesize that prescribe each fire. Relative humidity was 21 ± 3% woody plants are a primary component of burning these South Texas rangelands will (x ± SE), 20 ± 1%, and 32 ± 0%, tempera- white-tailed deer (Odocoileus virginianus reduce woody cover while maintaining ture was 23 ± 2° C, 22 ± 3° C, and 39 ± 0° Boddaert) diet (Arnold and Drawe 1979, woody plant densities and diversity and C, and wind speed was 13 ± 2 kph, 10 ± 2 Taylor et al. 1997). Woody plants are also that incorporation of summer fire will kph, and 8 ± 0 kph for 1997-1998 winter an important cover source for northern result in increased reductions of woody burns, 1999-2000 winter burn applica- bobwhite (Colinus virginianus L.) plants. tions, and 1999 summer burns, respective- (Guthrey 1986), and woody plants provide ly. Relative humidity and air temperature escape cover and thermal refugia to during winter fires followed recommended species of concern such as the Texas tor- Materials and Methods prescriptions for the vegetation type pre- toise (Gopherus berlandieri Stejneger) sent (Wright and Bailey 1982), while wind homed lizard speed was slightly below recommended (Kazamier 2000) and Texas The study area was on the Chaparral (Phrynosoma cornutum Gray) (Burrow et parameters. Little prescription data is Wildlife Management Area (28° 20' N, al. 2001). available for summer burning in South 99° 25' W) in the western South Texas Land ownership and land use practices Texas; however, 35° C is generally con- Plains. The study area was purchased by in South Texas have changed in recent sidered the upper limit for most prescribed the state of Texas in 1969, and is managed years. The size of individual landholdings burning (Wright and Bailey 1982). In by the Wildlife Division of the Texas and revenue derived from South Texas, adequate humidity levels for has decreased Parks and Wildlife Department (TPWD). ranch properties has become increasingly burning are generally not reached until air Climate is characterized by hot summers dependent on wildlife rather than tradi- temperatures are in excess of the recom- and mild winters with an average daily tional livestock operations (Wilkins et al, mended maximum. Wind direction was minimum winter (January) temperature of 2000). Fall and winter burning along the variable. Soil moisture was not recorded. 5° C, an average daily maximum summer transition zone between the South Texas No significant (> 0.4-cm) precipitation (July) temperature of 37° C, and a grow- Plains and Gulf Prairies and Marshes can events were recorded within a 30-day peri- ing season of 249 to 365 days depending reduce brush cover while maintaining od prior to winter burns and we considered on freezing temperatures (Stevens and woody plant diversity (Box and White soil moisture low. Summer burns were Arriaga 1985). Precipitation pattern is conducted 3 to 5 days following a 23-cm 1969) and increase herbaceous vegetation bimodal with peaks occurring in late con- preferred by wildlife (Box and White spring (May-June) and early fall rainfall event and soil moisture was 1969, Hansmire et al. 1988). Fire can also sidered high. Summer burns were con- (September-October). The 11-year (1989- provide an economical means of control- ducted under moist soil conditions because 2000) average precipitation amount is 54 ling woody species and maintaining the summer burning with little soil moisture cm (TPWD, unpubl. data). life of mechanical and chemical brush can result in significant mortality of peren- Treatments consisted of 2 winter pre- management treatments (Scifres and nial grasses (Scifres and Duncan 1982). scribed burns (winter burn), a winter burn Hamilton 1993). Burning during late win- Because of variable wind speed and direc- followed by a summer fire (winter-sum- ter and early spring is recommended for tion during all burns and uneven fuel mer burn), and nontreated rangeland sites. achieving many goals on South Texas loads, fire behavior was highly variable Each treatment was replicated 5 times and rangelands (Scifres and Hamilton 1993). and not recorded. Fuel loads appeared to arranged in a randomized block design. As a result of reported benefits, South vary within study plots. Previous work on All study plots were about 2-ha in size. Texas rangeland managers are beginning the study site in which fuel loads were Study plots were interspersed among 4 to utilize fire to enhance wildlife habitat; estimated by clipping above ground bio- pastures in the central portion of the study yet, little information is available on the mass indicate that adequate fuel loads for area and each plot separated by > 500-m. response of vegetation to fire in the more burning in western portions of South Pretreatment sampling was not conducted; xeric areas of the western Rio Grande Texas are >- 2,000 kg/ha (TPWD, unpubl. however, total woody plant canopy cover our visual estimations, Plains. Concerns have arisen that dormant- and composition of dominant woody data). Based on season fire may adversely affect species of fuel loads on study plots met adequate lev- species was considered similar among concern such as the Texas horned lizard, els. All burns were ignited as head fires study plots prior to application of burning which hibernate at shallow depths in with drip torches. treatments (Gabor 1997). Burned plots South Texas (Fair and Henke 1997) and Soils were similar among treatments and were located within larger (> 50-ha) areas consisted of Duval fine sandy loam, gently that burning during summer when species that were burned. Study plots and sur- such as the Texas horned lizard are active undulating, Duval loamy fine sand, 0-5% rounding rangeland subjected to fire were may direct mortality. Summer slopes, and Dilley fine sandy loam, gently reduce burned simultaneously during each burn-

160 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 undulating (Stevens and Arriaga 1985, mated in 25-x 1.5-m plots and used to esti- HSD was used to compare individual treat- Gabriel et al. 1994). The Duval series are mate species richness, diversity, and even- ments where univariate tests indicated sig- fine-loamy, mixed, hyperthermic Aridic ness. Woody species diversity (H') and nificant (P < 0.05) treatment effects on indi- Haplustalfs and the Dilley series are evenness (J') was quantified with vidual species. Total woody plant canopy loamy, mixed, hyperthermic shallow Shannon's Index (Pielou 1975). Because cover and total stem density, as well as Ustalfic Haplargids. Topography is nearly of their woody growth form in South species richness and Shannon's index of level to gently sloping and elevation Texas, succulents such as pricklypear species diversity and evenness were calcu- ranges between 168 and 180-m. (Opuntia spp. Miller) and suffrutescents lated at the treatment level and analyzed by Vegetation is characterized by a two- including lantana (Lantana spp. L.) were a 1-way analysis of variance (ANOVA) phase pattern of shrub clusters scattered considered woody plants in this study. with treatment as the main effect. Tukey's throughout a grassland/savanna (Whittaker Scientific names of Texas plants follow HSD was used to compare treatment means. et al. 1979, Archer et al. 1988). Plant com- Jones et al. (1997) and common names are Treatment differences were considered sig- munities belonged to the Honey mesquite- from Hatch et al. (1990). nificant at the P < 0.05 level. Spiny hackberry (Celtis pallida Torr.) Percent canopy cover, density, and fre- association (McLendon 1991). Within this quency of occurrence of woody species association were 2 primary communities, were evaluated at the transect level using Results the Honey mesquite-Lime pricklyash multivariate analysis of variance (MANO- [Zanthoxylum fagara (L.) Sarg.]/Spiny VA), with treatment as the main effect. Shannon's index of species diversity hackberry community, in which lime Two-treatment comparisons were not con- varied (P 0.0267) among treatments. pricklyash and brasil (Condalia hookeri M. ducted due to limited sample size. Tukey's = C. Johnst.) are the subdominants, and the Honey mesquite-Spiny hackberry/Hog- Table 1. Mean woody plant canopy cover (%) on winter burned', winter-summer burned2, and plum (Colubrina texana T. & G.) commu- nonburned sites3 on the Chaparral Wildlife Management Area, Dimmit and La Salle Counties, nity, in which hog-plum is the subdomi- Tex. 2001. nant. Prominent herbaceous species included Lehmann lovegrass (Eragrostis Species Winter burn Winter-summer burn Nonburned lehmanniana Nees), an introduced peren------(%)------nial, hooded windmillgrass (Chloris cucul- Guajillo 0.2a4 Oa lata Bisch.), hairy grama (Bouteloua hir- Catclaw Acacia 1.Oa O.la suta Lag.), partridge pea [Chamaecrista Huisache 0.05). rence of individual species was also esti- n = 5 sites/treatment.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 161 Table 2. Mean woody plant density (stems/ha) on winter burned', winter-summer burned2, and Density of narrowleaf forestiera was great- nonburned sites3 on the Chaparral Wildlife Management Area, Dimmit and La Salle Counties, est on nonburned plots, which did not dif- Tex. 2001. fer from winter burned plots. Density of whitebrush [Aloysia gratissima (Gill. & Species Winter burn Winter-summer burn Hook.) Tron.] was greatest on winter burned and nonburned plots. Frequency of Guajillo 2a4 Oa occurrence values varied between treat- Catclaw Acacia 121a 25a ments (MANOVA P = 0.0080) and fol- Huisache 2a Oa lowed similar trends as density estimates Blackbrush 16a Oa with the exception of lime pricklyash, Twisted Acacia 178a 94a Texas pricklypear, whitebrush, and spiny Whitebrush 14a Ob hackberry (Table 3). Lime pricklyash was Goatbush Oa Oa Sugar Hackberry Oa Oa most commonly encountered on winter Spiny Hackberry 116a 14Oab burned plots; Texas pricklypear had great- Hog-plum 731a 665a est frequency of occurrence on winter- Brasil 128a 87a summer burned and nonburned plots; Texas Persimmon 67a 113a whitebrush was most common on non- 5a Vine Ephedra Oa which did not differ from Narrowleaf Forestiera 16ab Oa burned plots, Guayacan 7a 7a winter burned plots; and spiny hackberry Tatalencho Oa Oa did not differ among treatments. Leatherstem 5a 23a Coyotillo 52a 14a Veinyleaf Lantana 4a Oa Discussion Common Lantana 485a 745a Wolfberry 27a 9a Texas Pricklypear 683a 1,00lab Our results indicate that combinations of Tasajillo 23a 30a dormant and growing-season burning of 242a 238a Honey Mesquite can Little-leaf Sumac 4a Oa South Texas rangelands effectively Desert Yaupon 5a 12a reduce woody plant cover and decrease Coma 44a 7a certain woody plant densities. It should be Lime Pricklyash 34a 2a noted that rangeland fire in southern Texas Lotebush 12a 20a typically produces a mosaic of burned and 3,237a Total woody plant densitys 3,018a nonburned areas as a result of uneven fuel Sites burned during winter 1997-1998 and winter 1999-2000. loads (Box and White 1969), and the 3Sites burned during winter 1997-1998 and summer 1999. 100% coverage by fire on our study sites n =15 transects/site/treatment. 4Cover values for a species across treatments followed by the same letter are not significantly different (P > 0.05). may not be typical of large-scale applica- n = 5 sites/treatment. tion of prescribed fire in much of South Texas. As edaphic characteristics can influence fire effects (Hansmire et al. Diversity did not differ between winter tocaulis Cand.) canopy cover was greatest 1988), care should be exercised when (2.33 ± 0.13) (x ± SE) and winter-summer on nonburned plots. Spiny hackberry extrapolating these results to other soil (2.13 ± 0.09) burning treatments whereas cover was greatest on nonburned and win- types in South Texas. Prescribed burning diversity on nonburned sites (2.55 ± 0.04) ter-summer burn plots. Guayacan studies conducted on clay soils in southern was greater than winter-summer burned (Guajacum angustifolium Engelm.) cover Texas (Box et al. 1967, Box and White treatments. Species richness and evenness was greatest on nonburned plots, which 1969) are the most in-depth studies upon did not differ among winter burned (16 ± did not differ from winter-summer burned which regional comparisons can be made. 2 species/treatment, 0.85 ± 0.02, respec- plots. Canopy cover of narrowleaf Our observed declines in many woody tively), winter-summer burned (13 ± 2, forestiera (Forestiera angustifolia Torr.) species following fire were similar to those 0.85 ± 0.01), and nonburned (18 ± 1, 0.88 was greatest on nonburned plots, which reported by Box et al. (1967) and Box and ± 0) treatments. Percent woody plant did not differ from winter burned plots. White (1969); however, Box et al. (1967) canopy cover differed between treatments Lime pricklyash canopy cover was highest also reported decreases in brasil following (MANOVA P <_ 0.0001). Percent total on winter burned plots. Density of woody September burns, which did not concur woody plant canopy cover was greatest (P species differed between treatments with the results of this study. The lack of a = 0.0002) on nonburned plots (Table 1). (MANOVA P = 0.0004). Total woody reduction of some woody species may be a Total woody plant cover did not differ plant densities varied (P = 0.0077) by result of less fuel load and variation in soil among winter and winter-summer burned treatment with nonburned having greater type or differences in grazing practices. plots. Honey mesquite, wolfberry (Lycium densities than burned plots (Table 2). Fuel loads on rangelands along the transi- berlandieri Dunal), twisted acacia (Acacia Density of wolfberry, lotebush, desert tion zone between the South Texas Plains schaffneri S. Wats.), Texas persimmon yaupon, guayacan, and tasajillo was high- and Gulf Prairies and Marshes can exceed (Diospyros texana Scheele), lotebush est on nonburned plots. Spiny hackberry 5, 476 kg/ha (Hansmire et al. 1988), which [Ziziphus obtusifolia (Hook.) T. & G.], and Texas pricklypear (Opuntia engel- can produce more intense fires (Stinson desert yaupon (Schaefferia cuneifolia mannii Salm-Reif.-Dyck) density was and Wright 1969, Britton and Wright Gray), coma [Sideroxylon celastrinum highest on nonburned plots, which did not 1971) and possibly higher top-kill and (Kunth) Penn.], and tasajillo (Opuntia lep- differ from winter-summer burned plots. mortality of woody species.

162 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 3. Frequency (%) on winter burnedl, winter-summer burned2, and nonburned sites3 on the maintenance of some form of apical domi- Chaparral Wildlife Management Area, Dimmit and La Salle Counties, Tex. 2001. nance. The lack of top-kill may have reduced basal sprouting leading to the Species Winter burn Winter-summer burn Nonburned observed maintenance of canopy cover ------(%)------reductions 2 years following the last bum Guajillo 1a4 Oa application. Catclaw Acacia 12a 4a Cacti (Opuntia spp. Miller) can also be a Huisache la Oa concern of livestock and wildlife man- Blackbrush Oa 4a agers. While cacti are an important food Twisted Acacia 36a 24a Whitebrush 4ab Oa source for species such as white-tailed Goatbush Oa Oa deer (Arnold and Drawe 1979) and javeli- Sugar Hackberry Oa Oa na (Tayassu tajacu L.) (Sowls 1997), large Spiny Hackberry 24a 29a colonies of Texas pricklypear can result in Hog-plum 67a 57a extensive areas virtually devoid of herba- Brasil 29a 19a ceous vegetation. The high susceptibility Texas Persimmon 17a 27a Vine Ephedra Oa la of tasajillo to fire was similar to other Narrowleaf Forestiera 3ab Oa studies (Box et al. 1967, Box and White Guayacan la la 1969, Bunting et al. 1980). Brownspine Tatalencho la Oa pricklypear (Opuntia phaeacantha Leatherstem la 2a Engelm.) suffers little direct mortality fol- Coyotillo 9a 3a lowing burning (Heirman and Wright Veinyleaf Lantana la Oa Common Lantana 55a 62a 1973, Bunting et al. 1980); however, tis- Wolfberry 5a 3a sue destruction caused by burning increas- Texas Pricklypear 76a 91b es vulnerability to damage from insects, Tasajillo 7a 7a rodents, and lagomorphs which can result Honey Mesquite 51a 51a in high mortality 2-3 years following Little-leaf Sumac la Oa burns. Moderate fuel loads combined with Desert Yaupon 2a 3a Coma 7a 2a the robust growth form of Texas prickly- a lb pear in South Texas may explain s gen- Lotebush 5a 6a eral resilience to fire and subsequent insect Sites burned during winter 1997-1998 and winter 1999-2000. damage. Reductions of Texas prick- 3Sites burned during winter 1997-1998 and summer 1999. lypear densities on winter burned plots n =15 transectslsiteltreatment. may in part be a result of herbivory by cat- 4Frequency values for a species across treatments followed by the same letter are not significantly different (P > 0.05). tle. Texas pricklypear can be an important portion of cattle diet during winter (Everitt Reductions of diversity on winter-sum- be significantly reduced following fire et al. 1981). Cattle were allowed access to mer burned plots compared to nonburned (Box and White 1969, Wright et al. 1976, pastures in which burned plots were locat- plots may not accurately reflect actual Cable 1965). Mesquite cover declined on ed immediately following winter burns. responses to fire. Shannon's index of both burning treatments but fire did not Cattle readily browsed freshly burned species diversity was considerably lower appear to induce significant mortality. As Texas pricklypear and observations were on 1 winter-summer burned plot compared with other woody species, moderate and made in which the entire above ground to the other 4 replications. This may uneven fuel loads coupled with varying biomass of some plants was consumed. reflect pretreatment differences among fire behavior may have produced less Winter burns were effective in reducing plots prior to burning. Previous vegetation intense fires resulting in low mortality. canopy cover of spiny hackberry, which mapping (Labor 1997) suggested that all Our results were similar to Heirman and was similar to responses observed in the study plots were similar in regards to dom- Wright (1973) in which living mesquites transition zone between the South Texas inant woody species and total woody plant were not killed following prescribed burns Plains and Gulf Prairies and Marshes (Box canopy cover before application of burn- in west Texas with fuel loads similar to et al. 1967). Spiny hackberry appears to ing treatments; however, pretreatment dif- our study. Wood boring insect infestations survive most fire damage but recovery ferences in species composition, especially may increase in trees damaged by fire rates of top-growth are slower than many less common (frequency of occurrence < facilitating burndown and mortality as a other woody species. Slow recovery rates 15%) species may have existed. With the result of secondary burns (Ueckert and may be compounded through herbivory by exception of lime pricklyash, declines in Wright 1974). It is unclear to what degree white-tailed deer and other herbivores. woody plants were limited to burned plots wood boring insects may have infested Top-removal of woody species can suggesting that declines were a result of honey mesquite following initial burns; increase nutrient content of regrowth the burning treatments. Sample sizes may however, little honey mesquite burndown (Everitt 1983), which may increase utiliza- also have been inadequate to fully assess was observed during secondary fire appli- tion by herbivores (Rasmussen et al. treatment effects on uncommon species cations. Crown sprouting by mesquite fol- 1983). Spiny hackberry and other woody such as lime pricklyash. lowing fire is greatest in trees that exceed plants such as guayacan, which declined Management of mesquite (Prosopis spp. 5-cm in stem diameter (Cable 1965). following burns, are considered highly L.) is a primary concern of range man- Although not quantified, stem diameter of preferred browse of white-tailed deer agers throughout the southwestern United most honey mesquite on our study sites (Taylor et al. 1997). Deer density on the States and northern Mexico. Mesquite can appear to exceed 5-cm resulting in the study area during the duration of the study

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 163 Appendix Table 1. Scientific and common names of woody species. periods of high soil moisture and may have resulted in cooler burning conditions Scientific names Common Name2 and less cambium destruction. Acacia berlandieri Benth. Guajillo Three common (frequency >_ 25%) Acacia greggii Gray Catclaw Acacia woody plants that appeared highly adapted Acacia minuata M. Jones Huisache to fire are twisted acacia, hog-plum, and Acacia rigidula Benth. Blackbrush common lantana (Lantana urticoides Acacia schaffneri S. Wats. Twisted Acacia Hayek). These species are generally low Whitebrush Aloysia gratissima (Gill. & Hook.) Tron. growing shrubs and because of litter accu- Castela erecta (Turpin) T. & G. Goatbush Celtis laevigata Willd. Sugar Hackbeny mulations beneath plants, especially hog- Celtis pallida Torr. Spiny Hackberry plum and common Lantana, are easily top- Colubrina texensis (T. & G.) Gray Hog-plum killed following fire. Observations were Condalia hookeri M. C. Johnst. Brasil made in which top growth of many hog- Diospyros texana Scheele Texas Persimmon plum and common lantana plants were Vine Ephedra Ephedra antisyphilitica (Ben.) Meyer entirely consumed by fires. Twisted acacia Forestiera angustifolia Ton. Narrowleaf Forestiera Guajacum angustifolium Engelm. Guayacan and hog-plum are most abundant in the Gymnosperma glutinosum (Spreng.) Less. Tatalencho interspace between honey mesquite-shrub Jatropha diocia Cerv. Leatherstem clusters, whereas lantana (Lantana spp. Karwinskia humboltiana (Schult.) Zucc. Coyotillo L.) is associated with honey mesquite Lantana achyranthifolia Desf. Veinyleaf Lantana clusters (Whittaker et al. 1979, Archer et Hayek Common Lantana Lantana urticoides al. 1988). Lack of competition between Lycium berlandieri Dunal Wolfberry Opuntia engelmannii Salm-Reif.-Dyck Texas Pricklypear twisted acacia and hog-plum and other Opuntia leptocaulis Cand. Tasajillo woody plants may explain their ability for Prosopis glandulosa Ton. Honey Mesquite rapid recovery following fire, while com- Rhus microphylla (Engelm.) Gray Little-leaf Sumac mon lantana appears to more effectively Schaefferia cuneifolia Gray Desert Yaupon compete with other woody plants within Coma Sideroxylon celastrinum (Kunth) Penn. clusters. The response of twisted acacia to Zanthoxylum fagara (L.) Sarg. Lime Pricklyash Ziziphus obtusifolia (Hook.) T. & G. Lotebush fire was comparable to that of huisache (Acacia minuata M. Jones) (Rasmussen et zJones et al. (1997) al. 1983), which tends to be replaced by Hatch et al. (1990) twisted acacia along the decreasing east to west precipitation gradient across the averaged one adult deer 18 ha' year', desert yaupon are closely associated with South Texas Plains. which we consider below carrying capaci- honey mesquite clusters (Whittaker et al. ty for these rangeland sites. It is unclear to 1979, Archer 1989). Mesquite clusters what degree herbivores such as white- result in concentrations of woody plants Conclusions and Management tailed deer, lagomorphs, rodents, and relative to the interspace between clusters. Implications arthropods may have impacted regrowth Brasil, which is also a common compo- of woody plants following fires. nent of shrub clusters and demonstrated treatments were effective Woody plants comprise a very small little decline following fires, appears to be Both burning portion of cattle diet in South Texas more efficient in competing with other at reducing honey mesquite cover, which the observed increases in (Everitt et al. 1981) and direct impacts of woody plants within clusters. Fire increas- may promote following fire on cattle grazing appear to have little impact es productivity of herbaceous plants, espe- herbaceous vegetation on woody plants following fire. High cially forbs on South Texas rangelands South Texas rangelands (Hansmire et al. 1993). intensity, low frequency grazing results in (Hansmire et al. 1988, Ruthven et al. 1988, Scifres and Hamilton significant greater consumption of less-preferred for- 2000), and increased competition between Summer burning following in managing age species (Drawe 1988), which can lead woody and herbaceous plants may also rainfall may be effective mesquite, while maintaining desir- to the more uniform grazing of herbaceous account for reductions of woody plants on honey such as spiny hackber- plants. The high intensity, low frequency burned plots. able woody species cover is a grazing system employed during the dor- Summer fire applied during periods of ry. If reducing total woody plant during winter mant-season on the study area appears to above normal soil moisture can result in management goal, burning little or no rainfall is reduce selective grazing and minimize any less damage and subsequent decline of following periods of effects selective grazing pressure may woody plants compared to dormant-season recommended. Further investigation into to summer have on competition between woody and fire applied under low soil moisture condi- the response of woody plants herbaceous plants following fire. tions (Adams et al. 1982). In part, reduc- fire under various environmental condi- low soil moisture is Deferment of grazing on the study area tions of spiny hackberry and Texas prick- tions, in particular, to evaluate the effects of during the majority of the growing-season lypear on winter burn treatments may be necessary fully may however, increase competition explained by soil moisture conditions at summer fire on woody plants. Where fea- can manage between herbaceous and woody plants. the time of burning. All winter burns were sible, prescribed burning without dramatic reduc- Competition among woody plants may conducted under relatively dry conditions, woody vegetation in plant diversity, which is also explain reductions in spiny hackberry which may have resulted in high tempera- tion woody traditional mechanical and other woody plants. Declining species tures at and just below the soil surface common with many (Fulbright and Beasom 1987, such as spiny hackberry, lotebush, Texas causing greater internal tissue damage. treatments pricklypear, tasajillo, wolfberry, and Summer burns were performed during Ruthven et al. 1993).

164 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 1987. If managing South Texas rangelands for Literature Cited Fulbright, T.E. and S.L. Beasom. effects of mechanical treatments deer is a desired goal, it may Long-term white-tailed white-tailed deer browse. Wildl. Soc. the use of fire in on be beneficial to limit Adams, D.E, R.C. Anderson, and S.L. Bull. 15:560-564. areas dominated by highly preferred Collins. 1982. Differential response of Gabor, T. 1997. Ecology and Interactions of species which decline following fire and woody and herbaceous species to summer sympatric collared peccaries and feral pigs. target areas dominated by vulnerable less and winter burning in an Oklahoma grass- Ph.D. Diss., Oklahoma State Univ., desired species and desirable fire-tolerant land. Southwest. Nat. 27:55-61. Stillwater, Okla. species. Woody plant species such as Archer, S. 1989. Have southern Texas savan- Gabriel, W.J., D Arriaga, and J.W. Stevens. twisted acacia and common lantana, which nas been converted to woodlands in recent 1994. Soil survey of LaSalle County, Texas. recover quickly following fire, are not history? Amer. Nat. 134:545-561. USDA. Wahington, D.C. S. 1994. Woody plant encroachment bob- generally considered important forage Archer, Guthery, F.S. 1986. Beef, brush and into southwestern grasslands and savannas: whites: quail management in cattle country. for white-tailed deer (Arnold and species rates, patterns and proximate causes, p Golden Banner Press, Inc., Corpus Christi, whereas Drawe 1979, Taylor et al. 1997), 13-68. In: M. Vavra et al. (eds.) Ecological Tex. hog-plum can be a preferred browse plant implications of livestock herbivory in the Hansmire, J.A., D.L. Drawe, D.B. Wester, (Arnold and Drawe 1979, Ruthven et al. west. Society for Range Management, and C.M. Britton. 1988. Effects of winter 1994). Although less desirable browse Denver, Colo. burns on forbs and grasses of the Texas species such as honey mesquite, wolfber- Archer, S., C. Scifres, C.R. Bassham, and R. Coastal Prairie. Southwest. Nat. 33:333-338. succession in a ry, and desert yaupon decline following Maggio. 1988. Autogenic Hatch, S.L., K.N. Gandhi, and L.E. Brown. fire, the importance of these species to subtropical savanna: conversion of grassland 1990. Check list of the vascular plants of thorn woodland. Ecol. Monogr. Sta. Misc. Pub. MP- other game and nongame wildlife is not to Texas. Tex. Agr. Exp. 58:111-127. 1655. well documented. The effects of the struc- Arnold, L.A. and D.L. Drawe.1979. Seasonal Heirman, A.L. and H.A. Wright. 1973. Fire communities tural changes of woody plant food habits of white-tailed deer in the south in medium fuels of West Texas. J. Range following fire on nongame wildlife war- Texas plains. J. Range Manage. 32:175-178. Manage. 26:331-335. rant further investigation. Box, T.W. and R.S. White 1969. Fall and Jones, S.D., J. K. Wipff, and P. M. We recommend that initial burns on winter burning of south Texas brush ranges. Montgomery. 1997. Vascular plants of native South Texas rangelands be conduct- J. Range Manage. 22:373-376. Texas. Univ. Texas Press, Austin, Tex. ed on a 2-year interval until the desired Box, T.W., J. Powell, and D.L. Drawe.1967. Kazmaier, R T. 2000. The effects of grazing structure of woody vegetation is achieved. Influence of fire on south Texas chaparral on ecology and demography of the Texas tor- Ecol. 48:955-960. the western Once desired goals are met, maintenance communities. toise (Gopherus berlandieri) in Britton, C.M. and H.A. Wright. 1971. Rio Grande Plains. Ph.D. Diss., Oklahoma on a 3-5 year frequency may be burning Correlation of weather and fuel variables to Stillwater, Okla. that allow for State Univ., adequate. Grazing strategies mesquite damage by fire. J. Range Manage. Lehmann, V.W. 1969. Forgotten legions: substantial deferment to produce adequate 24:136-141. sheep in the Rio Grande Plain of Texas. fuels to carry fire are critical to the suc- Bunting, S.C., H.A. Wright, and L.F. Texas Western Press, El Paso, Tex. cessful application of fire on South Texas Neuenschwander. 1980. Long-term effects McLendon, T. 1991. Preliminary description rangelands. High intensity, low frequency of fire on cactus in the southern mixed prairie of the vegetation of south Texas exclusive of grazing systems, such as that employed on of Texas. J. Range Manage. 33:85-88. coastal saline zones. Tex. J. Sci. 43:13-32. the study area, appear to be the most com- Burrow, A.L, R.T. Kazamier, E.C. Hellgren, Norwine, J. and R. Bingham. 1985. III. 2001. Microhabitat in south patible for the incorporation of fire into a and D.C. Ruthven, Frequency and severity of drought program (Drawe 1988). selection by Texas horned lizards in southern Texas, p. 1-19 In: R. Brown (ed.) Livestock management J. Wildl. Manage. 65:645-652. during drought. to be taken into Texas. and wildlife management Other considerations Cable, D.R. 1965. Damage to mesquite, Caesar Kleberg Wildlife Research Institute, account are the highly unpredictable Lehmann lovegrass, and black grama by a Kingsville, Tex. weather patterns in South Texas. Short- hot June fire. J. Range Manage. 18:326-329. Pielou, E.C. 1975. Ecological diversity. John term periods of drought are common and Canfield, R.H. 1941. Application of line inter- Wiley & Sons, New York, N.Y. rainfall can be highly variable between ception in sampling range vegetation. J. Rasmussen, G.A., C.J. Scifres, and D.L. locations (Norwine and Bingham 1985). Forest. 39:388-394. Drawe. 1983. Huisache growth, browse Drought can severely impact production of Drawe, L. D. 1988. Effects of three grazing quality, and use following burning. J. Range production, fine fuels necessary to carry fire and may treatments on vegetation, cattle Manage. 36:337-342. the Welder Wildlife in burning schedules. and wildlife on Ruthven, D.C., III, J.F. Gallagher, and D.R. require flexibility 1974-1982. Welder woody Foundation Refuge, Synatzske. 2000. Effect of fire and grazing Honey mesquite and associated Contribution B-8, Sinton, plains. throughout Wildl. Foundation on forbs in the western south Texas plants dominate rangelands Tex. Southwest. Nat. 445:89-94. South Texas; yet, productivity and species Everitt, J.H. 1983. Effects of plant shredding Ruthven, D.C., III, E.C. Hellgren, and S.L. composition of herbaceous plant commu- on nutrient content of four south Texas deer Beasom. 1994. Effects of root plowing on nities can vary greatly along the decreas- browse species. J. Range Manage. 36:779-781. white-tailed deer condition, population sta- ing east to west precipitation gradient Everitt, J.H., C.L. Gonzales, G. Scott, and tus, and diet. J. Wildl. Manage. 58:59-70. across the region. Decreases in herbaceous B.E. Dahl. 1981. Seasonal food preferences Ruthven, D.C.III, T.E. Fulbright, S.L. vegetation may increase the utilization of of cattle on native range in the south Texas Beasom, and E.C. Hellgren. 1993. Long- woody plants by herbivores. Further plains. J. Range Manage. 34:384-388. term effects of root plowing on vegetation in W.S. and S.E. Henke. 1997. Effects of plains. J. Range the interaction between burn- Fair, the eastern south Texas research into habitat manipulations on Texas horned to fully assess Manage. 46:351-354. ing and herbivory is needed lizards and their prey. J. Wildl. Manage. Scifres, C.J. and K.W. Duncan. 1982. as a manage- the effects of fire and its use 61:1366-1370. Brownseed paspalum response to season of ment tool on South Texas rangelands. burning. J. Range Manage. 35:251-253.

165 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Scifres, C.J. and W.T. Hamilton. 1993. Taylor, R.B., J. Rutledge, and J.G. Herrera. Wilkins, N., R.D. Brown, R.J. Conner, J. Prescribed burning for brushland manage- 1997. A field guide to common south Texas Engle, C. Gilliland, A. Hays, R.D. Slack , ment: the south Texas example. Texas A&M shrubs. Texas Parks and Wildlife Press, and D.W. Steinbach. 2000. Fragmented University Press, College Station, Tex. Austin, Tex. lands: changing land ownership in Texas . Sowls, L.K. 1997. Javelinas and other pecca- Ueckert, D.N. and H.A. Wright. 1974. Wood Texas A&M Univ , Agr . Commun . Pub . ries: their biology, management , and use . boring insect infestations in relation to , College Station , Tex . Texas A&M University Press , College Station, Tex. mesquite control practices. J. Range Manage. H.A. and A.W. Bailey. 1982. Fire 27:383-386. Stevens, J.W. and D. Arriaga.1985. Soil sur- Ecology. John Wiley and Sons, New York, vey of Dimmit and Zavala Counties, Texas. Whittaker, R.H., L.E. Gilbert, and J.H. N.Y.. USDA. Wahington, D.C. Connell. 1979. Analysis of two-phase pat- Wright, H.A., S.C. Bunting, and L.F. Stinson, K.J. and H.A. Wright. 1969. tern in a mesquite grassland, Texas. J. Ecol. Neuenschwander. 1976. Effects of fire on Temperature of headfires on the southern 67:935-952. honey mesquite. J. Range Manage. mixed prairie of Texas. J. Range Manage. 29:467-471 22:169-174.

166 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 J. Range Manage. 56:167-173 March 2003 Stocking rate effects on goats: A research observation

MIGUEL MELLADO, RAUL VALDEZ, LAURA M. LARA, AND RAMIRO LOPEZ

Authors are Professor of Reproductive physiology, Graduate Student, Research Biologist, and Professor of Ruminant nutrition, Department of Nutrition and Feeds, Universidad Autonoma Agraria Antonio Narro, Saltillo, Mexico.

Abstract Resumen

El impacto ecologico del pastoreo de las cabras en los Knowledge on the ecological effects of goat grazing on arid agostaderos aridos continua siendo un tema de debate, y rangeland is far from complete, and specifically there is little sci- todavia no existe suficiente informacion cientifica sobre el entific information on effects of heavy goat grazing on arid impacto del pastoreo de las cabras sobre ecosistemas aridos. ecosystems. One objective of this study was to determine botani- Mediante la tecnica microhistologica se estudio la composi- cal composition of dairy-type goat diets on heavily (1.5 ha per cion botanica de la dieta de cabras mestizas lecheras man- goat) and lightly (15 ha per goat) grazed Chihuahuan desert tenidas en agostaderos con alta (1.5 ha por cabra) baja (15 by fecal microhistological analysis. A second objective was y range ha por cabra) presion de pastoreo, en un matorral parvifolio to determine whether vegetation cover, some blood metabolites inerme. Tambien se determine el efecto de la carga animal and mineral levels, as well as fertility of goats were sensitive to sobre la cobertura vegetal, niveles de metabolitos y min- high grazing pressure. The lightly grazed site had more (P < erales en la sangre la fertilidad de las cabras. El sitio con 0.05) total foliage cover (38.6 vs 30.4%) than the overstocked y baja densidad de cabras presento mayor (P < 0.05) cobertu- pasture. Total shrubs in diets of goats was greater (86.4 vs 72.4 in ra de vegetacion (38.6 vs 30.4%). El porcentaje de arbustos the late-dry period, 78.6 vs 42.1 in late-wet period; P < 0.05) on fue mayor (86.4 vs 72.4 al final del periodo de sequia, 78.6 vs the heavily stocked pasture than the lightly stocked pasture. 42.1 al final del periodo lluvioso; P < 0.05) en la dieta de las Forbs in the diets were lower (P < 0.10) in the late-dry (11.4 vs en el terreno con sobrepastoreo, comparado con el early-wet (55.4 vs 64.0%) and late-wet period (15.0 vs cabras 21.5%), terreno de baja densidad de cabras. El contenido de her- 45.8%) on the heavily stocked pasture than the lightly stocked baceas fue menor (P < 0.10) en la dieta de las cabras en el Substantially lower (P < 0.01) serum glucose, urea nitro- pasture. con sobrepastoreo en comparacion con el sito con Mg concentration at the onset of the breeding period terreno gen, Zn and baja densidad de cabras al final del periodo de sequia. in goats on the heavily stocked pasture, compared to goats on the Niveles mas bajos (P < 0.01) de glucosa, urea, Zn y Mg en el lightly grazed pasture resulted in a higher (P < 0.01) abortion suero sanguineo en las cabras en el terreno con alta presion vs 12%) and consequently a lower (P < 0.05) kidding rate (22 de pastoreo, en comparacion con el terreno con baja carga (42 vs 55%). We concluded that overstocking with goats rate en un mayor (P < 0.01) porcentaje de abortos shrub and grass cover. Also, decades of continu- animal resulto greatly reduced (P < porcentaje de pariciones forced goats to alter diet selec- (22 vs 12%), y un menor 0.05) ously high grazing pressure has una alta presion de pastoreo ha more resinous, toxic, and coarse (42 vs 55%). Se concluyo que tion pattern by consuming la utilizacion por las cabras de plantas This switch was associated with a lower nutritional sta- incrementado species. resinosas, toxicas fibrosas. Este cambio se refleja en un negative daily weight gain, lower body condition score in y tus, a estatus nutricional mas bajo, to cual conduce a perdidas de the late-wet period, and lower fertility on heavily grazed range. peso y condicion corporal en el otono, to cual a su vez provo- ca una menor fertilidad de las cabras.

Key Words: Diets, blood chemistry, fertility, vegetation cover, Chihuahuan desert

grazed ranges, periodic differences in dietary botanical composi- and semi-arid zones of Mexico, goats are managed In arid tion resulted from the 2 grazing treatments (Malechek and extensive village systems and are grazed on nat- under traditional Leinweber 1972). Owens (1991) indicated that as goat density range throughout the year with no supplements. ural communal increased, utilization in the 0.75-1.5 m zone of Acacia shrubs typical goat production system involves concentration Thus, the community increased at a faster rate than in any other canopy large herds of goats in the communities and often sur- of several zones. Provenza and Malechek (1984) found that goats showed carrying capacity of these rangelands. The result is a passes the preference for basal twigs of blackbrush in heavily stocked pas- range in poor condition. An increase in severely overgrazed tures, whereas no preference for terminal or basal twigs was pressure generally represents a decrease in quantity stocking observed in lightly and moderately stocked pastures. Knowledge available to the grazing animals. Few and/or quality of forage of dairy-type goat diets under different stocking pressures the relation between goat stocking and diet studies have described incomplete, therefore the objective of the study was to Studies with Angora goats have shown that remains composition. describe forage selection and botanical composition of goat diets average annual diets were similar on lightly and heavily although and relate disparities to stocking pressure. An additional aim of this study was to examine whether blood chemistry and fertility of Financial support was received from World Wildlife Foundation (Project PN59). goats and vegetation cover were sensitive to stocking level. Manuscript accepted 27 May 02.

167 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Materials and Methods sure has been maintained for several Fecal samples collection and analysis decades. Feces samples were collected from the All Study Area does of both herds were exposed to rectum of 6 randomly-selected adult goats crossbred (milk-type) bucks (3% bucks The study was conducted on a 2,250 during 5 consecutive days in late-dry and per herd) during 4 weeks in 1,050 ha site January 2000. (spring), early-wet (summer) and late-wet associated with 2 adjacent Body condition score (5-point rural communities (2 km apart, scale; (fall) periods. The same goats were used in with a Santucci and Maestrini fence dividing the sites) in northeast 1985) were record- each collection period. Samples were ed at mating. Abortions and kiddings Mexico (25° 30' N, 101° 02 W). The com- were pooled across days within periods for each also recorded. munities had similar types of vegetation, animal. The samples were oven-dried and In January (5 soil, terrain and precipitation and were 2000, blood samples ml) ground through a 1-mm screen. Small sub- from selected based on different grazing pres- all adult goats in both flocks (n= samples were taken from the ground mate- sure of goats and the absence of other 330) were collected from the jugular vein rial and mounted in microscopic slides. (vacutainer stock. Vegetation cover was 8 percentage system), in the morning before Five slides from each sample were ana- points higher in the lightly stocked pas- grazing (14-16 hours from the last feed- lyzed. Composition of goat diets was ture. The rainy season extends from June ing). The blood samples were obtained in determined by counting the number of epi- to October with mean annual precipitation non-heparinized vacutainer tubes. Within dermal fragments of each species recog- being 326 mm. Mean elevation of the 2 hours of collection the samples were nized in 100 microscope fields at 125x study area is 1,700 m and average annual centrifuged at 3,000 g for 15 minutes at (Sparks and Malechek 1968). Some temperature is 18.2°C. Soils in the study room temperature. Serum was collected authors recommend the use of correction site were silty and depth overlaying a and stored at -20°C until analyzed for factors for forbs when the microhistologi- limestone substrate ranged from 250 to cholesterol, glucose, creatinine, urea, uric cal analysis is used, due to higher 650 mm. The plant community was domi- acid, total proteins, calcium, phosphorus, digestibility of forbs compared to grasses nated by creosotebush (Larrea tridentata copper, magnesium and zinc. All blood (Holechek et al. 1982) and their more frag- (DC.) Coy.). Other important browse metabolites analyses were carried out with ile epidermal layers (Bartolome et al. species were resin-bush (Viguiera greggii a Coleman Junior II spectrophotometer 1995). However, other authors dismiss the (Gray) Blake) and lechuguilla (A gave following protocols supplied by the kit use of correction factors on the grounds lechuguilla Torr.). Major grasses were manufacturer. All minerals, except phos- that phenological variation in digestibility blue grama (Bouteloua gracilis H.B.K.) phorus, were determined by atomic is often more significant than taxonomic and buffalo grass (Buchloe dactyloides absorption spectrophotometry. Phosphorus variation (Hansen et al. 1976). In this study (Nutt) Engelm). Major forbs were globe- was determined by the method of Fiske correction factors for forbs were not used. mallow (Sphaeralcea angustifolia (Cav.) and Subbarow (1925). Selection values for individual plants D. Don.), silver-leaf nightshade (Solanum Thirty randomly selected adult lactating were calculated as the ratio of each class elaeagnifolium Cav.) and rosval (Croton goats of each herd were weighed twice in percentage in dioicus Cav.). the diet to its percentage the fall 1999 (22 October and 16 availability (percentage cover) in the range December). Animals were weighed after (Taylor et al. 1980). The formula used in Goat characteristics and manage- feed and water were withheld overnight. developing these ratings was: ment (% Two flocks of goats (n 163 in diet - % cover of total vegetation) = and 167 Selection value= for high and low stocking rate, respective- x 10 (% in diet + % cover of total vegetation) ly) were used in this study. Goats were of undefined genotype (milk-type) of different ages and parity with 35 to 45 kg adult live weight. All animals were reared in the Botanical composition of the vegeta- An index value of 0 indicated nonselec- same areas where they were tested, had no tion tive use of a forage class; values >0 or <0 health intervention, and did not receive Vegetation sampling was carried out indicated grazing selectivity for or against feed or mineral supplements throughout just before the start of feces sample collec- a particular species, respectively. the year. tions for each of the 3 seasonal periods. Similarity of diets was calculated using In and around settlements, land use is Five transects, 500 m in length, were Kulczyniski's similarity index (Oosting predominantly by goats, which are man- established in sites frequently grazed by 1956). aged under a traditional extensive village goats in both the lightly and heavily system and are grazed on natural commu- grazed areas. Transects were positioned to Statistical analyses nal range throughout the year. The grazing avoid watering points and roads. T-tests were used to compare shrub, period is approximately 8 hours daily Vegetation (foliage) cover data were col- forbs, grass and total vegetation cover (from 1000 to 1800 hours) and animals are lected along these transects in the late-dry between pastures within seasons (Steel tended by herdsmen. Goats are penned (April), early-wet (July) and late-wet sea- and Torrie 1980). The numbers of epider- near the household at night without access sons (November) of 2000. The line-inter- mal fragments of each species were con- to feed and water. cept method (Canfield 1942) was used to verted to percentages and transformed to The lightly grazed area had been grazed determine percent cover. In each transect aresin (angular) prior to statistical analy- continuously by goats at the rate of 15 ha the intercept of each plant species was sis. For each individual species as well as per goat for several decades (one 150-herd summed and divided by the total length of life forms, t-tests at P < 0.10 were used to goat in 2,250 ha). In the heavily grazed the transect to obtain the percent cover per compare area the annual stocking rate averaged 1.5 sampling unit (transect). goat diets between pastures with- ha per goat (5 different flocks totaling 700 in seasons. Significant differences in selec- animals in 1050 ha), and this grazing pres- tion indices were assessed using the

168 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 minor amounts of creosotebush, but goats 40 on the heavily grazed area relied heavily on this shrub. This was a surprising finding; as it was expected that goats would not be able to cope with the chemical defenses in creosotebush. The heavy use of this resinous shrub by goats was of interest because creosotebush is generally considered unimportant and undesirable livestock forage. These results are contrary to other reports where creosote bush has been a minor diet component of Angora (Warren et al. 1984a) and Spanish goats (Mellado et al. 1991, Warren et al. 1984b). FOR OR TOT SIR FOR OR TOT SIR OR TOT These results reaffirm that goats can live LATEORY EARLY-WEr LAT&WE? on forages rich in secondary metabolites and lignin (Provenza et al. 1990). The relatively high consumption of poisonous plants (e.g. creosotebush, cover for lightly Fig. 1. Shrubs (SHR), forbs (FOR), grasses (GR) and total (TOT) vegetation Asclepias brachystephana Torr. and sil- and heavily stocked ranges in the Chihuahuan desert for 3 seasons. Paired bars with dif- ver-leaf nightshade) by goats on the heavi- ferent letters differ (P < 0.05) between ranges within each sampling period. ly stocked range merits discussion. The tolerance for high levels of poisonous Kruskal-Wallis test (SAS 1990). grazed site was also likely due to high plants in goat diets probably was due to Differences in serum metabolites and min- grazing pressure. Forb cover was very low the utilization of the less-toxic portion of erals as well as body weight change in both sites and remained nearly the same these plants (ruminants possess the ability between localities were assessed using the on the two treatments. Major forbs in the to select plant parts of low toxin concen- t-test procedure (Steel and Torrie 1980). study site were short-lived and dynamics trations; Provenza 1995, Pfister 1999). Percentage of aborting does and proportion were largely controlled by rainfall, rather However, observations on the heavily of does pregnant was compared using chi- than by grazing. In common with other stocked pasture showed that creosotebush square procedures (Steel and Torrie 1980). studies in arid environments (Severson was uniformly and intensively used, to the Statistical significance was assumed at P < and Debano 1991) forbs did not increase point that the bark of these plants was con- 0.05). Finally, these findings depict in response to decreasing shrub cover in sumed. Another explanation is that goats unreplicated research (only one flock from the heavily grazed pasture. In terms of may have the ability to detoxify some each site) and results should be applied to management these results indicate that plant poisons. Studies with desert other areas with some trepidation. goat production would be favoured by woodrats (Mangione et al. 2000) pigmy conservative continuous grazing in this rabbits (White et al. 1982) and goats type of vegetation. (Duncan et al. 2000) have documented the Results and Discussion development of mechanisms to adapt to Botanical composition of diets plant secondary metabolites. Vegetation cover At the 3 seasons sampled the number of Under the heavy grazing intensity, The lightly grazed site had significantly species detected in the goat diet were 34 goats consumed agrito (Berberis trifoliola- (P < 0.05) more shrub cover throughout vs 28, 30 vs 24, and 34 vs 29 on heavily ta Moric.) and Spanish dagger (Yucca the year than the heavily stocked pasture and lightly grazed areas in the late-dry, carnerosana (Trel.) Mckelvey), whereas (Fig. 1). Because the study site was not early-wet and late-wet periods, respective- in the lightly grazed area these species accessible to other domestic ungulates and ly. The contribution of shrubs to the diet were not used. The high consumption of shrubs were generally palatable to goats, it was greater (P < 0.05) on the heavily fibrous, resinous and poisonous plants by is thought that the marked decrease of grazed range than the lightly grazed range goats on the heavily stocked range reaf- shrubs, particularly creosotebush (136% in the late-dry and late-wet periods (Table firm the notion that goats are very flexible decrease cover of this shrub in the heavily 1). The predominance of shrubs (above with respect to the plants they consume. grazed pasture compared to the lightly 80%) in diets during the dry season on Another shrub of singular importance grazed pasture) on the pasture heavily similar types of vegetation is consistent was mesquite (Prosopis glandulosa Ton), grazed, arose from intensive goat grazing. with Mellado et al. (1991). They found which contributed 7.8% to total composi- This finding is consistent with that of diets contained up to 93% shrubs on an tion of the diets of goats in the heavily Manzano and Navar (2000) on the reduced overgrazed Larrea-dominated range. stocked site in the late-dry period. In the cover of shrubs as a result of goat over- Higher dietary shrub content in goat diets late-wet period a significant grazing inten- grazing. For most of the year the grass in both the heavily and lightly stocked sity effect (P < 0.05) was detected in the cover was less (P < 0.05) on the heavily sites in the late-dry period was a function percentage of this shrub in the goat diets. grazed compared to the lightly grazed site. of decreased cover of grasses and forbs in The level of mesquite in the goat diets in Although it is known that non-Angora this period. High stocking rates resulted in the present study is much higher than that goats make limited use of grasses in this increased (P < 0.05) utilization of cre- reported in other studies (Lopez-Trujillo type of vegetation (Mellado et. al. 1991), osotebush by goats in the late-dry and late- and Garcia-Elizondo 1995; Mellado et al. the decrease of graminoids in the heavily wet periods. Goats on the lightly stocked 1991). Goats on both ranges made consis- site tended to avoid or consume only

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 169 Table 1. Percentage of shrubs, forbs and grasses in goats' diets on heavily and lightly stocked nent was reduced in the late-dry period ranges. causing more dependence on shrubs. Regardless of stocking level, utilization Species Late-dry of grasses by goats in our study was low, Stocking rate Stocking rate Stocking rate which agrees with Lopez-Trujillo and Shrubs High Low Garcia-Elizondo (1995) and Mellado et al. Acacia farnesiana 4.6 3.1 (1991) in the same type of vegetation. Agave lechuguilla 2.3 6.3 Other Chihuahuan desert studies have Agave striata 0.6+ 3.3 shown much higher proportions of grasses Atriplex canescens 9.8 5.3 in Spanish goat diets. Warren et al. Berberis trifoliolata 1.6 0.0 Buddleja scordioides 1.1+ 5.7 (1984a) reported that grasses contributed Condalia warnockii 4.3 2.0 over half of the diets during spring, but Cowania plicata 5.8 2.3 17% in autumn. Warren et al. (1984b) Dalea bicolor 3.4 1.5 found grass utilization by Spanish goats Dasylirion palmeri 1.2* 6.3 was between 17 and 68%. Ephedra aspera 2.0 2.6 In the Chihuahuan desert rangelands it is Larrea tridentata 15.3 7.6 reported that high grazing pressure has lit- Opuntia leptocaulis 0.4* 3.0 tle effect on diet selection by Angora goats Opuntia rastrera 6.8 4.8 (Malechek and Leinweber 1972, Taylor Parthenium incanum 7.9** 13.6 and Kothmann 1990). It is believed that Prosopis glandulosa 7.8 4.0 the outstanding foraging skills of goats Yucca carnerosana 5.9 0.0 (upper mobile lip, bipedal grazing stance Other shrubs 5.6 1.0 and prehensile tongue) allows selection of Total shrubs 86.4* preferred forages even at excessive graz- Forbs ing pressure. Our data with milk-type Asclepias brachystephana 4.2 0.0 goats do not support such a contention Sida abutifolia 0.9 1.6 because high grazing pressure forced goats Solanum elaeagnifolium 2.6* 6.9 to have a more flexible foraging strategy. Sphaeralcea angustifolia 1.5** 9.7 Illius et al. (1999) indicated that goats Other forbs 2.2 3.3 select diets which tend to maximize their Total forbs 11.4+ Grasses rate of intake, rather than expressing pref- Aristida arizonica 1.2 0.8 erences that are specific to plant species. Bouteloua curtipendula 0.2 0.7 Moreover, hunger (Grote and Brown Bouteloua gracilis 0.3 1.8 1973) and social facilitation (Ralphs and Other grasses 0.4 2.8 Provenza 1999) rapidly extinguish food Total grasses 2.1+ 6.1 aversion. Thus, the different diet strategy **p showed by on the site with high +p = < 0.10; *p = < 0.05; = < 0.01 for comparisons between stocking rates within periods. goats grazing pressure apparently was an adap- tent use of huizache (Acacia farnesiana season, globe-mallow constituted one third tation (behavioiral and physiological) to (L.) Willd.) during all periods. Use of this of the diets. The abundance of forbs in the the condition imposed by grazing pres- shrub as well as mesquite by goats was diets is explained by the fact that actively sure. In other trials (Baptista and somewhat limited because animals could growing forbs in the Chihuahuan desert Launchbaugh 2001; Warren et al 1984b) a not reach the higher canopies. Butterfly- have higher protein, phosphorous, and cell considerable variation between individuals bush (Buddleja scordioides H.B.K.) was soluble materials than grasses during with respect to voluntary consumption of significantly more important in goat diets active growth (Nelson et al. 1970). unpalatable species has been found, there- on the lightly grazed area in the late-dry A forb of singular importance was milk- fore the flock on the overstocked site pos- and late-wet period, compared to goats on weed (Asclepias brachystephana Ton), a sibly developed an enhanced detoxification the heavily grazed area. poisonous plant which made up 4.2% of or tolerance ability. If this is true, goats Differences (P < 0.10) occurred in total the diet during the late-dry period on the with these capabilities could constitute a forbs in goat diets on the heavily and heavily stocked range, but was not present management strategy to increase the use of lightly grazed areas during the late-dry in diets on the lightly grazed site. plants containing deterrents or toxicants. and late-wet seasons (Table 1). Ralphs et There were no differences in percent- al. (1986) also found that forbs declined in ages of grasses in diets between stocking Dietary preferences sheep diets as the stocking rate increased. levels in summer and fall, but a trend (P < The influence of stocking level on Major forbs in the goat diets in the heav- 0.08) was noted indicating a higher pro- dietary selection values is shown in Table ily and lightly stocked ranges included sil- portion of grasses in goat diets on the low 2. In general, goats on the heavily stocked ver-leaf nightshade and globe-mallow. stocked range in the late-dry period. site were less selective (proportions of for- These forbs were consistently higher (P < Arizona three-awn (Aristida arizonica age species in their diets more closely 0.05) in diets of goats on the low stocked Vasey.) was significantly (P < 0.07) more matched the proportions available), than range compared to that of heavily stocked important in diets on the lightly grazed goats on the lightly stocked range, regard- range. Of the nearly 30 species represent- range than the heavily stocked site in the less of period of the year. ed in goat diets in the heavy and lightly early-wet and late-wet periods. Under the Among shrubs, huizache, fourwing salt- stocked sites at the beginning of the rainy heavy grazing intensity the grass compo- bush (Atriplex canescens (Pursh) Nut.),

170 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 2. Selection values for shrubs, forbs and grasses of goats on heavily and lightly stocked Goats in both the heavily and lightly ranges. stocked pastures showed a moderate preference for Arizona three-awn, but all oth- Species Late-dry ers grasses were consumed in proportion Stocking rate Stocking rate Stocking rate to availability. The marked differences in Shrubs High Low selectivity in the lightly and heavily Acacia farnesiana 9.7 9.5 stocked ranges stress the fact that prefer- Agave lechuguilla -4.6** 5.8 ences can not be generalized because Agave striata -2.8** 4.1 selection depends on the available choice Atriplex canescens 8.9 9.1 of feed items. Berberis trifoliolata 4.9 8.4 Diet similarities between the heavily and Buddleja scordioides 8.8 9.0 lightly grazed ranges were 99, 72, and 40% Condalia warnockii 7.7 8.2 for the late-dry, early-wet, and late-wet Cowania plicata 8.2* 6.5 periods, respectively. The high similarity Dalea bicolor 8.7 7.1 indices in the late-dry period were expected Dasylirion palmeri 8.4* 9.1 because warm-season species had not yet Ephedra aspera 43** 9.0 begun rapid growth and the least amount of Larrea tridentata -4.6 -7.7 forage was available. With higher forage Opuntia leptocaulis 2.1* 4.3 availability (late-wet period) the similarity Opuntia rastrera 8.6 9.4 index was low, reflecting the effect of Parthenium incanum 8.3 9.3 stocking density. Low overlap during wet Prosopis glandulosa 9.2* 6.4 times occurred when goats on the lightly Yucca carnerosana 0.4 stocked site concentrated more on forbs Forbs than goats on the heavily stocked site. Asclepias brachystephana 8.7 5.5 Sida abutifolia 7.5 8.6 Blood chemistry and reproductive Solanum elaeagnifolium 8.0 9.2 performance angustifolia 8.0 Sphaeralcea The effect of stocking level on average Grasses daily gain, body condition score (BCS), Aristida arizonica 6.4+ 3.7 reproductive performance and blood Bouteloua curtipendula -1.3 2.3 chemistry is presented in Table 3. A sig- Bouteloua gracilis -2.0 -0.6 nificant stocking level effect was detected **p +p = < 0.10; *p = < 0.05; = < 0.01 for comparisons between stocking rates within periods. for average daily gain and BCS in the late- wet season. The weight loss and lower butterfly-bush, Condalia warnockii M.C. several goats. Signs of intoxication were BCS for goats on the heavily stocked site Johnst, Cowania plicata D. Don., Dalea mandibular tremors and uncontrolled indicate a lack of nutrients in this pasture. bicolor H. & B. and mariola (Parthenium chewing, which resulted in the death of all The substantially higher (P < 0.01) serum incanum H.B.K.) in both treatments and in affected animals (7 out of 163 adult glucose levels in goats on the lightly the different seasons were eaten in propor- goats). These clinical signs from mesquite grazed area compared to the heavily tions much higher than their relative avail- ingestion are caused by a selective toxicity grazed site is additional evidence of short- ability would suggest. In both the heavily to neurons of some cranial nerve nuclei age of forage in the heavily stocked site. and lightly stocked sites creosotebush was (Tabosa et al. 2000). Glucose is a relatively good indicator of utilized less than expected during the 3 Less selective grazing by goats on the energy balance, decreasing as goats are seasons, although during the early-wet and heavily stock site was probably due to subjected to energy restriction (Hussain et late-wet seasons this shrub was signifi- depletion of accessible foliage of favored al. 1996) or fed lower levels of concen- cantly (P < 0.05) less preferred by goats species which forced the goats to turn to trate supplementation (Landau et al. on the lightly stocked site. Despite the other less palatable choices. Walker et al. 1993). avoidance shown for creosotebush, this (1994) have shown that as vegetation was Serum urea nitrogen was substantially shrub still made an important contribution progressively defoliated, goats were less higher (P < 0.01) in goats on the lightly to diets on the lightly grazed area. selective. Additionally, our goats were grazed area than in the heavily grazed site. Goats on the heavily stocked pasture restricted overnight to a pen and only able This metabolite originates either from 8 This showed a consistently higher (P < 0.05) to forage for hours each day. catabolism of amino acids to spare glucose time probably also preference for mesquite compared to goats restricted foraging oxidation, or from ammonia absorbed by the goats to favor some of the less on the lightly stocked site. Because caused the rumen (Oldham 1984). Thus, high urea palatable species. mesquite causes conditioned flavor aver- levels could indicate undernutrition, Forbs were highly preferred on both in (Baptista and reflecting increased gluconeogenesis from sion ruminants ranges, notably after summer rains. Of 2001), it is likely that the amino acids, or high protein intake. In this Launchbaugh note, milkweed and silver-leaf nightshade, intake of mesquite exhibited by goats on study the higher serum urea levels in goats species considered toxic for livestock, on the lightly stocked site is believed to be the heavily stocking pasture was not a true in both were highly preferred by goats due to a higher protein intake. Serum crea- reflection of animal preference, but a had higher selection val- treatments. Forbs tinine levels were lower (P < 0.01) in product of forced utilization caused by a and shrubs, indicating ues than grasses these animals, than goats on the heavily shortage of browse. The relatively high they would have selected more forbs had stocked site, and this metabolite is an indi- use of this shrub provoked intoxication in they been available.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 171 Table 3. Daily gain, body condition score, reproductive performance and blood chemistry of goats Bartolome, J., J. Franch, M. Gutman, and on heavily (n =167) and lightly (n =163) stocked ranges. Values following ± are standard error N.G. Seligman. 1995. Physical factors that of the mean. influence fecal analysis estimates of herbivore diets. J. Range Manage. 48:267-270. Canfield, R.H. 1942. Application of the line Item High stocking rate Low stocking rate interception method of sampling range vege- Reproductive traits tation. J. Forest. 39:388-394. Abortion (%) 22 (371167) ** 12 (20/163) Duncan, A.J., P. Frutos, and S.A. Young. Percentage kidding 42 (74/167) * 55 (90/163) 2000. The effect of rumen adaptation to oxal- Does pregnant (%) 60 (101/167) 67 (110/163) ic acid on selection of oxalic-acid rich plants Mortality due to toxic plants (%) 4.2 (7/167) (0/163) by goats. British J. Nutr. 83:59-65. Fiske, Y. The col- Blood chemistry C.H. and Subbarow.1925. orimetric determination of phosphorus. J. Glucose (mg/100 ml) 48.9 1.3** 1.1 Biol. Chem. 66:371-375. Urea (mg/100 ml) 10.8 ± 0.33** 0.29 Grote, F.W. Jr. and R.T. Brown. 1973. Creatinine (mg/100 ml) 0.73 ± .05** .04 Deprivation level affects extinction of condi- Cholesterol (mg/l00 ml) 66 ± 3.2+ 2.2 tioned taste aversion. Learn. Motiv. Total proteins (g/100 ml) 6.5 ± .13 .06 4:314-319. Calcium (mg/100 ml) 11.2 ± .16 .08 Hansen, R.M., T.M. Foppe, M.B. Gilbert, Phosphorus (mg/100 ml) 4.0 ± .11+ .09 R.C. Clark, and H.W. Reynolds. 1976. The Copper (ppm) 1.79 ± 0.06 0.06 microhistological analysis of feces as an esti- mator of herbivore diets. Unpublished report, Magnesium (ppm) 1.93 ± 0.06** 0.07 AAFAB Composition Analysis Laboratory, Zinc (ppm) 1.66 ± 0.05** 0.05 Inc., Ft. Collins, Colo. Body traits Holechek, J.L., M. Vavra, and R.D. Pieper. Body condition score (0-5 scale) 2.1 0.04** 0.04 1982. Botanical composition determination Average daily gain (gld) -19 ± 2.3** ±3.7 of range herbivore diets: a review. J. Range Manage. 35:309-315. +p=<0.10; *p=<0.05; **p=<0.01 Hussain, Q., 0. Harevoll, L.O. Eik, and E. Ropstadt. 1996. Effect of energy intake on plasma glucose, non-sterified fatty acids and cator of muscle break down. Conclusions aceto-acetate concentration in pregnant A trend (P < 0.07) was noted for higher goats. Small Rumin. Res. 21:89-96. serum cholesterol levels in goats on the Illius, A.W., I.J. Gordon, D.A. Elston, and The results of this study show that goat lightly stocked site than in goats on the J.D. Milne. 1999. Diet selection in goats: a herbivory reduces shrub and grass cover in heavily stocked range. This metabolite, in test of intake-rate maximization. Ecol. the Chihuahuan desert range the absence of excess dietary energy intake, when goat 80:1008-1018. density is high. These results also demon- is considered to reflect the capacity of the Ingraham, R.H., and L.C. Kappel. 1988. animal to mobilize body fat reserves strate that decades of continuously high Metabolic profile testing. Vet. Clin. North grazing pressure has led to divergent diet America: Food Anim. Prac. 4:391-411. (Ingraham and Kappel 1988). This suggests selection. Landau, S., J. Vecht, and A. Perevolotsky. that during winter, goats on the lightly Goats on the high stocking rate were forced to alter diet selection 1993. Effects of two level of concentrate sup- stocked site catabolized more fat reserves pattern by consuming more resinous (creosote- plementation on milk production of dairy than goats on the heavily stocked site. goats browsing Mediterranean shrubland. bush), toxic (mesquite, milkweed) and Serum phosphorus tended (P < 0.10) to Small Rumin. Res. 11:227-237. coarse (agrito, be higher in goats on the lightly grazed Spanish dagger) species. Lopez-Trujillo, R. and R. Garcia-Elizondo. This switch was associated with a lower area than heavily stocked site. Both serum 1995. Botanical composition and diet quality nutritional status (lower serum glucose, Zn and Mg were significantly higher (P < of goats grazing natural and grass reseeded urea, cholesterol, magnesium 0.01) in goats on the lightly stocked site and zinc lev- shrublands. Small Rumin. Res. 16:37-47. els), weight Malechek, J.C. and C.L. Leinweber. 1972. compared to goats in the heavily stocked loss, lower body condition Forage selectivity by goats on lightly and site. score (BCS) in the late-wet season, and lower fertility. These results highlight heavily grazed ranges. J. Range Manage. Percent of pregnant does was not signif- the necessity of establishing conservative and 25:105-111. icantly affected by stocking rate, but goats flexible (seasonally) stocking densities Mangione, A.M., M.D. Dearing, and W.H. on the heavily stocked pasture had higher Karasov. 2000. Interpopulation differences because herbivory would reduce forage (P < 0.01) abortion rates and consequently in tolerance to creosote bush resin in desert resources in areas where goat density is lower (P < 0.05) kidding rates. Of the pos- woodrats (Neotoma lepida). Ecol. great, thereby sible mechanisms inducing non-infectious setting the stage for nutri- 81:2067-2076. tional stress and low productivity abortions in the goats, the hypoglycemic of goats. Manzano, M.G. and J. Navar. 2000. condition in the mother (Wentzel 1982) Processes of desertification by goats overgrazing in the Tamaulipan thornscrub was the most likely cause of gestation fail- Literature cited (matorral) in north-eastern Mexico. J. Arid ure. Mean serum glucose of goats on the Env. 44:1-17. lightly stocked site was 31% higher than Mellado, M., R.H. Foote, A. Rodriguez, and goats on the heavily stocked site. Serum Baptista, R. and K.L. Launchbaugh. 2001. P. Zarate. 1991. Botanical composition and Mg levels as low as ours have been associ- Nutritive value and aversion of honey nutrient content of diets selected by goats ated with massive abortions in goats mesquite leaves to sheep. J. Range Manage. grazing on desert grassland in northern (Mellado et al. 2002, Unanian and 54:82-88. Mexico. Small Rumin. Res. 6:141-150. Feliciano-Silva 1984).

172 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Mellado, M., R. Valdez, L.M. Lara, and J.E. Provenza, F.D., E.A. Burritt, T.P. Clausen, Taylor, C.A. and M.M. Kothmann. 1990. Garcia. 2002. Risk factors for conception, J.P. Bryant, P.B. Reichardt, and R.A. Diet composition of Angora goats in a short- abortion, and kidding rates of goats under Distel. 1990. Conditioned flavor aversion: a duration grazing system. J. Range Manage. extensive conditions. Small Rumin. Res. (In mechanism for goats to avoid condensed tan- 43:123-126. press). nins in blackbrush. Amer. Nat. 136:810-828. Taylor, C.A., M.M. Kothmann, L.B. Merrill, Nelson, A.B., C.H. Herbel, and H.M. Ralphs, M.H. and F.D. Provenza. 1999. and D. Elledge.1980. Diet selection by cat- Jackson. 1970. Chemical composition of Conditioned food aversions: principles and tle under high-intensity, low frequency, short forage species selected by cattle on an arid practices, with special reference to social duration, and Merrill grazing systems. J. New Mexico Range. New Mexico Agr. Exp. facilitation. Proc. Nutr. Soc. 58:813-820. Range Manage. 33:428-434. L.B. Bull, 561. Ralphs, M.H., M.M. Kothman, and Unanian M.D.S. and E.D.E. Feliciano-Silva. Merrill. 1986. Cattle and sheep diets under , Oldham, J.D. 1984. Protein-energy interrela- 1984. Trace elements deficiency: Association short-duration grazing. J. Range Manage. tionships in dairy cows. J. Dairy Sci. with early abortion in goats . Int . Goat and 39:217-223. 67:1090-1114. Santucci, P.M. and Maestrini.1985. Body Sheep Res . 2:129-134 . Oosting, H.J. 1956. The study of plant com- 0. conditions of dairy goats in extensive sys- Walker , J . W ., S . L . Kronber g, S . L . Al- W.H. Freeman & Co. San munities. tems of production: method of estimation. Rowaily, and N.E. West. 1994. Comparison Francisco, Calif. Ann. Zootech. 34:473-474. of sheep and goat preference for leafy by Owens, M.K. 1991. Utilization patterns SAS. 1990. SAS procedure users guide version spurge. J. Range Manage. 47:429-434. canopies of Angora goats within the plant 6. Third edition. SAS Institute Inc., Cary, Warren, L.E., D.N. Ueckert, and J.M. two Acacia shrubs. J. Range Manage. NC., USA. Shelton. 1984a. Comparative diets of 44:456-461. Severson, K.H. and L.F. Debano. 1991. Rambouillet, Barbado, and Karakul sheep Pfister, J.A. 1999. Behavioral strategies for Influence of Spanish goats on vegetation and and Spanish and Angora goats. J. Range coping with poisonous plants. P 45-59. 1999. soils in Arizona chaparral. J. Range Manage. Manage 37:172-180. In: K.L. Launchbaugh, K.D. Sanders, and 44:111-117. Warren, L.E., D.N. Ueckert, J.M. Shelton, J.C. Mosley. (eds) Grazing behavior of live- Sparks, D.L. and J.C. Malechek. 1968. and A.D. Chamrad. 1984b. Spanish goat stock and wildlife. Idaho Forest, Wildlife, Estimating percentage dry weight in diets diet on mixed-brush range-land in the south Exp. Sta. Bull. No. 70. Moscow, and Range using a microscopic technique. J. Range Texas plains. J. Range Manage. 37:340342. Ida. Manage. 21:264-265. Wentzel , D. 1982. Non-infection abortion in . . feedback . . . Provenza , F D 1995. Postingestive Steel, R . G. and J .H Torrie 1980 goats. In: Proc. III International as an elementary determinant of food prefer- and procedures of statistics . Second Edition. Conference on Goat Production and Disease. ence and intake in ruminants. J. Range McGraw-Hill Book Co. , New York, N.Y. Univ . Arizona , Tucson , Ariz .. pp 155-161. Manage . 48:2-17 Tabosa I.M., J.C.D. Souza, D.L. Graca, J.M. , S.M., B.L. Welch, and J.T. Flinders. F.D. and J.C. Malechek. 1984. Barbosa, R.N. Almeida, and F. Riet- Provenza , Monoterpenoid content of pigmy rab- Diet selection by domestic goats in relation Correa. 2000. Neuronal vacuolation of the bit stomach ingesta. J. Range Manage. to blackbrush twig chemistry. J. Appl. Ecol. trigeminal nuclei in goats caused by inges- 35:107-109. 21:831-837. tion of Prosopis juliflora pods (Mesquite beans). Vet. Human Toxicol. 42:155-158.

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 173 J. Range Manage. 56:174-184 March 2003 Mineral concentration dynamics among 7 northern Great Basin grasses

DAVE GANSKOPP AND DAVE BOHNERT

Authors are Rangeland Scientist, USDA-ARS, Eastern Oregon Agricultural Research Center', 67826-A Hwy. 205, Burns, Ore. 97720; and Range Animal Nutritionist, Oregon State University, E.O.A.R.C., 67826-A Hwy. 205, Burns, Ore. 97720. 'The Eastern Oregon Agricultural Research Center is jointly operated by the USDA-Agricultural Research Service and the Oregon Agr. Exp. Sta. of Oregon State University.

Abstract Resumen

Livestock and wildlife managers must be aware of the Los manejadores de fauna y ganado deben estar enterados de nutritional dynamics of forages to sustain satisfactory growth and las dinamicas nutricionales de los forrajes para sostener un reproduction of their animals and assure fair value for pasture. crecimiento y reproduccion satisfactorios de sus animales y ase- Despite a history of livestock grazing in the northern Great Basin, gurar el valor favorable de sus praderas. A pesar del historial de annual and seasonal mineral concentrations of many of the apacentamiento de ganado en la Gran Cuenca del Norte las con- region's prominent grasses have not been measured. We centraciones anuales y estacionales de minerales de muchos de addressed this problem with monthly sampling (April-November) los zacates importantes de esta region no habian sido medidas. of 7 cool-season grasses at 6 sites during 1992, a drier than average Nosotros abordamos este problema con muestreos mensuales year (86% of mean precipitation), and 1993 when precipitation (Abril - Noveembre) de 7 zacates de estacion fria en 6 sitios, was 167% of average (255 mm). Grasses included: Poa sandbergii durante 1992, un ano mas seco que el promedio (86% de la pre- Vasey, Bromus tectorum L., Sitanion hystrix (Nutt.) Smith, cipitacion promedio) y en 1993 cuando la precipitacion fue 167% Agropyron spicatum (Pursh) Scribn. & Smith, Festuca idahoensis de la precipitacion promedio (255 mm). Los zacates muestreados Elmer, Stipa thurberiana Piper, and Elymus cinereus Scribn. & fueron: Poa sandbergii Vasey, Bromus tectorum L., Sitanion hys- Merr. Phosphorus, K, Ca, Mg, Mn, Fe, Cu, Zn, and Na were trix (Nutt.) Smith, Agropyron spicatum (Pursh) Scribn. & Smith, assayed, and initial statistical analysis was a split-split-plot with Festuca idahoensis Elmer, Stipa thurberiana Piper, and Elymus main effects of species, years, and months and all possible inter- cinereus Scribn. & Merr. y los minerales determinados fueron P, actions. For a preponderance of the minerals, (Zn and Na exclud- K, Ca, Mg, Mn, Fe, Cu, Zn y Na. El analysis estadistico inicial fue ed) the 3-way year x month x species interactions were significant un diseno experimental de parcelas subdivididas, en el que los (P < 0.05) indicating that main effects did not function indepen- efectos principales fueron las especies, anos y meses y todas sus dently. Generally, mineral concentrations averaged about 41 % posibles interacciones. Para la mayoria de los minerales higher among the grasses for the drier of the 2 years (1992). (excluyendo Zn y Na) las interacciones triples ano x mes x Copper, Zn, and Na concentrations were below required levels especie fueron significativas (p < 0.05) indicando que los efectos for beef cattle (9.9, 28.8, and 672 mg kg', respectively) among all principales no actuan en forma independiente. Generalmente las the grasses for all sampling periods. Seasonally deficient miner- concentraciones de minerales promediaron aproximadamente als included Ca, Mg, P, K, and Mn. Calcium and Mn were largely 41% mas en los zacates del ano mas seco (1992). Las concentra- deficient (< 3.2 and 1.15 g kg'', respectively) for beef cattle early ciones de Cu, Zn y Na estuvieron por abajo de los niveles in the growing season with levels rising as grasses matured. requeridos por el ganado para carne (9.9, 28.8 y 672 mg Seasonal patterns of Mg were variable among the grasses, respectivamente), esto fue para todos los zacates y todos los increasing in some as the season progressed, remaining stable periodos de muestreo. Los minerales estacionalmente defi- among others, and declining with maturity in yet others. cientes incluyeron al Ca, Mg, P, K y Mn. El Ca y Mn fueron Phosphorus and K levels were typically adequate (>1.94 and 5.76 muy deficientes (< 3.2 y 1.15 g kg"1, respectivamente) para el g kg'1, respectively) for beef cattle early in the growing season ganado para carne al inicio de la estacion de crecimiento y los and declined to deficient levels by July and August. Iron was of niveles fueron incrementando conforme los zacates madu- no concern, because concentrations were more than adequate for raron. Los patrones estacionales de Mg variaron entre las cattle (> 48 mg kg 1) among all the grasses for all seasons. While a especies de zacates, incrementando en algunos de ellos con- mixed stand of forages can extend the period of adequate mineral forme la estacion de crecimiento avanzo, permaneciendo nutrition for cattle in some instances, we suggest that a supple- estable en otros y disminuyendo con la madurez en otros. Al ment be available season-long on northern Great Basin range- inicio de le estacion de crecimiento los niveles de P y K fueron lands and that the formulation include at least Ca, Mg, P, K, Cu, adecuados (>1.94 y 5.76 g kg 1, respectivamente) para el ganado Zn, Mn, and Na in available forms and proper ratios. para carne y disminuyeron a niveles deficientes en Julio y Agosto. Respecto al Fe no hubo preocupacion porque las con centraciones fueron mas que adecuadas para le ganado (>48 mg kg'), esto fue para todas las especies y todas las estaciones. Mientras que en algunas situaciones una comunidad con Technical Paper No. 11899 Oregon Agr. Exp. Sta. Manuscript accepted 16 May 02.

174 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 diversas especies forrajeras puede exten- been intensively studied in this region, and environmental factors (Daubenmire der el periodo de una nutricion adecuada however, Murray et al. (1978) reported 1970, Hironaka et al. 1983). de minerales para el ganado, nosotros sug- enhanced April-December gains in cattle We evaluated 7 grasses in this study. erimos que en los pastizales de la Gran supplemented with protein, phosphorus, Among these were Sandberg's bluegrass, Cuenca del norte se tenga una suple- and sulfur and slightly elevated gains of a small stature, early maturing, caespitose, mentacion de minerales disponible supplemented calves from August to perennial grass and cheatgrass, an early durante toda la estacion y que la formu- December. Later, Mayland et al (1980) maturing winter annual. The remaining 5 lacion incluya al menos Ca, Mg, P, K, Cu, documented increased weight gains by perennial grasses included bottlebrush Zn, Mn y Na en formas disponibles y rela- grazing cattle supplemented with addition- squirreltail, bluebunch wheatgrass, Idaho ciones adecuadas. al zinc. fescue, Thurber's needlegrass, and giant Our objective was to characterize sea- wildrye. With the exception of giant sonal and annual mineral concentrations of wildrye, which can attain heights of up to Key Words: copper, zinc, manganese, 7 of the region's most prominent range- 2 m, these are mid-size caespitose grasses. sodium, iron, phosphorus, magnesium, land grasses. This was accomplished via In both years (1992 and 1993) we visited potassium, calcium, Poa sandbergii, monthly sampling at 6 sites during 1992, a all 6 locations and sampled each within a 3- Bromus tectorum, Sitan ion hystrix, drier than average year, and 1993 when day interval at the end of each month. spicatum, Festuca idahoensis, Agropyron above average precipitation occurred. Months sampled included April-November Elymus cinereus Stipa thurberiana, in both years. Yearly precipitation compiled on a cal- Stockmen and wildlife managers must Materials and Methods endar year basis at the Northern Great be aware of the nutritional concentrations Basin Experimental Range was 106 and of forages on rangelands to assure ade- 140% of the long-term average (x = 283 Six collection sites were identified near quate growth and reproduction of their mm, n = 41) for calendar years 1992 and Bums, Ore. with each supporting a broad animals and make informed decisions 1993, respectively (N.O.A.A. 1992-1993). array of forages. Specific locations, eleva- when purchasing supplements. Similarly, Sneva (1982), however, discovered that tions, and soil classifications for each site those marketing pasture should also be yearly forage yield in the region was most were reported by Ganskopp and Bohnert aware of the nutritional characteristics of closely correlated with precipitation accu- (2001). Among the 6 sites, mean soil their forages to assure reasonable and fair mulated on a crop year or September depth was 69 cm (se 5.1), and elevation returns. Despite a history of livestock = through June basis. Based on that premise, ranged from 1,375 to 1,472 m (x =1,429). grazing in the northern Great Basin, there accumulations for the 1992 and 1993 Along an east/west line, the sites spanned have been few efforts to quantify the sea- growing seasons at the Northern Great 118 km, and north/south extremes encom- sonal and annual mineral concentrations of Basin Experimental Range were 86 and passed 75 km. Climatological data report- the region's most prominent rangeland 167% of the crop year average (255 mm), ed herein were acquired at the Northern grasses (Murray et al. 1978, Mayland and respectively (Fig. 1). Both years of the Great Basin Experimental Range (119° 42' Shewmaker 1997). study, however, experienced higher than 30"W 43° 29' 37"N) with the location The northern Great Basin experiences average mid- to late-growing season rain- identified as the Squaw Butte Experiment an arid Mediterranean climate with about fall. June-July precipitation totaled 68 mm Station in N.O.A.A. records (N.O.A.A. 80% of the annual precipitation occurring in 1992 compared to an average of 34 mm. 1991 through 1994). in the fall, winter, and spring months when In 1993 pooled July- August precipitation The shrub layer at each site was domi- low temperatures hinder plant growth. nated by Wyoming big sagebrush Rangeland grasses typically begin growth (Artemisia tridentata subsp. wyomingen- with warming temperatures in the spring, sis Beetle) with occasional occurrences and herbage accumulation stops upon (<10% relative cover) of mountain big of soil moisture in mid- to late- depletion (Artemisia tridentata subsp. July (Sneva 1982, Ganskopp 1988). sagebrush vaseyana Beetle) at upper eleva- Hickman (1975) presented Ca and P pro- (Rydb.) tions. Dominant perennial grasses were files for 4 grasses common to the region, typically bluebunch wheatgrass (Agropyron and Murray et al. (1978) and Mayland and spicatum (Pursh) Scribn. & Smith) or Idaho described the seasonal Shewmaker (1997) fescue (Festuca idahoensis Elmer). dynamics of 9 elements for 2 introduced Subordinate grasses included Sandberg's and 5 native grasses in southern Idaho. bluegrass (Poa sandbergii Vasey), bottle- Neither source, however, addressed the brush squirreltail (Sitanion hystrix (Nutt.) to year disparities in nutritive value year Smith), Thurber's needlegrass (Stipa associated with precipitation differences. thurberiana Piper), giant wildrye Rates of gain for livestock reflect the (Elymus cinereus Scribn. & Merr.), of the region's for- Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug nutritional dynamics prairie Junegrass (Koeleria cristata ages with mature cows gaining up to 1.86 Pers.), and in disturbed areas, the intro- kg day' early in the growing season and Fig. 1. Monthly precipitation for 1992 (Sept. duced annual cheatgrass (Bromus tecto- losing 0.4 kg day' by mid- to late-August 1991-June 1992) and 1993 (Sept. rum L.). All of these grasses are common 1992-June 1993) crop-years, plus the (Raleigh and Wallace 1965, Turner and in the sagebrush steppe, and with the months of July and August in 1992 and DelCurto 1991). Within the same period, exception of prairie Junegrass, one or 1993, and mean monthly accumulations (n calf gains may range from 0.7 to as little another may dominate the herbaceous = 41) for the Northern Great Basin as 0.1 kg day' (Turner and DelCurto layer depending on site specific conditions Experimental Range near Burns, Ore. 1991). Mineral supplementation has not

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 175 Table 1. P-values derived from a split-split-plot analyses of variance for mineral content of 7 grasses commonly found on northern Great Basin rangelands near Burns, Ore. Main effects were species of forage (N = 7), years (1992 and 1993), and months (Apr.-Nov.). Shaded values are not statistically significant (P> 0.05).

Mineral SOURCE DF P K CA MG MN FE ZN NA

Species (S) 6 0.001 0.001 0.001 0.001 0.001 0.001 0.019

Year (Y) 1 0.003 0.045 0.001 0.001 0.001 0.001 0.001 0.001 0.001 .zpJ#3i`ii Y x S 6 0.001: 0.018 0.002 0.006 0.003 s

Month (M) 7 0.001 0.001 0.001 0.001 0.004 0.001 0.001 0.001 0.001

MxS 42 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.002 0.036

Y x M 7 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001

Y x M x S 42 0.009 0.001 0.001 0.001 0.010 0.002 0.046 was 97 mm exceeding the long term mean sodium (Na), and iron (Fe). Assays were species x year x month interactions. of 27.2 mm by nearly 70 mm. Mean performed by The Research Extension Because all year main effects and 21 of 27 April-July temperatures were 2.5° C Analytical Laboratory at The Ohio State interactions involving year effects (Table 1) warmer than average (x = 12.4° C) in University using Inductive Coupled were significant P < 0.05), data were sorted 1992 and 1.6° cooler than average in Plasma (ICP) spectroscopy (Vela et al. by year, and each year analyzed separately 1993. A model based on precipitation 1993, Sirois et al. 1994). The specific employing a split-plot analysis of variance accumulations for predicting annual instrument was an Applied Research with species serving as whole plots and herbage yields in the region furnished pro- Laboratory ICP Model 137'. Plant samples months as subplots. Mean separations duction estimates of 542 kg ha' for 1992 were dry-asked, treated with HNO3, dis- between adjacent months within a species and 1,257 kg ha' for 1993 (Sneva 1982). solved in HCL, and aspirated via an argon and year were obtained with Fisher's pro- For the most part the grasses exerted few carrier into the ICP plasma chamber where tected LSD procedures (Fisher 1966) with reproductive stems in 1992 and exhibited ionized elements were quantified by emis- statistical significance accepted at P < a wealth of reproductive effort in 1993. sion spectroscopy (AOAC 1990). 0.05. All "+" symbols within this manu- At each site, samples were harvested To establish a reference point for the script are associated with standard errors from at least 6 plants per species by clip- various minerals, the required levels in derived from 6 observations. ping to a 2.5-cm stubble and compositing forages for beef cattle were extracted from materials by species. Greater numbers of NRC (1996) requirements. The reference plants were used for small stature grasses animal was a 5-yr old 454 kg Hereford x Results and Discussion like Sandberg's bluegrass. Plants at each Angus cross cow, trailing its third 120-day site were sampled as they were encoun- old calf, 60 days post breeding, with a Statistical analyses tered along a pace 5, transect until adequate body condition score of and consuming With our initial split-split-plot analyses amounts of material were obtained. Each 2.5% (11.4 kg dry matter basis) of body of variance, the 3-way year x month x site experienced light (< 40% utilization) weight per day. species interaction was significant (P < summer/fall grazing by cattle, but only Experimental design was a randomized 0.05) for 7 of 9 elements (exceptions were ungrazed plants were included in our col- complete block with 6 replications (sites) Zn and Na (Table 1)). That being the case, lections. Samples harvested before spring and 3 factors (years (n months (n = 2), = the 3 main effects did not function inde- growth consisted of leaves and culms pro- and species (n 8), of forage = 7)). Initial pendently, and a majority of the data must duced in the previous growing season. analyses a employed split-split-plot analy- be presented in a 3-way format. To briefly After growth initiated in sis the spring, of variance with species as whole- communicate the relative amounts of vari- crowns of the caespitose grasses were plots, years as subplots, and months as ation associated with our initial split-split- lightly crushed and the brittle and broken sub-subplots (Petersen 1985). The replica- plot analyses, component variances were old-growth brushed aside before samples tion x species error term (30 df) tested for totaled across all 9 elements. Year, month, were collected. Plant materials were stored species effects. The replication x year x and species main effects respectively in labeled, paper bags in the field and species error term (35 df) tested the main accounted for 42, 29, and 9% of the total transported to Eastern Oregon Agricultural effect of years and year x species interac- variation (data not shown). Among the 2- Research Center headquarters where they tion. The species x x year replication x way interactions year x species constituted were oven-dried at 60° C for 48 hours, month error term df) tested main (490 for 2.1% month x species 1.8%, and year x ground to pass a 1-mm screen, and stored effects months, x of species month, and month 9.6%. The 3-way year x month x in plastic bags at room temperature for species interaction contributed 0.6%. subsequent chemical assays. Samples were Among error terms, number 1 (rep x analyzed for calcium (Ca), phosphorus 'Tradenames are supplied for information only and species) contributed 1.0%, number 2 (P), magnesium (Mg), potassium (K), cop- do not constitute endorsement by USDA-ARS of any (species (rep x year)) contributed 0.7%, per (Cu), zinc (Zn), manganese (Mn), product to the exclusion of others that may be suitable. and error term 3 approximately 0.4%.

176 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 2. P-values derived from separate split-plot analyses of variance for 1992 and 1993 mineral content of 7 grasses found on northern Great Basin rangelands near Burns, Ore. Main effects are species of forage (N = 7) and months (Apr.-Nov.). Shaded values are not statistically significant (P> 0.05).

1992

Minerals SOURCE DF P K CA MG MN FE CU NA

Species (S) 6 0.004 0.001 0.001 0.001 0.001 0.001 0.001 0.014

Month (M) 7 0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001

M x S 42 0.001 0.001 0.001 0.001 0.001 0.001 0.001

1993

Minerals SOURCE DF P K CA MG MN FE CU ZN NA

Species (S) 6 0.001 0.001 0.001 0.001 0.001 0.002 0.001 0.001 0.012

Month (M) 7 0.001 0.001 0.001 0.001 0.002 0.001 0.001 0.001 0.001

M x S 42 0.001 0.001 0.001 0.001 0.001 0.003 0.001 0.001

Replications or sites accounted for 3.7% Calcium ditions is frequently seen and attributed to of the total variation. Among grasses and for the duration of accumulation of more carbon under opti- When year effects were removed from the study, mean Ca concentration was 2.99 mal conditions. Conversely, growth the model and each year analyzed separate- g kg' (± 0.08). Given that all interactions restriction during drought is accompanied ly, 15 of 18 month x species interactions were significant (P < 0.05) among all by mineral concentration. With minimal (Table 2) were significant (P < 0.01). That analyses, there was great variability selective grazing among these forages, being the case, species did not react simi- among forages and months and between cattle could have easily ingested sufficient larly as we progressed from month to years (Fig. 2). With the exception of the Ca during the full 8-month sampling peri- month within years and again data must be winter annual, cheatgrass, most of the od of 1992. By selecting for cheatgrass presented for each species at monthly inter- grasses began the growing season with a early in the growing season and bluebunch vals for the majority of the data. Two of the Ca content just at or below 3.2 g kg', the wheatgrass later in the year, cattle could 3 month x species exceptions involved Na concentration necessary to meet NRC have potentially acquired sufficient Ca for (P > 0.13) in both 1992 and 1993. Main (1996) requirements for beef cattle. 7 of the 8 months sampled in 1993. All of effects of species and month were signifi- Sandberg's bluegrass, however, never the grasses, however, were Ca deficient cant (P < 0.02), however, suggesting that exceeded the recommended Ca level for for cattle during the late June-July inter- while Na concentrations differed among the cattle in any month of either year. val of 1993. Typically though, unless cat- grasses, they responded similarly as we Sandberg's bluegrass responded to tle are heavily lactating or foraging on progressed from month to month. The third November precipitation in 1992 with ele- rapidly growing herbage on acid, sandy, or exception to a significant (P = 0.23) month vated (P < 0.05) Ca levels that approached organic soils, clinical signs of Ca deficien- x species interaction was Zn in 1992. the required level, and cheatgrass respond- cies are rare among grazing beef cattle Species effects were not significant (P = ed with significant increases (P < 0.05) to (Underwood 1981). 0.32) either in that analysis suggesting that November rains in both years. Among the monthly means compiled across species medium and large stature perennial grass- Magnesium would adequately describe Zn dynamics of es (bottlebrush squirreltail, bluebunch Pooled across years, months, and for- 1 the grasses in 1992. Both main effects and wheatgrass, Idaho fescue, Thurber's ages mean Mg content was 0.94 g kg (± the month x species interaction were signif- needlegrass and giant wildrye), however, 0.006), and the concentration necessary to icant (P < 0.01) for Zn in 1993, however. Ca content typically peaked well above meet NRC (1996) requirements was 1,15 g When components of these analyses of recommended levels in late-July and then kg'. Like Ca, all interactions were signifi- variance were totaled across years and min- declined as these grasses matured and cant (P < 0.01) among all 3 analyses of erals, month effects accounted for 46% of entered quiescence. None of these 5 grass- variance (Tables 1 and 2) again indicative the total variation, species of grass about es, however, exhibited significant increas- that annual and seasonal Mg contents of 31%, replications or locations about 12.5%, es in Ca levels with the arrival of fall the grasses were quite variable among the species x month interaction 4.9%, error (October/November) moisture. year, month, and species combinations. term number 1 about 3.8%, and error term A general pattern evident among all the Cheatgrass exceeded the required level two, 1,9%. Again, given the preponderance grasses was that Ca levels were typically of Mg from early May to late June of both of significant 2 and 3-way interactions in higher (P < 0.05) during the dry 1992 years (Fig. 3). Idaho fescue also equaled our analyses, data will be presented at growing season than the more moist 1993 or exceeded the Mg requirements of cattle monthly resolutions for each year for each months. The dilution of mineral concen- in the early growing season, and both bot- species of grass. trations with more favorable growing con- tlebrush squirreltail and giant wildrye fur-

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 177 5 . 0 5.0 4.5 4.5 1 Sandberg's 1992 4.0 - 1993 4.0 bluegrass 3.5 i Y 3.5 3.0 D1 DC 3.0

2.5 1 tQ CS 2.5

U 2.0 ai U 2.0 1.5 10 Apt May Jun Jut Aug Sep Oct Nov Sec Apt May Jun Jut Aug Sep Oct Nov Sec Apt May Jun Jut Aug Sep Oct Nov Sec Apt May Jun Jut Aug Sep Oct Nov Sec 5.0 1.8 4.5 4.0 . 1.5 3.5 Y 1.2 DC 3.0 DC 2.5 0.9 CS DC U 2.0 Bluebunch -o-- 1992 0.0 15 wheatgrass -- 1993 10 10 Apt May Jun Jut Aug Sep Oct Nec Sec Apt May Jun Jut Aug Sep Oct Nov Sec Apt May Jun Jut Aug Sep Oct Nov Sec 5.0 4.5 4.0 YDC 3.5 DC 3.0 -0- 1992 a -..- 1993 CS 2.5

U 2 . 0 , Thurber ' s 1 . 5 0 T needlegra,ss r Apt May Jun Jut Aug Sep Oct Nov Sec Apt May Jun Jut Aug Sep Oct Nov Sec 5.0 1.8 Giant wildrye -- 1992 1993 . Lb T

0.0

0.3 1.0 a .- Apt May Jun Jut Aug Sep Oct Nov Sec Apt May Jun Jut Aug Sep Oct Nov Sec

Fig. 2. Mean calcium content (±SE bars) of 7 grasses sampled over 8 Fig. 3. Mean magnesium content (±SE bars) of 7 grasses sampled months at 6 sites in the sagebrush steppe near Burns, Ore. in 1992 over 8 months at 6 sites in the sagebrush steppe near Burns, Ore. and 1993. The dashed horizontal line denotes required Ca content of in 1992 and 1993. The dashed horizontal line denotes required forages for a 454 kg cow. Adjacent monthly means within a year Mg content forages for a 454 kg cow. Adjacent monthly means sharing a common letter are not significantly (P > 0.05) different. within a year sharing a common letter are not significantly (P> Fisher's protected LSD (P = 0.05) = 0.85 and 0.84 g kg'1, respective- 0.05) different. Fisher's protected LSD (P = 0.05) = 0.34 and 0.24 ly, for 1992 and 1993. g kg', respectively, for 1992 and 1993. nished adequate levels of Mg from July April into late June. Bottlebrush squir- 1992, cheatgrass sustained adequate levels into November of 1992. Both Sandberg's reltail, Idaho fescue, and giant wildrye fur- of phosphorus until mid-July, while all of bluegrass and cheatgrass responded to nished adequate Mg for the July-October the remaining grasses were largely defi- October rains with significant increases in period in 1992, but nearly all of the grass- cient by mid-June. Sandberg's bluegrass Mg that exceeded requirements of cattle in es were deficient beyond late-June for the and cheatgrass responded to October 1992 1992. Magnesium content of bluebunch 1993 growing season. precipitation with significant (P < 0.05) wheatgrass marginally approached the increases in P during November, but ade- required level for cattle for only 1-sam- quate P levels for cattle were not sustained pling period each year. Both bluebunch Phosphorus into December. The other medium stature wheatgrass and Thurber's needlegrass Mean P concentration among grasses grasses showed no significant (P > 0.05) (± were notable because neither species over the trials was 1.42 g kg' 0.09), responses to fall moisture in either year. exhibited a significant (P > 0.05) month to and the concentration necessary to meet On a worldwide basis, the most preva- month change in Mg content for either NRC (1996) requirements was 1.94 g kg 1. lent mineral deficiency among livestock is year. When pooled across species within Maximum and minimum values detected probably phosphorus (Underwood 1981). years, selective grazing by cattle could were 5.04 and 0.28 g kg', respectively. Phosphorus deficiencies are more promi- have furnished adequate Mg level for all While all interactions were again signifi- nent among tropical grasses than temper- cant (P < 3 but the last 15 days of November in 1992. 0.01) among all analyses of ate forages (McDowell et al. 1984), and In 1993, however, none of the grasses pro- variance (Tables 1 and 2), P did not exhib- the most devastating result of deficiency vided adequate Mg from mid-August it the same degree of year to year variabil- among cattle is reproductive failure 4) through late-November. Grass tetany gen- ity among species and months (Fig. as (McDowell and Valle 2000). Where seri- erally occurs during early spring, when Ca and Mg. The general pattern among the ous deficiencies occur, lactating cows may grasses are exhibiting rapid vegetative grasses was that P levels equaled or not enter oestrus until they cease milking growth (McDowell and Valle 2000), and exceeded cattle requirements early in the or are supplemented with phosphorus lactation demands of cattle are peaking. growing season and rapidly declined to (Lammond, 1970). Olson (1971) noted With the exception of cheatgrass, which inadequate levels by late July or early increased gains by cattle on southern was found primarily on disturbed sites, August in both years. Phosphorus levels Idaho cheatgrass ranges when phosphorus (P and giant wildrye early in the 1993 grow- were consistently higher < 0.05) for the was supplemented. ing season, our grasses were marginally early portions of the 1993 growing season satisfactory or deficient for Mg from late- than for the early months of 1992. In

178 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 ; 5.0 40 1992 4.5 -«- 1992 1993 4.0 Sandberg's Sand berg's 3.5 bluegrass 1993 3.0 bluegrass 0) 2.5 2.0

1,5 1 1,0 5 0.5 iJ 0.0 0 Apr May Jun Jut Aug Sep Oot Nov Deo Apr May Jun Jut Aug Sep Oot Nov Do Apr May Jun Jul Aug Sep Oot Nov Deo

1992 4.0 Bluebunch 1993 3.5 C) 7i wheatgrass Y 3.0 O) 2.5 2.0 a 1.5 1.0 0.5 Apr May Jun Jut Aug Sep Oot Nov Dec Apr May Jun Jut Aug Sep Oot Nov Dec Apr May Jun

5.0 i 40 4.5 4.0 1 3.5 +I 3.0 1 2.5 2.0 1,5 1.0 5 0.5 Apr May Jun Jut Aug Sep 001 Nov Deo Aug Sep 00) Nov Apr May Jun Jut Aug Sep Oot Nov Deo Apr May Jun Jut Aug Sep Oot Nov Deo Apr May Jun Jut Do

5.0 1 40 4,5 35 4.0 30 3.5 C) Y 3.0 Y 25 C) 2.5 C 20 2.0 Y 15 a. 1.5 10 1.0 5 0.5 0.0 T Apr May Jun Jut Aug Sep Oot Nov Dec Apr May Jun Jut Aug Sep Oot Nov Do

Fig. 4. Mean phosphorus content (±SE bars) of 7 grasses sampled Fig. 5. Mean potassium content (±SE bars) of 7 grasses sampled over over 8 months at 6 sites in the sagebrush steppe near Burns, Ore. 8 months at 6 sites in the sagebrush steppe near Burns, Ore. in 1992 in 1992 and 1993. The dashed horizontal line denotes required P and 1993. The dashed horizontal line denotes required K content of content of forages for a 454 kg cow. Adjacent monthly means forages for a 454 kg cow. Adjacent monthly means within a year within a year sharing a common letter are not significantly (P> sharing a common letter are not significantly (P > 0.05) different. (P 1, 0.05) different. Fisher's protected LSD (P = 0.05) = 0.66 and 0.34 g Fisher's protected LSD = 0.05) = 3.77 and 2.56 g kg respective- kg"1, respectively, for 1992 and 1993. ly, for 1992 and 1993.

Potassium itation, and all of the grasses were K defi- tie interest, because their Cu content never Mean K content of all forages was 9.0 g cient by about mid- November. approached the NRC (1996) requirement kg' (± 0.74), and seasonal values ranged McDowell and Valle (2000) found few for cattle forages of 9.6 mg kg' (Fig. 6). from a high of 42.9 to a low of 1.39 g kg' reports of a K deficiency among ruminants Cheatgrass and giant wildrye consistently (Fig. 5). The concentration in forages foraging under natural conditions. began the growing seasons with higher needed to meet NRC (1996) requirements Potassium typically exists as a cellular levels of Cu than the other grasses, but all only 10 to 25% of for beef cattle was 5.76 g kg'. Again the constituent among animals, but it occurs at of the grasses provided significant 3-way year x month x species much higher levels in milk than sodium. the Cu required for cattle by about July or interaction implied that K concentrations That being the case, lactating cattle can August. Idaho fescue was notable in that did not respond similarly between years not conserve supplies as effectively as no significant (P > 0.05) month to month and that species were not similar across their nonlactating counterparts (Maynard changes in Cu content occurred in 1992 or months. In both years, all of the grasses and Loosli 1969). When deficiencies do 1993. Seasonal dynamics were not sub- began the growing season with adequate K occur, general symptoms include: slow stantial either for bottlebrush squirreltail, levels for beef cattle. Sandberg's bluegrass growth, reduced feed and water intake, bluebunch wheatgrass or Thurber's and cheatgrass were the first forages to lowered feed efficiency, weakness, ner- needlegrass. exhibit inadequate K concentrations for vous disorders, and degeneration of vital Availability of dietary copper to animals cattle becoming deficient by approximate- organs. Potassium deficiencies are usually is affected by interactions with Mo, S, and ly late-July in both years. Bluebunch not seen, however, because associated Fe (Suttle 1981), with excesses of either wheatgrass, bottlebrush squirreltail, Idaho nutrients are typically even more deficient retarding copper availability. A wide array fescue, and Thurber's needlegrass, howev- and their symptoms manifested much of symptoms accompany copper deficien- er, sustained adequate levels of K into sooner than those of potassium cies among cattle, and their diversity may mid-August or early September. Giant (McDowell and Valle 2000). be linked to complex interactions involv- wildrye was notable because it sustained ing other minerals. Some of the clinical adequate amounts of K until about mid- Copper signs include: bleaching of hair, nervous in calves whose dams October of both years. Both Sandberg's Copper concentration of the 7 grasses symptoms (ataxia) pregnancy, bluegrass and cheatgrass responded to averaged 1.75 mg kg' (± 0.11). While all experienced deficiency during October 1992 precipitation with increased interactions were significant for each lameness, and swelling of joints (Maynard levels (P < 0.05) of K. None of the other analysis of variance (P < 0.46), the sea- and Loosli 1969). Serum assays of beef forages displayed a response to fall precip- sonal dynamics of these grasses are of lit- cattle by Raleigh (1988) in south-east

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 179 30 30 I 25 Cheatgrass j[ 20 at E 15 NC 10 5 5 -1- 1992 --- 1993 Apt May Jun Jul Aug Sep Oat Not Dec Apr May Jun Jul Aug Sep Oct Nov Dec 30 30 B 1992 25 5 1993 20 4 Y at E 15 c 10 a N 5

Apr May Jun Jul Aug Sep Oat Nov Dec Apt May Jun Jul Aug Sep Oat Nov Dec Apt May Jun Jul Aug Sep Oat Nov Dec Apt May Jun Jul Aug Sep Oat Nov Dec 30 8 25 5 Y 20 Y m E 15 E c 1a 10 N N 5 5

Apt May Jun Jut Aug Sep Oat Nov Dec Apt May Jun Jut Aug Sep Oat Nov Dec Apt May Jun Jul Aug Sep Oat Nov Dec Apt May Jun Jul Aug SeP Oat Nov Dec

Y at E C N

Apr May Jun Jul Aug Sep Oat Nov Dec

Fig. 6. Mean copper content (±SE bars) of 7 grasses sampled over 8 Fig. 7. Mean zinc content (±SE bars) of 7 grasses sampled over 8 months at 6 sites in the sagebrush steppe near Burns, Ore. in months at 6 sites in the sagebrush steppe near Burns, Ore. in 1992 1992 and 1993. The required copper content of forages for a 454 and 1993. The required zinc content of forages for a 454 kg cow is kg cow is 9.6 mg kg 1. Adjacent monthly means within a year 28.8 mg kgl (dashed horizontal line). Adjacent monthly means sharing a common letter are not significantly (P > 0.05) different. within a year sharing a common letter are not significantly differ- Fisher's protected LSD (P = 0.05) = 1.19 and 0.73 mg kg'1, ent (P > 0.05). Fisher's protected LSD (P = 0.05) = 7.82 and 3.75 respectively, for 1992 and 1993. mg kg'', respectively, for 1992 and 1993.

Oregon revealed marginal Cu levels, and the grasses sampled met the NRC (1996) changes in Mn concentration were diffi- up until 1988 clinical symptoms had not required concentration of Zn for beef cat- cult to establish within each species, espe- been noted. Recently, however, some tle forage (28.8 mg kg') for any sampling cially in 1993 (Fig. 8). herds in southeast Oregon have developed period in either year. Fleming (1963) With the exception of the beginning of health and reproductive disorders attrib- noted that zinc content varies considerably the growing season (late-April), monthly uted to Cu deficiency. Consequently, among components of grass plants and Mn levels were generally higher for 1992 many producers have begun monitoring found leaf/stem/flower concentrations of than for 1993. The NRC (1996) require- the Cu status of their animals and become 20, 15, and 36 mg kg', respectively. Zinc ment for beef cattle was 38 mg kg', and more attentive to management of their deficiencies can cause parakeratosis mean Mn content of forages over the 2 mineral programs. (inflamed skin around nose and mouth), sampling seasons was 38.6 mg kg' (± stiffness of joints, alopecia, breaks in skin 1.3). Only cheatgrass contained adequate Zinc around the hoof, and retarded growth. concentrations of Mn early in the growing Mean Zn content of the grasses was 12.1 Deficiencies have been induced experi- season, and it exceeded beef cattle require- mg kg' (± 0.44). With our full ANOVA mentally in calves (Miller and Miller ments for all but the last months of both model, species effects (P = 0.09) and the 1960), and while no applied reports of Zn years. Other grasses, however, like blue- 3-way interaction (P = 0.26) were not sig- deficiencies have occurred in sheep or cat- bunch wheatgrass or bottlebrush squir- nificant (Table 1). When years were ana- tle (Maynard and Loosli 1969), Mayland reltail, consistently supported adequate lyzed separately, however, all 7 grasses et al. (1980) saw improved gains among levels of Mn late in the year, so cattle con- responded similarly as we advanced from calves supplemented with zinc in southern suming a diverse diet could probably month to month in 1992, but both main Idaho. ingest sufficient Mn on a season long effects (Table 2) and the month x species basis. Except for giant wildrye, which interaction were significant (P < 0.01) for Manganese only contained adequate Mn levels in 1993. In 1993, all 7 grasses exhibited a Manganese concentrations varied signif- early May of both years, the other perenni- nearly linear or curvilinear decline of Zn icantly (P < 0.01) between years, and for- al bunchgrasses displayed more than as the seasons progressed (Fig. 7). ages did not exhibit similar patterns as we required Mn levels from early to mid-June Sandberg's bluegrass was the only forage advanced from month to month (Table 1) through early December in 1992. to show a significant (P < 0.05) increase in either year. Variability among samples Sandberg'5 bluegrass, bluebunch wheat- of Zn with the advent of fall precipitation within months was high, however, and grass, Idaho fescue, and Thurber's needle- in 1993. Despite these dynamics, none of significant (P < 0.05) month to month grass were largely Mn deficient for beef

180 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 800 800 70 Sandberg's ,., 700 --o-- 1992 loo 60 'C) 000 -- 1993 bluegrass 000 C) 50 500 Y 500 Y C) w C) aD E 405 E aoa E a) 300 a) 300 30 U- c LL. 205 200 20 top too to pr Apr May Jun Jul Aug Sep Oct Nov Sec Apr Mey Jun Jul Aug Sep Oct Nov Sec Ape May Jun Jul Aug Sep Oct Nov Dec Apr May Jun Jul Aug Sep Oct Nov Dec 800 700 -o- 1992 Bluebunch 800 1993 wheatgrass YC) C)Y 'C) 500 Y C) C) C) E E 400 E 300 N a) C U- U- 2ao 100 a8 a 0 Ape May Jun Jul Aug Sep Oct Nov Dec Apr May Jun Jul Aug Sep Oct Nov Dec Ape May Jun Jul Aug Sep Oct Nov Dec

70 80 C) C) Y Y Y 50 a C) 40 E E E 30 a) a) C U- U- 20 -j'--- 1992 20 Thurber'`s -'-- 1993 10 to needlegrass Apt May Jun Jul Aug Sep Oct Nov Dec Apr May Jun Jul Aug Sep Oct Nov Dec Ape May Jun Jul Aug Sep Oct Nov Dec Ape May Jun Jul Aug Sep Oct Nov Dec

C)Y C) E C

Apr May Jun Jul Aug Sep Oct Nov Dec Apt May Jun Jul Aug Sep Oct Nov Sec (±SE bars) of 7 grasses sampled over 8 Fig. 8. Mean manganese content (±SE bars) of 7 grasses sampled Fig. 9. Mean iron content 6 sites in the sagebrush steppe near Burns, Ore. in 1992 over 8 months at 6 sites in the sagebrush steppe near Burns, Ore. months at horizontal line denotes required Fe content in 1992 and 1993. The dashed horizontal line denotes the required and 1993. The dashed kg cow. Adjacent monthly means within a Mn content of forages for a 454 kg cow. Adjacent monthly means of forages for a 454 significantly (P > 0.05) dif- within a year sharing a common letter are not significantly (P> year sharing a common letter are not protected LSD (P 0.05) =191.7 and 79.8 mg kg 1, 0.05) different. Fisher's protected LSD (P = 0.05) =16.5 and 13.3 ferent. Fisher's = respectively, for 1992 and 1993. mg kg 1, respectively, for 1992 and 1993.

cattle in the 1993 growing season and fall appears that above average June rainfall in Among animals, Na is found primarily months. Giant wildrye cycled between 1992 stimulated Fe uptake (P < 0.05) by in extracellular fluids. In conjunction with being adequate and marginally deficient in all of the grasses except Sandberg's blue- K and Cl, it assists with maintaining 1992 and was largely deficient for the bulk grass, while above average precipitation in osmotic pressure, acid-base equilibrium, of the 1993 sampling period. July 1993 had no significant (P > 0.05) nutrient passage into cells, and water effect on Fe uptake. metabolism in general (Maynard and Iron Loosli 1969). Animals have considerable Na, but again that luxu- With the exception of the year x species ability to conserve Sodium to lactating individuals interaction (P 0.186) in our initial analy- ry is not available = Sodium concentrations varied consider- (McDowell and Valle 2000) suffering ses, (Table 1) most components of our with substantial ably among the grasses a lack of salt in the diet. Prolonged analyses for Fe were significant (P < 0.01) (P from monthly differences between years < deficiencies cause loss of appetite, (Table 2). Over the trial, mean Fe content 0.01) (Table 1). While species and month- of the grasses was 194.4 mg kg', and each decreased growth or weight loss, unthrifty ly differences (P < 0.05) occurred within appearance and reduced milking of the grasses exceeded or equaled beef years, all the grasses responded similarly mg (McDowell and Valle 2000), but supple- cattle forage requirements of 48 kg' (P > 0.13) as we advanced from month to (Fig. 9). With the mental salt can also stimulate weight gains for all months sampled month (Table 2 and Fig. 10) within each beginning of the growing (McDowell et al. 1984) among animals exception of the year. With the exception of the beginning the grasses was that are not showing signs of deficiencies. season, Fe content of of the growing season, monthly concentra- (P in 1993. higher < 0.01) in 1992 than tions were typically higher for all the sampling period also exhibited The 1992 grasses in 1992 than in 1993. The NRC Management implications seasonal dynamics within each grass typically more recommended sodium content for beef cat- Cattle in the sagebrush/steppe than 1993 (Fig. and we have some sus- their annual rangeland 9), tle forages was 672 mg kg', and all of our derive 85 to 90% of picion that rainfall and associated soil con- 1978, forages were deficient throughout both diet from grasses (Vavra and Sneva tamination of samples may have elevated Vavra 1987). Seasonally, years. Mean Na content of forages for the McInnis and Fe concentrations of our July 1992 period spring/summer forbs may contribute as study was only 61.3 mg kg'. The highest (Mayland and Sneva 1983). In 1993, the much as 6-9% of the diet, and shrubs may Na content attained by any of the grasses only significant (P < 0.05) month to month account for 12-15% of intake during fall was Sandberg's bluegrass in late-October change in Fe concentration involved and winter (McInnis and Vavra 1987). of 1992, and even then, it averaged only decreasing levels in bottlebrush squir- Given the preponderance of grasses in cat- 177 mg kg' (Fig. 10). reltail between late-April and late-June. It tle diets, however, many of the patterns

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 181 190 190 crested wheatgrass at equivalent dates and 170 170 iII 150 150 i stages of morphological development 0) 130 I Y 130 1 0) o o among years, and our data also exhibit E 90 90 some substantial disparities within months z(0 70 70 50 50 and between years for each species. 30 30 ,o Calcium, Mg, and Mn concentrations fluc- Apt May Jun Jul Aup Sep Oct Nov Dec Apt May Jun Jut Aup Sep Oct Nov Dec 190 tuated greatly between years, and indeed 190 - h 170 Bluebunch 1992 some grasses supported deficient levels for 150 SB q u irrre ltail (50 wheatgrass 1993 130 1 sampling season and more than adequate G) 110 E 90 levels for beef cattle during a second. The CO 70 z 50 clearest and most radical example was 30 Mn 10 Thurber's needlegrass with Ca and Apr May Jun Jut Aup Sep Oct Nec Dec between 1992 and 1993 (Figs. 2 and 8). 180 Idaho fescue 170 1992 Thurber's Both minerals were more than adequate in 150 1993 needlegrass .0) 1992 and at deficient concentrations for all Y 130 O) 1,0 of the 1993 sampling period. Other grasses E 90 Co 70 like bottlebrush squirreltail, bluebunch z 50 30 a 30 wheatgrass, and Idaho fescue approached to to Apr May Jun Jul Aup Sep Oct Nov Dec Apr this same pattern, but there were periods in both years when they contained deficient concentrations. Indeed plant mineral dynamics and animal mineral nutrition are complex issues, and our most prominent forages are clearly deficient for one or more minerals for much of the year. A few items, however, Fig. 10. Mean sodium content (±SE bars) of 7 grasses sampled over 8 months at 6 sites in the should be mentioned to illustrate that sagebrush steppe near Burns, Ore. in 1992 and 1993. The required sodium content of for- nutritive value, or perhaps more correctly, ages for a 454 kg cow is 672 mg kg'1. Adjacent monthly means within a year sharing a range quality, may be better than indicated common letter are not significantly (P > 0.05) different. Fisher's protected LSD (P 0.05) = by our data. First, we should point out that = 61.5 and 30.8 mg kg"1, respectively, for 1992 and 1993. our analyzed materials included whole plant samples taken above a 2.5-cm stub- and generalizations from this study are potentially lower Cu availability and exas- ble. Large herbivores typically harvest especially relevant to beef cattle manage- perate management problems associated diets of higher quality than hand-com- ment and animal performance. with copper deficiencies. pounded rations or whole-plant samples Of major interest were those minerals Also of interest was the year to year by selecting specific plant parts or por- that occurred at deficient levels among dynamics that occurred among minerals tions (Kiesling et al. 1969, Mclnnis and grasses on a year around basis. These within a species. With the exception of a Vavra 1987, Cruz and Ganskopp 1998). included Cu, Zn, and Na, and their defi- few early and late season reversals, a gross That being the case, adequate mineral con- ciencies should most definitely be given generalization was that mineral concentra- centrations in cattle diets probably extend some consideration by stockman (Figs. 6, tions were higher during the 1992 growing for longer periods of time than suggested 7, and 10). Seasonally deficient minerals season than for 1993. We did not quantify by our data. On the downside, however, included Ca, Mg, P, K, and Mn. Among leaf/stem/flower ratios of our samples, but the actual availability of forage minerals the caespitose grasses, Ca and Mn were our field notes clearly indicated that very to ruminants also fluctuates seasonally largely deficient for cattle early in the few grasses completed the reproductive (Peeler 1972). growing season with levels increasing as phases of phenology during the 1992 Second, cattle do forage from a variety the grasses advanced into summer (Figs. 2 growing season. Abundant early moisture of available forages. Early in the growing and 8). Magnesium patterns were less gen- in 1993, however, allowed great numbers season cattle may select up to 80% of their eralized (Fig. 4) because some grasses dis- of tillers to fully complete their reproduc- diet from a single grass, but their diets played increasing concentrations as the tive efforts before going dormant. That become more diverse as forages mature seasons progressed (giant wildrye and bot- being the case, the summer/fall 1992 sam- (Cruz and Ganskopp 1998). Thus a sub- tlebrush squirreltail), some remained rela- ples were largely vegetative in nature, stantial portion of their intake is derived tively stable through the seasons (blue- while the 1993 samples contained a wealth from other sources, and there may be forbs bunch wheatgrass and Thurber's needle- of reproductive stems. These reproductive and shrubs available that can also help rec- grass), and others declined (Sandberg's stems do add biomass, as 1993 standing tify deitary deficiencies. Clearly, our data bluegrass and cheatgrass). Phosphorus and crop was more than twice that of 1992 show a diversity of grasses can extend the K levels were typically adequate early in (Ganskopp and Bohnert 2001), but they period of adequate nutrition, and some of the growing season and declined to defi- dilute nutritive value (Ganskopp et al. the less desirable grasses like cheatgrass cient levels by July and August, respec- 1992, Mayland and Shewmaker 1997), and Sandberg's bluegrass do furnish tively (Figs. 4 and 5). Iron was of no con- because they are largely generated for excellent early forage and respond to cern, because levels were more than ade- structural support of reproductive parts. small amounts of precipitation with nutri- quate among all the grasses for all periods Angell et al. (1990) noted significant tious vegetative growth late in the season sampled. However, high levels of Fe could disparities in crude protein content of (Figs. 2, 3, and 8).

182 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Third, there may be natural licks or min- opment are arrested by drought. Literature Cited eral sources in the area. In some of our McDowell and Valle (2000) list several intensive, grazing behavior studies at the desirable characteristics of mineral supple- Allen, W.M. and P.R. Moore.1983. Parenteral Northern Great Basin Experimental Range ment formulations and also warn that min- methods of trace element supplementation, p. Ganskopp et al. (Cruz and Ganskopp 1998, eral excesses are capable of inducing other 87-92. In: N.F. Suttle, R.G. Gunn, W.M 1997) we observed instances where cattle deficiencies. When formulating mineral Allen, K.A. Linklater, and G. Weiner (eds.) ceased grazing and licked bare soil for up supplements for cattle pasturing in the Trace elements in animal production and to 1 minute. The nutritional value of these northern sagebrush steppe, we recommend Veterinary Practice. Occasional Pub. of the events, however, were not quantified. that 8 of the 9 minerals evaluated in this British Soc. of Anim. Prod., No 7, Edinburgh, Lastly, livestock have many mechanisms study be added to the mix. These included Scotland. for either conserving, recycling, mobilizing P, K, Zn, Mn, and Na. Angell, R.F., R.F. Miller, and M.R. Haferkamp. Ca, Mg, Cu, 1990. Variability of crude protein in crested or buffering mineral or nutrient balances Adequate concentrations of iron were within their systems, and these mecha- wheatgrass at defined stages of phenology. J. available on a year-round basis in all the Manage. 43:186-189. them to endure short term Range nisms allow grasses studied. AOAC. 1990. Official Methods of Analysis deficiencies without ill effects (Maynard (15th Ed.) Assoc. of Off. Anal. Chem, and Loosli 1969). Again though, lactating Washington D.C. animals can not avail themselves of many Conclusions Ares, R.N. 1953. Better cattle distribution of these mechanisms (Underwood 1981) through the use of meal-salt mix. J. Range when grazing deficient forages. Manage. 6:341-346. Addressing a variety of nutritional defi- Year-to-year, month-to-month, and Bailey, D.W. and G.R. Welling. 1999. ciencies is a vexing problem for produc- species specific patterns of mineral con- Modification of cattle grazing distribution molasses supplement. J. ers. In intensively managed pasture, many centrations were quite variable among the with dehydrated Range Manage. 52:575-582. mineral deficiencies can be rectified by Basin rangeland grasses 7 northern Great Cruz, R. and D. Ganskopp. 1998. Seasonal treating the land with a required element sampled. Minerals that were deficient on a preferences of steers for prominent northern or altering pH of the soil to enhance min- year-round basis included Cu, Zn, and Na, Great Basin grasses. J. Range Manage. eral availability for growing forages and these should of course be additives to 51:557-565. (MacPherson 2000). Other options include any rangeland supplement formulation for Daubenmire, R. 1970. Steppe vegetation of either oral treatments or injections for beef cattle. Minerals which were seasonal- Washington. Washington Agr. Exp. Sta., of Agr. Washington State Univ. Bull stock (Allen and Moore 1983). With largely deficient, and in some instances, defi- College 62. scale feedlot situations, frequent ration cient though out an entire growing season sampling and custom supplement formula- Fisher, R.A. 1966. The design of experiments. were: Ca, Mg, P, K, and Mn. Month to 8t' ed., Hafner, New York. tions may be mixed on even a daily basis month patterns among these found low Flemming, G.A. 1963. Distribution of major the dynamics of variable to accommodate levels of Ca, Mg, and Mn early in the and trace elements in some common pasture quality in feed supplies. growing season with concentrations species. J. Sci. Food and Agr. 14:203-208. In most extensive rangeland systems, increasing as forages matured and concen- Ganskopp, D. 1988. Defoliation of Thurber however, these solutions are economically needlegrass: herbage and root responses. J. trations declining as plants became weath- logistically impossible, and the Range Manage. 41:472-476. and/or ered and dormant. Phosphorus and K lev- only recourse is to supply free-access sup- Ganskopp, D. and D. Bohnert. 2001. els were elevated early in the growing sea- 7 Great plements in either block or loose form. Nutritional dynamics of northern son and declined to deficient levels by Basin grasses. J. Range Manage. 54:640-647. (2000) suggests these formu- MacPherson early July or August. Iron was the only Ganskopp, D., R. Angell, and J. Rose. 1992. have greatest efficacy if positioned lations mineral assayed with adequate year-round Response of cattle to cured reproductive near watering points with 1 station for concentrations in all forages. Contrary to stems in a caespitose grass. J. Range 25-40 cows. Conventional wisdom, Manage. 45:401-404. every intuitive thinking, mineral concentrations suggest though that rangeland Ganskopp, D., B. Myers, S. Lambert, and R. however, were generally higher among the grasses managers should try to disperse their cattle Cruz. 1997. Preferences and behavior of cat- when soil moisture levels were restricted 8 varieties of grasses. J. Range uniformly across pastures by positioning tle grazing and plants could not fully advance though Manage. 50:578-586. from the areas supplements some distance their reproductive stages of growth. Hickman, O.E. 1975. Seasonal trends in the their animals typically concentrate where Shallow rooted grasses like Sandberg's nutritive content of important range forage 1953, Bailey and Welling 1999). (Ares bluegrass, bottlebrush squirreltail, or the species near Silver Lake, Oregon. USDA of ascer- For. Serv. Res. Pap. PNW-187. Given the logistical demands winter annual cheatgrass can quickly taining forage nutritive value and supple- Hironaka, M., M.A. Fosberg, and A.H. respond to mid-summer or fall precipita- ment delivery in extensive pastures, ranch- Winward. 1983. Sagebrush-grass habitat tion and furnish additional high quality types of southern Idaho. Wildl. and Range ers for the most part can not respond to herbage. Cattle can possibly extend their Exp. Sta., Univ. of Idaho, Moscow. Bull 35. seasonal nutritional dynamics of their for- periods of adequate mineral nutrition by Kiesling, H.E., A.B. Nelson, and C. H. the best approach is to ages. Most likely, selectively grazing among mixtures of Herbel. 1969. Chemical composition of use a supplement formulation with the by hand-plucking and these grasses. We suggest, however, that a tobosa grass collected to rectify all known year-round and esophageal-fistulated steers. J. Range ability mineral supplement be available season- potential seasonal deficiencies of their for- Manage. 22:155-159. long on northern Great Basin rangelands ages. Based on our findings, mineral sup- Lammond, D.R. 1970. The influence of under- and that the formulation include at least nutrition on reproduction in the cow. Animal plementation is probably more of an issue Ca, Mg, P, K, Cu, Zn, Mn, and Na in Breeding Abstracts. 38: 354-372. during what we perceive as good forage available forms and proper ratios. years than when plant growth and devel-

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 183 MacPherson, A. 2000. Trace-mineral status of Miller, J.K., and W.J. Miller. 1960. Sirois, P.K., M.J. Reuter, C.M. Laughlin, forages, p. 345-371. In: D.I. Givens, E. Development of zinc deficiency in Holstein and P. J. Lockwood. 1994. A method for Owen, R.F.E. Axford, and H.M. Omed calves fed a purified diet. J. Dairy Sci. determining macro and micro elements in (eds.) Forage evaluation in ruminant nutri- 43:1854-1856. forages and feeds by inductively coupled tion. CABI Publishing. New York, N.Y. Murray, R. B., H.F. Mayland, and P.J. Van plasma atomic emission spectrometry. The Mayland, H.F. and G.E. Shewmaker.1997. Soest.1978. Growth and nutritional value to Spectroscopist 3:6-9. Seasonal changes in forage quality of C-3 cattle of grasses on cheatgrass range in Sneva, F.A. 1982. Relation of precipitation and grasses on sagebrush grasslands, p. 17-17.. southern Idaho. USDA For. Ser. Res. Pap. temperature with yield of herbaceous plants In: B.R. Christie (Chr.) XVII Grassld. Congr. INT-199. Intermountain For. and Range Exp. in eastern Oregon. Int. J. Biometeor. 8-18 June 1997. Winnepeg, Manitoba, and Sta., Ogden, Utah. 4:263-276. Saskatoon, Saskatchewan Canada. (CD- N.O.A.A. (National Oceanic and Suttle, N.F. 1981. Predicting the effects of Atmospheric ROM). Administration). 1991-1994. molybdenum and sulphur concentrations on Climatological data annual summary, Mayland, H.F. and F.A. Sneva. 1983. Effect Oregon the absorbability of copper in grass and for- 97-100:(13). of soil contamination on the mineral compo- age crops to ruminants. p. 545-548. In: NRC. (National Research Council). 1996. sition of forage fertilized with nitrogen. J. J.McC. Howell, J.M. Gawthorne, and C.L. Nutrients Requirements of Beef Cattle (7`h Range Manage. 36:286-288. White. (eds.) Trace elements in man and ani- Mayland, H.F., R.C. Rosenau, and A.R. Ed.). National Academy Press, Washington, D.C. mals - 4. Australian Acad. of Sci., Canberra, Florence. 1980. Grazing cow and calf Olson, T.E. 1971. Utilization and supplementa- Australia. responses to zinc supplementation. J. Anim. tion of cheatgrass in southern Idaho. M.S. Turner, H.A. and T. DelCurto. 1991. Sci. 51:966-974 Thesis, Univ. of Idaho. Moscow, Ida. Nutritional and managerial considerations for Maynard, L.A. and J.K. Loosli.1969. Animal Peeler, H.T. 1972. Biological availability of range beef cattle production. Beef Cattle nutrition, 6`h ed. McGraw-Hill Book Co., San nutrients in feeds: availability of major min- Nutr. 7:95-125. Francisco, Calif. eral ions. J. Anim. Sci. 35:695-712. Underwood, E. J. 1981. The mineral nutrition McDowell, L.R., G.L. Ellis, and J.H. Petersen, R.G. 1985. Design and analysis of of livestock. Commonwealth Agr. Bureaux, Conrad. 1984. Mineral supplementation for experiments. Marcel Dekker, Inc., New London, 180 p. grazing cattle in tropical regions. World York. Vavra, M. and F. Sneva. 1978. Seasonal diets Anim. Rev. 52:1-12. Raleigh, R.J. 1988. Joint HVDC Agricultural of five ungulates grazing the cold desert McDowell, L.R. and G. Valle. 2000. Major Study-Final report. Oregon State Univ., biome, p. 435-437. In: D.N. Hyder (ed.) minerals in forages, p. 373-397. In: D.I. Corvallis, Oregon and Dept. of Energy- Proc. First Internat. Rangeland Cong., Soc. Givens, E. Owen, R.F.E. Axford, and H.M. Bonneville Power Administration, Portland, Range Manage. Denver, Colo. Omed (eds.) Forage evaluation in ruminant Ore. Vela, N.P., L.K. Olson, and J.A. Caruso. nutrition. CABI Publishing. New York, N.Y. Raleigh, R.J. and J.D. Wallace. 1965. 1993. Elemental speciation with plasma mass McInnis, M.L. and M. Vavra. 1987. Dietary Nutritive value of range forage and its effect spectrometry. Analy. Chem. 65 (13) 585A- relationships among feral horses, cattle, and on animal performance. Research in beef cat- 597A. pronghorn in southeastern Oregon. J. Range tle nutrition and management. Special Report Manage. 40:60-66. 189. Agr. Exp. Sta., Ore. State Univ., Corvallis, Ore.

184 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 J. Range Manage. 56:185-192 March 2003 Vegetation dynamics from annually burning tallgrass prairie in different seasons

E. GENE TOWNE AND KEN E. KEMP

Authors are Research Associate, Division of Biology ([email protected]), and Professor, Department of Statistics, Kansas State University, Manhattan, Kans. 66506-4901

Abstract Resumen

Traditional perception of how tallgrass prairie responds to fire La percepcion tradicional de como la pradera de zacates altos at times other than late spring is either anecdotal or extrapolated responde al fuego en tiempos distintos a fines de primavera es from studies that lack spatial or temporal variability. Therefore, anecdotica o extrapolada de estudios que carecen de variabilidad we evaluated patterns of change in vegetation cover, species rich- espacial y temporal. Por to tanto, nosotros evaluamos los ness, diversity, and aboveground biomass production on 2 differ- patrones de cambio en la cobertura vegetal, riqueza de especies, ent topographic positions from ungrazed watersheds that were diversidad y produccion de biomasa en 2 posiciones topograficas burned annually for 8 years in either autumn (November), win- diferentes de cuencas hidrologicas sin apacentamiento que ter (February), or spring (April). Topoedaphic factors influenced fueron quemadas anualmente por 8 anos tanto en otono the response patterns of some species to seasonal fire, although (Noviembre), invierno (Febrero) o primavera (Abril). Los fac- differences were primarily in the rate of change. Annual burning tores topoedaficos influenciaron los patrones de respuesta de in autumn and winter produced similar trends through time for algunas especies al fuego estacional, aunque las diferencias most species. Big bluestem (Andropogon gerardii Vitman) cover fueron principalmente en la tasa de cambio. La quema anual en increased with all burn regimes, whereas indiangrass otono a invierno produjo tendencias similares a traves del tiempo [Sorghastrum nutans (L.) Nash] increased only with spring burn- para la mayoria de las especies. La cobertura de "Big bluestem" ing. Repeated autumn and winter burning eventually increased (Andropogon gerardii Vitman) se incremento con todos los perennial forb cover, with the largest increases occurring in regimenes de quema mientras que la del "Indiangrass" heath aster [Symphyotrichum ericoides (L.) Nesom], aromatic [Sorghastrum nutans (L.) Nash] se incremento solo con la quema aster [S. oblognifolium (Nutt.) Nesom], tall goldenrod (Solidago de primavera.. Las quemas repetidas en otono a invierno even- canadensis L.), and legumes. Species richness increased (P < tualmente aumentaron la cobertura de hierbas perennes y el 0.001) through time with spring and winter burning, but was mayor aumento ocurrio en "Heath aster" [Symphyotrichum eri- similar among all burn treatments after 8 years of annual fire. coides (L.) Nesom], "Aromatic aster" [S. oblognifolium (Nutt.) Average grass and forb biomass did not differ among burn sea- Nesom], "Tall goldenrod" (Solidago canadensis L.) y legumi- sons on either topographic position, although interannual bio- nosas. La riqueza de especies se incremento (P < 0.001) a traves mass production fluctuated inconsistently with time of burn. Our del tiempo con las quemas de primavera a invierno, pero despues findings contrast with many of the conventional views of how de 8 anos de quemas anuales fue similar en todos los tratamien- tallgrass prairie vegetation responds to seasonal fire and chal- tos de quema. La biomasa promedio de zacates y hierbas no lenges traditional recommendations that burning should only difirio entre las epocas de quema en cualquiera de las posiciones occur in late spring. topograficas, aunque la produccion interanual de biomasa fluc- tuo inconsistentemente con el tiempo de quema. Nuestros hallaz- gos contrastan con muchos de los puntos de vista convencionales de como la vegetacion de las praderas de zacates altos responde Key Words: burn season, fire ecology, grassland vegetation al fuego estacional y reta a las recomendaciones tradicionales de que la quema debe ocurrir solo a fines de primavera. Fire is an integral component of prairie development (Axelrod 1985), and for more than 7,000 years vegetation patterns have been influenced by anthropogenic burning practices (Sauer 1944, in Stewart 1951, Woodcock and Wells 1994, Kimmerer and Lake occur at any time of the year (Bragg 1982), intentional burning 2001). Although presettlement prairie fires potentially could autumn and late winter was a frequent ritual of most native American tribes (Catlin 1973, Pyne 1982, McClain and Elzinga 1994). Prairie fires were suppressed during European settlement, Earlier drafts of this manuscript were improved by comments from David and accidental or lightning-caused wildfires were the primary Hartnett and John Blair. We also thank Amanda Kuhl and the LTER clipping crew source of burning (Hoy 1989, McClain and Elzinga 1994). After for the biomass collections. Konza Prairie Biological Station is a preserve of The Nature Conservancy managed by the Division of Biology at Kansas State the influx of transient cattle to the Kansas Flint Hills in the late University. The National Science Foundation Long-Term Ecological Research 1800's, however, the incentive for prairie burning renewed and Program and the Kansas Agricultural Experiment Station provided support for this pastures were ignited annually in February or March to improve project. This paper is contribution No. 02-263-J from the Kansas Agricultural livestock gains (Kollmorgen and Simonett 1965, Isern 1985). Experiment Station. Manuscript accepted 2 Jul. 02. Traditional burn season shifted gradually to mid- or late-April,

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 185 because fire at that time favored the warm- grass prairie located in the Flint Hills of since 1972. Topographically, the water- season perennial grasses that are the main- northeastern Kansas (39°05' N, 96°35' W). sheds are comprised of upland plateaus, stay of livestock grazing (McMurphy and This site is the largest tract of taligrass rocky hillsides, and fertile lowlands. The Anderson 1965, Anderson et al. 1970). In prairie in the United States that is specifi- upland topographic positions are relatively addition, burning Kansas tallgrass prairie cally managed for ecological research. To shallow, silty clay loams overlying lime- at times other than late spring has been study how fire affects the structure and stone and shale layers (Udic Argiustolls, staunchly discouraged because of reputed function of grassland vegetation, Konza Florence series), whereas the lowland adverse effects on vegetation composition Prairie is parceled into 52 watersheds that positions are deeper colluvial and alluvial and productivity (Hanks and Anderson provide large replicated experimental units deposits (Pachic Argiustolls, Tully series). 1957, Anderson 1961, 1965, McMurphy subjected to different fire regimes. Seasonal burning began in November and Anderson 1963, 1965, Owensby and Vegetation is typical of native tallgrass 1993, when 2 watersheds were burned for Anderson 1967, Anderson et al. 1970). prairie and is dominated by warm-season the autumn treatment. Subsequent fire The impetus of fire research in tallgrass perennial grasses, primarily big bluestem treatments included 2 watersheds that prairie has focused on the vegetation (Andropogon gerardii Vitman), indian- were burned in February 1994 and 2 in responses imposed by fire frequency, with grass [Sorghastrum nutans (L.) Nash], and April 1994 for the winter and spring treat- the bums occurring in April (Abrams and little bluestem [Schizachyrium scoparium ments, respectively. The same 2 water- Hulbert 1987, Gibson and Hulbert 1987, (Michx.) Nash]. Forb species are wide- sheds were burned in the same season Gibson 1988, Collins 1992). Perceptions spread and constitute more than 75% of throughout the study. Average burn dates of how tallgrass prairie responds to the species richness (Towne 2002). for the 8-year period were 26 November, autumn or winter fires are derived either The climate for the area is characterized 17 February, and 24 April. All burns were from small plots (Aldous 1934, by hot summers, cold winters, and moder- conducted under conditions of moderate McMurphy and Anderson 1963, 1965, ately strong surface winds. Annual precip- wind speed and humidity, producing rela- Bragg 1982, Lovell et al. 1982, Towne itation averages 859 mm, with 75% of this tively intense head fires. and Owensby 1984), or from single burn occurring in the April to September grow- events (Penfound and Kelting 1950, ing season. Between 1994 and 2001, annu- Data Collection Kelting 1957, Adams and Anderson 1978, al precipitation exceeded the long term Species composition sampling began in Adams et al. 1982). Topographic position, average on 4 occasions, although rainfall 1994 after four, 50-m long transects, each soil texture, and climatic factors, however, during the growing season was above with 5 permanent plots, were established can affect how plants respond to fire average in only 3 years (Fig. 1). The aver- on both upland and lowland topographic (Abrams and Hulbert 1987, Gibson and age frost-free season lasts 180 days. positions in all watersheds (n = 20 plots Hulbert 1987), and documentation of spa- Six watersheds that have not been for each topographic position). The tial and temporal trends from repeated sea- grazed by cattle for more than 30 years canopy cover of every species in a 10-m2 sonal burning is lacking. Additionally, were selected for a long-term seasonal circular area around each plot was estimat- diversity indices and the response of most burning study. The watersheds ranged in ed and assigned to a percentage category subdominant species to seasonal fire are size from 11 to 39 ha and had been burned (Bailey and Poulton 1968). Cover of indi- anecdotal or speculative. Understanding previously every 3 or 4 years in the spring vidual species was determined by averag- the effects of seasonal burning on the ing the midpoint of the cover categories dynamics of tallgrass prairie plants is (i.e., 0.5, 3, 15, 37.5, 62.5, 85, and 97.5%) important in formulating rational manage- across the 20 plots for each topographic ment, and conventional generalizations of position. We also calculated frequency of how most species respond to season of fire occurrence (the proportion of plots where may be misleading (Engle and Bidwell an individual species occurred) as an alter- 2001). Thus, our objectives were to assess native indication of how species respond vegetation trends from an ongoing long- to seasonal fire. All plots were surveyed term study of annual burning in different each year in June and August. seasons. Specific questions considered Aboveground biomass production was were: (1) What species are differentially measured at the end of each growing sea- affected by repeated autumn, winter, and son by clipping 5 randomly selected spring fire, and does their response vary quadrats (0.1 m2) adjacent to each plant between topoeraphic sites? (2) How does composition transect (n = 20 plots per species richness and diversity change in topographic position). Vegetation in the response to annual burning in different sea- plots was clipped at ground level, separat- 0 sons? and (3) Are the purported adverse ed into graminoid, forb, and woody com- effects on biomass production from ponents, oven-dried at 60°C, and weighed. autumn and winter burning consistent 1994 1995 1996 1997 1998 1999 2000 2001 positions? Year across time and topographic Data Analysis Fig. 1. Total annual precipitation and grow- A total of 148 species were encountered ing season (Apr-Sep) precipitation (lower with > Materials and Methods hatched bar) for the years 1994-2001 at in this study, but only those species Konza Prairie. The solid horizontal line 2% mean canopy cover in any year or represents the 30 year annual precipita- treatment were analyzed individually. Study Area tion mean. The dashed horizontal line rep- Canopy cover of individual species and The study was conducted on Konza resents the 30 year growing season precip- the summed cover of species in similar Prairie Biological Station, a 3,487-ha tall- itation mean. taxonomic and life-form groups (e.g.,

186 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 sedges, legumes, annual forbs, and woody (a) Big bluestem--uplands species) were aresine square-root trans- Autumn 0 Winter A Spring formed and analyzed as a split-split plot over time. The full model contained terms for burn season, topographic position, year, and their interactions. To evaluate patterns of change through time from sea- -*=' . $-- sonal burning, we used the annual deviation from the mean year value for the 8- year study period as a linear covariate and 94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01 the deviation squared as a quadratic Year Year (c) 50 covariate. We initially fit a full covariance Indiangrass--uplands Indiangrass--lowlands a a model to each independent variable and 45 Autumn Winter Spring Autumn 0 Wine, 0 Spring 40 the nonsignificant (P > 0.05) effects were e /d deleted systematically. Inferences 35 / then 30 / trends through time were based regarding j 25 e 00 .tea on the respective regression coefficients. 20 0 If a species response to season of fire dif- 15 fered between topographic positions, the 10 a regression slopes were tested for each site 5 5 0 using appropriate contrasts. Otherwise, the 94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01 Year Year response was combined across topograph- (e) (t) ic positions and the slopes for each burn season compared. Linear and quadratic response curves were fitted using least square regression. Aboveground biomass was analyzed as a split-split plot with burn season as the whole plot factor, topographic position as the subplot factor, and year as the sub-subplot factor. The effect of burn season was r 94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01 tested using the variation between repli- Year Year cate watersheds (nested within burn seasons) as the whole plot error term; topographic effects were tested using the topo- Fig. 2. Canopy cover changes through time in response to annual autumn, winter, and spring b) on and lowland topographic locations, d) indian- graphic position x watershed (nested with- burning: (a, big bluestem upland (c, grass on upland and lowland topographic locations, (e) little bluestem on upland sites, and in burn season) mean square as the subplot (d) switchgrass on lowland sites. error term; and year effects were tested with the residual mean square. Species richness (the cumulative num- rate of change. In general, most species than on lowland positions (Fig. 2b). ber of plant species detected in the 20 that responded to time of burning followed Indiangrass increased only with spring plots for each topographic position) and gradual curvilinear shifts through time. burning, and exhibited the largest response the Shannon diversity index (H' = x In The quadratic downward trend exhibited of any warm-season grass to fire. by some species indicated sensitivity to However, it required 7 consecutive burns pi, where pi is the canopy cover of each species) were square-root transformed and fire, after which the population maintained before the upturn, with the greatest analyzed using the split-split plot over its presence in the community at a lower increase occurring on lowland topographic time covariance model. Frequency values level. In contrast, the concave temporal sites (Fig. 2c, 2d). Little bluestem cover were aresine square-root transformed and trend of some species indicated a negative increased on upland locations with autumn analyzed using the split-split plot model, short-term impact to fire, followed by a and winter burning, but did not change (P with burn season as the whole-plot factor period where the species recovered to pre- = 0.13) in response to spring burning (Fig. and topographic position and year as the existing levels. The canopy cover of most 2e). On lowland sites, little bluestem cover split factors. We used SAS procedures species, however, remained stable through remained stable through time with all burn (SAS Institute 1999) to analyze the data, time, suggesting tolerance to annual burn- treatments (Table 1). Switchgrass (Panicum with 0.05 as the probability level to estab- ing in any season. virgatum L.) increased in response to fire in lish statistical significance. any season on lowland sites (Fig. 2f); but Warm-season Grasses on upland sites, switchgrass cover did not Burning in autumn and winter produced change significantly through time with any Results and Discussion similar response patterns through time in burn treatment (Table 1). all warm-season grass species. Big The collective canopy cover of all warm-season grasses increased from burn- Topoedaphic factors influenced the bluestem cover increased in response to annual burning in any season, although the ing in any season, although the rate of response patterns of some species to change differed between topographic posi- repeated seasonal burning, although differ- changes through time were smaller on upland topographic positions (Fig. 2a) tions. On upland sites, the increases ences were primarily from variation in the through time were similar for all burn sea-

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 187 Table 1. Average percent cover of graminoid species after S years of annual burning in different erated or benefitted from annual burning. seasons on upland and lowland topographic positions. A positive superscript indicates cover Tall dropseed [Sporobolus composites increased (P < 0.05) from 1994. A negative superscript indicates cover declined significantly (Poir.) Men.] decreased in response to fire from 1994. in any season (Table 1). Cover of side-oats grama [Bouteloua curtipendula (Michx.) Lowlands Uplands also was reduced with spring burn- Species Autumn Winter Ton.] ing, but not from autumn or winter burning (Table 1). Andropogon gerardii 51.7+ 46.9+ Bouteloua curtipendula 3.5 2.6 Panicum virgatum 1.8 2.3 Cool-season Graminoids Schizachyrium scoparium 24.6+ 27.5+ Prairie junegrass [Koeleria macrantha Sorghastrum nutans 6.7 10.2 (Ledeb.) Schult.], a species predominant Sporobolus composites 0.8- 2.0- on upland sites, increased linearly in Sporobolus heterolepis 3.7+ 1.7+ response to autumn burning, but remained All warm-season grasses 93.1+ 93.7+ stable through time with winter and spring Carex spp. 17.5+ 17.1+ burning (Fig. 3c). All other cool-season Dichanthelium oligosanthes 2.9 8.3 graminoids declined significantly with Koeleria macrantha 9.3+ 2.7 0.1 repeated spring burning, whereas Poa pratensis 0.1- 0.1- Kentucky bluegrass (Poa pratensis L.) All cool-season graminoids 30.2+ 28.3+ was the only cool-season species that declined in response to autumn and winter 1). On upland sites, Not all burning (Table sons (Fig. 3a); but on lowland sites, spring warm-season grass cover (Fig. 3b). Scribner's panicum [Dichanthelium burning produced the greatest increase in warm-season grass species, however, tol-

(a) (b) (a) (b) 100 50 150 150 All perennial (orbs j 140 140 90 Autumn o Winter Spring 45 130 130 120 120 80 40 110 110 70 35 100 100 a 90 90 2 y 30 80 , p 0 80 (525 U 70 U 70 0 0 60 6` 60 20 50 50 e 15 40 40 30 30 10 20 20 5 10 10 _- -_ O T T T T 0 0 94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01 Year Year Year (c) (d) 25 Prairie junegrass--uplands 25 l Scribner's panicum--uplands Tall goldenrod--lowlands Autumn Winter A Spring Autumn ° Winter a Spring Autumn o Winter n Spring 20

15 15 as y 0 0 U U 40 10 0 /a \ 0 a' 5 5 e \ 0 5 0 - C , r 0 94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01 0 94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01 Year Year Year Year (e) (e)

0 94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 01 Year Year 94 95 96 97 98 99 00 01 94 95 96 97 98 99 00 Year Year

Fig. 3. Canopy cover changes through time in response to annual Fig. 4. Canopy cover changes through time in response to annual autumn, winter, and spring burning: (a, b) total cover of all warm- autumn, winter, and spring burning for various perennial forbs on season grass species on upland and lowland topographic locations, different topographic positions: (a) total cover of all perennial (c) prairie junegrass on upland sites, (d) Scribner's panicum on forbs for both topographic sites combined, (b) heath aster for both upland sites, (e) sedges for both topographic sites combined, and topographic sites combined, (c) aromatic aster on upland sites, (d) (f) total cover of all cool-season graminoids on upland sites. tall goldenrod on lowland sites, (d) total cover of all legume species on lowland sites, and (f) prairie lespedeza on lowland sites.

188 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 2. Average percent cover of various forb species after 8 years of annual burning in different aster [Symphyotrichum ericoides (L.) seasons on upland and lowland topographic positions. A positive superscript indicates cover Nesom], aromatic aster [S. oblongifolium increased (P < 0.05) from 1994. A negative superscript indicates cover declined significantly (Nutt.) Nesom], and tall goldenrod from 1994. (Solidago canadensis L.) (Figs. 4b-4d). Western ragweed (Ambrosia psilostachya Uplands Lowlands Species Autumn Winter DC), a dominant forb in tallgrass prairie, was not affected (P > 0.10) by time of burning on either topographic position psilostachya 2.0 1.3 Ambrosia (Table 2). However, dynamic interannual Artemisia ludoviciana 0.2- 0.7- 0.1- fluctuations in western ragweed cover sug- Brickellia eupatorioides 0.8 1.0 gest that factors other than season of burn Physalis pumila 0.5 0.4- Ruellia humilis 0.3- 0.4- influenced temporal patterns. Although Salvia azurea 5.8+ 4.0 0.1 0 burning in autumn or early-spring will Solidago canadensis 0 0 0 putatively increase forbs or "weedy" Solidago missouriensis 0.3 0.4- species (Anderson 1961, Anderson et al. Symphyotrichum ericoides 17.9+ 23.1+ 1970, Schwegman and McClain 1985), it Symphyotrichum oblongifolium 16.0+ 12.8+ 0 0 0 required repeated burning in autumn or All perennial forbs 51.9+ 51.8+ winter before forb canopy cover eventual- Amorpha canescens 2.8 0.4 ly increased. Dalea canadia 0.2 0.7 Total legume cover increased on low- Lespedeza capitata 0.2 0.7 land sites in response to burning in any Lespedeza violaceae < 0.1 < 0.1 0.1 season, but the greatest changes through All legume species 7.3 5.4 time occurred with autumn and winter All annual forbs 0.2 0.1 burning (Fig. 4e). Prairie lespedeza All woody species 0.2 0.2 [Lespedeza violaceae (L.) Pers.], a species occurring predominantly on lowland sites, exhibited the most prominent increase of oligosanthes (Nash) Gould], the most time with spring burning (Fig. 4a). Species all legumes to annual burning, and was the common cool-season grass on Konza that were primarily responsible for the only forb species that increased with Prairie, declined and then recovered to increase in forb cover from repeated spring burning (Fig. 4f). Cover of lead- beginning levels in response to winter autumn and winter burning were heath plant (Amorpha canescens Pursh), the burning (Fig. 3d); but on lowland sites, the temporal patterns remained stable with Table 3. Average percent frequency of species that changed significantly after 8 years of annual both winter and autumn burning (Table 1). burning in different seasons on upland and lowland topographic positions (10-m2 plots; n = 40). Sedges [primarily Carex mops Bailey, C. A positive superscript indicates the frequency of occurrence increased (P < 0.05) from 1994. A 1994. meadii Dewey, C. brevior (Dewey) Mack., negative superscript indicates the frequency declined significantly from and Cyperus lupulinus (Spreng.) Marcks] also declined initially in response to both Uplands Lowlands Autumn Winter autumn and winter burning before diverg- Species ------(%)------ing upward and attaining the highest cover Grasses: ------(%)------values after 8 years of fire (Fig. 3e). The Bouteloua curtipendula 77.5 97.5+ collective cover of all cool-season Bouteloua gracilis 22.5+ 30.0 0 0 0 graminoids followed concave patterns Dichanthelium acuminatum 5.0 0 7.5 through time with autumn and winter Dichanthelium oligosanthes 100 100 Eragrostis spectabilis 7.5 30.0 7.5 burning. Although the trends were similar Schizachyrium scoparium 87.5 95.0 on both topographic positions, cool-season Sporobolus compositus 85.0 95.0+ eventually surpassed ini- graminoid cover Poa pratensis 15.0- 32.5- tial values on upland sites (Fig. 30, but only recovered to the original levels on Forbs: Asclepias verticillata 32.5 lowland sites (Table 1). The transitory Asclepias viridis 20.0- 35.0 of most cool-season graminoid decline Artemisia ludoviciana 35.0 67.5 species in response to autumn and winter 2.5 0 0 burning coincided with 3 consecutive Dalea candida 27.5+ 37.5+ years of below normal precipitation during Dalea purpurea 65.0+ 52.5+ 0 the growing season, suggesting that mois- Lespedeza capitata 32.5+ 32.5+ ture availability may be crucial in mediat- Lespedeza violacea 2.5 2.5 2.5 ing their response patterns to seasonal fire. Oxalis stricta 15.0 2.5 0 2.5 2.5 Perennial Forbs Physalis pumila 47.5 42.5 The combined cover of all perennial Ruellia humilis 67.5+ 72.5+ forbs responded with similar curvilinear Solidago canadensis 0 0 0 upward trends in response to autumn and Solidago missouriensis 27.5 47.5 winter burning, but did not change through Vernonia baldwinii 55.0 67.5

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 189 most common legume species on Konza (a) Prairie, did not change significantly on either topographic position after 8 years of burning in any season (Table 2). Densities of many legume species are higher in tallgrass prairie burned annually in the spring than in prairie that is not burned (Towne and Knapp 1996). Trends in canopy cover, however, suggest that most legume species tolerate annual spring fire rather than benefit directly from it. Tr - Annual Species 94 95 96 97 98 99 00 01 Annual forb species were too sparse to Year Year analyze individually, but the combined (c) canopy cover of all annual forbs did not change through time with any burn treatment (Table 2). Collectively, annual forbs averaged 0.6% cover the first 2 years of burning, after which levels dropped and stabilized in all burn treatments. In contrast, extremely low levels of 2 annual grass species [common witchgrass (Panicum capillare L.) and little barley (Hordeum pusillum Nutt.)] were detected 0.4 in the first 2 years of the study, but subse- 0.0 quently disappeared and never reappeared 94 95 96 97 98 99 00 01 Year in any bum treatment. Year Annual plants are potentially susceptible to fire, but some forb species [e.g., daisy Fig 5. Changes in diversity indices through time in response to annual autumn, winter, and b) fleabane (Erigeron strigosus Muhl. ex spring burning on different topographic positions: (a, species richness for upland and lowland topographic positions, d) the Shannon diversity index for upland and lowland Willd.) and grooved flax (Linum sulcatum (c, topographic positions. Riddell)] persisted under all burn regimes. A few annual species [e.g., annual ragweed (Ambrosia artemisiifolia L.), snow-on-the- grass prairie, woody species are controlled ruellia (Ruellia humilis)] increased signifi- mountain (Euphorbia marginata Pursh), with fire (Adams et al. 1982, Hulbert cantly in response to different burning common pepperweed (Lepidium densiflo- 1986), and the lack of a significant change regimes without an accompanying change rum Schrad.), and smooth-seed wildbean of woody cover in this study is likely due in canopy cover. This suggests that bum- (Strophostyles leiosperma (Tory. & A. to low initial levels. Although annual fire ing may be important in the colonization Gray) Piper)] appeared sporadically in suppresses canopy cover of woody species of these species, but their density or some years. However, most other annual by removing accumulated top growth, 8 stature is sufficiently low that changes in forb species [e.g., rough false-penny-royal years of burning in autumn, winter, or canopy cover are not detectable. (Hedeoma hispida Pursh), prickly lettuce spring did not eliminate any shrub species. (Lactuca serriola L.), red-seed plantain Diversity Indices (Plantago rhodosperma Decne.), clasping Frequency of Occurrence Species richness increased in response Venus'-looking-glass [Triodanis perfoliata Depending upon topographic position, to spring and winter burning, but declined (L.) Nieuwl.], and field pansy (Viola bicol- burning in any season increased the fre- and then recovered to initial levels with or Pursh)] disappeared after the second quencies of 4 species [prairie lespedeza, autumn burning on both topographic posi- burn and never reappeared, indicating round-head lespedeza (Lespedeza capitata tions (Fig. 5a, 5b). After 8 years of annual intolerance to fire in any season. Michx.), white prairie-clover (Dalea can- fire, however, the number of species on dida Willd.), and purple prairie-clover either topographic position was similar for Woody Species (Dalea purpurea Vent.)] (Table 3). Eight all burn seasons. In tallgrass prairie, Woody shrubs [e.g., rough-leaf dogwood years of spring burning increased the fre- species richness declines as fire frequency (Corpus drummondii Mey.), smooth sumac quency of occurrence in 15 species, com- increases (Collins et al. 1995). The trends (Rhus glabra L.), New Jersey tea pared with 10 species increasing from in species richness observed in this study, (Ceanothus herbaceous Raf.), buckbrush winter burning and 9 species increasing in however, indicated the eventual downturn (Symphoricarpos orbiculatus Moench), and response to autumn burning. requires more than 8 consecutive burns. Arkansas rose (Rosa arkansana Porter)] Changes in the frequency of occurrence The effect of seasonal burning on the also occurred too sporadically to analyze were not always associated with concomi- Shannon diversity index varied with topo- individually. Average canopy cover of all tant changes in canopy cover. For exam- graphic position. On upland sites, diversity woody species (excluding leadplant) did not ple, the frequency of some species [white declined progressively in response to change significantly through time in any prairie-clover, purple prairie-clover, annual spring burning, and declined but bum treatment (Table 2). In ungrazed tall- round-head lespedeza, and fringe-leaf then recovered with both autumn and win-

190 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 tions, particularly on the xeric uplands. (a) (b) paradigms of reduced grass 700 700 Consequently, ---Autumn -Winter -E Spring -U-- -& e - Autumn Winter - - Spring production from autumn or winter burning 600 600 may be anomalous events from inoppor- tune precipitation patterns, site-specific 500 500 occurrences, artifacts from inadequate N 400 N 400 sampling, or confounded with livestock E E grazing; but they are not axiomatic for the 300 300 Kansas Flint Hills.

200 200

100 100 Summary and Implications

0 0 94 95 96 97 98 99 00 94 95 96 97 98 99 00 The changes in vegetation due to repeated seasonal burning documented in this (c) study occurred in ungrazed prairie where fire uniformly consumes the area. Response patterns may differ in grazed prairie because grazing produces a patchy burned landscape that creates numerous protective niches for species sensitive to fire. Thus, grazing can interact with seasonal burning to increase species richness and diversity (Coppedge et al. 1998). The mosaic burn patterns in grazed prairie will additionally buffer trends of many species

0 to seasonal fire. Objectives for utilizing 94 95 96 97 98 99 00 94 95 96 97 98 99 00 as a tool may Year Year fire season management vary between grazed and ungrazed prairie, but annual burning of ungrazed prairie at Fig. 6. Changes in biomass production in response to annual autumn, winter, and spring times other than late spring is apparently a burning on different topographic positions: (a, b) grass biomass on upland and lowland sustainable option that does not degrade sites, (c, d) forb biomass on upland and lowland sites. Inset in each graph represents the 7 the integrity of tallgrass prairie. (± year mean SE). Our findings contrast with many of the conventional views of how tallgrass ter burning (Fig. 5c). On lowland sites, among burn seasons, although production prairie vegetation responds to seasonal fire however, diversity decreased in response was usually lowest in response to spring and challenges traditional recommenda- to both spring and autumn burning, and burning on both topographic positions tions that burning should only occur in late did not change through time with winter (Figs. 6c and 6d). Woody biomass aver- spring. Based on these data, current deci- burning (Fig 5d). aged < 2 g/m2 on uplands and < 7 g/m2 on sions on managing tallgrass prairie that is lowlands, and did not differ (P > 0.78) burned at times other than late spring Biomass among burn treatments. needs to be objectively reevaluated. The effect of bum season on grass bio- Interactions among burn seasons, topo- Opposition to autumn, winter, or early mass varied inconsistently through time. graphic positions, and years suggest that spring burning is primarily an indoctrinat- Compared to spring burning, autumn fire biomass production was likely mediated ed tenet from anti-burn campaigns in earli- never reduced (P > 0.10) grass biomass on by soil moisture availability. Burning taller decades (Hoy and Isern 1995) and infer- upland sites (Fig. 6a), and winter burning grass prairie during winter or early spring ences extrapolated from other ecosystems reduced grass production only once (24% is traditionally denounced because bare (Wright and Bailey 1980). In addition, fire in 1999). In contrast, autumn and winter ground that is exposed for extended peri- season is often mistakenly blamed for the burning increased grass biomass 22% and ods could potentially increase surface adverse effects from concentrated live- 28%, respectively above spring burning on runoff and evaporation losses, thereby stock grazing in pastures that have been upland locations in 1995. On lowland lowering soil moisture and subsequent partially burned by wildfires (Engle and sites, spring burning produced significant- biomass production (Hanks and Anderson Bidwell 2001). Tallgrass prairie is resilient ly higher grass biomass than autumn burn- 1957, Bieber and Anderson 1961, to change, and although cover of some ing only in 1998, and produced more bio- Anderson 1965, McMurphy and Anderson indigenous perennial forb species eventu- mass than winter burning only in 1998 and 1965, Owensby and Anderson 1967). ally increased in response to autumn and 1999 (Fig. 6b). Average grass production Precipitation during the growing season winter burning, that effect required repeat- did not differ (P > 0.80) among burn sea- was below normal in 5 years of this study, ed fire and did not come at the expense of sons on either uplands (Fig. 6a inset) or and if early-season burning unequivocally warm-season grasses. lowlands (Fig. 6b inset). Average forb bio- reduced grass biomass, it should have mass also was not different (P > 0.30) been apparent under these droughty condi-

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 191 Literature Cited Hanks, R. J. and K. L. Anderson. 1957. Towne, G. and C. Owensby.1984. Long-term Pasture burning and moisture conservation. J. effects of annual burning at different dates in Soil and Water Conserv.12:228-229. ungrazed Kansas tallgrass prairie. J. Range Abrams, M. D. and L. C. Hulbert. 1987. Hoy, J. 1989. Controlled pasture burning in the Manage. 37:392-397. Effect of topographic position and fire on folklife of the Kansas flint hills. Great Plains Woodcock, D. W. and P. V. Wells. 1994. The species composition in tallgrass prairie in Quart. 9:231-238. burning of the new world: the extent and sig- northeast Kansas. Amer. Midi. Natur. Hoy, J. F. and T. D. Isern. 1995. Bluestem nificance of broadcast burning by early 117:442-445. and tussock: fire and pastoralism in the flint humans. Chemosphere 29:935-948. Adams, D. E. and R. C. Anderson. 1978. The hills of Kansas and the tussock grasslands of Wright, H. A. and A. W. Bailey. 1980. Fire response of a central Oklahoma grassland to New Zealand. Great Plains Quart. ecology and prescribed burning in the Great burning. Southwest. Natur. 23:623-632. 15:169-184. Plains-a research review. USDA General Adams, D. E., R. C. Anderson, and S. L. Hulbert, L. C. 1986. Fire effects on tallgrass Tech. Rep. INT-77. Collins. 1982. Differential response of prairie. Proc. Ninth No. Amer. Prairie Conf. woody and herbaceous species to summer 9:138-142. and winter burning in an Oklahoma grass- Isern, T. D. 1985. Farmers, ranchers, and land. Southwest. Natur. 27:55-61. stockmen of the flint hills. Western Aldous, A. E. 1934. Effect of burning on Historical Quart. 16:253-264. Kansas bluestem pastures. Kansas Agr. Exp. Kelting, R. W. 1957. Winter burning in central Sta. Bull. 38. Oklahoma grassland. Ecol. 38:520-522. Anderson, K. L. 1961. Burning bluestem Kimmerer, R. W. and F. K. Lake. 2001. The ranges. Crops and Soils 13:13-14. role of indigenous burning in land manage- Anderson, K. L. 1965. Time of burning as it ment. J. Forestry 99:36-41. affects soil moisture in an ordinary upland Kollmorgen, W. M. and D. S. Simonett. bluestem prairie in the flint hills. J. Range 1965. Grazing operations in the flint hills- Manage. 18:311-316. bluestem pastures of Chase County, Kansas. Anderson, K. L., E. F. Smith, and C. E. Ann. Assoc. Amer. Geog. 55:260-290. Owensby. 1970. Burning bluestem range. J. Lovell, D. L., R. A. Henderson, and E. A. Range Manage. 23:81-92. Howell. 1982. The response of forb species Axelrod, D. I. 1985. Rise of the grassland to seasonal timing of prescribed burns in biome, central North America. Bot. Rev. remnant Wisconsin prairies. Proc. Eighth No. 51:163-201. Amer. Prairie Conf. 8:11-15. Bailey, A. W. and C. E. Poulton.1968. Plant McClain, W. E. and S. L. Elzinga.1994. The communities and environmental interrela- occurrence of prairie and forest fires in tionships in a portion of the Tillamook Burn, Illinois and other midwestern states, 1679 to northwestern Oregon. Ecol. 49:1-13. 1854. Erigenia 13:79-90. Bieber, G. L. and K. L. Anderson. 1961. Soil McMurphy, W. E. and K. L. Anderson. moisture in bluestem grassland following 1963. Burning bluestem range--forage yields. burning. J. Soil and Water Conserv. Trans. Kansas Acad. Sci. 66: 49-51. 16:186-187. McMurphy, W. E. and K. L. Anderson. Bragg, T. B. 1982. Seasonal variations in fuel 1965. Burning flint hills range. J. Range and fuel consumption by fires in a bluestem Manage. 18:265-269. prairie. Ecol. 63:7-11. Owensby, C. E. and K. L. Anderson. 1967. Catlin, G. 1973. Letters and notes on the man- Yield responses to time of burning in the ners, customs, and conditions of the North Kansas flint hills. J. Range Manage. American Indians. Dover Publications, Inc., 20:12-16. New York, N.Y. Penfound, W. T. and R. W. Kelting. 1950. Collins, S. L. 1992. Fire frequency and com- Some effects of winter burning on a moder- munity heterogeneity in tallgrass prairie veg- ately grazed pasture. Ecol. 31:554-560. etation. Ecol. 73:2001-2006. Pyne, S. J. 1982. Fire in America. A cultural Collins, S. L., S. M. Glenn, and D. J. Gibson. history of wildland and rural fire. Princeton 1995. Experimental analysis of intermediate University Press, Princeton, N.J. disturbance and initial floristic composition: SAS Institute. 1999. SAS system for windows, decoupling cause and effect. Ecol. version 8.01. SAS Institute, Cary, N.C. 76:486-492. Sauer, C. 0. 1944. A geographic sketch of Coppedge, B. R., D. M. Engle, C. S. Toepfer, early man in America. Geog. Rev. and J. H. Shaw. 1998. Effects of seasonal 34:529-573. fire, bison grazing and climatic variation on Schwegman, J. E. and W. E. McClain.1985. tallgrass prairie vegetation. Plant Ecol. Vegetative effects and management implica- 139:235-246. tions of a fall prescribed burn on an Illinois Engle, D. M. and T. G. Bidwell. 2001. hill prairie. Natural Areas J. 5:4-8. Viewpoint: the response of central North Stewart, 0. C. 1951. Burning and natural veg- American prairies to seasonal fire. J. Range etation in the United States. Geog. Rev. Manage. 54:2-10. 41:317-320. Gibson, D. J. 1988. Regeneration and fluctua- Towne, E. G. 2002. Vascular plants of Konza tion of tallgrass prairie vegetation in response Prairie Biological Station: an annotated to burning frequency. Bull. Torrey Bot. Club checklist of species in a Kansas tallgrass 115:1-12. prairie. Sida 20:269-294. Gibson, D. J. and L. C. Hulbert. 1987. Towne, E. G. and A. K. Knapp. 1996. Effects of fire, topography and year-to-year Biomass and density responses in tallgrass climatic variation on species composition in prairie legumes to annual fire and topograph- tallgrass prairie. Vegetatio 72:175-185. ic position. Amer. J. Bot. 83:175-179.

192 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 J. Range Manage. 56:193-197 March 2003 Prescribed fire effects on dalmation toadflax

JAMES S. JACOBS AND ROGER L. SHELEYI

'Authors are Research Assistant Professor and Associate Professor, Department of Land Resources and Environmental Sciences, Montana State University, Bozeman, Mont. 59717.

Abstract Resumen

Los fuegos prescritos son importantes para la restauracion de Prescribed fires are important for rangeland restoration and los pastizales afectan a la composicion de las comunidades de affect plant community composition and species interactions. y plantas la interaccion entre especies. Muchas comunidades Many rangeland plant communities have been, or are under the y vegetales de pastizal han sido invadidas o estan bajo la amenaza threat of noxious weed invasion, however there is little informa- de ser invadidas por malezas nocivas, sin embargo, hay poca tion on how fire effects weeds. Our objective was to determine informacion sobre como el fuego afecta las malezas. Nuestro the effects of prescribed rangeland fire on dalmatian toadflax objetivo fue determinar los efectos del fuego prescrito en el pasti- [Linaria dalmatica (L.) Miller] density, cover, biomass, and seed zal sobre la densidad, cobertura, biomasa produccion de semil- production. These plant characteristics, as well as density, cover, y la de "Dalmatian toadflax" [Linaria dalmatica (L.) Miller]. Estas and biomass of perennial grasses and forbs were measured with- caracteristicas de la planta, asi como la densidad, cobertura in burned and adjacent not-burned areas on 3 Artemisia tridenta- y biomasa de zacates perennes hierbas, se midieron dentro de ta/Agropyron spicatum habitat types in Montana. Areas were y areas quemadas adyacentes a areas no quemadas en 3 habitats burned in the spring and measured in the fall 1999. Comparisons del tipo Artemisia tridentata/Agropyron spicatum de Montana. of plant characteristics between the burned and not-burned sites Las areas fueron quemadas en primavera medidas en otoi o de were made using t-tests and non-parametric Wilcoxon Rank Sum y 1999. Las comparaciones de las caracteristicas de las plantas tests. After 1 growing season, fire did not affect density or cover entre los sitios quemados no quemados se realizaron mediante of dalmatian toadflax. Burning increased dalmatian toadflax bio- y pruebas de t la prueba no parametricas de la Suma del Rango mass per square meter at 2 sites, and per plant biomass at all 3 y de Wilcoxon. Despues de una estacion de crecimiento, el fuego no sites. Seed production of dalmatian toadflax was increased by afecto la densidad o cobertura del "Dalmatian toadflax". La cover at 1 site and increased fire at all 3 sites. Fire reduced forb quema incremento la biomasa por metro cuadrado de grass biomass at 2 sites. The increases in dalmatian toadflax bio- "Dalmatian toadflax" en 2 sitios la biomasa por planta en los 3 seed production suggest that fire used to restore y mass and sitios. La produccion de semilla de "Dalmatian toadflax" se healthy plant communities may increase dalmatian toadflax incremento con el fuego en todos los sitios. El fuego redujo la dominance. We recommend weed management procedures, such cobertura de hierbas en un sitio a incremento la biomasa de as herbicide control and seeding desirable species, be integrated zacates en 2 sitios. El incremento de biomasa produccion de with prescribed fire where dalmatian toadflax is present in the y semilla de "Dalmatian toadflax" sugieren que el fuego utilizado plant community. para restaurar comunidades vegetales saludables puede incre- mentar la dominancia de "Dalmatian toadflax". Nosotros recomendamos que donde el "Dalmatian toadflax" este presente Key words: Linarea dalmatica, noxious rangelands weeds, inte- en la comunidad vegetal se integren procedimientos de manejo grated weed management. de maleza, tales como control con herbicidas y siembras de especies deseables, con el fuego Dalmatian toadflax [Linaria dalmatica (L.) Miller] is a rhi- zomatous, perennial weed native to the Mediterranean region. It was brought to the west coast of North America as an ornamental toadflax about 1874 and has spread throughout the western states, British of its high seed production. Infestations of dalmatian Columbia and Alberta (Alex 1962). Because of high genetic vari- may reduce desirable plants causing loss of winter forage for ability, dalmatian toadflax is adapted to a wide variety of habitat wildlife (Lajeunesse 1999), reduce cattle-carrying capacity, and conditions but favors well-drained, relatively coarse-textured reduced appraised rangeland value (Lacey and Olsen 1991). soils (Lajeunesse 1999). Dalmatian toadflax is an aggressive Cattle occasionally browse flowering shoots, but usually avoid competitor because of early spring regeneration from vegetative this weed (Harris and Carter 1971). buds on root stocks and rhizomes buried in the soil, and because Fire is a natural disturbance of ecosystems that can change community structure and function, at least in the short-term (Ueckert et al. 1978, Sharrow and Wright 1977, Whittaker 1961). Funding for this study was provided by the USDI Bureau of Land Management Prescribed burning has been used to increase diversity, reduce and the Rocky Mountain Elk Foundation. Study sites were contributed by the shrub and tree encroachment, and control excess litter buildup on USD1 Bureau of Land Management and USDA Forest Service. Personnel and rangeland for centuries. In most cases, the target species for man- resources for conducting the burns were contributed by the USDI Bureau of Land Management and USDA Forest Service. agement with fire have been indigenous, including junipers Manuscript accepted 29 May 2002. (Juniperus) and pines (Pinaceae). During the last 50 years, many

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 193 North American rangelands have been Table 1. Monthly summary from 1998, 1999, Maiden-Lap-rock outcrop complex con- invaded by, or are under the threat of inva- and the 30-year average of precipitation sisting of loamy-skeletal, carbonatic Typic sion by non-indigenous species. There are (mm) for Elk Horn, Mont., the area of the 3 Calciborolls. The predominant grasses at sites. few studies, however, that quantify the Wood Gulch were bluebunch wheatgrass effects of fire on noxious rangeland (Agropyron spicatum (Prush) Scribn. & 1998 1999 30 Yr weeds, and no studies on fire effects on Average Smith) and Idaho fescue (Festuca ida- dalmatian toadflax. hoensis Elmer). Site 2 was in Upper Horse (mm) ------Alterations in community structure and Gulch (46° 12.140' N, 111 ° 55.408' W) at function caused by rangeland fire may January 6 4 an elevation of 1,825 m. Slope ranged February 2 4 9 affect invasibility of an ecosystem or the from 15 to 45% with an eastern aspect. ability of an existing invader to reach March 3 1 Upper Horse Gulch soils were Whitcow, April 12 14 dominant status. On Artemisia steppe Bouldery-Shawmut, very bouldery-rock May 34 39 communities, outcrop complex consisting of loamy- establishment of cheatgrass June 87 65 and (Bromus tectorum L.) communities can be July 32 9 skeletal, mixed Typic Argiborolls enhanced by fire. Increased build-up of the August 13 63 loamy-skeletal, carbonatic, frigid Calcic continuity of fine fuels from cheatgrass September 29 6 Ustochrepts. Grasses were predominantly causes self perpetuating, larger, and more October 9 1 bluebunch wheatgrass and prairie frequent fires that favors cheatgrass repro- November 1.6 6 Junegrass (Koeleria macrantha (Ledeb.) duction and suppresses native perennial December 12 4 Schultes). Site 3 was in Lower Horse ° grasses (Pellant 1990, Peters and Bunting Annual 256 214 Gulch (46° 13.348' N, 111 55.111' W) at 1994). A single low intensity fire an elevation of 1,635 m. Slope averaged increased the cover and density of diffuse 15% and the aspect was southeast. Soils (Pseudotsuga menziesii (Mirb.) Franco), (Centaurea diffusa Lam.) and spotted were Windham-Lap very cobbly loamy big sagebrush (Artemisia tridentata Nutt.) knapweeds (C. maculosa Lam.) in north- consisting of loamy-skeletal, carbonatic and rubber rabbitbrush (Chrysothamnus em Washington (Sheley and Roche 1982). Typic Calciborolls. Grasses were blue- nauseosus (Pallas ex Pursh) Britt). In Montana, there is evidence that spotted bunch wheatgrass, prairie Junegrass, blue Monthly summaries for precipitation in the knapweed increased sixfold within 2 years grama (Bouteloua gracilis (H.B.K.) Lag. area encompassing all 3 sites are summa- after a controlled fire on a forested site Ex Griffiths), and red threeawn (Aristida rized for the year before the burn (1998), (Rice and Sacco 1995). Nutrients released prupurea Nutt). Plant species present at the year of the burn (1999), and the 30 year from plant litter, and the disturbance creat- the time of sampling for the 3 sites are list- (1961-1990) average (Table 1). ed by fire may favor growth and reproduc- ed in Table 2. Site 1 was located in Wood Gulch (46° tion of weedy species with early germina- N, ° tion and rapid growth characteristics 13.002' 111 5.094' W ) at an elevation of 1,750 m. Slope ranged from 15 to 35% Experimental Design and Sampling (Sheley et al. 1998). The objectives of this with a southern aspect. Soils were The experimental design was a compar- study were to determine the short-term ison of 2 treatments, burned and not- effects of a single spring prescribed fire on the cover, density, biomass, and seed pro- Table 2. Plant species present (% of 40 plots) at Wood Gulch, Upper Horse Gulch and Lower duction of dalmatian toadflax. We hypoth- Horse Gulch sites at the time of sampling. esized that dalmatian toadflax cover, density, biomass and seed production would Plant species Wood Upper Horse Lower Horse increase after burning. ------(%of40 plots) ------Achillea millefolium (L.) 5 23 Agropyron spicatum (Pursh) Gould 97 100 Methods Antennaria parvifolia (Nutt.) 13 28 Arnica diversifolia (Greene) 0 8 0 Study Sites Aristida purpurea (Nutt.) 3 0 8 The study was conducted within an Artemesia tridentata (Nutt.) 5 20 3 Astragalus 0 5 5 Artemisia tridentata/Agropyron spicatum spp. Bouteloua gracilis (H.B.K.) Lag. 0 0 5 (Mueggler and Stewart 1980) habitat type Chrysothamnus viscidiflorus (Hook) 13 0 on 3 sites located in the foothills of the Cirsium undulatum (Nutt.) 0 3 3 Elkhorn Mountains, southeast of Boulder, Erigeron ochroleucus (Nutt.) 20 20 5 Mont. The area was managed for elk Festuca idahoensis (Elmer) 35 93 (Cervus elaphus) range and had not been 0 0 3 grazed by livestock for 10 years. Linaria dalmatica (L.) 40 100 Prescribed burns in the area were designed Linum lewisii (Pursh) 15 18 5 to improve wildlife habitat by creating a Opuntia polyacantha (Haw.) 0 0 8 diverse mosaic of open parks, and the pre- Oryzopsis hymenoides (Roem. and Schult.) 25 10 scription to burn was based on the pres- Phlox hoodii (Richardson) 10 40 ence of woody species. Targeted Poa secunda (Presl) 80 72 encroachment tree and shrub species Stipa comata (Trip. and Rupr.) 5 15 5 including limber pine (Pinus flexilis Taraxacum officinale (Weber) 5 8 James), Rocky Mountain juniper 3 0 (Juniperus scopulorum Sarg.), Douglas fir Tragopogon dubius (Stop.) 10 43

194 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 3. Prescribed burn conditions at the 3 sites. below the 30-year average and 28% lower in 1999, the year of the burns. Site Size Date of burn temp. Precipitation during June and August of length of spread 1999 were 29 and 79% above the 30-year (ha) (km hr) (°C) hr) average, for those months, respectively Wood Gulch 80 11/3199 Upper Horse Gulch 60 16/3/99 Density Lower Horse Gulch 40 16/3/99 Burning did not affect dalmatian toadflax density the season following burning at any of the 3 sites (P > 0.05, Table 4). burned. The bum treatments for the 3 sites Data Analysis Rosette and mature stem density responses are described in Table 3. Comparisons Sites differed in dalmatian toadflax similar were made by sampling 20 randomly density and were analyzed separately. were to total density and therefore only total density is presented. At Wood selected 0.44 m2 circular plots within the Differences between burned and non- Gulch, dalmatian toadflax was present in burns and 20 independent randomly burned treatments in density, cover, bio- 40% of the plots sampled with mean den- selected plots within immediately adjacent mass per square meter, per plant biomass, sity of 20 plants m 2 in the burned plots non-burned areas at each of the 3 sites. per plant seed production of dalmatian and 43 plants m 2 in the not-burned plots. Soil and species composition in the burned toadflax, and density, cover and biomass Dalmatian toadflax was present in 100% and non-burned areas were identical at of perennial grass and perennial forbs of the plots at Upper Horse Gulch and its each site. Sites were sampled from 29 other than dalmatian toadflax were deter- density was 141 and 154 plants m2 in the September through 8 October 1999. The mined using a 2 sample t-test procedure burned and not-burned plots, respectively. experiment was not replicated in time (SAS Institute Inc. 1990). An F' (folded) At Lower Horse Gulch, dalmatian toad- because of the aggressive herbicide con- statistic was calculated to test for equality flax was present in 60% of the plots sam- trol of the dalmatian toadflax, and there- of the 2 variances. When normality or pled with mean density of 30 plants m 2 in fore it was replicated in space by using 3 equality of variances were not reasonable the burned plots and 25 plants m 2 in the sites. All species were counted within (Prob. F < 0.05), data were analyzed using not-burned plots. There was no difference sample plots and the area of ground cov- the non-parametric Wilcoxon Rank Sum between the burn treatments in the density ered by species was visually estimated as a test (SAS Institute Inc. 1990). of perennial grass or perennial forbs percent of the total area. Each dalmatian (Table 4). toadflax rosette or flowering stem was counted as an individual. Each bunchgrass Results was counted as an individual. All plants Cover Burning did not affect dalmatian toad- within the circle were then clipped to Precipitation by species, oven dried (60°C) flax cover the season following burning at ground level The 30 year average annual precipita- to constant weight and weighed. Seed any of the 3 sites (P > 0.05, Table 4). At tion for the area was 297 mm (Table 1). Wood Gulch, dalmatian toadflax cover pods of dalmatian toadflax were separated Annual precipitation in 1998, the year material and seeds averaged 2.8% in the burned plots and from other vegetative before the prescribed burns was 14 % were extracted and counted. 2.2% in the not-burned plots. Dalmatian

Table 4. Means of density (m 2), cover (%), biomass (m2 and per plant for toadflax), and toadflax seed production (per plant) for dalmatian toadflax, perennial grass, and perennial forbs at Wood Gulch, Upper Horse Gulch and Lower Horse Gulch for the burn and not-burn treatments. Differences between burn and not-burn treatments are indicated by an asterisk before the means.

Site Variable Toadflax burn burn burn burn burn burn Wood Density 20 43.0 (plants m 2) Wood Cover (%) Wood Biomass (g m 2) 17.0 Wood Per plant biomass *4.2 (g plant') Wood Seed production (no. plant') Upper Horse Density Upper Horse Cover 7.8 Upper Horse Biomass *53.0 Upper Horse Per plant biomass *7.1 Upper Horse Seed production *1580 Lower Horse Density 30.0 Lower Horse Cover 5.7 Lower Horse Biomass *38.0 Lower Horse Per plant biomass *7.3 Lower Horse Seed production *1,328.0

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 195 toadflax cover at Upper Horse Gulch was plant at Wood Gulch, from 79 to 158 grass present to respond to increased nutri- 6.2 and 7.8% in the not-burned and burned seeds per plant at Upper Horse Gulch, and ents, or dalmatian toadflax was dense plots, respectively. Cover at Lower Horse from 29 to 1,328 seeds per plant at Lower enough to overwhelm the grass. Nitrate is Gulch averaged 2.7 and 5.7% in the not- Horse Gulch. mobile and rapidly used by plants and burned and burned plots, respectively. therefore high nitrate conditions may favor Burning did not affect perennial grass weeds with early and rapid growth charac- cover at any site. Perennial forb cover was Discussion teristics (Sheley et al. 1998) over native reduced by burning from 5.9% in the plots grasses that tend to have less aggressive that were not burned to 2.8 in the burned growth characteristics. It is not surprising that there was no at (Table 4). The more robust dalmatian toadflax plots Upper Horse Gulch short-term burning effect on dalmatian plants produced 2- to 10-fold more seeds toadflax, perennial grass, or perennial forb in the burned areas compared to areas not- Biomass density. The heat intensities of the burns at burned. While we did not detect increases The influence of fire on dalmatian toad- the 3 sites were considered normal for in dalmatian toadflax density in the season flax biomass per square meter varied spring fires by technicians conducting the following fire, increased seed production depending on site (Table 4). At Wood burns (USDA Forest service, personal that we can expect increases in Gulch where dalmatian toadflax was communication). Many bunchgrass suggests toadflax density and spread on these sites absent from 60% of the plots, the fire clumps and dalmatian toadflax rosettes in the future. Increases in the density of effect was not significant even though the dalmat- were still green after burning and other invasive non-indigenous species biomass was twice as high on plots that ian rhizomes were generally toadflax have been reported. Sheley and Roche burned (17 m 2) compared to those that below 10 mm of soil and not damaged by g (1982) found increased cover and density did not (8 m 2). On Upper Horse Gulch fire generated heat (Hayward 1938). It is g of diffuse and spotted knapweeds in north- where dalmatian toadflax occurred on all that the thermal death point for doubtful ern Washington, and Rice and Sacco plots, fire increased dalmatian toadflax plant tissue was reached during the fires 2 (1995) report a sixfold spotted knapweed biomass from 31 g m where there was no (Write 1970, Wright et al. 1976). 2 increase in Montana within 2 years after a fire to 53 g m where there was fire. Fire Maximum temperatures in shrubland controlled fire on a forested site. increased dalmatian toadflax biomass fires are always near the top of the vegeta- 2 from 8 to 38 g m on the not-burned plots tion (Bailey and Anderson 1980, compared to burned plots at Lower Horse Whittaker 1961), killing shrub and tree Gulch where dalmatian toadflax occurred species with perennating buds in this zone. Conclusions on 60% of the plots. The removal of big sagebrush and pine Fire increased dalmatian toadflax per species by the prescribed burns in this Our results suggest fire will not reduce plant biomass at all sites (Table 4). At study opened niches once occupied by dalmatian toadflax populations on range- Wood Gulch, dalmatian toadflax biomass these species and made available resources land and may increase the dominance of was 1.0 g per plant in the plots that were they used. The reduced competition for this weed. We recommend implementing not burned and 4.2 g per plant in the resources could partly explain the increase an integrated dalmatian toadflax manage- burned plots. Dalmatian toadflax increased in biomass of dalmatian toadflax and ment program when conducting a pre- from 1.1 to 1.7 g per plant in the not- perennial grass, and the increased seed scribed burn for restoring native plant burned and burned plots, respectively, at production of dalmatian toadflax in the communities and controlling encroaching Upper Horse Gulch. Dalmatian toadflax burned areas. We expect future increases undesirable plants. Integrated practices biomass in the not-burned plots at Lower of dalmatian toadflax because large areas could include herbicide application to Horse Gulch was 2.0 g per plant compared once occupied by trees and shrubs were reduce dalmatian toadflax density and seed to 7.3 g per plant in the burned plots. left open to invasion by high seed produc- production, and seeding with desirable Fire increased perennial grass biomass ing dalmatian toadflax especially on the plant species to fill niches opened by fire. at Wood and Lower Horse Gulches, but Upper Horse Gulch site where there were there was no difference at Upper Horse fewer perennial grasses to fill open niches. Gulch. Perennial grass biomass production On semi-arid rangelands, nutrients Literature Cited 2 at Wood Gulch was 40.8 g m where there bound in living and dead biomass are was no burning and 70.5 m 2 where it g made available for plant growth by fire Alex, J.F. 1962. The taxonomy, history, and was burned. Upper Horse Gulch perennial (DeBano and Conrad 1978). In addition, 2 distribution of Linaria dalmatica. Can. J. grass biomass was 28.9 and 48.4 g m in soil conditions after burning often favor Bot. 40:295-307. the burned compared to not-burned plots, nitrification (Sharrow and Wright 1977). Bailey, A.W. and M.L. Anderson. 1980. Fire respectively. At Lower Horse Gulch, Increases in dalmatian toadflax and grass temperatures in grass, shrub, and aspen forest perennial grass biomass increased from biomass in the burned areas may have communities of Central Alberta. J. Range 2 45.1 to 73.0 g m in the not-burned com- occurred because of increased nutrients. Manage. 3:37-40. pared to the burned treatments. Fire did not On a different burn under similar condi- DeBano, L.F. and C.E. Conrad. 1978. The affect perennial forb biomass (Table 4). tions, nitrate increased (P 0.002) from effect of fire on nutrients in a chaparral = ecosystem. Ecol. 59:489-497. 3.3 mg on plots that did not burn to kg' Harris, P. and A.C. Carter. 1971. Linaria vul- Dalmatian Toadflax Seed 7.6 mg kg' on plots that burned (Jacobs garis mill., yellow toadflax, and L. dalmatica Production unpublished data). This offers another (L.) Mill., broad-leaved toadflax Burning dramatically increased dalmat- explanation to the increase in dalmatian (Scrophulariaceae). In: Biological control ian toadflax seed production per plant at toadflax and grass biomass in the burned programmes against insects and weeds in Canada 1959-1968. Commonwealth Agr. all 3 sites (Table 4). Burning increased areas. The lack of grass response at Upper Bureaux, Slough, England. 94-97. seed production from 7 to 355 seeds per Horse Gulch suggests that there was less

196 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Hayward, F. 1938. Soil temperatures during Peters, E.F. and S.C. Bunting. 1994. Fire con- Ueckert, D.N., T.L. Whigham, and B.M. forest fires in longleaf pine region. J. For. ditions pre- and postoccurrence of annual Spears. 1978. Effect of burning on infiltra- 36:478-491. grasses on the Snake River Plain. In: S.B. tion, sediment, and other soil properties in a Lacey, J. and B. Olsen. 1991. Environmental Monson and S.G. Kitchen (eds.) Proc. mesquite-tobosagrass community. J. Range and economic impacts of noxious rangeland Ecology and management of annual range- Manage. 31:420-425. weeds. In: L.F. James, J.O. Evans, M.H. lands. USDA For. Ser. Gen. Tech. Rep. INT- Whittaker, E. 1961. Temperature in heath Ralphs, R.D. Child (eds.) Noxious GTR-313, 31-36. fires. J. Ecol. 49:709-715. Rangeland Weeds. Westview Press, Boulder, Rice, P. and R. Sacco. 1995. Appendix E. Write, H.A. 1970. A method to determine Colo. 5-16. Spotted knapweed response to Burning. heat-caused mortality in bunchgrass. Ecol. Lajeunesse, S. 1999. Dalmatian and Yellow Manuscript. Division of Biology, University 51:582-587. Toadflax. In. Sheley, R.L. and J.K. Petroff of Montana, Missoula. p. 100-102. Wright, H.A., S.C. Bunting, and L.F. (eds.). Biology and Management of Noxious SAS Institute Inc. 1990. SAS/STAT user's Neunschwander. 1976. Effect of fire on Rangeland Weeds. Oregon State Press, guide, ver 6. Vol. 2. Cary, NC. 920 p. honey mesquite. J. Range Manage. Corvallis, Ore.. p.249-260. Sharrow, S.H. and H.A. Wright. 1977. 29:467-471. Mueggler, W.F. and W.L. Stewart. 1980. Effects of fire, ash, and litter on soil nitrate, Grassland and shrubland habitat types of temperature, moisture and tobosagrass pro- Western Montana. Intermountain Forest and duction in the Rolling Plains. J. Range Range Experiment Station. USDA-FS. Manage. 30:266-270. Ogden, Utah. 154 pp. Sheley, R.L. and BF. Roche, Jr. 1982. Pellant, M. 1990. The cheatgrass-wildfire Rehabilitation of spotted knapweed infested cycle-are there any solutions? In: E.D. rangeland in northeastern Washington. Abstr. McArthur, E.M. Romney, S.D. Smith, and of papers, W. Soc. Weed Sci., Denver, Colo. P.T. Tueller (eds.) Proc. Symp. on cheatgrass Sheley, R.L., J.S. Jacobs, and M.L. invasion, shrub die-off, and other aspects of Carpinelli. 1998. Distribution, biology, and shrub biology and management. USDA For. management of diffuse (Centaurea diffusa) Ser. Gen. Tech. Rep. INT-276,11-18. and spotted knapweed (Centaurea maculosa). Weed Technol.12:353-362.

197 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 J. Range Manage. 56:198-202 March 2003 Overcoming dormancy in New Mexico mountain mahogany seed collections

LEE S. ROSNER, JOHN T. HARRINGTON, DAVID R. DREESEN, AND LEIGH MURRAY

Authors are Science Specialist, Department ofAgronomy and Horticulture, New Mexico State University, Las Cruces, N.M. 88003; Associate Professor, New Mexico State University Mora Research Station, Mora, N.M. 87732; Agronomist, USDA Natural Resource Conservation Service, Los Lunas, NM., 87031; and Professor, Department of Experimental Statistics, New Mexico State University, Las Cruces, N.M. 88003

Abstract Resumen

Mountain mahogany (Cercocarpus montanus Raf) is a useful El "Mountain mahogany" (Cercocarpus montanus Rat) es una reclamation species because it can occupy and improve poor especie util para la restauracion de pastizales porque puede ocu- soils. Literature regarding seed propagation of this species is var- par y mejorar suelos pobres. La literatura respecto a la propa- ied and often contradictory, recommending stratification dura- gacion de semilla de esta especie es variada y a menudo es con- tions of 14 to 90 days, and sulfuric acid scarification durations of tradictoria, recomendando periodos de estratificacion de 14 a 90 none to 60 minutes. To assess variability in propagation require- dias y tiempos de escarificacion con acido sulfiirico de nada a 60 ments among seed sources, 8 New Mexico seed sources were test- minutos. Para evaluar la variabilidad en requerimientos de ed with factorial combinations of scarification and stratification propagacion entre fuentes de semilla se evaluaron 8 fuentes de treatments. Sources were selected to encompass both a range of semilla provenientes de New Mexico en combinacion factorial de latitudes throughout New Mexico and a range of elevations at tratamientos de escarificacion y estratificacion. Las fuentes de Questa, N. M. Seeds were scarified 5 or 10 minutes in concentrat- semilla fueron seleccionadas para abarcar un rango de altitudes ed sulfuric acid, tumbled 5 or 10 days in course grit, or unscari- a traves de New Mexico y un rango de elevaciones en Questa, N. fied (control). Seeds underwent subsequent stratification for 0 M. Las semillas fueron escarificadas por 5 o 10 minutos en acido (control), 30, or 60 days. Averaged across scarification treat- sulfiirico concentrado, agitadas 5 o 10 dias en una cama de arena ments, the 2 southernmost sources lacked a stratification o sin escarificar (control). Las semillas recibieron una estratifi- requirement, while northern seed sources achieved their highest cacion subsecuente de 0 (control), 30 o 60 dias. Promediando a germination following the longest stratification duration (60 traves de los tratamientos de escarificacion, se detecto que las days). Improvement in germination due to stratification was dos fuentes de semilla de mas al sur no requirieron de estratifi- greatest for the 2 highest elevation Questa sources. Scarification cacion, mientras que las fuentes de semilla del norte alcanzaron treatments were less effective in improving germination than los mas altos porcentajes de germinacion con el tratamiento de stratification treatments, and produced more variable results. A estratificacion mas largo (60 dias). La mejoraa en la germinacion 5-minute soak in sulfuric acid was the most effective scarification debida a la estratificacion fue mayor para las 2 fuentes de semil- treatment, but for 2 sources, this treatment reduced germination. la de mayor altitud en Questa. Los tratamientos de escarificacion Variability in the stratification requirement appears to be an fueron menos efectivos en mejorar la germinacion que los adaptation to macroclimatic differences among seed sources, tratamientos de estratificacion y produjeron resultados mas whereas differential response to scarification may be a response variables. Un remojo de 5 minutos en acido sulfurico fue el to microclimatic differences. tratamiento de escarificacion mas efectivo, pero para en dos fuentes de semilla este tratamiento redujo la germinacion. La variabilidad en los requerimientos de estratificacion parece ser una adaptacion las diferencias macroclimaticas entre las fuentes Key Words: acid scarification, stratification, ecotypic variability, Cercocarpus montanus de semilla, mientras que la respuesta diferencial a la escarificacion puede ser una respuesta a diferencias microclimaticas.

Mountain mahogany (Cercocarpus montanus Rat) occupies sites that are dry, unstable, erosive, and of low fertility (Brotherson 1992), and is actinorhizal, forming a nitrogen-fixing Published literature recommends stratification at 0 to 5°C, but symbiosis with Frankia bacteria (Paschke 1997). These charac- recommended treatment durations vary from 14 to 90 days teristics make it an excellent shrub species for reclamation in the (Hervey 1955, Heit 1970, Smith 1971, Deitschman et al. 1974, western United States. Poor germinability has proven to be an Kitchen and Meyer 1990). Kitchen and Meyer (1990), studying obstacle to production of seedlings for revegetation, and propaga- Utah and Colorado seed sources, found the stratification require- tion literature is inconsistent and often contradictory. ment to vary from 42 to 84 days depending on source. Although the seed coat of mountain mahogany is permeable to water (Heit This research was funded, in part, through grants from McIntire Stennis, 1970), scarification treatments have been effective in promoting Molycorp Inc. of Questa, N.M., and the New Mexico Agriculture Experiment germination. Hervey (1955) reported that a 60-minute soak in Station. acid improved germination. Dreesen (unpublished data) found Manuscript accepted 12 Jun. 02 incubation in acid for as short as 5 minutes to result in complete

198 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 1. Sources of mountain mahogany seed collected in September and October 1997 and used in in the canister drum, so that each can germination studies. rotated 360 degrees for each 360-degree rotation of the drum. Source Latitude Location Before undergoing stratification treat- Capulin 36°42'N Molycorp Mind-Questa, N.M. m ment (or germination testing for stratifica- Blind Gulch 36°42'N Molycorp Mind-Questa, N.M. m tion-control seeds) all seeds were soaked Mahogany Hill 36°42'N Molycorp Mind-Questa, N.M. m for 1 minute in hydrogen peroxide (VWR Headframe Hill 36°42'N Molycorp Mind-Questa, N.M. m 3% Stabilized) for surface sterilization and Boxcar 36°42'N Molycorp Mind-Questa, N.M. m thoroughly rinsed under running tap water. Rociada 35°50'N Rociada, N.M. m This step was undertaken to minimize Sandia 35°10N Cibola National Forest, N.M. m potential differences in seed contamina- Sacramento 32°58'N Lincoln National Forest, N.M. m tion between acid-treated and non-acid- treated seeds, which could have biased the results. Hydrogen peroxide can affect ger- seed destruction. Heit (1970) found that a sources used in this study. As a result, the mination. However, when used as a seed 10-minute soak in acid increased germina- study consisted of 2 overlapping experi- pretreatment, much longer soak durations tion speed but not percentage, while a 20- ments generating 2 data sets. The first and/or higher concentrations are typically minute soak injured seeds. experiment tested a factorial combination used (Young and Evans 1981, Rosner Inconsistencies among propagation pro- of all 8 seed sources, 3 scarification treat- 2000). tocols for mountain mahogany may be ments (unscarified control, 5-minute sulfu- Seeds undergoing stratification were related, in part, to extensive ecotypic vari- ric acid soak, and 5-day dry-tumble treat- spread evenly over half of a 20 x 20 cm ability (Young et al. 1978). This study was ment), and 3 stratification lengths (0, 30, piece of cotton cloth that had been saturat- undertaken to assess variability in response or 60 days). The second experiment evalu- ed with distilled water. The cloth was to stratification and scarification treatments ated a factorial combination of 5 seed folded to enclose the seeds, placed within among 8 New Mexico collections of sources, 5 scarification treatments (unscar- 15 x 16 cm self-sealing poly bags, and mahogany selected across latitu- ified control, 5-minute sulfuric acid soak, mountain ml peat moss, which had 10-minute sulfuric acid soak, 5-day dry- covered with 200 dinal and elevational gradients. with distilled water and tumble treatment, and 10-day dry tumble been saturated squeezed by hand to remove excess water. treatment) and 3 stratification lengths (0, were placed in a walk-in cooler 30, or 60 days). Results for the second Poly bags Materials and Methods 5°C for the duration of experiment are presented only in regards maintained near stratification treatment. to the 2 additional scarification treatments. from 8 New germination Seeds were collected Each treatment combination was tested Seeds were tested for 1). Sources were cm filter papers (VWR Mexico sources (Table with four, 100-seed replications. Due to between two, 20.5 both a range of lati- Qualitative) saturated with dis- selected to encompass missing data, the Sacramento source was Grade 413 and a range sealed in 1- tudes throughout New Mexico not included in statistical comparisons. tilled water. Filter papers were Mine in Questa, poly bags, which were of elevations at Molycorp However, some germination data are pre- gallon self-sealing were collected from a mini- on greenhouse benches. Thermostat N.M. Seeds sented for reference. placed Seeds adjusted to mum of 5 plants at each source. Seeds undergoing acid scarification settings in the greenhouse were ripe when easily removed highs near 30°C and were considered were soaked in concentrated sulfuric acid maintain daytime from from the plant. Styles were separated (Reagent ACS, 95.0 to 98.0%, VWR), nighttime lows near 15°C. Although test- box, and seeds were over a wide range of con- seeds in a rubbing then thoroughly rinsed under running tap ing germination debris in a Dakota temperature regimes is necessary to then separated from water. Seeds underwent dry-tumble scari- trolled Cleaned seeds were stored at 5°C fully characterize dormancy, germination blower. fication mixed with 10 course grit (True start of the study. g in this manner to for over 2 years until the Square Metal Products) in a 4-ounce testing was conducted seeds retain viability of dormancy in At this temperature, round ointment tin (US Can), which was characterize the expression (Springfield 1973). greenhouse setting. Optimal ger- for at least 6 years placed within the canister drum of a No. a typical seed availability restricted the temperature for this species Limited 140 Model B High Speed Tumbler (Tru- mination acid and dry-tumble seed sources (Platt 1976, number of sulfuric square Metal Products) turning at approxi- varies among imposed on 3 of 8 and Bass 1973, scarification treatments mately 40 rpm. Tins were tightly packed Smith 1971, Smith

Table 2. Categorical analysis of variance tables for effect of scarification, stratification, seed source, and interactions of these factors on mountain mahogany germination.

8-Source Experiment' Component df Chi-Square Observed Experiment Observed Significance Significance

-Control, 5-minute acid soak, and 5-day Scare 2 scarification treatments (5). 4 dry-tumble scar fication treatments. Strat 2 2 6 Source Blind Gulch, Mahogany Hill, 4 Hill, Scar by Strat Hill, Boxcar, by Source 12 seed sources. Source and Sandia seed sources. Strat by Source 12 by Source 8 Scarby Strat by Source 24 121.3 <.0001 by Strat by Source

'Sacramento source was dropped from this analysis leaving 7 sources. 2Scar=Scarfcation. Strat=Stratification

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 199 Table 3. Mean germination percentages and standard errors for data averaged across 90 0 stratification and scarification (control, 5- o No Scarification minute acid soak, 5-day dry-tumble) treat- 80 ® 5-Minute Acid Soak ments for each mountain mahogany seed ® 10-Minute Acid Soak source. ® 5-Day Dry-Tumble 8 10-Day Dry-Tumble Seed Source Mean 0 0 Germination Error 0 D (%) 0 Capulin 19.6 0.7 C 40 Blind Gulch 21.6 0.7 0 Mahogany Hill 44.6 0.8 C 30 Headframe Hill 82.1 0.6 Boxcar 64.8 0.8 20 Rociada 61.9 0.8 10 o1) Sandia 53.8 0.8 C) Sacramento 28.8 0 0 0.8 D C Overall Capulin Blind Gulch Mahogany Headframe Boxcar Rociada Sandia (5-sources) Hill Hill Kitchen et al. 1989). Greenhouse temperatures during the germination test ranged Seed Source * from 28.2°C 0.5°C for mean daytime Fig. 2. Effect of scarification treatment and its interaction with seed source (averaged across highs to 13.9°C * 0.3°C for mean night- stratification treatments) on mountain mahogany germination. Means labeled with the time lows. Paper moisture level was main- same letter are not significantly different at a = (0.05)/3-where 3 treatments are com- tained by application of distilled water as pared and a = (0.05)/10-where 5 treatments are compared. needed. Seeds were checked for germination after 0, 7, 14, 21, and 28 days. Seeds data with equal variances, but germination homogeneity. The maximum-likelihood were considered germinated when the rad- percentage data has unequal variances technique was used to calculate X2 test ical had emerged through the seed coat. between treatments and is frequently statistics. Observed significance levels less Categorical analysis of variance (SAS skewed and, therefore, non-normal. than 0.05 were considered significant. Proc CATMOD, SAS Institute 1989) was Usually percentage data are aresine trans- Percentages and standard errors were cal- used to determine treatment differences in formed to achieve normality and then ana- culated for main effects and interaction germination using a factorial treatment lyzed by ANOVA. Until, the advent of combinations. Approximate pairwise Z- structure. Analysis was also run separately high-speed microcomputers, use of more statistics were used to conduct pairwise by seed source. (Traditionally, analysis of appropriate categorical models for analy- comparisons of main treatment effects variance [ANOVA] has been used to ana- sis of germination data was impractical.) using a conservative alpha value of 0.05 lyze germination data. The ANOVA Categorical analysis of variance is a gen- divided by the number of comparisons. assumes continuous, normally distributed eralization of the chi-square (X2) test of Results 100 = 0-Day Stratification 90 ® 30-Day Stratification In both the 8-source and the 5-source ® 60-Day Stratification experiments, scarification, stratification, 80 a seed source, and all interactions of factors 70 a affected germination (Table 2). Seed 60 source had the greatest effect on germination. Germination percentages were highly 501 variable among sources (Table 3) and 401 were improved by stratification (for data averaged over seed sources and scarifica- 30 H tion treatments). A 60-day treatment was

20 H most effective (Fig. 1). Differences in response to stratification among seed 10H sources (for data averaged across scarifi- 0 cation treatments) conformed to a latitudi- Overall Capulin Blind Mahogany Headframe Boxcar Rociada Sandia Sacramento nal gradient with germination improving Gulch Hill Hill in response to increasing stratification Seed Source duration only for the 6 northern sources (all except Sandia and Sacramento) (Fig. Fig. 1. Effect of stratification and its interaction with seed source (averaged across 3 scarifi- 1). Percentage improvement due to strati- cation treatments) on mountain mahogany germination. Pairwise comparisons were omit- fication was greatest for the 2 Questa seed ted for the Sacramento source because they would be biased by a missing treatment combi- sources from highest elevation (Capulin nation. Germination means labeled with the same letter are not significantly different at a and Blind Gulch) and least for the 3 south- = (0.05)/3. ernmost sources (Rociada, Sandia, and

200 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Table 4. Improvement in mountain mahogany germination (percentage and SE) by seed source due to stratification treatment, for data averaged across 3 scarification treatments (control, 5-minute acid soak, 5-day dry-tumble).

Stratification Duration Capulin Blind Mahogany Headframe Boxcar Rociada Sandia Sacramento Gulch Hill Hill (%) (%) (%) (%) (%) (%) (%) (%) 0 5.9 (0.7) 14.8 (1.0) 34.6 (1.4) 77.8 (1.2) 55.4 (1.4) 57.6 (1.4) 54.0 (1.4) 27.5 (1.6) 30 days 22.7 (1.2) 20.6 (1.7) 46.3 (1.4) 83.2 (1.1) 64.5 (1.4) 60.9 (1.4) 51.5 (1.4) 27.7 (1.3) 60 days 310.1(1.3) 29.4 (1.3) 52.9 (1.4) 85.3 (1.0) 74.4 (1.3) 67.1(1.4) 55.8 (1.4) 30.8 (1.3) Improvement following 410% 99% 53% 10% 34% 16% 3%1 12%2 60-day stratification relative to control Germination percentages were not significantly different. 2Significance of improvement not tested because of missing treatment combinations.

Sacramento) and one northern source only when no stratification was used. The fication requirements of mountain (Headframe Hill) (Table 4). interaction between stratification and scar- mahogany (Hervey 1955, Heit 1970, Scarification also affected germination ification also varied among seed sources Smith 1971, Deitschman et al. 1974, percentage. Although no scarification (Fig. 4). For 5 of 8 seed sources, benefit of Kitchen and Meyer 1990). treatment improved germination relative scarification was lost when stratification Kitchen and Meyer (1990) found vari- to control seeds for data averaged over was increased to 60 days. However, for ability in stratification requirements of 9 seed sources, some treatments were effec- the remaining 3 sources, a 5-minute acid Utah and Colorado mountain mahogany tive for individual sources (Fig. 2). Acid soak improved germination even when seed sources to be related to environmen- scarification improved germination for 4 combined with the longest (60 days) strati- tal factors such as winter precipitation and of 7 seed sources tested, but reduced ger- fication treatment. the probability of spring drought (but not mination for 2 seed sources. Dry-tumble elevation), in addition to winter tempera- scarification (averaged across stratifica- ture. These environmental factors may tion treatments) had little influence on Discussion also account for some of the variability germination, increasing germination for among New Mexico sources, where precipitation varies along latitudinal and ele- only 1 seed source. For the 5 sources eval- Overall germinability and response to increased vational gradients. The data in this study uated, as scarification duration scarification treatments was highly vari- suggest a latitudinal gradient in the degree from either a 5-minute to a 10-minute acid able among the New Mexico sources of 10-day tumble of stratification requirement between soak or from a 5-day to a mountain mahogany studied. Considerable germination was either northern and southern New Mexico seed scarification, total ecotypic differences exist within mountain sources and an elevational gradient in unaffected or reduced. mahogany species (Young et al. 1978). stratification requirement between upper Response to scarification varied across Ecotypic and environmental differences (Fig. 3). Averaged and lower elevational seed sources at a stratification treatments among populations interact to affect seed the best acid soak (5 single location. across seed sources, dormancy (Baskin and Baskin 1973) and (5-day) treat- Scarification treatments improved ger- minutes) and dry-tumble viability, and could account for the vari- germination appreciably mination to a limited extent in this study. ments improved ability in reported stratification and scari- As was the case in previous studies, scarification treatments were less effective than stratification in promoting germination 80 (Hervey 1955, Heit 1970, Smith 1971). No Scarification 5-Day Dry-Tumble 0 ® Scarification in concentrated sulfuric acid 5-Minute Acid Soak I 10-Day Dry-Tumble 70 H ® was a more effective treatment than dry- 10-Minute Acid Soak ® tumble scarification for 4 of 7 sources, but r 60 for 2 seed sources, all levels of acid scarification significantly reduced germination. 50 H Variability in reported optimal acid-soak duration in mountain mahogany (Hervey

40 HI 1955, Heit 1970, Dreesen, unpublished data) is likely the result of variability in

30 H seed coat thickness among seed lots. Seed coat thickness is a major factor related to 20 both the requirement for and the sensitivity to scarification treatments. 10- Although acid scarification is often employed to improve seed coat permeability or soften restrictive seed coats, neither of these dormancy mechanisms is docu- 0 30 60 mented to occur in mountain mahogany Fig. 3. Effect of interaction between stratification and scarification (averaged across seed (Heft 1970). However, mountain mahogany source) on mountain mahogany germination for the 5-source experiment. seeds contain a water-extractable inhibitor

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 201 Brotherson, J.D. 1992. Mineral nutrient con- 90 Capulin Blind Gulch 80 centrations in the mountain mahogany 70 species Cercocarpus montanus and 60 Cercocarpus intricatus and in their associat- 50 ed soils. J. Plant Nutr. 15(1):49-67. A.P. 40 ti Deitschman, G.H., K.R. Jorgenson, and 30 Plummer. 1974. Cercocarpus H.B.K. p. 20 309-312. In: C.S. Schopmeyer (coord.) Seeds 10 of Woody Plants in the United States. Agr. 0 Handb. No. 450. USDA, U.S. Government Printing Office, Washington, D.C. 90 - Mahogany Hill Heit, C.E. 1970. Germination characteristics 80 and optimum testing methods for twelve 70 western shrub species. Proc. Assoc. Official so Seed Analysts. 60:197-205. 50 Hervey, D.F. 1955. Factors which influence 40 the reseeding of certain browse species in 30 Colorado. Ph.D. Diss., A&M College of 20 Texas. College Station, Tex. 10 Kitchen, S.G. and S.E. Meyer. 1990. Seed 0 dormancy in two species of mountain mahogany (Cercocarpus ledifolius and Cercocarpus montanus). In: K.L. Johnson (ed.) Proc. Fifth Utah Shrub Ecology Workshop. Utah State Univ., Logan, Ut. Kitchen, S.G., S.E. Meyer, G.R.Wilson, and R. Stevens. 1989. Addition of Cercocarpus montanus-true mountain-mahogany-to the rules. Assoc. Official Seed Analysts Newsletter 63:28-30. Moore, T.C. 1963. A germination inhibitor in achenes of Cercocarpus montanus. Ecol. 90 Sandia Sacramento 80 42(2):406-408. 70 Paschke, M.W. 1997. Actinorhizal plants in 60 rangelands of the western United States. J. 50 Range Manage. 50 (1):62-67. 40 Piatt, J.R. 1976. Effect of water stress and 30 temperature on germination of true mountain 20 mahogany. J. Range Manage. 29(2):138-140. 10 Rosner, L.R. 2000. Variability in seed propa- 0 gation requirements of two co-occurring montane shrub species-Symphoricarpos D m 60 0 30 60 oreophilus (mountain snowberry) and Ribes Stratification Duration (days) cereum (wax currant). Masters Thesis, New Mexico State Univ. Las Cruces, N.M. No Scarification p SAS Institute Inc. 1989. SAS/STAT Users 5-Minute Acid Soak ® Guide, Version 6, Fourth Edition, Volume 10-Minute Acid Soak ® 11. SAS Inst. Inc., Cary, N.C. ® 5-Day Dry-Tumble Smith, D.R. 1971. Growth Responses of true B 10-Day Dry-Tumble mountain mahogany on four soil types within the front range of Colorado. Ph.D. Diss., Fig. 4. Effect of interaction between seed source, stratification duration, and scarification Utah State Univ. Logan, Ut. treatment on mountain mahogany germination. Smith, D.R. and L.N. Bass. 1973. Germinability of true mountain mahogany that may be hydrocyanic acid or a macroclimate may influence seed source achenes as influenced by soil and other envi- cyanogenic compound (Moore 1963). Acid response to scarification treatments. This ronmental factors. Proc. Assoc. Official Seed scarification may chemically alter or pattern of variability contrasts sharply Analysts 63:26-134. Springfield, H.W. 1973. Cliffrose and moun- remove these germination inhibitors from with variability in stratification require- tain mahogany seeds retain viability 6 years the seed coat. Dry-tumble scarification, on ment, which delineates along macrocli- in cold storage. USDA For. Serv. Res. Note the other hand, would be expected to be matic gradients. RM-236. less effective in reducing seed coat- Young, J.A. and R.A. Evans. 1981. inhibitor content. Variable response to acid Germination of seeds of antelope bitterbrush, scarification among seed sources may be Literature Cited desert bitterbrush, and cliff rose. Agricultural explained by variablity in inhibitor content. Research Results. ARR-W-17. USDA Sci. There was no elevational or latitudinal and Educ. Admin. Baskin, C.C. and J.M. Baskin. 1973. Plant Young, J.A., R.A. Evans, and D.L. Neal. trend among sources in response to scarifi- population differences in dormancy and ger- cation treatments in this study. 1978. Treatment of curlleaf Cercocarpus mination characteristics of seed: heredity or J. Range Microclimatic, edaphic, and other local seeds to enhance germination. environment? Amer. Midl. Nat. 90(2):493-498. Manage. 42(3): 614-620. biotic and abiotic factors rather than

202 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 Book Reviews

The Politics of Precaution: Genetically try's response in each of the five areas is the analysis of these 2 nations together. In Modified Crops in Developing either promotional, permissive, precau- India, IPR and international trade policies Countries. By Robert L. Paarlberg. tionary, or preventive. The 4 developing have been preventive. In contrast, in 2001. Johns Hopkins University Press, countries that are studied in this book are China, IPR policies have been precaution- Baltimore and London. 180p. US$19.95. Brazil in Latin America, China and India ary and international trade policies have Paper. ISBN 0-8018-6823-8. in Asia, and Kenya in Africa. been permissive. In India, biosafety poli- In the early 1970s, scientists-almost Governmental caution and weak capaci- cies have been precautionary and food exclusively in the western world-began ty in Kenya is the subject of Chapter 3. As safety and consumer choice policies have orchestrating actual recombinations of far as IPRs, biosafety, international trade, been permissive. In contrast, in China, DNA molecules by moving particular and public research investment are con- biosafety and food safety and consumer genes carrying desired characteristics from cerned, Kenya's policy choices have been choice policies have both been permissive. a source life-form into the DNA of a liv- precautionary. Only in the area of food Only in the area of public research invest- ing target organism. Why was this done? safety and consumer choice have Kenya's ment have policies in India and in China The answer is that in comparison with the policies been promotional in the sense that been promotional. slow and often inaccurate techniques that this nation's food safety and labeling laws What accounts for these differences? In accompanied conventional plant breeding, have not made any distinction between India, non-governmental organizations recombinant DNA gave scientists a quick, GM and non-GM foods. What explains (NGOs) have successfully fomented a more potent, and a possibly more accurate this highly precautionary state of affairs? number of anxieties. For instance, one technique. This novel technique has According to this book, the "explanation NGO used "biosafety rules and procedures increasingly been referred to as genetic for Kenya's caution can be found in its to try to block Bt cotton, but the core of its modification or simply as GM. With the weak governmental capacity and its high argument against this GM crop was that it availability of these GM techniques, scien- dependence on the donor community" (p. was being introduced from abroad by the tists have been able to produce a variety of 63). Although this is a plausible explana- Monsanto company, which owned rights agricultural plants with a number of tion, 2 questions arise. First, is Kenya's to the dreaded "terminator" technology"(p. acceptable attributes. dependence on the international donor 106). In stark contrast, in China, there is Given the salience of agriculture in the community excessive relative to the other virtually no political space for independent developed and in the developing world, countries being studied in this book? critics to challenge state policy. As this the subject of genetically modified or GM Second, what about Kenya's dependence book rightly points out, "a lack of open crops has aroused great interest across the relative to that of other African countries? political space for civil society to chal- world. Are GM crops really the panacea Regrettably, these questions are not lenge government policy in China has that their protagonists claim they are? addressed in any significant detail in this been one reason for China's ability to go What are the real and hypothetical dangers chapter. ahead with some GM crops while others in associated with the planting of GM crops? What about Brazil? A perusal of the developing world have not" (p. 155). Why is it that dissimilar countries have Chapter 4 tells us that with regard to IPRs Therefore, this book reasons that China's reacted so differently to the adoption of and food safety and consumer choice, this "policy insulation approach should not and GM crops? Finally, in the context of GM nation's policies have been permissive. In probably could not be imitated by others" crops, what lessons do we learn by study- contrast, as far as biosafety and interna- (p.155). ing the disparate reactions of 4 important tional trade are concerned, this country's Let me conclude this review by noting developing countries? The general pur- policies have been precautionary. Only in that, in general, this is a topical and a well pose of this book is to shed light on these the area of public research investment researched book. Despite a few errors of sorts of questions. The specific "purpose have Brazil's policies been promotional. omission, this book makes a number of of this book is to document and explain Although it is helpful to know this, my points that are both correct and thought [the] emerging pattern of policy resistance only concern here relates to the length of provoking. Consequently, I recommend to GM crops among some developing- the author's study period. As pointed out this book to all readers who wish to learn country government authorities" (p. 4). on p. 157, the length of the study period is more about the ontogenesis of policies How does one begin a meaningful study 1999-2001 for the other 3 countries. pertaining to the adoption of GM crops in of the policy opposition to GM crops by However, for Brazil (see p. 69) it is 4 prominent developing countries.- dissimilar developing countries? Chapter 2 1999-2000. Given the brevity of this Amitrajeet A. Batabyal, Rochester proposes one way. As this chapter helpful- study period, the inferences drawn in this Institute of Technology, Rochester, New ly points out, any purposeful analysis of chapter are fundamentally "snapshot" York. the above question must adequately dis- inferences. Do such inferences tell us any- cuss a nation's response in 5 policy areas. thing substantive about the future? On a Invasive Exotic Species in the Sonoran These areas are intellectual property rights related note, why focus on Brazil specifi- Region. Edited by Barbara Tellman (IPRs), biosafety, international trade, food cally and not on, say, Argentina? I suspect with 29 text contributors. 2002. The safety and consumer choice, and finally, that many readers will ask these sorts of University of Arizona Press and the public research investment. Having speci- questions. Arizona-Sonora Desert Museum. 424 p. fied these 5 areas, this chapter uses a 4- Chapters 5 and 6 focus on the policy US$75.00 cloth. ISBN 0-8165-2178-6. part classificatory scheme to study nation- responses of India and China. Since these The distinctive flora and fauna of the al responses in each of the 5 policy areas 2 populous developing countries are often Sonoran region inspire above average identified above. In this scheme, a coun- compared, in this review, I shall consider interest in the potential, ongoing and his-

JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 203 torical effects of invasive species in that Sonoran Region immigrations on a geo- a brief glossary, and 3 admirable appen landscape. And deleterious invaders, both logical time scale. Chapter 2 by the editor dices, including a summary of laws, agree- producers and consumers, are not in short explores the notably frequent and dubious- ments and executive orders related to supply. In Invasive Exotic Species in the ly strategic introductions by humans in exotic species, and lists of species of flora Sonoran Region, we are provided a broad recent times. Chapter 3 examines the and fauna that have been introduced to the survey of the ecology and ecological man- Sonoran floristic province and Chapter 4 Sonoran Region. The list of cited literature agement of diverse invasive species in the discusses barriers to plant naturalizations in extensive; it is compiled at the end of Sonoran Region of the American and invasions. the book rather than at the end of each Southwest and Northern Mexico. The Chapters 5 through 13 constitute Part 2, chapter. A general index and brief bios of book is the product of a symposium held 4 and these chapters cover a diverse group the contributors complete the book. years earlier in May 1998 at the Arizona- of specialized topics involving sub-regions The contributed chapters of Dr. Tellman's Sonora Desert Museum near Tucson, within the Sonoran Region, particular edition are semi-technical in character, typi- Arizona. From my non-specialist perspec- kinds of invasive species, or special topics cal of a university press book. Some chap- tive, 4 years from symposium to publica- related to invasions. Sample topics are ters, even within sections, are more special- tion seems a leisurely publishing schedule diverse, and include: Invasive Vertebrates ized than others, and some are sufficiently on a topic as dynamic and virulent as inva- on Islands of the Sea of Cortez; Exotic surfeited with references that their readabili- sive species, where a few years can be an Species in Grasslands; Mexican Grasslands, ty is slightly impaired. Collectively, though, eternity. In any case, among the contribu- Thornscrub, and the Transformation of the Invasive Exotic Species of the Sonoran tors to the book are some notable regional, Sonoran Desert by Invasive Exotic Region is a good survey of the ecology and national and international specialists in the Buffelgrass (Pennisetum ciliare). the issues involved with floral and faunal ecology of and lands. Part 3 addresses exotic species manage- invasions of the Sonoran communities. It An interesting preface by Gary Paul ment in 6 chapters, covering such topics as should be of interest to students, Nabhan and informative introduction by plant protection and quarantine, biological researchers, land managers, and environ- the editor commence a text that is divided control, fence-line contrasts, and case mentally aware citizens, especially in that into 3 major sections. The 4 chapters of studies involving control. Chapter 19 by region, who will surely find value in some Part 1 examine invasive species from a Jeff Lovich is a summary of the sympo- aspects of its diverse content. - David L. broad perspective. Chapter 1 by Thomas sium's major messages, and contains a Scarnecchia, Washington State University, Van Devender examines the history of strategic call to action. The book contains Pullman, Washington.

204 JOURNAL OF RANGE MANAGEMENT 56(2) March 2003 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 are expected to pay current page charges. Since most research observations in the title. Authors 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. paper should be written accurately, clearly, and concisely. It Manuscripts that do not follow the directives and style in Journal hand- Every should lead the reader from a clear statement of purpose through materi- book will be returned to the authors before being reviewed. A manuscript methods, results, and to discussion. The data should be reported number and submission date will be assigned when the paper is received als and in coherent sequence, with a sufficient number of tables, drawings, and in the appropriate format. 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 accept- ability of the revision. Revisions not returned within 6 months, are considered terminated. Authors who consider that their manuscript has