Diet Selection of Mountain Lions in Southeastern Arizona

J. Range Manage. 52:202Ð207 May 1999 Diet selection of mountain lions in southeastern Arizona

STAN C. CUNNINGHAM, CARL R. GUSTAVSON, AND WARREN B. BALLARD

Stan Cunningham and Carl Gustavson, at the time of the research, were research biologists, and Warren Ballard was research program supervi- sor, Research Branch, Arizona Game and Fish Department, 2221 W. Greenway Road, Phoenix Ariz. 85023.

Abstract Resumen

Prey selection by mountain lions (Puma concolor) in the Presa seleccionada por el León de Montaña (Puma Aravaipa-Klondyke area in southeastern Arizona was studied concolor) en el Sureste del Estado de Arizona en el area de from February 1991 to September 1993. Overall diet as deter- Aravaipa Klondyke. El estudio fue conducido en el mes de mined from frequency of occurrence in 370 scats was 48% Febrero del año 1991 a Septiembre de 1993. El alimento deer (Odocoileus virginianus cousi and O. hemionus com- preferido fue determinado basado en la frecuencia que ocur- bined), 34% cattle, 17% javelina (Tayassu tajacu), 6% rabbit rio en 370 ejemplos de escrementos y los resultados fueron los (Sylvilagus spp. and Lepus californicus combined), 4% siguientes porcentajes: 48% Venado (Cola-Blanca y Venado rodent, and 2% desert bighorn sheep (Ovis canadensis mexi- Mula Odocoileus spp.), 34% Ganado Vacuno, 17% Cochi canus). With respect to biomass consumed, cattle composed Javalin (Tayassu tajacu), 6% Conejo (Sylvilagus spp) y Liebre 44%, deer 40%, javelina 10.9%, rabbits 2.9%, and rodents Lepus californicus), 4% Roedor y 2% Borrego del Decierto 0.02%. Based on mean weights of prey consumed, the propor- (Ovis canadensis mexicanus). Tambien fue estimado el portion of individuals killed and eaten changed to rabbits 52.7%, centaje de biomasa y la proporción de especies casadas y con- deer 16.3%, rodents 12%, javelina 10%, cattle 8%, and sumidas por León de Montaña En cuanto al respecto de bio- desert bighorn sheep 0.5%. Mountain lions selected deer less masa consumida, Ganado Vacuno fue 44%, Venado 40%, frequently than their availability would suggest, selected Cochi Javalin 10.9%,conejos 2.9%, y roedores 0.02%. De calves slightly more than their availability, and javelina as acuerdo con el promedio de peso de presa consumida por el expected. We speculated that lions selected calves because León de Montaña la proporción de especies consumidas cam- they were more vulnerable to predation than deer. bio a lo siguiente: 52.7% Conejo, 16.3% Venado, 12% Roedores, 10% Cochi Javalin, 8% Ganado Vacuno, 0.5% Borrego del Decierto. Tambien fue estimada la cantidad de Key Words: prey availability, cattle, diet selection, Puma con- diferente especies que pudieran existir en esta misma area color, javelina, mule deer, white-tailed deer. usando un Helicoptero en 4 diferente vuelos, las siguientes especies fueron encontradas: (Venado, Ganado Vacuno, y Cochi Javalin ). En los resultados observamos que el León de Livestock have been preyed upon by mountain lions since Montaña selecciono Venado con menos frequencia, compara- they were first introduced from Europe (Barnes 1960), and do con la fiques sugerida, Ganado Vacuno (Crias ) fue poco livestock loss remains a major rationale for controlling them. mas seleccionado de lo esperado, y Cochi Javalin como se Arizona reportedly has some of the highest mountain lion kill esperaba. Especulamos que Leónes de Montaña selec- rates on cattle in the United States (Christensen and Fischer cionarón Ganado Vacuno (Crias), porque fue mas vulnerable 1976, Nowak 1976, Anderson 1983). Killing mountain lions a la casa del León de Montaña que de lo que fue el Venado. that prey on livestock (i.e., depredation control) remains a legal, though controversial practice, and accounts for a substantial portion of the human-caused mortalities of Arizona mountain lions (Cunningham et al. 1995). During 1987, the Aravaipa-Bonita Cattle Growers mountain lions. From 1987 through 1989, 57 mountain lions Association of southeast Arizona reported that mountain lions were killed within a 100,000 ha area. Ensuing public articles had killed more calves than historically. Consequently, they resulted in negative criticism of the control operation (B. contracted with the Animal and Plant Health Inspection Burkhart: Rancher, U.S. hunters kill wildlife without rein by Service (APHIS), U.S. Department of Agriculture, to control State. Arizona Republic, June 15, 1989). We initiated an investigation of mountain lion-livestock interactions in the Aravaipa-Klondyke area of southeast Arizona. A main objec- This manuscript is dedicated to the memory of Dr. Carl R. Gustavson, an excel- lent biologist and better friend. This study was funded by the Federal Aid in tive was to determine if mountain lions in the area selected Wildlife Restoration Act Project W-78-R. Constructive criticism on earlier drafts livestock and other prey in proportion to availability. was provided by J. Truett, B. Wakeling, P. Beier, R. Schwiensburg, K. Logan, and J. deVos. We thank K. Sargent, J. Yarchin, and J. Simmons who were involved in data collection. We also thank the San Carlos Apache Nation for allowing us to conduct research on their land. Manuscript accepted 24 Aug. 1998.

202 Journal of Range Management (52)3, May 1999 Study Area

The Aravaipa-Klondyke study area (AKSA) consisted of 95,100 ha which included Aravaipa Canyon, portions of the Aravaipa Creek watershed, Galiuro Mountains, and the San Carlos Apache Indian Reservation (Fig. 1). Elevations on the study area ranged from 750 to 2,300 m. Topography within all mountain ranges was steep and broken, with stone pinnacles, narrow and deep canyons, and rugged cliffs common at higher elevations. These for- mations merged downslope to compara- tively level terraces near the bajadas of the Aravaipa Valley to the east, the Gila River Basin to the north, and the San Pedro River to the west. The study area climate typified that of southeastern Arizona (Lowe 1964). Mean temperatures ranged from 8 to 26 C, and precipitation averaged 31.6 cm per year. Annual spring droughts were usually followed by late summer thun- derstorms (monsoons) produced by moisture from the Gulf of Mexico. Winter rains were less episodic and more widespread, originating from Pacific fronts. Snowfall in the area was infrequent at higher elevations and rare in the lower valley areas. The study area had a wide diversity of vegetation types. Seven of Brown and Lowe's (1974) biotic communities occurred in the study area; semidesert grassland (40% of area), Arizona upland Fig. 1. Boundaries of the Aravaipa-Klondyke study area where mountain lion diet selection was Sonoran desert scrub (30%), interior studied, 1991-1993. chaparral (14%), Madrean evergreen woodland (13%), Great Basin conifer leading to removal of most goats by od from the 1940s to the late 1960s than woodland (2%), petran montane conifer 1940. The Federal government passed they were afterward. forest (<1%), and riparian strips along the Taylor Grazing Act in 1934, which Javelina populations have recently streams. called for removal of wild horses, fenc- declined in the study area, but reasons ing of public lands, and reductions for the decline were not well understood Livestock and Prey Occurrence (>50% in some areas) in the numbers of (Cunningham et al. 1995). Desert Both historic and current livestock cattle grazed. During our study, there bighorn sheep were extirpated in the operations on the study area could affect were 22 cattle grazing allotments on l930’s. In 1973, 22 bighorn sheep were mountain lion diets. Beginning in the public lands (Commercial livestock reintroduced into the area and the popu- 1870s, >50,000 cattle (>125/km2) annu- area) with no restrictions on the San lation increased to >100 by 1985 ally occupied the Aravaipa Valley. Carlos Indian Reservation (Fig. 1). (Cunningham et al. 1993). Cattle numbers were reduced after sig- Wildlife has also varied in abundance on the study area during the past centu- nificant drought in the late 1890's, but Methods were still grazed in larger numbers than ry. Hadley et al. (1991) reported that today. By the 1920's, an additional Apaches complained about the scarcity 40,000 angora goats grazed the rougher of game shortly after 1900. Large wild Prey Availability parts of the Aravaipa watershed and mammals remained scarce for another Densities of deer, calves and adult cat- 2,000 burros and 1,800 horses ran free 2Ð3 decades, then began to increase. tle, and javelina within the study area (Hadley et al. 1991). Based on interviews with local residents were estimated 4 times: October 1991, Grazing pressure declined during the and natural resource managers, deer April and October 1992, and April 1993. 1930's. The value of mohair decreased, populations were larger during the peri- We conducted aerial surveys for 4 days

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 203 each estimate from a Bell Jet-Ranger which excluded most coyote (Canis differences between prey use and avail- helicopter along 52 10-km nonoverlap- latrans) and bobcat (Felis rufus) scats ability by season and overall. If differ- ping transects following procedures of (Danner and Dodd 1982). Scats were ences (P < 0.05) existed, Bonferroni Anderson et. al. (1979) and Burnham et judged to be <1 month old when dark confidence intervals were used to indi- al. (1980). Transects were oriented at and moist and were labeled as fresh; all cate in which instances use did not equal right angles to major drainages and were others were labeled as old. While search- availability (Neu et al. 1974, Byers et al. flown at an airspeed of 48 km/hr at 30 m ing for scat, we also identified fresh (<1 1984). For significant Bonferroni confi- altitude. Front and rear observers posi- month old) mountain lion kills by crite- dence intervals we used Jacobs' D selec- tioned on the right side of the helicopter ria described by Shaw (1989). tivity index (Jacobs 1974) to indicate made independent counts of prey Prey items in scats were identified on direction and magnitude of selection or species by group size up to 180 m from basis of hair and skeletal remains. Scats avoidance of prey. the helicopter. Cattle were divided into were soaked in water and washed 2 groups based on size; calves (<180 kg) through 1- and 3-mm sieves. Skeletal versus adults (>180 kg). material was compared with known ref- Results Based on line transect procedures, we erence materials. Hair was examined had to assume equal visibility of the 3 microscopically and compared with ref- Prey Availability key prey species (deer, javelina, and cat- erence slides and photomicrographs Overall, deer were the most abundant tle). However our qualitative observa- from Moore et al. (1974). We did not prey species surveyed in the study area tions were that deer and javelina were differentiate between white-tailed and (Table 1). Deer numbers were higher in not as visible as cattle in thicker vegeta- mule deer hair, or between cottontail fall (post-fawning) than spring. Calf tion at distances >100 m from the heli- and jackrabbit, or among rodent species. numbers were higher in spring than fall copter. This bias could result in lower We used frequency of occurrence data as most calving occurred from deer and javelina density estimates and to estimate the relative number of each December to February. Javelina num- higher cattle density estimates than real- taxonomic group that were eaten using bers were relatively stable among the ly existed. procedures described by Ackerman et al. first 3 surveys, but were lower in April (1984). Weights of white-tailed and Population density estimates were cal- 1993. The number of javelina groups mule deer were estimated from average culated using TRANSECT II software seen on the San Carlos Indian live weights measured at deer hunt check Reservation were too small to allow (Laake et al. 1979) as recommended by stations near the study area (R. Olding, density estimation. Burnham et al. (1980). Prey numbers Ariz. Game and Fish Dept., unpubl. Prey densities (all species combined) among surveys and between the com- data). Weights of other species were were higher on the commercial livestock mercial livestock and San Carlos Indian from Burt and Grossenheider (1964). portion of the study area than on the Reservation portions of the study area Kill estimates were divided into 2 sea- Reservation (P = 0.043). However, there were compared using MANOVA. sons; MarchÐAugust (springÐsummer) were no differences in relative propor- and SeptemberÐFebruary (fallÐwinter). tions of prey species between the com- Prey Consumption and Selection We then analyzed diet versus availabili- mercial and reservation areas and data Two mountain lion scats were collect- ty when calf numbers were high and sets were pooled. ed each month within 8 equal-sized sub- low. Scats considered fresh were pooled divisions of the study area. We searched across years to increase sample size. Prey Selection for scat by hiking or riding horseback We compared estimates of abundance We found 41 mountain lion kills along ridges and washes in each subdivi- (i.e., availability) of deer, calves, and which included 16 white-tailed deer, 7 sion. We identified mountain lion scats javelina with numbers estimated eaten mule deer, 10 cattle, 4 javelina, and 4 by size and shape (Murie 1954); only by mountain lions (i.e.,use). We used desert bighorn sheep. Male:female ratio scats >30 mm in diameter were collected Chi-square contingency tables to test for

Table 1. Large prey densities within the Aravaipa-Klondyke study area, Arizona, 1991Ð93, as determined by helicopter surveys.

Area Calves Adult cattle Deer Javelina Date (<180 kg) (>180 kg) 95% CI 95% CI 95% CI 95% CI Commercial ------(No./km2)------livestock portion Oct 1991 0.6 0.4 - 1.0 1.7 1.0 - 2.8 6.1 0.5 - 8.1 2.5 1.7 - 3.7 Apr 1992 1.2 0.9 - 1.8 5.1 4.3 - 6.0 5.5 4.3 - 7.1 2.7 1.9 - 3.7 Oct 1992 0.6 0.0 - 2.0 1.7 0.0 - 62.9 6.9 5.4 - 8.8 2.9 0.1 - 77.0 Apr 1993 2.0 1.6 - 2.4 4.8 4.2 - 5.5 2.7 2.1 - 3.5 0.6 0.4 - 1.0 xÐ 1.1 3.3 5.3 2.2 San Carlos Indian Reservation Oct 1991 0 0 - 0 0.4 0.2 - 0.9 3.6 2.6 - 4.9 0 0 - 0 Apr 1992 2.3 0.0 - 56.5 1.8 1.2 - 2.8 2.5 1.5 - 4.1 27.5 0.3 - 2.7 Oct 1992 0.03 0.0 - 0.3 0.1 0.0 - 0.4 2.5 1.6 - 3.8 0.6 0.3 - 1.0 Apr 1993 0.05 0.0 - 0.9 1.8 1.0 - 3.3 6.3 4.3 - 9.4 0 0 - 0 xÐ 0.6 1.0 3.7

204 Journal of Range Management (52)3, May 1999 Table 2. Diet of mountain lions, based on the analysis of 370 scats, from the Aravaipa-Klondyke Over the entire study, mountain lions Study Area, Arizona, 1991-93. Percent biomass and relative number of individuals estimated in the study area ate fewer deer than consumed by mountain lions are calculated according to Ackerman et al. (1984). expected based on deer availability, slightly more calves than expected, and (A) (B) (C) (D) (E) javelina in proportion to availability Relative Frequency Estimated number of (Table 4). During the fall-winter season, of weights of Correction Biomass individuals mountain lions consumed fewer deer Prey species occurrence individuals factor consumed2 consumed than expected, but consumed calves and (%) (kg) (kg/scat)1 (%) (%)3 javelina in proportion to availability. During the spring-summer season, all 3 Deer 0.48 44 3.52 40.1 16.6 Calves 0.34 100 5.48 44.2 8.0 species were consumed as expected Javelina 0.17 20 2.68 10.9 10.0 based on availability. Rabbit 0.06 1 2.02 2.9 52.8 Rodent 0.04 0.03 0.03 0.02 12.0 Desert bighorn 0.02 50 3.73 1.8 0.5 Discussion 1Estimated weight of prey consumed per collectible scat produced, when such prey is the only item in the scat (C = 1.98 + 0.035 B). Prey Use 2D = (A x C)/∑ (A x C). In our study, rabbits were the most 3 E = (D÷B)/∑ (D÷B). numerous food item. Deer are the principal prey of mountain lions in many was almost equal for white-tailed deer prised 10.9% of the biomass, and rab- other areas based on frequency of occur- kills, but all mule deer kills were bits, rodents, and bighorn sheep all com- rence in scats (Anderson 1983). Our females. We found no deer fawn or prised <3%. data suggest that the importance of rab- juvenile javelina kills, but 3 of 4 bighorn Neither frequency of prey nor estimat- bits, rodents, or other small prey are sheep kills were lambs. Nine of 10 ed biomass accurately reflected number masked by just using frequency of killed cattle were calves, and one was an of individual prey items eaten. Rabbits occurrence without a conversion factor. adult cow; evidence indicated she was were the most frequently killed prey Studies where mountain lion kills are killed at the same time as her calf. Of item, followed by deer, rodents, javeli- used for diet determination would also the calves we found and 7 additional na, cattle, and desert bighorn. underrate small animals as most moun- ones reported by ranchers, 8 were esti- During both seasons, rabbits were tain lions would consume the whole ani- mated <4 months of age, and 8 were killed and eaten more than any other mal. Also, Johnson and Aldred (1982) estimated >4 months of age. prey, but calves composed the majority found that bobcats digested more bones Deer remains were found in 48% of of the biomass eaten (Table 3). During and hair of smaller prey over larger 370 scats and cattle remains occurred in MarchÐAugust (spring-summer), small prey. Mountain lions may digest almost 34% (Table 2). Javelina was the next prey (rabbits and rodents) were killed all bones and hair of small species. most commonly-encountered food item most frequently and were a greater pro- Based on the Ackerman et al. (1986) (i.e., 17%), followed by rabbits, rodent, portion of the mountain lion diet than caloric needs model, and individual prey and desert bighorn sheep. Other prey during September-February. The number proportions eaten by mountain lions on such as birds, badger (Taxidea taxus), of individual deer, calves, and javelina the study area, we estimated the num- black bear (Ursus americanus), porcu- consumed in September-February (fall- bers of prey eaten each year. A resident pine (Erethizon dorsatum), and skunk winter) all increased more than 100% female with 3 kittens would eat 35Ð40 (Mephitis spp.) occurred in very few over spring-summer. Rabbits were still deer, 17Ð19 calves, 21Ð24 javelina, scats. Converted to biomass, diet of the most frequently killed prey in fall- 90Ð100 rabbits, 20Ð23 rodents, and 1Ð2 mountain lions was dominated by cattle winter, but the estimate of number of desert bighorn sheep. A resident female (44%) and deer (40%). Javelina com- individuals consumed decreased 31.6%. without kittens would eat 9Ð11 deer,

Table 3. Seasonal diet based on frequency of occurrence, percent biomass, and relative numbers of individuals consumed by mountain lions in the Aravaipa-Klondyke study area, Arizona, 1991Ð93.

MarchÐAugust SeptemberÐFebruary (n = 82) (n = 54) Frequency Number of Frequency Number of of Biomass individuals of Biomass individuals occurrence consumed consumed occurrence consumed consumed ------(%) ------Deer 36.6 33.2 8.6 44.4 37.9 21.0 Calves 36.6 51.7 5.9 33.3 44.3 10.8 Javelina 11.0 7.5 4.4 22.2 14.3 17.7 Rabbit 12.2 6.5 73.3 3.7 1.7 41.7 Rodent 4.9 0.02 7.5 1.9 0.01 8.0 Bighorn 1.2 1.0 0.2 1.9 1.7 0.7

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 205 Table 4. Selection of 3 prey species by mountain lions based on frequency of occurrence in moun- Shaw (1981) suggested that increasing tain lion scats and proportion available as determined from aerial surveys, Aravaipa-Klondyke the deer:calf ratio may reduce cattle pre- Study Area, Arizona, 1991Ð93. (Rabbits, rodents, and desert bighorn sheep were not recorded dation by mountain lions. However, we during surveys). suggest that calf vulnerability was the major factor influencing kill rates, and Relative Proportion number of of prey increases in the deer prey base may not Season Frequency of individuals species Bonferroni affect kill rates on calves. Also, changes Species occurrence consumed available 95% CI Jacobs’ D1 in deer and javelina populations are Year-long diet highly influenced by climatic factors Deer 48 47.3 58.0 0.41 Ð 0.54 -0.22 which are impossible for wildlife man- Calves 34 23.2 17.8 0.18 Ð 0.28 0.16 agers to control. Javelina 17 29.5 24.1 0.24 Ð 0.35 Shaw (1981) reported that the majori- Fall diet ty of calves killed were <4-months old Deer 47 42.6 63.1 0.27 Ð 0.59 -0.40 and that calf predation was seasonal. We Calves 35 21.8 10.7 0.08 Ð 0.35 Javelina 24 35.6 26.2 0.2 Ð 0.51 found mountain lions killed and ate calves of all ages year round. This also Spring diet Deer 44 45.8 52.7 0.33 Ð 0.59 suggests that moderate increases in deer Calves 44 31.2 25.8 0.19 Ð 0.43 numbers (fall fawning period in our Javelina 13 22.9 21.5 0.12 Ð 0.34 study) are unlikely to cause appreciable 1A positive number means the item was selected for; a negative number means the item was selected against (Range = reduction in mountain lion predation on 1.0 to +1.0). calves. In our study area, many allottees have nowhere to graze cattle except in rugged 5Ð6 calves, 8Ð11 javelina, 7Ð9 rabbits, availability and vulnerability (Sunquist terrain with relatively dense vegetation 5Ð7 rodents, and <1 desert bighorn and Sunquist 1989, Iriate et al. 1990). cover. Those with sufficient flat, open sheep per year, while a resident male Only in Arizona have mountain lions pasture hold their younger calves out of would eat 14Ð18 deer, 7Ð9 calves, 9Ð11 been reported to prey heavily (>30% rugged areas as long as possible and javelina, 36Ð46 rabbits, 8Ð11 rodents, diet) on cattle (Tully 1991, Cunningham generally experience fewer losses to and <1 bighorn sheep per year. et al. 1995). Most reports come from mountain lions. Those without lower Ackerman et al. (1986) estimated that mid-elevation chaparral and pine-oak pastures experience greater losses of young female or male transients would woodlands in central Arizona; few cases calves to mountain lions. consume approximately 50% the prey of have been documented in high-elevation We believe that mature male moun- a resident female without kittens. or low desert areas (Shaw et al. 1988). tain lions, rather than females and We estimated that mountain lions In central Arizona, Shaw (1977) found immatures, caused most of the livestock killed >600 deer and >225 calves each cattle to comprise at least 37% of the losses. During the study and statewide year on the study area (including both mountain lion kills, and cattle remains since 1990, females were killed in the commercial livestock and San occurred in 34% of the scats he ana- depredation control cases only during Carlos Indian Reservation portions). If lyzed; cattle kills peaked in spring. periods when small calves were abun- our deer kill estimates were accurate, we Estimates of prey densities compared dant, and only 1 mountain lion <24 should have observed a reduction in with mountain lion diet suggested that months of age was killed in connection deer numbers, but this did not occur. calves were selected by mountain lions with a depredation case (Cunningham et Conversations with ranchers suggested in preference to deer. Deer densities in al. 1995). Although males are easier to they believed livestock losses, though chaparral and forest vegetation types catch than females (Anderson 1983), substantial, were not >225. were probably higher than we estimated current strict controls on snaring and Several studies (see Ackerman et al. because of poor visibility during heli- allowing dog hunting only in the vicini- 1986 for review) have estimated that copter surveys. If correct, this bias ty of the kill increase the chance of mountain lions kill a deer at 4 to 20 day would make mountain lion selection for killing the offending animal. intervals, depending on sex and age of calves over deer greater than our calcu- Anderson (1983) reported that male prey. Although mountain lions have the lations suggested. mountain lions weighed approximately ability to kill ungulate prey frequently, We speculate that calves were more 1.4 times more than females. Iriate et al. we suggest they may not do so if they vulnerable to mountain lion predation (1990) found that there was a positive opportunistically feed on smaller prey. than deer. Hereford or crossbreed calves correlation (r = 0.875) between moun- However, this would be difficult to eval- are more visible in thick or open terrain tain lion body size and prey size select- uate, and would require intense 24-hour while deer are cryptically colored, quiet, ed. Kruuk (1986) reported that felids are tracking without affecting lion behavior. and spend most of the day hiding. Deer strictly carnivorous and select prey com- are more alert and wary than calves. mensurate with their own body size. If Interaction with Cattle When we followed mountain lion travel mature males killed most of the live- Cattle, except for calves, are larger routes we observed more calves than deer stock in this study, then our estimates of than most mountain lion prey. Prey or javelina while prey surveys suggested potential livestock losses to females selection is influenced by both prey deer and javelina densities were greater. with kittens were probably high. Also, additional bias may have occurred

206 Journal of Range Management (52)3, May 1999 because we restricted scat collections to Danner, D.A. and N.L. Dodd. 1982. Sunquist, M.E. and F.C. Sunquist. 1989. scats >30 mm in diameter; such scats Comparison of coyote and gray fox scat Ecological constraints on predation by may have been more representative of diameters. J. Wildl. Manage. 46:240Ð241. large felids. Pp 283Ð301 In: J. L. large lions. Regardless, our estimates Hadley, D., P. Warshall, and D. Bufkin. Gittleman, ed. Carnivore behavior, ecolo- indicate the potential for large economic 1991. Environmental change in Aravaipa, gy and evolution. Cornell Univ. Press. 1870Ð1970: an ethnoecological survey. Ithaca, N.Y. losses from mountain lion predation. U.S. Bur. Land Manage. Cultural Res. Ser. Tully, R.J. 1991. Results, 1991 question- No. 7. Safford, Ariz. 368 pp. naire on damage to livestock by mountain Iriate, J.A., W.L. Franklin, W.E. Johnson, lion. Pp. 68Ð74 In: C. E. Braun, ed. Proc. Literature Cited and K.H. Redford. 1990. Biogeographic mountain lion-human interaction symp., variation of food habits and body size of the Colo. Div. Wildl., Denver, Colo. Ackerman, B.B., F.G. Lindzey, and T.P. American puma. Oecologia 85:185Ð190. Hemker. 1984. Cougar food habits in Jacobs, J. 1974. Quantitative measurements southern Utah. J. Wildl. Manage. of food selection. Oecologia 1:413Ð417. 48:147Ð155. Johnson, M.K. and D.R. Aldred. 1982. Ackerman, B.B., F.G. Lindzey, and T.P. Mammalian prey digestibility by bobcats. Hemker. 1984. Predictive energetics J. Wildl. Manage. 46(2):1982. model for cougars. Pp 333Ð352 In: S.D. Kruuk, H. 1986. Interactions between Miller and D.D. Everett, eds. Cats of the Felidae and their prey species: a review. world: biology, conservation and manage- Pp 353Ð374 In: S.D. Miller and D.D. ment. Nat. Wildl. Fed., Washington, D.C. Everett, eds. Cats of the world: biology, Anderson, A.E. 1983. A critical review of lit- conservation and management. Natl. erature on puma (Felis concolor). Colo. Div. Wildl. Fed., Washington D.C. Wildl. Spec. Rep. 54, Denver, Colo. 91pp. Laake, J.L., K.P. Burnham, and D.R. Anderson, D.R., J.L. Laake, B.R. Crain, Anderson. 1979. User's manual for pro- and K.P. Burnham. 1979. Guidelines for gram TRANSECT. Utah State Univ. line transect sampling of biological popu- Press, Logan, Ut. 26 pp. lations. J. Wildl. Manage. 43:70Ð78. Lowe, C.H. 1964. The vertebrates of Barnes, C.T. 1960. The cougar or mountain Arizona. Univ. Arizona Press, Tucson, lion. The Ralton Co., Salt Lake City, Ut. Ariz. 270 pp. 176pp. Moore, T.O., L.D. Spence, C.E. Dugnolle, Brown, D.E. and C.H. Lowe. 1974. A digi- and W.G. Hopworth. 1974. Identification tized computer-compatible classification of the dorsal guard hairs of some mam- for natural and potential vegetation in the mals of Wyoming. Wyo. Game and Fish Southwest with particular reference to Dept. Bull. No. 14. 117 pp. Arizona. J. Ariz.-Nev. Acad. Sci. 9 (suppl. Murie, O.J. 1954. A field guide to animal 2): 1Ð11. tracks. Houghton-Mifflin Co., Boston, Burt, W.H. and R.P. Grossenheider. 1964. Mass. 374 pp. A field guide to mammals. Houghton Nowak, R.M. 1976. The cougar in the Mifflin Co., Boston, Mass. 284 pp. United States and Canada. U.S. Fish and Burnham, K.P., D.R. Anderson, and J.L. Wildl. Serv., Washington, D.C., and New Laake. 1980. Estimation of density from York Zool. Soc., New York, N. Y. 190 pp. line transect sampling of biological popu- Neu, C.W., C.R. Byers, and J.M. Peek. lations. Wildl. Monogr. 72. 202 pp. 1974. A technique for analysis of utiliza- Byers, C.R., R.K. Steinhorst, and P.R. tion-availability data. J. Wildl. Manage. Krausman. 1984. Clarification of a tech- 38:541Ð545. nique for analysis of utilization-availabili- Shaw, H.G. 1977. Impacts of mountain lion ty data. J. Wildl. Manage. 48:1050Ð1053. on mule deer and cattle in northwestern Christensen, G.L. and R.S. Fisher, editors. Arizona. Pp 306Ð318 In: R.L. Phillips and 1976. Transactions of the mountain lion C. Jonkel, eds. Proc. 1975 predator symp., workshop. U.S. Fish and Wild. Serv. and Univ. Montana, Missoula, Mont. Nev. Div. Wild., Sparks. 213pp. Shaw, H.G. 1981. Comparison of mountain Cunningham, S.C., N. Dodd, and R. lion predation on two study areas in Olding. 1993. Arizona's bighorn sheep Arizona. Pp 306Ð318 In: Proc. Wildl. live- reintroduction program. Pp. 203-239 In: stock relationships symp. Univ. Idaho R. Lee, ed. The desert bighorn sheep in Forest Wildl., and Range Exp. Sta., Arizona. Ariz. Game and Fish Dept., Moscow, Ida. Phoenix, Ariz. Shaw, H.G. 1989. Mountain lion field guide. Cunningham, S.C., L.A. Haynes, C. Fourth ed. Ariz. Game and Fish Dept. Gustavson, and D.D. Haywood. 1995. Spec. Rep. 9, Phoenix, Ariz 47 pp. Evaluation of the interaction between Shaw, H.G., N.G. Woolsey, J.R. Wegge, mountain lions and cattle in the Aravaipa- and R.L. Day, Jr. 1988. Factors affecting Klondyke area of southeast Arizona. Ariz. mountain lion densities and cattle depreda- Game and Fish Dep. Tech. Rep. 17, tion in Arizona. Ariz. Game and Fish Phoenix, Ariz. 64 pp. Dept. Final Report, P-R Proj. W-78-R, Work Plan 2, Job 29, Phoenix, Ariz. 16 pp.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 207 J. Range Manage. 52:208Ð217 May 1999 Is supplementation justified to compensate pastoral calves for milk restriction?

D. LAYNE COPPOCK AND SANDRO SOVANI

Authors are associate professor, Department of Rangeland Resources, Utah State University, Logan, Utah 84322-5230, and research veterinarian, v. Irpina 23, 58100 Grosseto, Italy. At the time of research, the authors were research scientist and research associate, respectively, at the International Livestock Center for Africa (ILCA), P.O. Box 5689, Addis Ababa, Ethiopia.

Abstract Resumen

Competition for milk between calves and pastoral herders La competencia por leche entre becerros y los pastores de sis- may reduce weaning weights, retard growth, and delay temas pastoriles puede reducir los pesos al destete, retardar puberty in cattle. Calf supplementation could over-ride such el crecimiento y retrasar la pubertad del ganado. La suple- effects and improve pastoral economies. To examine these mentación a los becerros puede anular o aminorar tales efec- issues in semiarid Ethiopia, 266 Boran calves (Bos indicus) tos y mejorar las economías pastoriles. Para examinar estos were used in a 2 X 3 plus 1 factorial design contrasting grad- problemas en la Etiopía semiárida, se utilizaron 266 becerros ed levels of supplemental alfalfa hay (i.e., Medicago sativa L. Boran (Bos indicus) en un diseño factorial de 2 x 3 mas 1 para with mean intakes of 0, 344, and 557 g head-1 day-1 on a DM contrastar niveles graduales de suplementación de alfalfa basis) and supplemental water (i.e., with mean intakes of 0 (Medicago sativa L. con consumos medios de 0, 344, y 557 g cabeza-1 dia-1 en base a materia seca) y agua suplementaria and 3.8 liters head-1 day-1). The trial was repeated for animals (con consumos medios de 0 y 3.8 litros cabeza-1 dia-1). El born in 2 consecutive years. Treatments occurred over a ensayo se repitió para animales nacidos en 2 años consecu- background of simulated traditional management in which tivos. Los tratamientos se diseñaron basado en antecedentes calves had limited access to grazing and water and were de un manejo tradicional simulado en el cual los becerros allowed to suckle about two-thirds of their dams' daily milk tenían acceso limitado a apacentar y al agua y se les permitió yield. Traditionally managed controls received no supple- amamantar aproximadamente dos terceras partes de la pro- ments while other (positive) controls received no supplements ducción diaria de leche de sus madres. Los controles fueron but had greater access to milk. After 10 months of treatment animales que recibieron el manejo tradicional sin recibir calves were weaned and monitored. Supplementation with the suplemento, mientras otros animales control (positivo) no high level of hay plus water markedly enhanced (P<0.01) all recibieron suplemento pero tuvieron un mayor acceso a la productive features of calves at weaning compared to tradi- leche. Después de 10 meses de tratamiento los becerros tionally managed controls, and was an effective substitute for fueron destetados y monitoreados. La suplementación con un milk forgone in both years. Despite high variability in milk alto nivel de heno mas agua incremento marcadamente intake, access to supplements, and weaned body size as calves, (P<0.01) todas las características productivas de los becerros all male cattle converged in liveweight and other productive al momento del destete en comparación con los becerros que features by 3.5 years of age, largely due to compensatory recibieron el manejo tradicional (control) y esta suple- growth of traditionally managed controls. Heifers also con- mentación fue un substituto efectivo de la leche de la que verged in various attributes at maturity, but those which had fueron privados los animales en ambos años. A pesar de la received hay plus extra water as calves still conceived 2.6 to alta variabilidad en el consumo de leche, acceso a suplemen- 4.3 months earlier (P<0.05) than traditionally managed con- tos y tamaño corporal de los becerros destetados, a los 3.5 trols. We concluded that supplementation with hay and water años, todos los machos fueron similares en peso vivo y otras características productivas. Esto debido principalmente al can indeed compensate a young calf for typical levels of milk crecimiento compensatorio de los becerros que recibieron el restriction here. Carry-over effects, however, were insuffi- manejo tradicional (control). Al llegar a la madurez, las cient to justify large investments in supplementation consid- vaquillas también fueron similares en varios atributos, pero aquellas que recibieron heno mas agua extra cuando eran Research was conducted under the auspices of the Ethiopian rangelands pro- becerras concibieron de 2.6 a 4.3 meses antes de las mane- gram at ILCA. The authors thank Eshetu Zerihun and his assistants for implement- jadas tradicionalmente. Concluimos que la suplementación ing the research. Mesfin Shibre, Mr. J. Sherington, and Dr. E. W. Richardson helped with statistical analyses. The Southern Rangelands Development Unit con heno y agua puede compensar a los becerros jóvenes las (SORDU) of the Ethiopian Third Livestock Development Project (TLDP) gave restricciones típicas de leche a las que se someten. Sin embar- permission to use Demballa Wachu ranch for initial phases of research. The go, los efectos totales fueron insuficientes para justificar las SORDU veterinary service assisted with health management of our cattle. grandes inversiones en suplementación considerando los altos Assistance from CARE-Ethiopia with cattle procurement is greatly appreciated. Finally, we acknowledge the 2 anonymous reviewers who helped us improve the riesgos inherentes de la producción y tradiciones de mer- manuscript. cadeo de animales maduros. Manuscript accepted: 26 Jul. 1998.

208 Journal of Range Management (52)3, May 1999 ering the high inherent risks of produc- managed facility on the Borana Plateau from 477 mm in the first year (i.e., from tion and traditions of marketing mature 610 km south of Addis Ababa near the April, 1986, through March, 1987) to animals. Kenya border. A semiarid climate sup- 300 mm in the second year (i.e., from ports perennial savanna vegetation. April, 1987, through March, 1988; Key Words: Cattle nutrition, compen- Annual rainfall averages 500 to 600 Unpublished data, Coppock). satory growth, puberty, seasonal envi- mm, with about 60% received during Six of the 7 treatments were incorpo- ronments, watering frequency, Borana the long rains from April to May and rated into a factorial design involving pastoralists, Ethiopia. 30% during the short rains from October supplementation with graded levels of to November. Rainy periods are separat- water and alfalfa hay, offered to calves ed by a cool dry season (June to over a background of grazing and Traditional African pastoralists rely on September) and a warm dry season restricted access to milk and water. milk from livestock as their dietary (December to March). The dominant Treatments included 2 levels of supple- mainstay. In cattle production systems ethnic group is the Boran, who are semi- mental water intake (i.e., ad-libitum and pastoralists may take up to 65% of total nomadic and herd cattle. Forage zero) and 3 levels of supplemental hay milk yield, with the remainder left for resources are dynamic. Standing crops intake [i.e, ad-libitum (high), 60% of calves (Dahl and Hjort 1976, Holden et of herbaceous material can vary from high (medium), and zero]. The high al. 1991). It has been speculated that 1.5 to 4.5 tonnes DM ha-1 in dry and wet level of hay offered was set at 20% competition between calves and people periods, respectively (Menwyelet 1990). above the previous day's intake while for milk is a major contributor to lower Grazed portions of calf diets, on a DM the medium level offered was 60% of weaning weights, stunting, and delays in basis, can have 17% CP and 73% in the high level. Alfalfa is not traditional- time to puberty for cattle (Dahl and Hjort vitro digestibility in wet periods, and ly grown in these rangelands. It was 1976, Waters-Bayer 1988, Preston this can decline to <7% CP and 55% selected because its feeding value is 1989). Given that reliance on milk by digestibility in dry seasons (Unpub- well-recognized and it offered the best pastoralists is unlikely to change, it has lished data, Coppock). Calving is highly possibility of eliciting improved cattle been hypothesized that a major produc- seasonal, as 60% of births occur during performance. Local legumes would have tion intervention would be to compen- the long rains. This birth peak is accom- to substitute for alfalfa in new feeding sate calves for nutrients lost as a result of panied by a flush of milk production systems for Boran calves (Coppock and milk restriction. This strategy could have which subsequently declines in dry peri- Reed 1992). Alfalfa was harvested at major implications for improving animal ods. Cattle drink from ponds in rainy 50% flowering, baled after sun drying, production if carry-over effects occurred periods. In dry periods cattle drink water and chopped to 25-cm lengths before as sustained improvements in mature lifted from deep wells using human feeding. Chemical content (DM basis) weights and reductions in time to puber- labor. Watering frequency for adult cat- was 21.2% CP, 65.6% IVDMD, and ty for heifers (Cossins and Upton 1988). tle can vary from daily (rainy periods) to 42% NDF (Coppock and Reed 1992). Improved nutritional management for once every 4 days (at the end of dry Supplements were offered nightly at calves could be implemented using cut- periods). Limitations of water and for- 1800 hours in buckets to individual and-carry feeding with improved forages age, therefore, can constrain animal pro- calves under confinement. The 6 factori- and perhaps additional water in semiarid, ductivity. Water restriction is permitted al treatments were superimposed over a pastoral production systems. by the relatively cool climate and is background that simulated traditional The core objectives of this work were implemented to minimize labor calf management from birth to weaning. to experimentally test hypotheses that (Nicholson 1987). As with other pastoralists, calf manage- supplementation of nursing calves with ment by the Boran involves intensive alfalfa hay ( Medicago sativa L.) and/or hand-rearing, and, except for brief bouts water could: (1) compensate animals for Animals, Diets, and Experimental of suckling, calves are kept from their moderate levels of milk restriction until Design dams to ensure sufficient milk for peo- weaning, and (2) elicit sustained and A growth trial was repeated for 2 ple. The background in our trial consist- significant improvements in the subse- groups of local Boran calves (Bos indi- ed of: (1) individually housing calves at quent performance of animals up to sex- cus) born in consecutive years. Alberro night in small mud-and-thatch huts, (2) ual maturity. A third objective was to (1986) describes the dual-purpose Boran limiting daily milk intake for calves to use the experimental results to help breed. around half of production by allowing answer the question of whether or not Early in 1986, 147 Boran cows in the calves to suckle 2 of 4 quarters from calf supplementation could be an eco- third trimester of pregnancy were pur- their dams for a few minutes in the nomically viable means to improve the chased from local pastoralists. Calves morning and evening, (3) after 2 months production system of Borana pastoralists were born during March to May, 1986, of age, allowing calves to graze daily in in southern Ethiopia. stratified by sex and birthdate, and ran- a calf herd on Pennisetum mezianum domly allocated among 7 treatments Leeke range, and (4) altering watering (i.e., n = 21 for year 1). Cows were sub- frequency as water availability changed Materials and Methods sequently re-bred and produced 119 with season. Background watering fre- more calves during March to May, quency for all cows and calves in the 1987. These calves were similarly Study Area trial varied from daily at the height of assigned to the same 7 treatments (i.e., n Research was conducted on and near the rainy season to once every 3 to 4 = 17 for year 2). Local rainfall varied Demballa Wachu ranch, a government- days at the height of the dry season. A

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 209 seventh treatment was not part of the Shoulder height was used to assess terone analyses were positive, the date factorial design and calves in this group frame size and to derive a weight:height of mating was considered to be the date did not receive supplements. The back- ratio as an index of body condition. of puberty. Routine rectal palpations ground for the seventh group only dif- After weaning, calves were no longer were also conducted for all animals on a fered in that these calves received about housed nightly in huts and managed by monthly basis to provide a backup in 12 hours of continuous access to milk research staff, but were kept at nearby case matings were unobserved. each night because they were corralled pastoral encampments. Research ani- with their dams. The seventh treatment mals were maintained under traditional Statistics was intended to illustrate production management until females reached Our experiment proceeded smoothly potential in the absence of milk compe- puberty 3 to 4 years later. Pastoral man- from April 1986, until June 1991. The tition with people. It would also provide agement of immatures included corral- first group of animals were successfully a comparative baseline to quantify ing for about 12 hours each night and monitored as planned. The trial was whether supplementation could off-set grazing by day in mixed savanna sites. scheduled to end by August 1991, after effects of milk restriction. Watering frequency again varied from the second group of heifers would have Supplemental hay or water was first daily (wet periods) to once every 3 to 4 all become puberal. Regional insecurity offered when calves began to graze, and days (dry periods). Travel to dry-season associated with an overthrow of the cen- supplementation continued until calves watering points required 20-km round- tral government, however, required that were weaned. The treatment period trips from encampments. work be abruptly terminated in mid- roughly occurred from July through At the encampments liveweights and 1991. This meant we had to use a piece- March each year, ending at the onset of shoulder heights of males were mea- meal approach for data analysis given the long rains in April of 1987 and sured monthly. Females were similarly that the first group of animals had been 1988. The treatment period thus covered monitored until they were 24 months fully evaluated and many in the second an average of 9 months in 1986Ð7 (year old, anticipated to be the earliest age at group had not. For example, because 1) and 7 months in 1987Ð8 (year 2). which they could become puberal. At 24 most heifers in the second group had not Daily intake of supplements was expect- months females were run with breeding been observed to reach puberty, only ed to fluctuate according to seasonal bulls at a ratio of 20:1. Every 10 days reproductive data from the first group of resource dynamics and increase as females were weighed and blood was heifers were analyzed. The main analy- calves grew. Range forage was senes- collected from jugular veins for analysis ses included factorial ANOVA to cent for about 5 to 7 months of the treat- of plasma progesterone. Blood was cen- assess: (1) effects of birth year, hay ment periods in 1986Ð7 and 1987Ð8, trifuged on site and plasma was stored at level, and water level on liveweight, respectively. Forage was green during Ð20¡C for future analysis using the shoulder height, and weight:height ratio and briefly after the short rains in Enzyme-Linked Immuno Assay for calves at weaning as well as average October and November of both years. (ELISA) technique (Mukasa-Mugerwa daily gain (ADG) and total gain from Hay and water refusals were removed et al. 1989) using Ovucheck test kits birth to weaning, (2) effects of birth and measured each morning, and intake (Cambridge Veterinary Sciences, UK). year, hay level, and water level on was estimated by difference. Milk intake Onset of puberty was verified using 3 liveweight, shoulder height, and was estimated bi-weekly for each calf complementary methods: (1) animals weight:height ratio for the first group of using a weigh-suckle-weigh method were observed for successful mounts, heifers at puberty, as well as on ADG (Coppock and Reed 1992). After suck- recorded as to date and hour, (2) rectal and total gain for these heifers from ling each cow was hand-milked and the palpations were conducted for pregnan- weaning to puberal age, and (3) effects residual yield was measured. This cy diagnosis (through detection of of the same factors on the same allowed estimation of total milk yields asymmetry of uterine horns and pres- response variables as just mentioned in and percent of total yields received by ence of enlarged embryonic vesicula) at (2), but using a terminal age of 3.5 years calves. Background (i.e., non-supple- 40 and 60 days post-mating, and (3) for the second group of heifers and both mental) water intake was measured bi- analysis of plasma progesterone was groups of males, with sexes analyzed weekly when all calves drank from conducted for 4 samples collected with- separately. Other factorial ANOVA known amounts in buckets. Daily water in 10 days before and 30 days after an were used to assess patterns of hay, intake was calculated by dividing observed mount. Plasma progesterone -1 water, and milk intake of calves during amounts consumed by watering frequen- levels over a threshold of 1.3 ng ml treatment periods to confirm that the cy. Intake of the grazed diet was not were considered indicative of a signifi- trial was adequately implemented and measured because logistics in this cant rise in hormone concentrations detect interactions among dietary com- remote area prohibited fecal collection associated with estrous and an active ponents and year. Factorial ANOVA for large numbers of nursing calves corpus luteum. This conservative limit were also used to examine ancillary fea- (Coppock and Reed 1992). We therefore was specifically tailored for our sample tures of the trial such as total milk yield focused on an empirical assessment of population (Personal communication, of cows, percent of total milk yield con- supplementation effects. Mukasa-Mugerwa). The value of 1.3 ng -1 sumed by calves, background and total Prior to weaning liveweights and ml represented the upper limit of 2 water intake, etc. Simple linear regres- shoulder heights of each calf were mea- standard deviations above the mean sion was used to calculate ADG per sured bi-weekly. Shoulder height was basal concentrations of plasma proges- head and to correlate hay intake with measured from the base of the right terone (Unpublished data, Sovani). If total water intake for calves in each front hoof to the top of the scapula. the pregnancy diagnosis and proges-

210 Journal of Range Management (52)3, May 1999 birth year. An ANOVA was used to Averaged across water levels and background of 1.9 liters head-1 day-1 (not contrast mean production responses of years, hay DM intake averaged 344 and tabulated). In drier year 2, targeted animals from each of the 6 factorial 557 g head-1 day-1 for the medium and calves consumed an average of 4.0 liters treatments with those from animals in high levels offered, respectively (Table head-1day-1 of supplemental water. This positive control groups for both years. 1). Hay intake was affected by year and increased intake of water 3.5-fold to 5.6 These analyses were conducted to eval- showed an interaction with supplemen- liters head-1 day-1 from a background of uate responses from birth to weaning tal water (both at P=0.0001). Averaged 1.6 liters head-1 day-1. and weaning to a terminal age of 3.5 across the 2 levels of hay offered, calves There were main effects of birth year years. All ANOVA employed least- ate 307 versus 595 g head-1 day-1 in year on calf age at weaning, total milk yield squared means (SAS 1987) and milk 1 and 2, respectively. Compared to year of cows, and daily and total milk intake intake (ml kg LW0.73) was used as a 1 and averaged across water levels, hay for calves (all at P < 0.01). Weaning age covariate in ANOVA when positive intake for animals on the medium level for the first and second groups of calves controls were omitted. Variation among in year 2 increased by 103% from 227 to averaged 324 and 270 days, respective- means was interpreted as significant 462 g head-1 day-1 while hay intake for ly. Daily milk intake by calves varied when P<0.05. Dunnett's procedure (Day animals on the high level increased in from an average of 1.12 to 1.05 liters and Quinn 1989) was used to make 6 year 2 by 87% from 387 to 727 g head-1 head-1 day-1 in years 1 and 2, respective- planned comparisons between the posi- day-1. Calves in year 1 that received sup- ly. Daily milk intake exhibited a signifi- tive control group and each of the 6 fac- plemental water ate 27% more hay than cant hay x water interaction (P = 0.002); torial treatments in both years. Contrasts those which did not receive supplemen- milk intake was higher for traditionally for linear trend were used to examine tal water (i.e., 271 versus 343 g head-1 managed controls over both years main effects of hay feeding level when day-1). In year 2 calves which received (Table 1). Total milk intake for calves higher-order interactions were absent supplemental water ate 20% more hay varied from 363 to 283 liters head-1. (Winer et al. 1991). than those which did not receive supple- Total milk yield (i.e., intake plus off- mental water (i.e., 541 versus 649 g take) averaged 1.79 to 1.67 liters cow-1 head-1 day-1). Linear regression results day-1 in years 1 and 2, respectively, or a Results for year 2 indicated that hay DM intake total yield of 580 to 472 liters cow-1 lac- (y) in g head-1 day-1 was related to total tation-1. Calf milk intake averaged 67% Trial Implementation water intake (x) in ml head-1 day-1 by the of total yield with no effect of year (P = Intakes of supplements and milk by equation y = 0.036x + 322.8 (P = 0.77). For supplementation to compen- calves according to factorial treatment 0.0062; r2 = 0.11; df = 66). No such rela- sate calves for milk restriction it would are illustrated in Table 1. There were tionship, however, was apparent for year have to offset foregone milk volumes some significant interactions involving 1 (P = 0.23). ranging from 189 to 217 liters head-1 year and intake of water, hay, and milk Intake of supplemental water was 25% over the 2 years. that merit brief review. It is useful to higher in year 2 compared to year 1 recall that year 1 had nearly 60% higher overall (Table 1). There was also a sig- Treatment Responses: Birth to rainfall than year 2, and this had a nificant hay x water x year interaction (P < 0.001) indicating that calves on the Weaning marked influence on most aspects of the The year x hay x water interaction (P trial. Compared to year 1, the relative high level of hay drank relatively more supplemental water than other groups, = 0.0001) for weaning weight is shown dryness of year 2 reduced the time that in Figure 1; this primarily illustrates the green forage was available for cows or especially in year 2. In wetter year 1, effect of supplemental water in allowing calves and reduced availability of drink- targeted calves consumed an average of -1 -1 animals to better utilize hay in both ing water. Calves therefore tended to 3.2 liters head day of supplemental years. One source of the interaction of consume more supplements, and less water. This increased intake of water -1 -1 feeding treatment with year was annual milk, in year 2 than year 1. 2.7-fold to 5.1 liters head day from a

Table 1. Average daily intake (x± SE) of supplemental hay (g head-1 day-1 on a DM basis), supplemental water (liters head-1 day-1), and milk (liters Kg LW-0.73 day-1) for calves in 6 factorial treatments over 2 years1.

Supplemental Hay Level Zero Medium High Year Supplemental water? Hay Water Milk Hay Water Milk Hay Water Milk (g head-1 (liters head-1 (liters kg (g head-1 liters head-1 (liters head-1 (g head-1 (liters head-1 (liters kg day-1) day-1)LW-0.73 day-1) day-1) day-1)LW-0.73 day-1) day-1) day-1)LW-0.73 day-1) No 0 0 1.3±0.07 207±3 0 1.0±0.07 333±5 0 1.1±0.07 1 Yes 0 3.1±0.02 0.9±0.08 245±5 3.1±0.01 1.1±0.08 441±4 3.5±0.01 1.2±0.09 No 0 0 1.3±0.07 460±4 0 1.0±0.09 621±5 0 0.9±0.09 2 Yes 0 3.5±0.02 0.9±0.09 464±3 3.9±0.03 0.9±0.09 833±5 4.7±0.03 1.0±0.09 1Based on n = 21 or 17 calves per treatment in years 1 and 2, respectively. Intake of supplements was measured daily while milk intake was measured bi-weekly. A seventh treatment of calves served as a “positive” control group. Calves in this treatment received greater access to milk but no supplemental hay or water. Milk intake was not measured for the seventh treatment because the calves had continuous access to their dam’s milk overnight.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 211 For animals born in year 1, the productive performance of positive controls (i.e., those which received over-night access to milk) exceeded that for all groups except those which had received the medium or high level of hay plus extra water (Table 3). For animals born in year 2, the performance of positive controls exceeded that for all groups except the one which received the high level of hay plus water. These results are interpreted to indicate that supplementation with hay and water indeed helped calves fully compensate for milk restriction. In the first year 116 kg of hay plus 924 liters of water helped offset 217 liters of milk foregone per calf. In the second year 175 kg of hay plus 987 liters of water helped offset 189 liters of milk foregone per calf.

Treatment Responses: Weaning to Maturity Males Overall, main effects of birth year had the most pervasive effects on growth features for males from weaning to 3.5 Fig. 1. Interaction of birth year, hay feeding level, and watering level on weaning weights of Boran years of age (Table 4). Compared to calves (both sexes combined; P = 0.0001). The medium level of alfalfa hay intake (DM basis) males born in the second year, those averaged from 227 to 462 g head-1 day-1 in 1986Ð7 and 1987Ð8, respectively. The high level of alfalfa hay intake averaged from 387 to 727 g head-1 day-1 in 1986Ð7 and 1987Ð8, respectively. born in the first year had growth rates Intake of supplemental water averaged from 3.2 to 4.0 liters head-1 day-1 in 1986Ð7 and 1987Ð8, that were 14% higher, with an advan- respectively. tage in total gain of 8%. At 3.5 years of age males born in the first year were variation in the relative impact of treat- interactions and these data are reported 20% larger in weight and 9% taller at ments compared to traditionally man- in Table 2. Compared to the respective, the shoulder compared to those born in aged controls. For example, the water traditionally managed controls in suc- the second year. plus high hay treatment yielded weaning cessive years, the water plus high hay Main effects of hay or water supple- weights 31% higher than those of the treatment yielded ADGs from 38 to mentation were not significant for controls in the first year (i.e., 108 versus 138% higher, total gains from 41 to liveweights of males at 3.5 years of age 82 kg head-1); in the second year this 108% higher, shoulder heights from 4 to (P>0.18). There was an interaction, how- difference increased to 63% (i.e., 80 6% higher, and height:weight ratios 25 ever, of water x hay for liveweights of versus 49 kg head-1). Other response to 54% higher. males at 3.5 years (P = 0.04). We inter- variables exhibited qualitatively similar preted this pattern to suggest that ani-

Table 2. Interaction of birth year with level of hay and water supplementation on calf growth features (x±SE) from birth to weaning in 6 factorial treatments.

Supplementation Group Growth Feature Average Daily Gain Total Gain Shoulder Height Weight:Shoulder Height Water level Hay Level 1986Ð7 1987Ð8 1986Ð7 1987Ð8 1986Ð7 1987Ð8 1986Ð7 1987Ð8 (g head-1 day-1) (kg head-1) (cm head-1)1 [(kg:cm) head-1]1 No Zero 193±9.6 93±10.4 62±3.7 28±4.0 93±1.1 87±1.2 0.88±0.03 0.56±0.03 No Medium 199±10.5 140±13.8 64±4.0 39±5.3 93±1.2 88±1.6 0.89±0.03 0.68±0.04 No High 217±10.7 158±13.8 70±4.1 43±5.3 93±1.2 88±1.6 0.97±0.03 0.71±0.04 Yes Zero 209±11.5 140±12.9 66±4.4 38±4.9 92±1.3 89±1.5 0.93±0.04 0.67±0.04 Yes Medium 261±10.9 168±12.1 81±4.2 43±4.7 95±1.2 89±1.4 1.00±0.03 0.72±0.04 Yes High 266±11.3 222±11.8 88±4.3 57±4.5 97±1.3 92±1.3 1.10±0.03 0.86±0.04 F-test probability for 3-way interaction2 *** ** ** ** **,***Significant at the 0.01 and 0.0001 levels respectively. 1Measurements taken at weaning. 2Where n = 38 calves per treatment over 2 years, with 224 error degrees of freedom in each ANOVA.

212 Journal of Range Management (52)3, May 1999 Table 3. Growth features (x±SE) from birth to weaning for cavles in “positive” control groups having overnight access to their dam’s milk1.

Growth Feature Birth Year n Birth Weight Weaning Weight Average Daily Gain Total Gain Shoulder Height Weight:Shoulder Height (kg head-1) (kg head-1) (g head-1 day-1) (kg head-1) (cm head-1) [kg:cm)head-1] 1986Ð7 21 19±1.4 111±7.8 253±20 97±7.9 99±2.4 1.10±0.06 Dunnett’s test probability of the positive control versus: High hay plus water NS NS NS NS NS NS Medium hay plus water NS NS NS NS NS NS All other factorial treatments NS ** ** ** ** ** 1987Ð8 17 22±0.9 85±5.3 201±14 63±5.4 98±1.6 0.88±0.04 Dunnett’s test probability of the positive control versus: High hay plus water NS NS NS NS NS NS All other factorial treatments NS ** ** ** ** ** **Significant at the 0.01 level using the 2-tailed Dunnett’s procedure. 1To compare means with those of respective factorial treatments also see Table 1. 2Where Dunnett’s procedure was used for pre-planned comparisons between the positive control and the response from 6 factorial treatments in each year. The high hay plus water group received 3.5 to 4.7 liters head-1 day-1 of supplemental water and 441 to 833 g head-1 day-1 of supplemental hay (DM basis) during years 1 and 2, respectively. The medium hay plus water group received 3.1 to 3.9 liters head-1 day-1 of supplemental water and 245 to 464 g head-1 day-1 of supplemental hay (DM basis) during years 1 and 2, respectively. mals which had received supplemental that tended to exceed those found for nificant differences in liveweight (i.e., hay and supplemental water as calves animals in the other 4 treatments. 225 versus 226 kg head-1, Fig. 2). achieved higher liveweights at 3.5 years Averaged over both birth years, the Convergence in liveweight was reflected of age compared to those which had only maximum range among treatments in in the main effects of hay level on received supplemental hay (Fig. 2). terms of weaning weights for males was growth after weaning. Growth rates and Another notable result was the trend for 28 kg head-1 [i.e., 66 kg head-1 (controls) total gain were inversely related to the traditionally reared controls to have sim- versus 94 kg head-1 (water plus the high amount of hay animals had consumed as ilar mean liveweights at 3.5 years of age level of hay), a relative increase of calves (Table 5). compared to animals in the high hay plus 42%]. By 3.5 years of age, however, water treatment; both of these groups males from the same 2 treatments, aver- Females had mean liveweights at 3.5 years of age aged over both years, exhibited no sig- The total number of heifers born in the first year was 67 across 7 treatments. Fifty four (81%) had become pregnant before the trial ended. Forty-two percent of heifers conceived at their first behav- ioral estrous. In 48% of the first behav- ioral estrouses, however, animals were physiologically non-puberal according to analysis of their plasma progesterone levels, which were <1.3 ng ml-1. These animals tended to be younger and lighter in liveweight than those which exhibited sexual behavior that coincided with plasma progesterone concentrations >1.3 ng ml-1 (Unpublished data, Sovani). Mating activity was a function of nutritional plane and was thus highly seasonal. Ninety-six percent of conceptions occurred during rainy or post-rainy periods. These were divided between the short and long rains in a bimodal pattern. Forty-two percent of all conceptions occurred during or just after the short rains in November and December, while 54% occurred during or just after the long rains in March through June. Fig. 2. Interaction of hay feeding level and watering level on liveweight of male Boran cattle at 3.5 years of age (P = 0.04). Animals were born over 2 consecutive years. The medium level of alfalfa Heifers first began to get pregnant hay intake (DM basis) averaged 345 g head-1 day-1 over 2 birth years while the high level of alfal- during the short rains (October to fa hay intake averaged 557 g head-1 day-1. Intake of supplemental water averaged 3.6 liters head-1 November) of 1989, when some were day-1 over the 2 birth years.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 213 Table 4. Main effects of birth year on growth features of male Boran cattle from weaning to 3.5 whether animals received hay or water years of age as averaged across 6 factorial treatments as calves (Table 6). Animals which did

1 not receive extra water had to gain 128 Birth Year kg head-1, 16% more than those which Growth Feature 1986Ð7 1987Ð8 F-Test Probability -1 -1 had not received extra water (i.e., 110 Average Daily Gain (g head day ) 151 132 ** kg head-1). Puberty among animals Total Gain (kg head-1) 144 133 * -1 which did not receive extra water was Weight at 3.5 years (kg head ) 236 196 *** achieved over a period of time which Shoulder Height at 3.5 years (cm head-1) 124 114 *** -1 was 6% longer than those which Weight:Shoulder Height [(kg:cm)head ) 1.90 1.72 *** received extra water. Animals which did *,**,***Significant at the 0.05, 0.01, and 0.0001 levels respectively. not receive hay had to gain 126 kg head-1 1Where n = 63 males born in 1986Ð7 and n = 51 born in 1987Ð8, with 98 error degrees of freedom in the ANOVAs. The milk intake covariate was not significant for any case (P>0.10). after weaning to become puberal. This exceeded the gain needed by animals 40 months old. Following a period of no received water plus the high level of hay which received the high level of hay by sexual activity during the subsequent exceeded the controls by 59% (i.e., 127 15% (i.e., 109 kg head-1). Animals which warm dry season, the next (and last) versus 80 kg head-1). At puberty, howev- did not receive hay, however, took 10% wave of pregnancies for the first group er, these weight differences had largely longer to achieve puberty. of heifers occurred during the long rains disappeared. Groups also tended to be Despite that an analysis of puberal (April to May) of 1990, when the similar in terms of shoulder height and features could not be conducted for remaining animals averaged 48 months weight:height ratio at puberty. females from the second cohort, data of age. The overall mean age (± SE) at That weight at puberty converged collected from weaning to termination conception was thus 43.4 ± 1.5 months among treatments was not due to com- of the trial allowed another factorial (or 3.6 years). The mean liveweight (± pensatory growth per se as observed for analysis of growth patterns from wean- SE) at conception was 217 ± 7.1 kg the males. In contrast to the males, ing to a common minimum age of 3.5 head-1, roughly 96% of the mature which were analyzed at a common years for these animals (n = 8 to 9 per weight for females (i.e., 225 kg head-1; chronological age, time to puberty var- treatment). Patterns again demonstrated Alberro, 1986). ied for females due to different factor convergence among treatments in It was expected that heifers which had levels and this complicated interpreta- liveweight, shoulder height, and been born in the second year would tion of results (Table 6). Heifers that had weight:height ratios (all at P>0.05; not begin to get pregnant starting during the received extra water as calves became illustrated). This occurred even though short rains (OctoberÐNovember) of puberal 2.7 months earlier on average hay and water treatments resulted in 1990. These rains, however, were than those under traditional watering greater impacts on animals relative to sparse. Animals remained on a low (i.e., 42.5 versus 45.2 months of age, respective controls in the second year plane of nutrition and few conceptions respectively). Heifers which received compared to the first (Fig. 1). were recorded (Unpublished data, the high level of hay became puberal 2.6 Sovani). The main wave of pregnancies and 4.3 months earlier than those which Contrasts of Factorial Treatments for the second cohort was thus expected received the medium and zero levels of with Positive Controls at Maturity to occur in April to May of 1991, but hay, respectively (i.e., means of 41.6, An ANOVA was conducted across all this was when security collapsed and the 44.2, and 45.9 months). 7 treatments to analyze effects of birth trial had to be terminated. Because of treatment variation in year and treatment on total liveweight Results from ANOVA which analyzed puberal age, groups differed in total gain gain and ADG from birth to 3.5 years of responses from weaning to puberty for from weaning to puberty, but did not age, and on liveweight, shoulder height, females which had been born in the first differ in terms of ADG. Animals thus and weight:height ratio at 3.5 years of year and conceived are shown in Table had to attain a similar target weight to age. Birth year was significant 6. These animals had exhibited the same become puberal, and this took different (P<0.005) for all response variables. interaction response to hay and water lengths of time depending on treatment. Considering the period from birth to 3.5 treatment in terms of weaning weight as Total gain needed to achieve puberty years of age, animals born in wetter year previously reported; the group which after weaning varied in relation to 1 had 25% higher mean (±SE) total liveweight gains (i.e., 204 ± 3.6 versus Table 5. Main effects of hay feeding level on growth of male Boran cattle from weaning to 3.5 years 163±3.3 kg head-1) and 26% higher of age as averaged for animals across 6 factorial treatments born in 2 successive years.1 mean ADG (i.e., 157 ± 3 versus 125 ± 3 -1 -1 Hay Feeding Level2 F-Test Probability g head day ). At 3.5 years of age ani- 3 mals born in wetter year 1 had a 21% Growth Feature Zero Medium High ANOVA Linear Trend higher mean liveweight (i.e., 223 ± 3.7 -1 -1 Average Daily Gain (g head day ) 153 142 129 * * versus 185±3.4 kg head-1), 10% greater Total Gain (kg head-1) 151 139 127 * * mean shoulder height (i.e., 124 ± 1 ver- *Significant at the 0.05 level. sus 114 ± 1 cm head-1), and a 12% 1Where each tabulated entry is the mean of n = 38 animals. Each ANOVA was based on 98 error degrees of freedom. The milk intake covariate was not significant for either case (P>0.10). greater weight:height ratio (i.e., 1.80 ± -1 2Where the medium level consumed was 227 and 462 g head-1 day-1 and the high level was 387 and 727 g head-1 day-1, 0.02 versus 1.62 ± 0.02 kg:cm head ). in the first and second years, rspectively, on a DM basis. In contrast, however, there were no sig- 3Linear contrast.

214 Journal of Range Management (52)3, May 1999 Table 6. Growth features from weaning to puberty, and time to puberty, for Boran heifers born in 1986 as influenced by supplementation in 6 factorial treatments.1

Feature Supplementation Group Weaning Puberal Average Daily Shoulder Time:Weaning Water Level Hay Level Weight Weight Total Gain Gain height3 Weight:Height3 to Puberty Age at Puberty (kg head-1) (kg head-1) (kg head-1)2 (g head-1 day-1)2 (cm head-1) [kg:cm)head-1] (days head-1) (months head-1) No Zero 80±6.0 213±7.4 133±7.2 119±7.6 123±2.1 1.7±0.05 1,129±53.0 48.1±1.61 No Medium 84±5.9 211±7.2 127±7.0 129±7.4 120±2.0 1.7±0.05 1.009±52.1 43.9±1.62 No High 88±4.9 212±6.0 124±5.9 124±6.1 119±1.7 1.8±0.04 993±43.3 43.7±1.31 Yes Zero 91±6.3 210±7.7 119±7.5 125±7.9 122±2.2 1.7±0.05 973±55.7 43.7±1.73 Yes Medium 112±5.6 230±6.9 118±6.7 121±7.0 124±1.9 1.8±0.05 974±49.4 44.5±1.54 Yes High 127±5.9 220±7.3 93±7.1 108±7.5 128±2.1 1.7±0.05 872±52.4 39.5±1.62 F-test probability: Hay ** NS * NS NS NS NS * Water *** NS ** NS * NS ** * Hay x Water * NS NS NS NS NS NS NS Linear Trend (Hay) — — — — — — — NS *,**,***Significant at the 0.05, 0.01, and 0.0001 levels respectively. 1Where n = 8 to 9 animals per treatment. Entries are x±SE. ANOVAs were based on 45 error degrees of freedom. See text for treatment details. The milk intake covariate was not significant in any case (P>0.13). 2From weaning to puberty. 3Measurements taken at puberty. nificant effects of treatment on mean taken by people, it could result in a 34% King (1983). Research intended to total liveweight gain (P = 0.48; range gross increase in the cash value of the enhance livestock performance in Africa 174 ± 5.9 to 199 ± 10.7 kg head-1), mean lifetime output per cow. It was assumed often has a sole focus on forage ADG (P = 0.35; range 134 ± 4 to 151 ± that major impacts of supplementation resources. Our findings indicate that 8 g head-1 day-1), or attributes at 3.5 would include a doubling of weaning benefits from improved forages would years including mean liveweight (P = weight from 47 to 94 kg head-1 and a not be realized in a dry region such as 0.14; range 195 ± 6.0 to 221 ± 10.9 kg reduction in time to puberty for heifers this if watering constraints were ignored. head-1) and mean weight:height ratio (P from 4 to 3 years. This analysis support- Including water supplementation in our = 0.55; range: 1.65 ± 0.04 to 1.75 ± 0.07 ed the theory that lifetime performance of trial was therefore extremely useful. kg:cm head-1). The only persistent effect cattle in pastoral systems is significantly Given our experimental results, an of treatment was on shoulder height, limited by competition between cattle important question is whether calf sup- which was significantly highest in the and people for milk (Dahl and Hjort plementation should be promoted here positive control group (P = 0.005). 1976, Pratt and Gwynne 1977, Wagenaar or in similar systems in attempts to Averaged over both years, animals et al. 1986, Waters-Bayer 1988). markedly improve weaning weights. which received over-night access to We interpret our results, however, to This could be a valid point of interven- milk as calves had a mean shoulder basically refute the proposition that tion if higher weaning weights were an height of 126 ± 2.4 cm head-1. This enhancing calf growth via supplementa- important production goal and offered exceeded means for the 6 factorial treat- tion is an appropriate means to trans- an economic advantage to the producers. ments by an average of 7%. The range form this particular pastoral economy. We have eventually learned, however, of mean shoulder heights for the factori- This perspective is based on a synthesis that neither of these conditions are gen- al treatments was 116 ± 1.4 to 121 ± 1.2 that includes empirical results from this eralizable to the Borana pastoral system. cm head-1. Overall, we interpret these trial, an improved understanding of pas- The critical point is that the traditional results to indicate that effects of birth toral production goals, and considera- production rationale in the Borana sys- year on cattle growth were much more tion of how risk and uncertainty affect tem is not the same as that of a western persistent compared to those from treat- production decisions. These are all con- cow-calf operator (Coppock 1994). In ments. This illustrates the powerful role sidered further below. the Borana system overhead costs of of environment on cattle productivity in production are relatively low, immature this system. Short-Term Effects: Birth to Weaning stock are commonly retained by produc- Our trial results supported the hypoth- ers to build-up herds, and mature cattle over 6 years years of age are the pre- Discussion esis that supplementation could compensate calves in terms of weaning weight ferred animal for most households to for a moderate loss of milk, but it is sell. In contrast to selling mature stock, Cossins and Upton (1988) used analyti- notable that extra water was required in selling immatures is more common cal modeling to evaluate potential eco- addition to high-quality forage to among poorer households who are nomic impacts of improved calf manage- achieve this effect. Traditional watering forced to sell immatures because they ment in the Borana pastoral system. They practices thus appear to limit dry matter have small and less-diverse inventories estimated that if supplemental forage intake for such calves, as noted in gener- of animals and a chronic need to fre- could fully compensate calves for milk al for pastoral livestock in Africa by quently buy cereal grains or other non-

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 215 pastoral foods. Selling a mature male Long-Term Effects ratios. We speculate that supplementation not only provides money for a house- Growth Dynamics resulted in subtle improvements in body hold to buy various commodities, but Our results falsified the hypothesis that condition, undetected by our monitoring also allows the seller to use the substantial increases in early growth rates system, that predisposed some animals to "change" to buy a couple replacement and weaning weights would result in sus- cycle one rainy season earlier during their calves. The sale of a mature male thus tained increases in liveweight or frame puberal year (Mukasa-Mugerwa 1989). simultaneously contributes to procure- features for these cattle. Regardless of Reductions in time to puberty ranged ment of basic necessities as well as tra- what level of milk, hay, or water animals from 2.6 to 4.3 months compared to tradi- ditional values of herd building. received as calves, and despite relative tionally managed animals, less than half If we assumed higher weaning weights differences in weaning weights that of the advantage assumed in analyses by were an important production and mar- exceeded 60% among treatments in some Cossins and Upton (1988). keting goal, could it be profitable to sup- cases, the animals were remarkably simi- While a reduction in time to puberty plement calves? Assuming input costs of lar by 3.5 years of age. Convergence was of 2 to 4 months would be very signifi- US $0.15 per kg DM of legume hay and typically due to compensatory growth of cant in western production systems US $0.03 per 10 liters of water based on traditionally managed animals. This is where beef heifers become puberal at 12 labor profiles and other criteria consistent with findings elsewhere which to 15 months of age, it is important to (Mulugeta 1990), and considering calf indicate that cattle can overcome early once again view the benefits, costs, and performance in the high hay and water nutritional deprivation as calves (Laster risks of intervening in an African con- treatment over both years, 28 kg of extra et al. 1976, Richardson et al. 1978, text where the typical heifer is sexually weight was produced per calf as a result Tawonezvi 1989). These authors found mature at 36 to 48 months of age and is of 146 kg of hay (valued at US $2.86) that differences in weaning weights of reared in a much more challenging envi- and 956 liters of water (valued at US calves due to variable consumption of ronment. One risk to a pastoral producer $21.90). The selling price for fattened milk and/or solid foods only persisted is the likelihood that hard-won gains in calves is US $1.00 per kg liveweight from 1.5 to 11 months. This is not to say, animal productivity due to supplementa- (Unpublished data, Coppock). A net however, that permanent stunting from a tion investment could be lost in the return of US $3.24 per calf does not higher degree of milk restriction could future due to uncontrollable events. appear very profitable, and importantly, not occur in some cases. Milk provides Such events include disease outbreaks, the costs estimated by Mulugeta (1990) important bypass nutrients, and if these predation, and thievery; even our trial did not incorporate the substantial risks are severely restricted permanent stunting had to be terminated by 1991 because of and problems of acquiring sufficient in cattle can result (Preston 1989). insecurity. Even if these events could be quantities of supplements. The true costs mitigated, the problem of variable rain- of supplementation were probably fall may be the most insidious. For Time to Puberty underestimated, further reducing profit. example, 1 simple way to lose a hard- The marked seasonality of conception We speculate that in contrast to won, 2 to 4 month advantage in time to observed in this study is common in improving weaning weights, supplemen- puberty is to merely have the first rainy pastoral Africa (de Leeuw and Wilson tation to mitigate calf mortality under season fail in a heifer's puberal year. As 1987). That Boran heifers conceived at severe milk-restriction regimes is a observed in our trial, failure of one rainy 96% of mature weight is higher than fig- more relevant production objective. This season could adversely affect nutrition ures reported elsewhere for Bos indicus of heifers and bulls and delay concep- would be especially important to poorer (Macfarlane and Worrall 1970, Wilson households which have lower ratios of tion until the next favorable rainy period 1987) or B. taurus (Dobson and 6 months (or more) later. Another aspect milking cows per person compared to Kamonpatana 1986). This suggests that middle-income or wealthier households. is the often calamitous impact of Boran cows achieve physiological and drought. Herd dynamics in southern The poor tend to deprive their calves of physical maturity at roughly the same milk to a critical extent in order to pro- Ethiopia indicate that up to 50% of the time (Fitzhugh 1976). The nonpuberal cows may die every 5 to 8 years, largely vide more milk for family members, estruses and prepuberal rises in plasma especially children (Holden et al. 1991, due to interactions among stocking rate, progesterone levels observed here have rainfall fluctuation, and bottlenecks in Coppock 1994). Although we do not been reported elsewhere (Nelsen et al. have empirical data on the amount of regional marketing systems (Coppock 1985, Rutter and Randel 1986), and jus- 1994). Such a situation clearly mitigates supplemental forage or water needed to tified use of several puberty detection mitigate calf mortality under extreme against substantive investments to methods. Bulls were also an important improve animal productivity over the cases of milk deprivation, we expect factor underlying seasonal patterns of long term. Such risk perspectives were that the benefit/cost ratio would be more conception. Bulls typically showed no not incorporated into the deterministic favorable compared to using supplemen- sexual behavior until they had recovered analyses of Cossins and Upton (1988). tation to stimulate gain. For example, if body condition during wet seasons a female calf can make it to 2 years of (Unpublished data, Sovani). age in the Borana system, the likelihood Our results are interpreted to suggest Conclusions of subsequent survival, and hence bene- that early supplementation can result in a fit of a lifetime of productive output, persistent reduction on time to puberty for dramatically increases (Coppock 1994). Our empirical results indicated that heifers, despite convergence among treat- nutritional supplementation could help ments in liveweight and weight:height Boran cattle compensate for moderate

216 Journal of Range Management (52)3, May 1999 levels of milk restriction experienced as Day, R.W. and G.P. Quinn. 1989. Pratt, D.J. and M.D. Gwynne (ed.). 1977. calves. Supplemental water was Comparisons of treatments after an analy- Rangeland management and ecology in required in addition to legume hay, sis of variance in ecology. Ecol. Monogr. East Africa. Hodder and Stoughton Co., however, and underscores the crucial 59(4):433Ð463. London. role of watering constraints in this and de Leeuw, P.N. and R.T. Wilson. 1987. Preston, T.R. 1989. The development of Comparative productivity of indigenous milk production systems in the tropics. similar systems. The compensatory cattle under traditional management in Tech. Center for Agr. and Rural effect of supplementation was most sub-Saharan Africa. Quart. J. Int. Agr. Cooperation, Wageningen. apparent at weaning, and there was also 26:377Ð390. Richardson, F.D., Oliver, J., and G.P.Y. a slight carry-over effect in terms of a 2 Dobson, H. and M. Kamonpatana. 1986. Clarke. 1978. The pre-weaning and post- to 4 month reduction in time to puberty A review of female cattle reproduction weaning growth of beef calves in relation for heifers. Most other production fea- with special reference to a comparison to the amounts of milk and solid food con- tures, however, had completely con- between buffaloes, cows, and zebus. J. sumed during the suckling period. verged by the time animals reached Reprod. Fertil. 77:1Ð36. Rhodesian J. Agr. Res. 16:97-108. maturity. This was largely due to com- Fitzhugh, H.A. 1976. Analysis of growth Rutter, L.M. and R.D. Randel. 1986. pensatory growth of control animals. curves and strategies for altering their Nonpuberal estrus in beef heifers. J. Anim. shape. J. Anim. Sci. 42:1036Ð1051. Sci. 63:1049Ð1053. Despite some of these results, further Holden, S.J., D.L. Coppock, and Mulugeta SAS. 1987. SAS/STAT TM Guide for per- consideration of the rationale and real- A. 1991. Pastoral dairy marketing and sonal computers, 6 ed. Statistical Analysis world risks of pastoral production has household wealth interactions and their System Institute, Inc., Cary, N.C. led us to conclude that supplementation implications for calves and humans in Tawonezvi, H.P.R. 1989. Growth of to enhance anything but animal sur- Ethiopia. Hum. Ecol. 19(1):35Ð59. Mashona cattle on range in Zimbabwe. I. vivorship is a highly risky, and a likely King, J.M. 1983. Livestock water needs in Environmental influences on liveweight wasteful, use of scarce resources. Low- pastoral Africa in relation to climate and and weight gain. Trop. Anim. Health. input management is therefore justified forage. Res. Rep. 7. International Prod. 21:37-42. when production risks are high. All Livestock Center for Africa, Addis Ababa. Wagenaar, K.T., Diallo, A., and A.R. things considered, our work also under- Laster, D.B., Smith, G.M., Cundiff, L.V., Sayers. 1986. Productivity of transhumant and K.E. Gregory. 1976. Characterization Fulani cattle in the inner Niger delta of mines the notion that moderate levels of of biological types of cattle. IV. Mali. Res. Rep. 13. International milk consumption by pastoralists consti- Postweaning growth and puberty of heifers. Livestock Center for Africa, Addis Ababa. tutes a significant loss for the animal J. Anim. Sci. 43:500Ð508. Waters-Bayer, A. 1988. Dairying by settled production system; in our case this was Macfarlane, J.S. and K. Worrall. 1970. Fulani agropastoralists in central Nigeria: on the order of 189 to 217 liters of milk Observations on the occurrence of puberty The role of women and implications for per calf. Finally, the future risks of los- in Bos indicus heifers. East Afr. Agr. For. dairy development. Farming Systems and ing benefits of past investments in ani- J. p. 409Ð410. Resource Economics in the Tropics, vol. 4. mal production must be considered and Menwyelet, A. 1990. Ecology of calf pas- Wissenschafts Verlag. Kiel. incorporated into stochastic economic tures and supplementary feeding by Wilson, R.T. 1987. Livestock production in analyses of production interventions. Borana pastoralists of southern Ethiopia. central Mali: Factors influencing growth Master's thesis, Colorado State Univ., Fort and liveweight in agro-pastoral cattle. Such risks are particularly important in Collins, Colo. Trop. Anim. Health. Prod. 19:103Ð114. traditional pastoral production systems. Mukasa-Mugerwa, E. 1989. A review of Winer, B.J., D.R. Brown, and K.M. reproductive performance of female Bos Michels. 1991. Statistical principles in indicus (zebu) cattle. Monogr. 6. experimental design. 3rd ed. McGraw-Hill, Literature Cited International Livestock Center for Africa, Inc., New York, N.Y. Addis Ababa. Alberro, M. 1986. The Boran cattle and Mukasa-Mugerwa, E., Tegegne, A., their tribal owners. World Anim. Rev. Mattoni, M., and G. Cecchini. 1989. 57:30Ð39. Effect of oestrous synchronization with Coppock, D.L. 1994. The Borana Plateau of prostaglandin F2 alpha in Ethiopian zebu Southern Ethiopia: Synthesis of pastoral (Bos indicus) cows. Anim. Prod. research, development, and change, 1980- 48:361Ð371. 91. Systems Study No. 5. International Mulugeta, A. 1990. Borana cattle herds: Livestock Center for Africa, Addis Ababa. Productivity, constraints, and possible Coppock, D.L. and J.D. Reed. 1992. interventions. Master's thesis, Colorado Cultivated and native browse legumes as State Univ., Fort Collins, Colo. calf supplements in Ethiopia. J. Range Nelsen, T.C., Short, R.E., Phelps, D.A., Manage. 45(3):231Ð238. and R.B. Staigmiller. 1985. Nonpuberal Cossins, N.J. and M. Upton. 1988. Options estrus and mature cow influences on for improvement of the Borana pastoral growth and puberty in heifers. J. Anim. system. Agr. Systems 27:251Ð278. Sci. 61:470Ð473. Dahl, G. and A. Hjort. 1976. Having herds: Nicholson, M.J.N. 1987. The effect of Pastoral herd growth and household econ- drinking frequency on some aspects of the omy. Stockholm Studies in Social productivity of zebu cattle. J. Agr. Sci. Anthropology. Univ. Stockholm. (Camb.) 108:119Ð128.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 217 J. Range Manage. 52:218Ð227 May 1999 Stream channel and vegetation responses to late spring cattle grazing

WARREN P. CLARY

Author is project leader, Forestry Sciences Laboratory, Rocky Mountain Research Station, USDA Forest Service, 316 E. Myrtle St., Boise, Ida. 83702.

Abstract Resumen

A 10-year riparian grazing study was conducted on a cold, Se condujo un estudio de apacentamiento de áreas mountain meadow riparian system in central Idaho in ribereñas con una duración de 10 años. El estudio se llevó a response to cattle grazing-salmonid fisheries conflicts. Six cabo en la región central de Idaho en un sistema ribereño frío pastures were established along Stanley Creek to study the de pradera de montaña y fue motivado por los conflictos effects on riparian habitat of no grazing, light grazing surgidos entre el apacentamiento de ganado y las pesqueras (20Ð25% utilization), and medium grazing (35Ð50%) during de salmones. Se establecieron 6 potreros a lo largo del arroyo late June. Stream channels narrowed, stream width-depth Stanley para estudiar los efectos que el no apacentamiento, ratios were reduced, and channel bottom embeddedness apacentamiento ligero (20–25% de utilización) y apacen- decreased under all 3 grazing treatments as the area respond- tamiento moderado (35Ð50%) a fines de junio tienen sobre el ed to changes from heavier historic grazing use. Streambank hábitat ribereño. Los canales del arroyo se estrecharon, las stability increased and streamside willow communities (Salix relaciones ancho-profundidad de los arroyos se redujeron y spp. L.) increased in both height and cover under all 3 treat- los depósitos en el fondo del canal decrecieron, esto ocurrió ments. Plant species richness increased on both streamside bajo los 3 tratamientos de apacentamiento conforme el área and dry meadow areas during the years of grazing and mod- cambio de un uso histórico de pastoreo fuerte. La estabilidad erate drought. The numbers of species receded to near origi- del banco del arroyo se incrementó y las comunidades de nal levels in the ungrazed and light grazed pastures in 1996, a "willow" (Salix spp. L.) presentes a los lados del arroyo se wet post-grazing year, primarily due to a decrease in forb incrementaron tanto en altura como cobertura, esta respues- species. Streamside graminoid height growth was similar ta se obtuvo en los 3 tratamientos de apacentamiento. among treatments after 1 year of rest. Most measurements of Durante los años de apacentamiento y sequía moderada, la streamside variables moved closer to those beneficial for riqueza de especies vegetales se incrementó tanto a lado del salmonid fisheries when pastures were grazed to 10 cm of arroyo como en las áreas de pradera seca. En 1996, un año graminoid stubble height; virtually all measurements húmedo después de la estación de apacentamiento, el número improved when pastures were grazed to 14 cm stubble height, de especies vegetales de los potreros que recibieron apacen- or when pastures were not grazed. Many improvements were tamiento ligero o no apacentamiento regresó a niveles cer- similar under all 3 treatments indicating these riparian habi- canos a los originales, esta respuesta se debió principalmente tats are compatible with light to medium late spring use by a una disminución de las especies de hierbas. Después de un cattle. año de descanso, el crecimiento en altura de las gramíneas de los lados del arroyo fue similar entre tratamientos. Cuando los potreros se apacentaron a 10 cm de altura del rastrojo Key Words: riparian, mountain meadow, streambank stabili- remanente, la mayoría de las mediciones de las variables ty, width/depth ratio, willow, species richness, salmonid, fish- tomadas a los lados del arroyo se movieron a niveles cercanos eries, livestock management de los que son benéficos para las pesqueras de salmones. Cuando los potreros se apacentaron a 14 cm de altura del Riparian areas are among the most important features of nat- rastrojo remanente o no se apacentaron, virtualmente todas ural landscapes. Their biotic productivity and diversity stand las mediciones mejoraron. Muchas de las mejorías fueron out within the surrounding mosaic of terrestrial habitats similares bajo todos los tratamientos, indicando que estos (Kondolf et al. 1996). They typically function to moderate hábitats son compatibles con el uso ligero o moderado por el flood intensity, store water, and maintain water quality by act- ganado durante la primavera. ing as nutrient and sediment sinks (Hawkins 1994). These ecological attributes make riparian areas and the included

Support by personnel of the Sawtooth National Recreation Area and the streams highly valued for many human uses. One of these Sawtooth National Forest made this study possible. Appreciation is expressed to uses is livestock grazing. Major concerns about the impacts of John W. Kinney and various field crew members assisting in field data collection, grazing on riparian areas have been raised in the last 2 and to Rodger L. Nelson for instruction and assistance in stream channel measurements. Livestock were provided by the Stanley Basin Grazing Association. decades (Swanson 1988, US GAO 1988, Armour et al. 1994). Manuscript accepted 8 Aug. 1998. In earlier years, livestock grazing practices rarely addressed

218 Journal of Range Management (52)3, May 1999 the needs of riparian areas (Winward m. Stanley Creek is a 3rd order, C4 and disturbance by European man. 1994). Conflicting reports over the stream (Rosgen 1994). Soils are Typic Placer gold was discovered in Stanley effects of livestock grazing on riparian Cryaquepts formed in alluvium and Basin in 1863 (Van Noy et al. 1986). areas have pointed to a critical need to lacustrine sediments derived from gran- Mining in the upper portion of Stanley examine grazing practices that can ite. They have moderately slow to Creek began in the early 1870’s by potentially permit livestock production moderate permeability. The upper 23 cm ground sluicing. Dredge mining was while simultaneously preserving the is typically a silty clay loam overlying a conducted from 1900 to 1914. Various riparian characteristics needed for sandy to coarse sandy clay loam. Below forms of placer mining occurred from wildlife habitat, native fisheries, and 76 cm the profile contains 60% pebbles, 1933 to 1938 (Choate 1962). Obvious water quality (Waters 1995). Despite the cobbles, and stones (Personal communi- signs of mining activity are still present need for objective management strate- cation, D.R. Gilman). National Weather immediately upstream from the study gies, most current recommendations for Service records are incomplete, but annu- pastures. Other indications are also pre- improvement of riparian grazing are al precipitation during the treatment years sent that suggest heavy use by early set- based on collective experiences and case (1987Ð1995) appeared to have been tlers. Water diversion ditches and trail studies. Experimental examination of approximately 20Ð25% below the 389 roads are apparent in a number of loca- specific management hypotheses has mm average. This below average precipi- tions. A log-supported stream crossing occurred at only a limited number of tation period is referred to as a drought in is still present within the boundaries of 1 sites, including locations in Colorado the current study. The post-grazing year study pasture. (Schulz and Leininger 1990), Montana (1996), when final measurements were Sheep grazing began by 1879 and up (Marlow and Pogacnik 1986), Oregon taken, precipitation was 570 mm or to 200,000 sheep grazed during sum- (Bryant 1985, Green and Kauffman approximately 46% above average. mers in the Sawtooth Valley, although 1995), and Wyoming (Siekert et al. Average temperature during the June there is no current sheep use of the study 1985). grazing period is 11¡C; average annual area. Cattle grazing in Stanley Basin The need for more grazing manage- temperature is 2¡C (Personal communi- apparently started in 1899, but records ment information has been apparent in cation, Idaho Climate Services). of grazing use are not available earlier the Pacific Northwest as concern for the The area is representative of the than 1939 and little attention was given dwindling numbers of anadromous mountain meadows ecosystem in the to grazing management until the 1970s salmonids heightened riparian habitat forest zone of the mountain West con- (Sawtooth mountain area study: history. issues. A review of the Sawtooth taining wet to intermittently wet sites 1965. Copy on file Sawtooth National National Recreation Area by the Chief (Garrison et al. 1977). Typical stream- Recreation Area, Ketchum, Ida.) of the Forest Service and representatives side plant species included: Kentucky (Environmental analysis report: Stanley of The Wildlife Society, American bluegrass (Poa pratensis L.), tufted hair- Basin revegetation and rehabilitation Fisheries Society, and other concerned grass (Deschampsia cespitosa [L.] project. 1974. Copy on file Sawtooth parties defined several courses of action Beauv.), water sedge (Carex aquatilis National Recreation Area, Stanley, Ida.). to improve fish reproduction and migra- Wahl.), beaked sedge (C. rostrata Until fenced at the beginning of this tion, including development of Stokes), Baltic rush (Juncus balticus study, the study area was grazed as part improved riparian grazing systems Willd.), foxtail (Alopecurus spp. L.), of the Stanley Basin cattle allotment. (Peek and Gebhardt 1980). The present timber danthonia (Danthonia interme- Limited utilization records suggested a study was initiated in response to the dia Vasey), thick-stemmed aster (Aster 60Ð65% utilization rate of the dry mead- identified concerns about grazing-fish- integrifolius Nutt.), cinquefoil ow, tufted hairgrass sites, but no infor- eries conflicts in the Sawtooth National (Potentilla spp. L.), gentian (Gentiana mation was available on utilization rates Recreation Area. This study spanned a spp. L.), Lemmon’s willow (Salix lem- for streamside locations (Personal com- 10-year period and examined the monii Bebb), and Drummond willow (S. munication, L. Burton). Streamside uti- response of a cold mountain meadow drummondiana Barratt). The streamside lization rates were assumed to have been riparian system to 3 intensities of con- area, 7% of the pastures (Clary and high because some cattle usually had trolled late June cattle grazing. Booth 1993), was incised an average of access to Stanley Creek throughout the 0.38 m below the surrounding drier summer (Personal communication, B. meadow and averaged 16 m in width. Webster). No significant utilization by Study Area Idaho fescue (Festuca idahoensis wild herbivores was apparent during the Elmer), western needlegrass (Stipa occi- study. The grazing study was initiated on dentalis Thurb.), and mountain big sage- Stanley Creek, Sawtooth National brush (Artemisia tridentata Nutt. ssp. Recreation Area, Sawtooth National vaseyana (Rydb.) Beetle) were common Methods Forest, central Idaho in 1987. The study in the portion of the area away from the area is about 6 km northwest of Stanley, stream. These areas, referred to as the Field Procedure Ida., in portions of sections 19, 29, and “dry meadow,” were typically dry from Six experimental pastures, 3.7 to 9.0 30 T11N, R13E (Lat 44¡15'46''N, Long about mid July into fall, but bog-like ha, were established along Stanley 114¡59'02''W) where Stanley Creek areas and other areas of excess moisture Creek in fall 1986 (Fig. 1). Grazing was flows through a broad, flat valley with a were present in all years. conducted annually with cow-calf pairs westerly aspect at an elevation of 1,950 Stanley Creek and the surrounding in the last half of June from 1987 to meadows have had a long history of use 1995; except for 1993 when concerns

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 219 Fig. 1. Layout of experimental pastures on Stanley Creek in central Idaho. Treatments are medium (M), light (L), and no grazing (N). Black area represents the slightly incised stream and streamside area. about federal listing of chinook salmon Plant canopy cover, by graminoids, (based on linear proportion of active as a threatened species precluded graz- forbs, and shrubs, and litter were visual- banks estimated to be slumped, broken, ing. The last half of June corresponded ly estimated (Daubenmire 1959); herba- or eroding), channel bottom embedded- in nearly all years to the period when the ceous plant height was measured in cen- ness (rated as the proportion of the aver- dry meadow vegetation had made sub- timeters, and number of species was age perimeter of individual gravel, rub- stantial growth, and yet had sufficient recorded by plant life form. Plant attrib- ble, and boulder particles covered by soil moisture remaining to maintain for- utes were determined in each 0.25-m2 fine (<4.7 mm diameter) sediment), and age succulence. Since the dry meadow plot in 1987, 1990, 1994, and 1996. channel bottom textural composition pre-study utilization rates of 60Ð65% Height of the willow closest to each plot (Platts et. al 1987, Zweygardt and exceeded the recommended 55% maxi- was measured in centimeters at the Buckhouse 1996). mum to maintain healthy tufted hair- beginning and at the end of the study grass communities (Reid and Pickford (1987 and 1996). Plant community-type 1946) and the grazing appeared to have classifications were made within a Analyses negatively impacted the riparian habi- radius of 3 m from each 0.25-m2 plot in Although the pastures appeared to tats, the target utilization rates on the 1988 and 1996 following the general have similar characteristics when dry meadow portions of the pastures approach of Tuhy and Jensen (1982). fenced; it became apparent that each one were 50, 25, and 0% for the medium, Plant and soil moisture contents and was somewhat unique. Therefore, to light, and no grazing grazing treatments. their relationship to grazing distribution partially compensate for these initial dif- Two pastures were assigned to each of 3 on the study area were reported in Clary ferences, analyses were based on com- treatments: medium grazing (average of and Booth (1993). parisons between the initial reading for a 2.20 animal unit months [AUM] ha-1), - Percentage utilization to the nearest variable (1986 or 1987) and later read- light grazing (average of 1.27 AUM ha 5% was determined by visual estimation ings (1990, 1994, or 1996). Stream pro- 1), and no grazing. Stocking was adjust- (Pechanec and Pickford 1937) by file variables were analyzed as propor- ed so all pastures were grazed for a sim- graminoid, forb, and shrub categories tional changes because stream channel iliar period (usually 14 days). A 4-ha, 100-point sampling grid was based on comparison with 6 reference width and width/depth ratio were physi- established within 5 pastures with inter- cages per pasture. The cages were relo- cally limited in their potential response. point distances of 20 m; the 6th pasture cated at the start of the grazing period Other variables were analyzed based on had interpoint distances of 17 m. At each year. Beginning in 1988, mean numeric differences between initial and each point a 0.25-m2 plot was sampled residual stubble heights were measured later readings. for various vegetation and soil attribut- to the nearest centimeter immediately Variables were transformed as neces- es. Distribution of the 100-plot grid after each grazing period. Autumn sary to normalize data distributions. between streamside and dry meadow remeasurements were initiated in 1989 Transformations used were logarithm, locations varied among pastures because to determine season-end heights. square root, and arc sine for continuous of the variable size and location of the Thirty-one channel cross-sections variables, counts, and percentages and streamside areas. A second set of forty were systematically located along the angles. Average values presented in 0.25-m2 plots was concentrated near the stream within the boundaries of the plot tables were transformed back into the stream to provide a more detailed sam- grid in each pasture and measured mid original data form. Analyses of treat- ple of the streamside area in each pas- summer in 1986, 1990, 1994, and 1996. ment effects were conducted by ture. Analyses were based on 140 plots Variables measured included wetted Analysis of Variance (ANOVA) using a per pasture (100-plot grid plus 40 addi- width, average wetted depth, bank sta- General Linear Model. Repeated mea- tional plots) with 45 to 59 of the plots bility (based on estimated protection sures analysis was used when data per pasture sampling the streamside area from erosion provided by vegetation or included more than 1 response year. and 81 to 95 sampling the dry meadow by boulders and rubble), bank alteration Plant community-type frequency of area.

220 Journal of Range Management (52)3, May 1999 occurrence was examined by Chi-square compared to 1986. The width/depth the least change occurred under medium analysis. Significant differences among ratio decreased under all treatments as grazing (Table 3). The surface area com- means in the ANOVA tests were identi- compared to pre-study conditions at posed of fine textured sediments fied using a protected Fisher’s Least study end; the ungrazed treatment pro- increased or showed no change with Significance Difference (LSD). Additional duced greater decreases than did either medium grazing. Both the light and T-tests were conducted to determine if grazed treatment (Table 1). ungrazed pastures showed little reduc- responses within individual treatments Ratings of streambank stability tion in surface fines during the grazed differed from the initial readings in 1986 improved at a similar rate for the 3 graz- years, although the lightly grazed pasture or 1987. This was used as an aid in ing treatments (Table 2). Ratings of illustrated a significant reduction from interpretations of treatment trends, even streambank alteration decreased under initial conditions by the end of the study when no significant differences were all treatments by the end of the study; (Table 3). defined among treatments. Probabilities the ungrazed treatment showed the most of 0.05 or less were considered signifi- reduction (Table 2). Riparian Vegetation cant in all analyses. The analysis of Streamside Willows changes for most data are presented in 2 Channel Bottom Willows in the streamside area were periods: 1990 and 1994 during the Embeddedness changed differently scattered along most the length of grazed drought years compared with ini- among treatments. Embeddedness had Stanley Creek included within the study tial year (1986 or 1987); and 1996 the decreased in all treatments at study end; area. Willow heights increased during post-grazing, high precipitation year compared with the initial year. Table 1. Proportional changes in channel profile characteristics, Stanley Creek pastures.1

Stream profile characteristics Results Width Depth Width/depth ratio Grazing treatment 1990&1994 1996 1990&1994 1996 1990&1994 1996 Graminoid utilization averaged 35.2% —————————(Proportion of initial measurement2)————————— at streamside and 51.8% in the dry Medium grazing 0.821 Cb3 0.856 Bb 0.812Ab 1.598 Ab 1.011 Ba 0.536 Bb meadow for the medium grazing treat- Light grazing 0.665 Bb 0.824 Bb 0.730 Ab 1.585 Ab 0.912 Ba 0.520 Bb ment; 21.6% at streamside and 25.0% in No grazing 0.591 Ab 0.687 Ab 0.861 Ab 2.336 Bb 0.686 Ab 0.294 Ab the dry meadow for the light grazing 11990 and 1994 measurements were taken during grazed, droughty years; 1996 measurements were taken in ungrazed, wet conditions 1 year following cessation of treatments. treatment. The residual stubble heights 2 Data presented are the proportional changes from initial measurements, therefore, the units are dimensionless. for graminoids immediately following 3Treatment means sharing an upper case letter within a characteristic and year are not different grazing were 10.5 cm (4.1 in) at stream- at P≤0.05. Lower case letters indicate: a=not different from initial reading, or b=significantly different from initial reading. side and 7.1 cm (2.8 in) on dry meadow for medium grazing and 14.1 cm (5.6 in) Table 2. Changes in streambank ratings, Stanley Creek pastures.1 at streamside and 13.4 cm (5.3 in) on dry meadow for light grazing. Season- Streambank characteristics end streamside stubble heights were Streambank stability Streambank alteration 12.9 cm (5.1 in) for medium, 16.4 cm Grazing treatment 1990&1994 1996 1990&1994 1996 (6.5 in) for light grazing, and 26.2 cm —————————-(Change from initial rating2)—————————— (10.3 in) for no grazing. These utiliza- Medium grazing 12.4 Ab3 12.0 Ab 3.3 Bb -17.4 Bb tion levels were less severe and the sea- Light grazing 11.2 Ab 11.8 Ab -1.3 Ba -22.6 Bb son of grazing more restricted than had No grazing 16.7 Ab 19.5 Ab -11.0 Ab -35.2 Ab been the situation on the study site for 11990 and 1994 measurements occurred during grazed, droughty years; 1996 measurements occurred in ungrazed, wet most of this century. conditions 1 year following cessation of treatments. 2Characteristics were rated on a scale of 0-100. Table values indicate direction and magnitude of rating change. 3Treatment means sharing an upper case letter within a characteristic and year are not different Stream Channel at P<0.05. Lower case letters indicate: a=not different from initial reading, or b=significantly different from initial reading. Stream Channel Profile and Streambank Ratings Table 3. Changes in channel bottom characteristics, Stanley Creek pastures.1 A decrease in stream width occurred under all treatment regimes from 1986 Channel bottom characteristics to 1996 (Table 1). The average amount Embeddedness Fine sediments of narrowing was inversely associated Grazing treatment 1990&1994 1996 1990&1994 1996 with grazing intensity. The change in (Coverage of large particles) (Proportion of channel bottom) depth was more erratic among years Medium grazing 6.7 Bb2 Ð16.5 Cb 15.0 Bb 6.9 Ba than the change in width. Depths Light grazing Ð2.6 Ab Ð33.3 Ab Ð1.0 Aa Ð14.0 Ab decreased during the drouthy grazing No grazing Ð8.6 Ab Ð20.0 Bb Ð2.8 Aa Ð4.0 Ba years, but had increased in the wet post- 11990 and 1994 measurements occurred during grazed, droughty years; 1996 measurements occurred in ungrazed, wet conditions 1 year following cessation of treatments. treatment year. The ungrazed pastures, 2 Treatment means sharing an upper case letter within a characteristic and year are not different at P≤0.05. Lower case which displayed the greatest narrowing, letters indicate: a=not different from initial reading, or b=significantly different from initial reading. showed the greatest increase in depth

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 221 Table 4. Streamside willow responses to grazing management from beginning to end of study, Stanley Creek pastures.

Willow characteristics Grazing Change in Change in treatment height cover1 (m) (%) Medium grazing 0.35 Ab2 28.89 Ab Light grazing 0.28 Ab 37.13 Ab No grazing 0.40 Ab 56.39 Bb 1Based only on plots containing willows throughout the study. 2Treatment means sharing an upper case letter within a characteristic are not different at P≤0.05. Lower case letters indicate: a=not different from initial reading, or b=significantly different from initial reading. the study period, but the changes did not differ among treatments (Table 4). Willow cover increased with all treatments; the greatest increase occurred in the absence of grazing. Fig. 2. Proportional change in several streamside plant community-type group frequencies during 1988 to 1996. The shift in community-types typified by strongly rooted, late seral graminoids Plant Species Richness compared to all other herbaceous plant community-types was significant at P=0.01. The grazed pastures increased in graminoid species in all years compared types in the streamside locations. A sig- treatment. Changes in total herbaceous to the initial reading (Table 5). The nificant change did occur in the frequency plant cover did not differ among treat- grazed pastures also showed an increase of the entire group of strongly-rooted, late ments in the grazed years. The no in forb species during the grazed years. seral species (beaked sedge, water sedge, grazed treatment ended in 1996 with a The year after grazing ceased and precipi- bluejoint reedgrass [Calamagrostis lower total plant cover than the initial tation was high all pastures lost forb canadensis (Michx.) Beauv.], and Baltic reading. Litter tended to decrease in the species richness. The average number of rush) (USDA-FS 1992)(P=0.01). An moderately grazed pastures, gain in the no shrub species per plot increased slightly increase in this group occurred in the grazing treatment, and change minimally from the initial readings; the response was ungrazed and lightly grazed pastures (Fig. in the light grazing treatment (Table 6). similar among treatments. Overall, the 2). This increase was nearly matched by a grazed treatments experienced a greater non-significant downward trend in the Plant Height increase in total plant species during the Kentucky bluegrass community-type Graminoid heights were similar period of grazing than did the ungrazed (P=0.07). among treatments after the cessation of treatment. In the year following the end of grazing for 1 year (P=0.56). Average grazing when a general reduction of forb Plant Cover and Litter streamside graminoid heights were 28.2, species occurred, only the medium grazed Little change in graminoid canopy 28.1, and 29.4 cm for medium, light, treatment maintained a significant cover occurred. No treatment differed and no grazing treatments, respectively. increase in total species richness com- significantly from the initial measure- These results suggest that similar pared to initial readings (Table 5). ments at the end of the study (Table 6). growth rates of herbaceous plants were The forb cover tended to increase vari- attained among treatments within 1 year Plant Community-Types ably in the grazed treatments and to after grazing stopped. No significant changes occurred in fre- decrease significantly in the ungrazed quencies of individual plant community-

Table 5. Changes in numbers of streamside plant species, Stanley Creek.1

Plant growth form Grazing Graminoid Forb Shrub Total Plant treatment 1990&1994 1996 1990&1994 1996 1990&1994 1996 1990&1994 1996 ——————————————————————Change in no. 0.25-m-2 ———————————————————-— Medium grazing 0.73 Bb2 0.64 Bb 0.59 Bb -0.01 Aa 0.08 Ab 0.09 Ab 1.40 Bb 0.72 Bb Light grazing 0.65 Bb 0.35 ABb 0.47 Bb -0.45 Ab 0.13 Ab 0.20 Ab 1.25 Bb 0.10 ABa No grazing 0.14 Aa 0.06 Aa 0.04 Aa -0.45 Ab 0.12 Ab 0.18 Ab 0.30 Ab -0.21 Aa 11990 and 1994 measurements were taken during grazed, droughty years; 1996 measurements were taken in ungrazed, wet conditions 1 year following cessation of treatments. 2Treatment means sharing an upper case letter within a characteristic and year are not different at P≤0.05. Lower case letters indicate: a=not different from initial reading, or b=significantly different from initial reading.

222 Journal of Range Management (52)3, May 1999 Table 6. Changes in streamside litter and herbaceous plant cover, Stanley Creek pastures.1

Plant growth form Grazing Litter Graminoid Forb Total herbaceous plant treatment 1990&1994 1996 1990&1994 1996 1990&1994 1996 1990&1994 1996 —————————————————————(Change in % cover) ————————————————————— Medium grazing Ð1.19 Aa2 Ð3.39 Ab Ð3.20 Aa Ð2.91 Aa 0.82 Ba 4.28 Bb Ð2.38 Aa 1.37 Ba Light grazing 1.50 Bb Ð0.66 Ba Ð1.95 Aa Ð4.89 Aa 3.62 Bb 0.02 Ba 1.67 Aa Ð4.87 ABa No grazing 2.82 Cb 2.77 Cb 8.44 Ab 1.93 Aa Ð6.36 Ab Ð7.83 Ab 2.08 Ab Ð5.90 Ab 11990 and 1994 measurements were taken during grazed, droughty years; 1996 measurements were taken in ungrazed, wet conditions 1 year following cessation of treatments. 2Treatment means sharing an upper case letter within a characteristic and year are not different at P≤0.05. Lower case letters indicate: a=not different from initial reading, or b=significantly different from initial reading.

Dry Meadow Vegetation primary forb type on the study area. The alteration as through changes in vegeta- Species Richness greatest change occurred in the tion biomass (Winward 1986). Overuse The graminoids were greater in both ungrazed pastures. A group of late seral by cattle can easily destabilize and break the grazed period and in the post-graz- graminoid community-types (beaked down streambanks as vegetation is ing year in comparison to initial values sedge, water sedge, bluejoint reed grass, weakened and the physical forces of (Table 7). The forb species richness and Baltic rush), that inhabited the more hoof impacts shear off bank segments exhibited a general increase across treat- moist locations in the dry meadow area, (Marlow and Pogacnik 1985, Trimble ments during the grazed years, but decreased in all treatments (P<0.01). and Mendel 1995). As grazing and tram- declined in the ungrazed treatment in pling damage are reduced, the residual 1996, a wet year. This indicates that Plant Cover and Litter vegetation aids in trapping of sediments both drought and grazing stresses pro- On the dry meadows graminoid cover that serve as base material to rebuild vided the opportunity for an increase in decreased during the period of study streambanks (Clary et al. 1996). The forb species. A slight increase in num- with no difference among treatments channel narrowing and the reduced ber of different shrubby species (Table 8). Forb cover increased in all width/depth ratio of all 3 treatments in occurred from initial readings in the treatments during the grazed years, but 1 this study suggest the grazing stress light and ungrazed treatments, although year after grazing stopped the medium applied during treatment was within the no differences among treatments were grazing treatment was the only treat- sites’ capabilities for annual recovery detected. During the grazed period, the ment different than the initial measure- and that the original degree of degrada- average number of total species record- ment. There were no changes in shrub tion did not preclude an improving trend ed increased for all treatments. In the cover. Total plant cover decreased on under these conditions. Because the year after grazing ceased, only the medi- the dry meadows during the grazed degree of change in these variables was um grazed treatment showed an increase years, but in 1996 only the no grazed associated with grazing intensity, this in number of total plant species com- treatment had less total cover than initial study illustrates that streambank and pared to the initial reading (Table 7). readings (Table 8). Generally, less litter aquatic habitat impacts can be con- was recorded during the study than in trolled through grazing management. Plant Community-Types the initial readings, although little differ- When streambanks rebuild and channels There was greater evidence of a ence occurred among treatments. narrow, the decreased width/depth ratio change in the frequencies of communi- improves the stream’s hydraulic and ty-types in the dry meadow portion of sediment transport efficiency (Morisawa the pastures than in the streamside areas. Discussion 1968, Olson-Rutz and Marlow 1992, The frequency of the tufted hairgrass Leopold 1994) and provides potential type (P<0.01) and the Kentucky blue- Streamside increases in fish hiding cover (Meehan grass type (P<0.01) increased inversely Grazing along streambanks probably et al. 1977, Kozel et al. 1989, Bjornn to grazing pressure at the expense of the does as much or more damage to and Reiser 1991, Overton et al. 1995). thick-stemmed aster type (P<0.01), the stream-riparian habitats through bank All treatments decreased in substrate

Table 7. Changes in numbers of dry meadow plant species, Stanley Creek.1

Plant growth form Grazing Graminoid Forb Shrub Total plant treatment 1990&1994 1996 1990&1994 1996 1990&1994 1996 1990&1994 1996 —————————————————————-Change in no. 0.25-m-2 ———————————————————— Medium grazing 0.98 Bb2 0.91 Bb 0.84 Ab 0.44 Bb 0.01 Aa 0.02 Aa 1.83 Bb 1.37 Bb Light grazing 0.58 Ab 0.34 Ab 0.54 Ab -0.16 Aa 0.03 Ab 0.04 Ab 1.15 Ab 0.22 Aa No grazing 0.46 Ab 0.37 Ab 0.52 Ab -0.23 Ab 0.03 Ab 0.05 Ab 1.01 Ab 0.19 Aa 11990 and 1994 measurements were taken during grazed, droughty years; 1996 measurements were taken in ungrazed, wet conditions 1 year following cessation of treatments. 2Treatment means sharing an upper case letter within a characteristic and year are not different at P≤0.05. Lower case letters indicate: a=not different from initial reading, or b=significantly different from initial reading.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 223 Table 8. Changes in dry meadow ground cover, Stanley Creek.1

Ground cover characteristics Grazing Litter Graminoid Forb Shrub Total plant treatment 1990&1994 1996 1990&1994 1996 1990&1994 1996 1990&1994 1996 1990&1994 1996 ————————————————————————Change in % cover———————————————————————— Medium grazing Ð3.15 Ab2 Ð3.60 Ab Ð14.38 Ab Ð11.43 Ab 5.00 Ab 11.09 Cb 0.01 Aa 0.01 Aa Ð9.37 Ab Ð0.33 Aa Light grazing 1.63 Bb Ð1.22 Ab Ð11.71 Ab Ð5.27 Ab 3.04 Ab 1.83 Ba 0.02 Aa 0.03 Aa Ð8.65 Ab Ð3.41 Aa No grazing 0.80 Aa Ð1.62 Ab Ð16.00 Ab Ð8.67 Ab 3.75 Ab Ð2.04 Aa 0.02 Aa 0.04 Aa Ð12.23 Ab Ð10.67 Ab 11990 and 1994 measurements were taken during grazed, droughty years; 1996 measurements were taken in ungrazed, wet conditions 1 year following cessation of treatments. 2Treatment means sharing an upper case letter within a characteristic and year are not different at P≤0.05. Lower case letters indicate: a=not different from initial reading, or b=significantly different from initial reading. embeddedness by the end of the study, and Kauffman 1995). In 1996, one year ticularly grazing stress, typically pro- but the decrease in proportion of the after cessation of grazing and a year of vide disturbance conditions favorable surface composed of fine sediments was above average precipitation, forb for increases in species richness and variable. This response may have been species numbers dropped to less than diversity (Dobson 1973, Hayes 1978, affected by downstream movement of initial counts for the lightly grazed and Green and Kauffman 1995). After pro- old dredge mining sediments. Channel no grazed treatments as grazing and tection from grazing, dry meadows char- bottom conditions are greatly affected moisture stresses were reduced. The acteristically experience a reduction in by sediments contributed by upstream increase in frequencies of strongly-root- the forb component while retaining the sources and may not respond rapidly to ed, late seral graminoid community- exotic Kentucky bluegrass component on-site management (Rinne 1988). The types in streamside locations under light (Kauffman et al. 1983, Green and channel substrate status is important to or no grazing was expected (Green and Kauffman 1995). Reduction in spawning and incubation of stream fish- Kauffman 1995). The extent and graminoid cover during the study was es, production of aquatic invertebrates strength of the roots and rhizomes of probably a result of drought because the for salmonid food, and cover for young these plants provide essential stability to ungrazed treatment decreased as well. fish (Bjornn and Reiser 1991). the banks (Kleinfelder et al. 1992, There were mixed directions of succes- Willow height and cover increased in Dunaway et al. 1994), thus allowing sional changes. The increase in tufted all treatments of this study. The mainte- undercuts to form as habitat segments hairgrass and a decrease in forb domi- nance of an adequate herbaceous forage for salmonids (Platts 1991). The lack of nance under reduced grazing pressure supply (Winward 1994, Pelster 1998) differences among treatments in height suggest improved meadow conditions. and control of season of grazing undoubt- of graminoids, after 1 year of rest from A decrease in wet-site rhizomatous edly reduced impacts on the willow com- grazing, suggests that any loss of vigor graminoids during drought suggests a munity as compared to historic grazing from the current grazing treatments was depletion of meadow conditions (Reid procedures (Kovalchik and Elmore 1992, largely recovered after 1 year. A similar and Pickford 1946, Hansen et al. 1995). Winward 1994). Some impact on willows response was suggested from studies of is typical even under managed grazing simulated grazing (Clary 1995). (Myers and Swanson 1995), thus the posi- Although substantial changes occurred Management Implications tive growth response of willows in this in a number of characteristics that bene- study exceeded expectations. The benefits fit the stream environment and aquatic No single management approach is of streamside vegetation canopies, partic- habitat, changes of herbaceous plant best for all situations, nor perhaps is ularly of various species of willow are: characteristics along the stream edge at even required for a given situation provision of hiding cover, modulation of Stanley Creek were limited. Overall, (Clary and Webster 1989, Ehrhart and stream temperatures, and contribution of stream channel characteristics of the Hansen 1997). There are, however, leaf detritus and terrestrial insects that Stanley Creek riparian area seemed to management approaches that work well expand food sources for fish (Meehan et respond more rapidly than the vegeta- in many circumstances. For instance, al. 1977, Murphy and Meehan 1991, tion characteristics—a sequence differ- several authors have emphasized the Kovalchik and Elmore 1992, Li et al. ing from studies in some other locations potential benefits of late summer graz- 1994). (Kondolf 1993, Knapp and Matthews ing (Kauffman et al. 1983, Marlow and Plant species richness in the grazed 1996). Pogacnik 1985). Alternatively, spring pastures increased in all categories dur- grazing has shown promise in many ing the treatment years. The continued Dry Meadow areas of the western United States grazing stress in the grazed pastures, During this study we were applying (Hayes 1978, Platts and Nelson 1985, together with the below normal precipi- differential grazing treatments and the Siekert et al. 1985, Goodman et al. tation that was present for the bulk of regional climate was applying drought 1989, Kovalchik and Elmore 1992, the study, apparently opened the stand stress for most of the years of grazing. Pelster 1998). The combination of suc- and allowed new plant establishment. These combined stresses opened the culent upland forage, cool temperatures, Changes in species richness in the stands sufficiently to provide an oppor- and wet soils near water sources acts to ungrazed treatment were noticeably less, tunity for additional species to thrive. encourage a more dispersed spring graz- as was expected (Hayes 1978, Green Stress conditions on dry meadows, par- ing pattern (Krueger 1983, Marlow and

224 Journal of Range Management (52)3, May 1999 Pogacnik 1986, Kovalchik 1987, Myers mid to late season grazing was eliminat- Chaney, E., W. Elmore, and W.S. Platts. 1989, Clary and Booth 1993). ed. This approach appears to have been 1993. Livestock grazing on western ripari- Grazing strategies employed in this successful. Most riparian area changes an areas. Rep. for U.S. Environ. Protection study were designed to stay within the in grazed pastures were in a similar Agency. Northwest Res. Inform. Center, annual tolerance of the site for plant or direction, but in different magnitudes, to Eagle, Ida. Choate, R. 1962. Geology and ore deposits streambank/channel impacts each year. those in the ungrazed treatment here and of the Stanley area. Idaho Bureau of Mines Even though many riparian forage in other ungrazed mountain meadows and Geology Pam. 126. Moscow, Ida. plants have season-long access to ade- (Knapp and Matthews 1996). At the end Clary, W.P. 1995. Vegetation and soil quate soil moisture, their ability to with- of the study the conditions on Stanley responses to grazing simulations on ripari- stand grazing has limits (Allen and Creek were continuing to improve, but it an meadows. J. Range Manage. 48:18Ð25. Marlow 1994, Lamman 1994, Clary was not known how much additional Clary, W.P. and G.D. Booth. 1993. Early 1995, Hall and Bryant 1995, Dovel change could have been expected under season utilization of mountain meadow 1996). Willows are notably vulnerable either carefully grazed or ungrazed con- riparian pastures. J. Range Manage. to cattle use in late summer (Kovalchik ditions. Stanley Creek appeared to be 46:493Ð497. and Elmore 1992, Lamman 1994, approaching relatively stable conditions Clary, W.P. and B.F. Webster. 1989. Managing grazing of riparian areas in the Winward 1994, Myers and Swanson when compared to undisturbed meadow Intermountain Region. USDA Forest Serv. 1995), particularly as the forage supply systems (Overton et al. 1995), . Gen. Tech. Rep. INT-263. Ogden, Ut. is reduced (Pelster 1998). Heavy tram- Although changes were slow in this Clary, W.P. and B.F. Webster. 1990. pling on streambanks is typically very cold mountain valley, these early season Riparian grazing guidelines for the damaging (Trimble and Mendel 1995), grazing regimes allowed improvements in Intermountain Region. Rangelands especially when the banks are moist stream channel conditions and streamside 12:209Ð212. (Marlow and Pogacnik 1985). The strat- vegetation characteristics. Most measure- Clary, W.P., C.I. Thornton, and S.R. Abt. egy in the Stanley Creek study was to ments improved to some degree under all 1996. Riparian stubble height and recov- limit grazing to the early season when 3 treatments; this suggests that early sea- ery of degraded streambanks. Rangelands less grazing use occurred near the son grazing practices that leave 10 to 14 18:137Ð140. Daubenmire, R. 1959. A canopy-coverage water’s edge. The Stanley Creek stream- cm of residual forage stubble height pro- method of vegetational analysis. bank substrate composition was vide an avenue for riparian habitat Northwest Sci. 33:43Ð64. amenable to this grazing approach; its improvement while maintaining substan- Dobson, A.T. 1973. Changes in the structure relatively permeable streamside soils tial livestock use of the meadow area. of a riparian community as the result of graz- were not likely as susceptible to spring Potential changes in other riparian mead- ing. Proc. New Zealand Ecol. Soc. 20:58Ð64. trampling damage as other more fine ow situations will vary depending on past Dovel, R. L. 1996. Cutting height effects on textured soils (Chaney et al. 1993) grazing management, streambank sub- wetland meadow forage yield and quality. (Personal communication, D. Dallas and strates, weather, and other factors. J. Range Manage. 49:151Ð156. C. Marlow). Cattle congregated on the Dunaway, D., S.R. Swanson, J. Wendel, dry meadows during those weeks the and W. Clary. 1994. The effect of herba- Literature Cited ceous plant communities and soil textures forage there was still green and succu- on particle erosion of alluvial stream- lent, rather than concentrating on the banks. Geomorph. 9:47Ð56. wetter streamside areas (Clary and Allen, D.R. and C.B. Marlow. 1994. Shoot Ehrhart, R.C. and P.L. Hansen. 1997. Booth 1993). Grazing was terminated population dynamics of beaked sedge fol- Effective cattle management in riparian each year before herbaceous vegetation lowing cattle grazing. J. Range Manage. zones: a field survey and literature review. on the dry meadows had matured. 47:64Ð69. USDI Bur. Land Manage. Mont. BLM Riparian grazing recommendations for Armour, C., D. Duff, and W. Elmore. Riparian Tech. Bull. 3. Missoula, Mont. the recovery of depleted meadow ripari- 1994. The effects of livestock grazing on Garrison, G.A., A.J. Bjugstad, D.A. western riparian and stream ecosystem. Duncan, M.E. Lewis, and D.R. Smith. an systems, presented after initiation of Fisheries 19:9Ð12. this study, suggested that 10Ð15 cm of 1977. Vegetation and environmental fea- Bjornn, T.C. and D.W. Reiser. 1991. tures of forest and range ecosystems. forage stubble height should remain on Habitat requirements of salmonids in USDA Forest Serv. Agr. Handb. 475. streamside areas at the end of the grow- streams, p. 83Ð138. In: W.R. Meehan Washington, D.C. ing season, or at the end of the grazing (ed.), Influences of forest and rangeland Goodman, T., G.B. Donart, H.E. Kiesling, season after fall frost, to limit potential management on salmonid fishes and their J.L. Holechek, J.P. Neel, D. impacts to the herbaceous plant commu- habitats. Amer. Fish. Soc. Spec. Pub. 19. Manzanares, and K.E. Severson. 1989. nity, the woody plant community, and Bethesda, Md. Cattle behavior with emphasis on time and streambank stability. Spring or early Bryant, L.D. 1985. Livestock management activity allocations between upland and summer grazing was recommended in the riparian ecosystem, p.285Ð289. In: riparian habitats, p. 95Ð102. In: R.E. R.R. Johnson, C.D. Ziebell, D.R. Patton, Gresswell, B.A. Barton, and J.L. Kershner where feasible (Clary and Webster P.F. Ffolliott, and R.H. Hamre (tech 1989, 1990). The grazing strategy on the (eds.), Practical approaches to riparian coords.), Riparian ecosystems and their resource management: an educational Stanley pastures closely paralleled these management: reconciling conflicting uses: workshop. USDI Bur. Land Manage. recommendations even though the graz- first North American conference. USDA Billings, Mont. ing rates were originally based on use of Forest Serv. Gen. Tech. Rep. RM-120. Green, D.M. and J.B. Kauffman. 1995. the dry meadow sites. Compared to his- Fort Collins, Colo. Succession and livestock grazing in a toric management patterns, the total for- northeastern Oregon riparian ecosystem. J. age utilization in this study was less, and Range Manage. 48:307Ð313.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 225 Hall, F.C. and L. Bryant. 1995. Kozel, S.J., W.A. Hubert, and M.G. Myers, T.J. and S. Swanson. 1995. Impact Herbaceous stubble height as a warning of Parsons. 1989. Habitat features and trout of deferred rotation grazing on stream impending cattle grazing damage to ripari- abundance relative to gradient in some characteristics in central Nevada: a case an areas. USDA Forest Serv. Gen. Tech. Wyoming streams. Northwest Sci. study. North Amer. J. Fish. Manage. Rep. PNW-GTR-362. Portland, Ore. 63:175Ð182. 15:428Ð439. Hansen, P.L., R.D. Pfister, K. Boggs, B.J. Krueger, W.C. 1983. Cattle grazing in man- Olson-Rutz, K.M. and C.B. Marlow. 1992. Cook, J. Joy, and D.K. Hinckley. 1995. aged forests, p. 29Ð41. In: B.F. Roche, Jr. Analysis and interpretation of stream Classification and management of and D.M. Baumgartner (comps.), channel cross-sectional data. North Amer. Montana’s riparian and wetland sites. Forestland grazing: symp. proc. Wash. J. Fish. Manage. 12:55Ð61. Mont. Forest and Conserv. Exp. Sta. Misc. State Univ. Spokane, Wash. Overton, C.K., J.D. McIntyre, R. Pub. 54. Missoula, Mont. Lamman, J.S. 1994. Effects of season and Armstrong, S.L. Whitwell, and K.A. Hawkins, C.P. 1994. What are riparian intensity of defoliation on two important Duncan. 1995. User’s guide to fish habi- ecosystems and why are we worried about montane riparian species. M.S. Thesis, tat: descriptions that represent natural con- them?, p. 1Ð9. In: G.A. Rasmussen and Colo. State Univ. Fort Collins, Colo. ditions in the Salmon River Basin, Idaho. J.P. Dobrowolski (eds.), Riparian Leopold, L.B. 1994. A view of the river. USDA Forest Serv. Gen. Tech. Rep. INT- resources: a symposium on the distur- Harvard University Press, Cambridge, Mass. GTR-322. Ogden, Ut. bances, management, economics, and con- Li, H.W., G.A. Lamberti, T.N. Pearsons, Pechanec, J.F. and G.D.Pickford. 1937. A flicts associated with riparian ecosystems. C.K. Tait, J.L. Li, and J.C. Buckhouse. comparison of some methods of determin- Utah State Univ. Logan, Ut. 1994. Cumulative effects of riparian dis- ing percentage utilization of range grasses. Hayes, F.A. 1978. Streambank stability and J. Agr. Res. 54:753Ð765. meadow condition in relation to livestock turbances along high desert trout streams of the John Day Basin, Oregon. Trans. Peek, J. and J. Gebhardt. 1980. Report on grazing in mountain meadows of central fieldtrip to riparian zones in Sawtooth Idaho. M.S. Thesis, Univ. Idaho. Moscow, Amer. Fish. Soc. 123:627Ð640. Marlow, C.B. and T.M. Pogacnik. 1985. National Recreation Area and vicinity, Ida. Idaho. The Wildlifer 183:56. Kauffman, J.B., W.C. Krueger, and M. Time of grazing and cattle-induced dam- Pelster, A. 1998. Steer diets and livestock Vavra. 1983. Effects of late season cattle age to streambanks, p. 279Ð284. In: R.R. management in a montane riparian zone. grazing on riparian plant communities. J. Johnson, C.D. Ziebell, D.R. Patton, P.F. M.S. Thesis, Colo. State Univ. Fort Range Manage. 36:685Ð691. Ffolliott, R.H. Hamre (tech. coords.), Kleinfelder, D., S. Swanson, G. Norris, Riparian ecosystems and their manage- Collins, Colo. and W. Clary. 1992. Unconfined com- ment: reconciling conflicting uses: first Platts, W.S. 1991. Livestock grazing, pressive strength of some streambank soils North American conference. USDA Forest p.389Ð423. In: W.R. Meehan (ed.), with herbaceous roots. Soil Sci. Soc. Serv. Gen. Tech. Rep. RM-120. Fort Influences of forest and rangeland man- Amer. J. 56:1920Ð1925. Collins, Colo. agement on salmonid fishes and their habi- Knapp, R.A. and K.R. Matthews. 1996. Marlow, C.B. and T.M. Pogacnik. 1986. tats. Amer. Fish. Soc. Spec. Publ. 19. Livestock grazing, golden trout, and Cattle feeding and resting patterns in a Bethesda, Md. streams in the Golden Trout Wilderness, foothills riparian zone. J. Range Manage. Platts, W.S. and R.L. Nelson. 1985. California: impacts and management 39:212Ð217. Impacts of rest-rotation grazing on stream implications. North Amer. J. Fish. Meehan, W.R., F.J. Swanson, and J.R. banks in forested watersheds in Idaho. Manage. 16:805Ð820. Sedell. 1977. Influences of riparian vege- North Amer. J. Fish. Manage. 5:547Ð556. Kondolf, G.M. 1993. Lag in stream channel tation on aquatic ecosystems with particu- Platts, W.S., C. Armour, G.D. Booth, M. adjustment to livestock exclosure, White lar reference to salmonid fishes and their Bryant, J.L. Bufford, P. Cuplin, S. Mountains, California. Restor. Ecol. food supply, p. 137Ð145. In: R.R. Johnson Jensen, G.W. Lienkaemper, G.W. 1:226Ð230. and D.A. Jones (tech. coords.), Impor- Minshall, S.B. Monsen, R.L. Nelson, Kondolf, G.M., R. Kattelmann, M. tance, preservation and management of J.R. Sedell, and J.S. Tuhy. 1987. Embury, and D.C. Erman. 1996. Status riparian habitat: a symposium. USDA Methods for evaluating riparian habitats of riparian habitat, p. 1009Ð1030. In: Forest Serv. Gen. Tech. Rep. RM-43. Fort with applications to management. USDA Sierra Nevada Ecosystem Project Final Collins, Colo. Forest Serv. Gen. Tech. Rep. INT-221. Report to Congress. Vol II. Wildland Morisawa, M. 1968. Streams: their dynam- Ogden, Ut. Resour. Center Rep. 37. Univ. Calif., ics and morphology. McGraw-Hill Book Reid, E.H. and G.D. Pickford. 1946. Davis, Calif. Co., New York, N.Y. Judging mountain meadow range condi- Kovalchik, B.L. 1987. Riparian zone associ- Murphy, M.L. and W.R. Meehan. 1991. tion in eastern Oregon and eastern ations: Deschutes, Ochoco, Fremont, and Stream ecosystems, p. 17-46. In: W.R. Washington. USDA Circ. 748. Winema National Forests. USDA Forest Meehan (ed.), Influences of forest and Washington, D.C. Serv. Reg. 6. Ecol. Tech. Pap. 279Ð87. Rinne, J.N. 1988. Grazing effects on stream Portland, Ore. rangeland management on salmonid fishes and their habitats. Amer. Fish. Soc. Spec. habitat and fishes: research design consid- Kovalchik, B.L. and W. Elmore. 1992. erations. North Amer. J. Fish. Manage. Effects of cattle grazing systems on wil- Pub. 19. Bethesda, Md. Myers, L.H. 1989. Grazing and riparian 8:240Ð247. low-dominated plant associations in central Rosgen, D.L. 1994. A classification of nat- Oregon, p. 111Ð119. In: W.P. Clary, E.D. management in southwestern Montana, ural rivers. Catena 22:169Ð199. McArthur, D. Bedunah, and C.L. Wambolt p.117Ð120. In: R.E. Gresswell, B.A. Schulz, T.T. and W.C. Leininger. 1990. (comps.), Proceedings—symposium on Barton, and J.L. Kershner (eds.), Practical Differences in riparian vegetation structure ecology and management of riparian shrub approaches to riparian resource manage- communities. USDA Forest Serv. Gen. ment: an educational workshop. USDI between grazed areas and exclosures. J. Tech. Rep. INT-289. Odgen, Ut. Bur. Land Manage. Billings, Mont. Range Manage. 43:295Ð299.

226 Journal of Range Management (52)3, May 1999 Siekert, R.E., Q.D. Skinner, M.A. Smith, Tuhy, J.S. and S. Jensen. 1982. Riparian Waters, T.F. 1995. Sediment in streams: J.L. Dodd, and J.D. Rodgers. 1985. classification for the Upper Salmon/Middle sources, biological effects and control. Channel response of an ephemeral stream Fork Salmon River Drainages, Idaho. Final Amer. Fish. Soc. Mongr. 7. Bethesda, Md. in Wyoming to selected grazing treat- Rep. to USDA Forest Serv., Region IV. Winward, A.H. 1986. Vegetation character- ments, p. 276Ð278. In: R.R. Johnson, C.D. White Horse Assoc., Smithfield, Ut. istics of riparian areas. Trans. West. Sect. Ziebell, D.R. Patton, P.F. Ffolliott, R.H. US General Accounting Office. 1988. Public The Wildl. Soc. 22:98Ð101. Hamre (tech. coords.), Riparian ecosys- rangelands: some riparian areas restored but Winward, A.H. 1994. Management of live- widespread improvement will be slow. tems and their management: reconciling stock in riparian areas, p.49Ð52. In: G.A. GAO/RCED-88-105. Gaithersburg, Md. conflicting uses: first North American con- Rasmussen and J.P. Dobrowolski (eds.), USDA Forest Service. 1992. Integrated ference. USDA Forest Serv. Gen. Tech. Riparian resources: a symposium on the riparian evaluation guide. USDA For. Rep. RM-120. Fort Collins, Colo. disturbances, management, economics, Serv., Intermountain Region. Ogden, Ut. Swanson, S. 1988. Riparian values as a and conflicts associated with riparian focus for range management and vegeta- Van Noy, R.M., J. Ridenour, N.T. Zilka, F.E. Federspiel, R.K. Evans, E.T. ecosystems. Utah State Univ. Logan, Ut. tion science, p. 425Ð445. In: P.T. Tueller Zweygardt, L.R. and J.C. Buckhouse. (ed.), Vegetation science applications for Tuchek, and A.B. McMahan. 1986. 1996. Evaluation of a riparian classifica- rangeland analysis and management. Economic appraisal of the eastern part of Kluwer Academic Publ., Boston, Mass. the Sawtooth National Recreation Area, tion protocol, p. 88. Private rangelands: Trimble, S.W. and A.C. Mendel. 1995. The ID. US Geol. Surv. Bull. 1545-E. stewardship and responsibility. Abstracts cow as a geomorphic agent—a critical Washington, D.C. 49th Annual Meeting, Soc. Range review. Geomorph. 13:233-253. Manage. Wichita, Kan.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 227 J. Range Manage. 52:228Ð234 May 1999 Late-summer forage on prairie sandreed dominated rangeland after spring defoliation

PATRICK E. REECE, THOMAS L. HOLMAN, AND KENNETH J. MOORE

Authors are associate professor, Panhandle Research and Extension Center, University of Nebraska, 4502 Avenue I, Scottsbluff, Nebr. 69361; extension educator, Scotts Bluff and Morrill Counties, University of Nebraska, Gering, Nebr. 69341; and professor, Agronomy Department, Iowa State University, Ames, Iowa 50011.

Abstract Resumen

The potential of using spring defoliation to improve late- Se estudio el potencial de utilizar la defoliación en primav- summer nutritive value of prairie sandreed [Calamovilfa era para mejorar el valor nutritivo a fines de verano del longifolia (Hook.) Scribn.] on rangeland was studied with a zacate "prairie sandreed" [Calamovilfa longifolia (Hook.) factorial array of replicated 1-year treatments that included Scribn]. El estudio se condujo en un pastizal bajo un arreglo clipping plots at ground level or at a 5 or 10 cm height on 1 factorial de tratamientos repetidos 1-año que incluyeron April, 26 April, 20 May, or 14 June. Vegetative tillers parcelas defoliadas a nivel del suelo, o a 5 y 10 cm de altura. accounted for 83% of prairie sandreed herbage on unclipped La defoliación se realizo el 1 de abril, 26 de abril, 20 de mayo control plots. After spring treatments, late-summer crude o 14 de junio. En las parcelas sin defoliar (control), los hijue- protein content (CP) in vegetative tillers of prairie sandreed los vegetativos aportaron el 83% del forraje producido por el ranged from 5.0 to 7.9% and in vitro dry matter digestibility "prairie sandreed". Después de los tratamientos de primav- (IVDMD) ranged from 45 to 52% compared to 5.0% CP and era, el contenido de proteína cruda (PC) de los hijuelos vege- 45% IVDMD for unclipped plots. Reductions in mean weight tativos a fines del verano fluctúo de 5.0 a 7.9% y la digestibil- of prairie sandreed vegetative tillers after April and May idad in vitro de la materia seca (DIVMS) estuvo en un rango treatments were offset by 20 to 30% increases in tiller densi- de 45 a 52% comparado con 5.0% de PC y 45 % DIVMS de ty. Treatments that increased tiller density had little or no las parcelas sin defoliar (control). Las reducciones del peso effect on forage nutritive value when applied more than 90 promedio de los hijuelos vegetativos del "prairie sandreed" days before herbage was sampled. Nutritive value of prairie ocurrida después de aplicar los tratamientos de abril y mayo sandreed and total yield from all species in mid-September fueron compensados por un incremento del 20 al 30% de la were unchanged after April treatments. After sandreed tillers densidad de hijuelos. Cuando los tratamientos que incremen- began to emerge in early May, late-summer nutritive value taron la densidad de hijuelos vegetativos fueron aplicados 90 improved as clipping was delayed and degree of defoliation o más días antes del muestreo de forraje estos tuvieron poco o increased during May and June, however, yield was inversely ningún efecto en el valor nutritivo del forraje. El valor nutri- related to nutritive value. While mid-September nutritive tivo del "prairie sandreed" y el rendimiento total, a mediados value of prairie sandreed was comparable to mid-summer de septiembre, de todas las especies no cambio después de los values after June treatments, clipping reduced projected, tratamientos de abril. Después de que los hijuelos de "prairie late-summer stocking rates by 58 to 100% compared to con- sandreed" comenzaron a emerger a inicios de mayo, el valor trol. It may be possible to improve mid-September forage nutritivo a fines del verano mejoro conforme la defoliación se nutritive value with moderate stocking rates in June with less retrasó y el grado de defoliación incrementó durante mayo y reduction of total late-summer herbage because of selective junio, sin embargo, el rendimiento fue inversamente rela- herbivory. Measurable increases in prairie sandreed yield cionado al valor nutritivo. Mientras el valor nutritivo del after complete defoliation of associated species in late April "prairie sandreed" a mediados de septiembre fue compara- indicated prairie sandreed populations might be increased by ble al de mediados del verano después de los tratamientos de concentrating cattle in selected pastures during late April. junio, la defoliación redujo la carga animal proyectada para fines del verano, la reducción fue del orden de 58 a 100% en comparación con el control. Puede ser posible mejorar el Key Words: Calamovilfa longifolia, clipping date, cutting valor nutritivo del forraje de mediados de septiembre uti- height, yield, crude protein, in vitro dry matter digestibility, lizando cargas animales moderadas en junio teniendo menos tiller demographics reducción de forraje total de fines de verano debido a la her- bívora selectiva. Incrementos medibles en el rendimento de "prairie sandreed" después de completar la defoliación de las Authors wish to thank Sharon Holman for preparation of text and graphics, and especies asociadas a fines de abril indica que las poblaciones Gordon Moeller for technical support. We also wish to thank Dr. Monte Rouquette, Jr. and anonymous reviewers for their comments and editorial work. de "prairie sandreed’ podrían ser incrementadas por la con- Published as Paper 11634, Journal Series, Nebraska Agricultural Research centración de ganado en potreros seleccionadas durante fines Division. de abril. Manuscript accepted 22 Aug. 1998.

228 Journal of Range Management (52)3, May 1999 The feasibility of using spring grazing Table 1. Average species composition based upon herbage biomass in control plots clipped in to reduce seasonal declines in forage September, 1994. nutritive value associated with plant 1 maturity has not been reported for semi- Species Buffalo Creek Wildcat Hills arid rangeland. Some studies have Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð(%)Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð Ð examined the effects of defoliation on Calamovilfa longifolia (Hook.) Scribn. 54 52 nutritive value in a following season, Stipa comata Trin. and Rupr. 29 4 Poa pratensis L. 0 22 but initial dates of defoliation have been 2 Carex spp. 311 confounded with additional clipping Bouteloua gracilis (H.B.K.) Lag. ex Griffiths 4 9 dates between first cutting and final har- Agropyron smithii Rydb. 2 1 vest (George and Obermann 1989, Forbs 8 1 Willms 1991, Willms and Beauchemin 1Nomenclature follows The Great Plains Flora Association (1986). 1991, Belsky and Fedders 1994, 1995). 2Carex species were predominantly C. filifolia Nutt. and C. heliophila Mack. In early spring, seasonally low forage allowances of green herbage for live- ranked by relative abundance of prairie mesic Typic Haplustolls and mixed, cal- stock can cause severe defoliation of ini- sandreed. Within comparable levels of careous, mesic, shallow Typic tial plant growth. Given the potential for abundance, sites were randomly Ustorthents. Precipitation during April a wide range in degree of spring defolia- assigned to years to include similar vari- to September 1993 was about 193 mm tion, improvement in late-summer nutri- ation in species composition each year. at both locations and 239 mm at Buffalo tive value could be caused by reduced All vegetation in 1.0 m2 plots was Creek and 282 mm at Wildcat Hills in stem development (Perry and Balten- clipped at ground level or at a 5 or 10 1994 compared to the long-term average sperger 1979), reduced average age of cm height on 1 April, 26 April, 20 May, of 293 mm for this 6-month period tiller populations (George and or 14 June. This provided deferment (NOAA 1994). Long-term average Obermann 1989, Culvenor 1993, periods of about 165, 140, 115 and 90 annual precipitation is 410 mm and the Bullock et al. 1994), and/or reduced days between spring defoliation and average frost-free period is 136 days. composition of reproductive tillers late-summer harvest in mid-September. Forage samples were analyzed to (Cook and Stoddart 1953, Murray 1984, Current-year prairie sandreed tillers determine concentrations of crude pro- Ganskopp et al. 1992, Brummer 1994). were counted and clipped from the inte- tein (CP) and in vitro dry matter In the semi-arid region of western rior 0.5 m2 of plots when spring defolia- digestibility (IVDMD) by near infrared Nebraska and adjoining states, the duration treatments were applied. All other reflectance spectroscopy (NIRS) using tion of temperature and soil moisture herbage clipped in the spring was dis- the protocol described by Windham et conditions required for rapid plant carded. Prairie sandreed tillers harvested al. (1989). Reflectance measurements growth is relatively short. Studies of from plots clipped at ground level were (log 1/R) were collected for all samples interactions between date and degree of cut into segments of 0Ð5, 5Ð10, and from 1,100 to 2,500 nm and recorded in spring defoliation are needed to under- above 10 cm to estimate degree of 4-nm intervals using a Pacific Scientific stand the mechanisms by which end-of- spring defoliation at the 5 and 10 cm 6250 (NIRS Systems, Silver Spring, season forage quality might be heights. In addition, each site included 3 Md) scanning monochromator. Based improved. Extensive rhizomes and deep control plots. Mean response of control upon spectral characteristics, a subset of roots enhance prairie sandreed's plots for each site was used for analysis 60 samples representing the entire range [Calamovilfa longifolia (Hook.) Scribn.] of variance. Late-summer, current-year of H values (Mahalanobis distance) was ability to tolerate drought and produce herbage was harvested at ground level selected for NIRS calibration (Shenk large quantities of herbage compared to from the interior 0.5 m2 of plots in mid- and Westerhaus 1991a). Estimates of CP associated plant species on sandy soils September, separated into prairie san- and IVDMD were determined for cali- throughout the northern Great Plains dreed vegetative and reproductive tillers bration samples using the micro- (Lodge 1963, Welch 1968, White 1977). and other species, and oven dried at Kjeldahl and rumen fermentation Our objective was to quantify the effects 60¡C to a constant weight. Tillers with (Marten and Barnes 1980) procedures, of spring defoliation on late-summer exposed seed heads were classified as respectively. All analyses were done in nutritive value of prairie sandreed and reproductive. Plant composition at each duplicate and averages of duplicate sam- yield of all species on rangeland domi- location was estimated in late summer ples were used as analytical values. nated by prairie sandreed. by clipping and weighing species indi- Calibrations were developed using mod- vidually from 12 control plots at each ified partial least squares regression location on 1994 study sites. (Shenk and Westerhaus 1991b). Material and Methods Locations were the Buffalo Creek Coefficients of determination and stan- Wildlife Management Unit, 9 km south- dard errors for calibration and cross val- A 3 x 4 factorial array of 1-year spring west of Melbeta, Nebr. and the Wildcat idation were, respectively, 0.93, 1.47 clipping treatments was replicated on Hills State Park, 11 km south of Gering, and 1.86 for IVDMD and 0.99, 0.09 and sandy range sites dominated by prairie Nebr. Livestock had been excluded from 0.16 for CP. Calibration statistics were sandreed at 2 locations in 1993 and both locations for 10 years or more. within acceptable limits for all variables 1994 (Table 1). Eight sites were selected Soils at Buffalo Creek are mixed, mesic (Windham et al. 1989). at each location in March 1993 and Entic Haplustolls. Soils at the Wildcat Experimental units were 0.5 m2 Hills location are a mosaic of mixed, quadrats centrally placed in 1.0 x 1.0 m

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 229 treated areas. Data were analyzed as a randomized complete block using the General Linear Models Procedure (SAS 1986). Level of probability selected for significance was P ≤ 0.05. Single degree of freedom orthogonal contrasts were used to select variables for equations to describe significant main effects and interactions. Equations for main effects were fit to treatment means from both study years, and equations for spring clipping date by cutting height interactions were fit to factorial means averaged over years using the Regression Procedure (SAS 1986). Dunnett's test was used to compare control with each clipped treatment (Dunnett 1955). The Least-Squares Means Procedure within Fig. 1. Effects of spring clipping date (D = days after 31 Mar) on late-summer density of SAS was used for mean separation prairie sandreed vegetative tillers. among defoliation treatments (Searle et al. 1980). generally eliminated reproductive tillers increased as the degree of spring defoli- which comprised about 25% of prairie ation increased and clipping date was sandreed herbage on control plots in delayed (Fig. 2a, Table 2). Removing 53 Results mid-September. In the following year to 100% of the herbage from prairie san- when growing-season precipitation was dreed in mid-June reduced late-summer Composition of prairie sandreed on near average, spring defoliation had no mean tiller weight by 32 to 55%, com- control plots in mid-September ranged effect on reproductive tillers regardless pared to control. from 24 to 93% among study sites. of cutting height. Levels of CP content in vegetative While the mean composition of prairie Clipping in April or May increased tillers within comparable weights were sandreed was comparable between loca- late-summer density of vegetative tillers higher under drought conditions in 1993 by 20 to 30% (Fig. 1) compared to 300 than in 1994 when near average precipi- tions, differences in the composition of -2 cool-season species occurred between m for control. Changes in vegetative tation occurred (Fig. 3a, Table 3). Crude locations. Needleandthread (Stipa coma- tiller density diminished as the amount protein declined rapidly as tiller weight ta Trin. and Rupr.) was the primary of herbage removed from prairie san- increased with about 75% of the change cool-season species at Buffalo Creek dreed increased from May to June (Fig. occurring between the low and mid- (Table 1). In contrast, the primary cool- 1, Table 2). Mean weight of vegetative range tiller weights. Average IVDMD of season species at Wildcat Hills were tillers in mid-September was reduced by prairie sandreed vegetative tillers was kentucky bluegrass (Poa pratensis L.) about 15%, compared to control, when not different between years, but the rate and sedges (Carex spp.). plots were clipped in April (Fig. 2a). at which IVDMD declined with increas- After prairie sandreed tillers emerged in ing tiller weight was about 2 times May, reductions in tiller weight greater when precipitation was near Prairie Sandreed Tiller Responses All prairie sandreed tillers emerged Table 2. Percent defoliation of prairie sandreed in the spring, crude protein (CP) content and after 26 April and growing points in digestibility (IVDMD) of vegetative tillers of prairie sandreed in mid-September, and projected reproductive tillers were elevated less late-summer stocking rate after clipping at 0, 5, or 10 cm on 20 May or 14 June. than 5 cm above the soil surface on 14 June at both locations in both years. Spring Cutting Height1 When growing-season precipitation was Spring Clipping Date 0 cm 5 cm 10 cm 34% below the long-term average in 20 May 1993, complete defoliation at Buffalo Spring Defoliation (%) 100 61 29 Creek increased the density of reproduc- Late-summer2 tive tillers by 82% after 1 April treat- CP (%) 6.0 (5.4) (5.1) IVDMD (%) 46.7 (45.4) (45.0) ments and 29% after clipping on 26 Projected Stocking Rate (AUD ha-1) 15 23 31 April compared to control (28 m-2). 14 June However, percent of late-summer Spring Defoliation (%) 100 78 53 herbage composed of reproductive Late-summer2 tillers was unchanged after April treat- CP (%) 7.9 6.1 5.6 ments because of concurrent increases IVDMD (%) 52.2 48.9 47.2 -1 in vegetative tiller density. After prairie Projected Stocking Rate (AUD ha ) 0 5 17 1 sandreed tillers emerged, complete defo- Values in parentheses are not significantly different from control based on Dunnett's test, P>0.05. 2Late-summer values for control were 5% CP, 45% IVDMD, and 40 AUD ha-1. Projected stocking rate was calculated liation in May or June at Buffalo Creek by dividing herbage available for livestock consumption by 11.8 kg to estimate potential animal unit days (AUD).

230 Journal of Range Management (52)3, May 1999 year herbage from prairie sandreed in stocking rate in mid-September for con- mid-June (Table 2) reduced mid- trol was about 40 AUD ha-1. Removing September yield of prairie sandreed an average of 29 and 61% (10 and 5 cm herbage from vegetative tillers by 35 to height) of current-year herbage from 52% (Fig. 2b). prairie sandreed in mid-May reduced About 75% of herbage from other projected, late-summer stocking rates by species was produced by cool-season 22 and 43%, respectively, but did not grasses or sedges (Table 1). Mean yield improve nutritive value of prairie san- of herbage from other species on control dreed (Table 2). Clipping prairie san- plots was 610 kg ha-1. When averaged dreed in mid-June at 10 or 5 cm heights over spring clipping dates, little change increased mean defoliation to 53 and occurred in late-summer yield of 78% causing measurable increases in herbage from other species after clip- late-summer nutritive value of prairie ping at 5 compared to 10 cm (Fig. 4a). sandreed, but reduced projected stock- However, predicted late-summer yield ing rates by 58 and 88%, respectively. declined rapidly as cutting height decreased from 5 to 0 cm. Late-summer herbage from species other than prairie Discussion sandreed also declined about 26 kg ha-1 for each week that spring clipping was While crude protein content was lower delayed after 1 April (Fig. 4b). in reproductive compared to vegetative Average, current-year herbage from tillers of prairie sandreed, the occur- all species that occurred on control plots rence of reproductive tillers had little in mid-September was about 1,900 kg effect on average IVDMD of late-sum- Fig. 2. Spring cutting height by clipping date -1 interaction effects on (a) mean tiller weight ha . Projected, late-summer stocking mer herbage. Density of reproductive and (b) yield of prairie sandreed vegetative rates were based on leaving 950 kg ha-1 tillers was also affected by spring clip- tillers in mid-September. Columns with (half of control herbage) for ecosystem ping at only one location in a single dots (¥) are different from control, P<0.05. functions and utilizing 50% of the year. Consequently, the risks of reduced remaining herbage (Table 2). The bal- livestock performance in May or June, average in 1994 compared to drought ance, 50% of herbage beyond the target damage to associated plant species, and conditions in 1993, e.g. 5.5 compared to level of 950 kg ha-1 in remaining reduced infiltration and site stability 2.8 percentage points decline in herbage, was an estimate of losses by caused by the severe defoliation needed IVDMD per 100 mg increase in mean factors other than cattle. Herbage avail- to reduce density of reproductive tillers tiller weight (Fig. 3b, Table 3). While able for livestock forage was divided by can not be justified. Additionally, carbo- CP content was generally higher in veg- 11.8 kg to calculate projected animal hydrates from the relatively large photo- etative than in reproductive tillers, the unit days (AUD). The average projected synthetically active surface areas of range in treatment means for IVDMD of each tiller type was comparable in both years. Mean weight, CP content and Table 3. Polynomial equations for spring cutting height (H) by clipping date (D = days after 31 IVDMD of reproductive tillers ranged Mar) interaction effects on late-summer vegetative tiller weight and yield of prairie sandreed -1 from vegetative tillers and regression equations for the effects of late-spring cutting height and from 465 to 1,970 mg tiller , 3.2 to late-summer tiller weight (W) on crude protein (CP) content and digestibility (IVDMD) of late- 4.6%, and 45 to 51%, respectively. summer herbage from vegetative tillers of prairie sandreed in mid-September.

Herbage Responses Late-summer Dependent Variable Equation1 R2 Vegetative tillers accounted for 77 to 98% of prairie sandreed herbage after Spring Cutting Height by Date (Figure 1) 2 spring clipping compared to 83% for Tiller Weight Y = 315 + 0.55 D - 0.032 D - 1.018 H + 0.12 DH .94 control. Yield of herbage from vegeta- Yield Y = 1090 + 8.54 D - 0.203 D2 + 0.27 DH .91 tive tillers in mid-September ranged Spring Cutting Height (Table 2) -1 from 520 to 1,330 kg ha (Fig. 2b) com- 2 CP YMay = 6.04 - 0.18 H + 0.0082 H .45 -1 2 pared to 1,080 kg ha for control. Late- Y Jun= 7.90 - 0.49 H + 0.0262 H .75 2 summer yield from vegetative tillers IVDMD YMay = 46.7 - 0.34 H + 0.017 H .63 2 was unchanged by clipping at any cut- YJun = 52.2 - 0.82 H + 0.032 H .99 ting height in April or May except for Late-summer Tiller Weight (Figure 2) complete defoliation in late April (Fig. -5 2 CP Y93 = 15.3 - 0.055 W + 7.6 x 10 W .93 2b). Complete defoliation of plots on 26 -5 2 Y94 = 16.3 - 0.077 W + 12.7 x 10 W .89 April, before prairie sandreed tillers IVDMD Y93= 55.6 - 0.028 W .81 emerged, increased prairie sandreed Y94 = 59.9 - 0.055 W .84 yield by 23% compared to control. 1Cutting height by clipping date treatment means (n=12) were used to compute equations for Figures 1 and 2. Year by Removing 53 to 100% of the current- cutting height treatment means within clipping dates (n=6) were used to compute equations for May and June.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 231 September were compara- Additionally, complete defoliation in ble to levels reported for mid-May is unnecessary because defer- prairie sandreed tillers har- ring until mid-June and removing only vested from ungrazed pas- 50% of current-year herbage from tures in mid-June to mid- prairie sandreed would provide similar July (Northup 1993, nutritive value and mass of forage in Hendrickson et al. 1997). mid-September with reduced risks (Fig. Rapid decline in nutritive 2b, Table 2). value during early devel- Differences in yield and quality opment of new cohorts of responses to spring defoliation may prairie sandreed tillers occur between clipping and grazing reported by Hendrickson because of selective herbivory. Seasonal et al. (1997) would changes in preferred plant species, explain the declining dif- which correspond to plant maturity ferences in nutritive value (Northup 1993), and selection for cur- between progressively ear- rent-year versus carry-over herbage may lier spring clipping treat- reduce the magnitude of changes to ments and control when micro-environment and plant competi- measurable increases in tion caused by clipping. Defoliation of tiller recruitment occurred. current-year herbage by livestock should The sum of late-summer be similar to clipping in April because herbage from prairie san- cool-season species are in vegetative dreed plus herbage from stages and highly palatable while prairie other species was not sandreed tiller emergence would be lim- affected by defoliation in ited. Selection for cool or warm-season April at any cutting species may also be similar in mid-May height. In contrast, total because most cool-season species are in Fig. 3. Relationship between mean weight of prairie sandreed vegetative tillers and (a) crude protein (CP) and (b) yield of herbage in mid- vegetative or boot stages and warm-sea- digestibility (IVDMD) in mid-September. September declined after son grass tillers are in early vegetative clipping in mid-May stages. In contrast, most cool-season reproductive tillers not grazed in late because of 30 and 20% species will initiate inflorescences in summer may be a significant source of yield reductions in other species after mid-June and cattle will selectively energy for root, rhizome, and/or bud complete defoliation and clipping at 5 graze prairie sandreed and other warm- development (Brejda et al. 1989, Nixon cm, respectively. Increases in prairie season grasses thus removing less 1993, Reece et al. 1996). sandreed tiller density compensated for herbage from cool-season species than Degree of defoliation of associated up to 15% reductions in mean weight of removed by clipping (Streeter et al. species increased and opportunity for vegetative tillers. After clipping in mid- 1968, Northup 1993). Consequently, cool-season species to recover before June, total late-summer yield was about competition from cool-season species rapid growth of prairie sandreed 30 to 70% below control because of may reduce growth of prairie sandreed declined as clipping date was delayed reductions in herbage from prairie san- and reduce the relative value of from 1 April to 20 May. However, dreed and other species. All reductions improved late-summer forage quality in increases in density of prairie sandreed in the yield of prairie sandreed corre- prairie sandreed after grazing in June. vegetative tillers after clipping in April sponded to 20% or more reductions in mean weight of vegetative tillers. or May appeared to be caused primarily Conclusion by micro-environmental changes While the largest increases in nutritive (Briske and Richards 1995), rather than value of prairie sandreed occurred after release from competition, because yield complete defoliation, projected stocking Precipitation, which occurs primarily of prairie sandreed increased only after rates for mid-September after clipping at during the summer for much of the complete defoliation in late April. ground level in May were 30% below rangeland dominated by prairie san- The inverse relationship of nutritive control and complete defoliation in June dreed, will have a measurable effect on value to mean weight of vegetative precluded projected grazing until the how vegetation responds to spring defo- tillers and measurable declines in mid- following growing season because of liation. Soil moisture deficits will reduce September mean tiller weight when clip- inadequate herbage (Table 2). Clearly, or stop plant growth after spring grazing ping treatments improved nutritive value defoliation at ground level in mid-May (Dahl 1963, Ganskopp 1998). Drought indicated that the average age or maturi- or mid-June was not a sustainable prac- is a common and unpredictable compo- ty of tiller populations was reduced tice when repeated in consecutive years nent of semi-arid ecosystems. Regional (Hendrickson et al. 1997, Northup and (Reece et al. 1996). However, this level drought has occurred in about 20% of Nichols 1998). In our study, CP content of defoliation documents the physiologi- the years since 1940 in the northern and IVDMD of tiller populations with cal upper limits in mean CP and Great Plains (Holechek et al. 1989). improved nutritive value in mid- IVDMD levels in prairie sandreed in Shortages in the quantity of herbage are late summer after spring defoliation.

232 Journal of Range Management (52)3, May 1999 season should allow prairie sandreed to fully benefit from late-April grazing when average or above average precipitation occurs. Herbage in these pastures could then be used for dormant-season grazing, or left for watershed or wildlife habitat management objectives.

Literature Cited

Belesky, D.P. and J.M. Fedders. 1994. Defoliation effects on seasonal production and growth rate of cool-season grasses. Agron. J. 86:38Ð45. Belesky, D.P. and J.M. Fedders. 1995. Warm- season grass productivity and growth rate as influenced by canopy management. Agron. J. 87:42Ð48. Brejda, J.J., L.E. Moser, and S.S. Waller. 1989. Rhizome and tiller development of three Nebraska Sandhills warm-season grasses. p. 211Ð215. In: T.B. Bragg and J. Stubbendieck (eds.) Proc. Eleventh North Amer. Prairie Conf., Lincoln, Nebr. Briske, D.D. and J.H. Richards. 1995. Plant response to defoliation: A physiologic, morphologic, and demographic evaluation. In: Physiological ecology and developmental morphology. (Eds. D.J. Bedunah and R.E. Sosebee). Soc. Range Manage., Denver, Colo. Brummer, J.E. 1994. Effect of carryover herbage on utilization of little bluestem. Ph.D. Diss., Univ. of Nebraska, Lincoln, Nebr. Fig. 4. Effects of (a) spring cutting height and (b) clipping date on yield of current-year Bullock, J.M., B. Clear Hill, and J. herbage from species other than prairie sandreed in mid-September. Silvertown. 1994. Tiller dynamics of two grasses-responses to grazing, density and weather. J. Ecol. 82:331Ð340. generally of greater economic impor- versely, if late-summer forage quality is Cook, C.W. and L.A. Stoddart. 1953. The tance than nutritive value under drought the primary concern, it may be possible quandary of utilization and preference. J. conditions. Additionally, CP content for cattle to select a diet of equal quality Range Manage. 6:329Ð335. was higher when drought occurred in for an equal number of AUD’s in pas- Culvenor, R.A. 1993. Effect of cutting during our study than when near average pre- tures that are deferred until late summer reproductive development on the regrowth and regenerative capacity of the perennial cipitation occurred. with no risk to rangeland resources. grass, Phalaris aquatica L., in a controlled A limited opportunity may exist in Prairie sandreed is commonly more environment. Ann. Bot. 72:559Ð568. June for some improvement in late-sum- abundant on spring-grazed compared to Dahl, B.E. 1963. Soil moisture as a predictive mer nutritive value of prairie sandreed summer-grazed pastures on sandy range index to forage yield for the Sandhills range with minimal risk to livestock perfor- sites throughout the northern Great type. J. Range Manage. 16:128Ð132. mance and rangeland resources. Small Plains. Heavily grazing pastures in late Dunnett, C.W. 1955. A multiple comparisons but measurable increases in CP content April before prairie sandreed tillers procedure for comparing several treatments with a control. J. Amer. Stat. Assoc. and IVDMD after 50% defoliation of emerge may be an effective method of 50:1096Ð1121. prairie sandreed in mid-June would not maintaining or increasing prairie sand- Ganskopp, D. 1998. Thurber needlegrass: fully meet nutritional requirements of reed populations. Cattle will selectively Seasonal defoliation effects on forage quanti- growing and/or lactating cattle (NRC graze cool-season species in the spring ty and quality. J. Range Manage. 51:276- 1984) and less than half of the potential (Streeter et al. 1968). Feeding relatively 281. number of animal unit days (AUD) of large numbers of cattle on selected pas- Ganskopp, D., R. Angell, and J. Rose. 1992. Response of cattle to cured reproductive grazing would be available in late sum- tures during late April should provide stems in a caespitose grass. J. Range mer compared to ungrazed control. adequate nutrition for livestock and Manage. 45:401Ð404. However, the sum of AUDs or gain in allow cattle to defoliate cool-season George, J.R. and D. Obermann. 1989. Spring livestock weight from mid-June and species to the physical limit of their for- defoliation to improve summer supply and mid-September grazing periods may aging ability. Excluding livestock from quality of switchgrass. Agron. J. 81:47Ð52. equal or exceed the total from a single these pastures from the time of tiller mid-September grazing period. Con- emergence to the end of the growing

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 233 Great Plains Flora Association. 1986. Atlas Nixon, J.L. 1993. Dynamics of prairie sand- Streeter, C.L., D.C. Clanton, and O.E. of the flora of the Great Plains. T.M. Barkley reed rhizome development. M.S. Thesis. Hoehne. 1968. Influence of advance in sea- (ed.) Iowa State Univ. Press, Ames, Iowa. Univ. of Nebraska, Lincoln, Nebr. son on nutritive value of forage consumed by Hendrickson, J.R., L.E. Moser, K.J. Moore, Northup, B.K. 1993. Utilization of native for- cattle grazing western Nebraska native range. and S.S. Waller. 1997. Leaf nutritive value ages of the Nebraska Sandhills by yearling Nebr. Exp. Sta. Bull. 227. related to tiller development in warm-season cattle. Ph.D. Diss. Univ. of Nebraska, Welch, T.G. 1968. Carbohydrate reserves of grasses. J. Range Manage. 50:116Ð122. Lincoln, Nebr. sandreed grass under different grazing inten- Holechek, J.L., R.D. Pieper, and C.H. Northup, B.K. and J.T. Nichols. 1998. sities. J. Range Manage. 21:216Ð220. Herbel. 1989. Range management principles Relationships between physical and chemical White, L.M. 1977. Perenniality and develop- characteristics of 3 Sandhills grasses. J. and practices. Prentice Hall Inc., Englewood ment of shoots of 12 forage species in Cliffs. N.J. Range Manage. 51:353Ð360. Perry, L.J. Jr. and D.D. Baltensperger. 1979. Montana. J. Range Manage. 30:107Ð110. Lodge, R.W. 1963. Complementary grazing Willms, W.D. 1991. Cutting frequency and systems for Sandhills of the northern Great Leaf and stem yields and forage quality of three N-fertilized warm-season grasses. cutting height effects on rough fescue and Plains. J. Range Manage. 16:240Ð244. Parry oat grass yields. J. Range Manage. Marten, G.C. and R.F. Barnes. 1980. Agron. J. 71:355Ð358. Reece, P.E., J.E. Brummer, R.K. Engel, B.K. 44:82Ð86. Prediction of energy digestibility of forages Northup, and J.T. Nichols. 1996. Grazing Willms, W.D. and K.A. Beauchemin. 1991. with in vitro rumen fermentation and fungal date and frequency effects on prairie sand- Cutting frequency and cutting height effects enzyme systems. p. 61Ð71. In: W.J. Pigden, reed and sand bluestem. J. Range Manage. on forage quality of rough fescue and Parry C.C. Balch, and M. Graham (ed.), 49:112-116. oat grass. Can. J. Anim. Sci. 71:87-96. Standardization of analytical methodology SAS. 1986. SAS/STAT users guide. Release Windham, W.R., D.R. Mertens, and F.E. for feeds. Intern. Develop. Res. Center, 6.03. SAS Inst., Inc., Cary, N.C. Barton, II. 1989. Supplement 1. Protocol for Ottawa, Can. Searle, S.R., F.M. Speed, and G.A. Milliken. NIBS calibration: sample selection and equa- Murray, R.B. 1984. Yield, nutrient quality, 1980. Populations marginal means in the lin- tion development and validation, p. 96-103. and palatability to sheep of fourteen grass ear model: an alternative to least squares In: G.C. Marten, J.S. Shenk, and F.E. Barton accessions for potential use on sagebrush- means. The Amer. Statistician 34:216-221. (ed.), Near infrared reflectance spectroscopy grass range in southeastern Idaho. J. Range Shenk, J.S. and M.O. Westerhaus. 1991a. (NIRS): analysis of forage quality. USDA Manage. 37:343Ð348. Population definition, sample selection, and Agr. Handb. 643, Washington, D.C. National Oceanic and Atmospheric calibration procedures for near infrared Administration. 1994. Climatological data reflectance spectroscopy. Crop Sci. (Nebraska). Asheville, N.C. 31:469Ð474. National Research Council. 1984. Nutrient Shenk, J.S. and M.O. Westerhaus. 1991b. requirements of domestic animals. No. 4. Population structuring of near infrared spec- Nutrient requirements of beef cattle. 6th rev. tra and modified partial least squares regres- ed. Nat. Acad. Press, Washington, D.C. sion. Crop Sci. 31:1548Ð1555.

234 Journal of Range Management (52)3, May 1999 J. Range Manage. 52:235Ð240 May 1999 Season-long grazing of seeded cool-season pastures in the Northern Great Plains

J. F. KARN, R. E. RIES, AND L. HOFMANN

J. F. Karn, research animal scientist; R. E. Ries, range scientist; and L. Hofmann, collaborator, are with the USDA, Agricultural Research Service, Northern Great Plains Research Laboratory, P.O. Box 459, Mandan, N. D. 58554.

Resumen Abstract En la región semiárida de las Grandes Planicies del Noreste, In the semi-arid Northern Great Plains, seeded cool-season los zacates de crecimiento invernal son recomendados princi- grasses are primarily recommended for spring and fall graz- palmente para el apacentamiento de primavera y otoño ing because their nutritive quality is perceived as too low to porque se percibe que su calidad nutritiva es muy baja para support acceptable animal weight gains during mid-summer. producir ganancias de peso animal aceptables a mediados del This perception is caused in part by traditional use of high verano. Esta percepción es causada, en parte, por el uso tradi- spring stocking rates, which leave little forage remaining for cional de altas cargas animal en primavera, lo cual deja poco mid-summer use. A study was conducted near Mandan, N.D. forraje remanente para utilizar a mediados del verano. Se con- -1 dujo un estudio cerca de Mandan, N.D. para determinar el to determine the effect of moderate (1.6 AUM ha ) and heavy -1 -1 efecto de la carga animal moderada (1.6 UAM ha ) y alta (2.4 (2.4 AUM ha ) stocking rates on weight gains of yearling UAM ha-1) en las ganancias de peso de novillos de año raza Hereford steers grazing crested wheatgrass (Agropyron deser- herford apacentando "crested wheatgrass" (Agropyron desertorum [Fisch. Ex Link] Schult.), western wheatgrass torum [Fish. Ex Link] Schult), "western wheatgrass" (Pascopyrum smithii [Rydb.] Löve), smooth bromegrass (Pascopyrum smithii [Rybd]Löve), "smooth bromegrass" (Bromus inermis Leyss.) and flat (class II and III) and rolling (Bromus inermis Leyss.) en pastizales nativos planos (clase II y (class IV and VI) native rangelands. Studies were conducted III) y ondulados (clase IV y VI). Los estudios se realizaron over a 140-day grazing season during 3 summers from durante 3 veranos, de 1992 a 1994, y la estación de apacen- 1992Ð1994. Grazing was initiated in mid-May and terminated tamiento fue de 140 días. El apacentamiento fue iniciado a the last week of September or the first week of October each mediados de mayo y terminado la última semana de septiem- year. At the end of each grazing season forage samples were bre o la primera de octubre. Al final de la estación de apacen- clipped inside and outside of cages randomly located in each tamiento se cortaron muestras de forraje fuera y dentro de pasture to estimate end of season standing crop and forage uti- jaulas localizadas aleatoriamente en cada potrero, esto con el lization. Animal activity data were collected for 9 days during fin de determinar la biomasa en pie al final de la estación de August and September 1994. Steer weight gains were not dif- crecimiento y la utilización del forraje. Durante 9 días, entre ferent among crested wheatgrass, western wheatgrass, smooth agosto y septiembre de 1994, se tomaron datos de actividad bromegrass and flat native pastures, but weight gains of steers animal. Las ganancias de peso de los novillos no difirieron grazing rolling native pastures were lower (P<0.05) than gains "crested wheatgrass", "western wheatgrass", "smooth on other pastures. Weight gains per steer were 8% higher bromegrass" y pastizales planos. Sin embargo, las ganancias (P<0.05) on moderately grazed pastures, but weight gains per de peso de los novillos apacentando pastizales ondulados fueron menores (P<0.05) que las ganancias obtenidas en otros hectare were 39% higher on heavy grazed pastures. Steers potreros. Las ganancias por novillo fueron 8% mayores spent more (P<0.05) time grazing on smooth bromegrass than (P<0.05) en los potreros con carga animal moderada, pero las western wheatgrass, crested wheatgrass, or flat native pas- ganancias por hectárea fueron 39% mayores en los potreros tures and they also spent more (P<0.05) time grazing on heavy con carga alta. Los novillos pasaron mas tiempo (P<0.05) than moderately grazed pastures. Seeded cool-season grasses apacentando "smooth bromegrass" que "western wheat- produced season-long yearling steer weight gains comparable grass", "crested wheatgrass o pastizales planos. Los novillos to flat native, and superior to rolling native pastures, even de la carga alta pasaron mas tiempo apacentando que los de la when grazed at a stocking rate that was 80% heavier than the carga moderada. Los zacates de crecimiento invernal produ- rate recommended for native rangeland by the USDA-SCS jeron ganancias de peso comparables a las obtenidas en pasti- (1984). These results suggest that seeded cool-season grasses zales planos y superiores a la de los ondulados, aun cuando can be successfully grazed season-long in the Northern Great fueron apacentados con una carga animal 80% mayor a la Plains where environmental conditions and precipitation pat- recomendada por el USDA-SCS (1984) para pastizales nativos. terns are comparable to central North Dakota. Estos resultados sugieren que en las Grandes Planicies de Noreste, donde las condiciones ambientales y patrones de pre- The authors wish to thank Mr. Richard Huppler and Mr. Gordon Jensen for their cipitación son comparables a los de la región central de North assistance in conducting this study and Dr. Gary V. Richardson for advice and Dakota, los zacates invernales pueden ser exitosamente pas- assistance with statistical procedures. toreados a lo largo de la estación. Mention of a trade name is solely to identify materials used and does not consti- tute endorsement by the U.S. Department of Agriculture. Manuscript accepted 27 Aug. 1998.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 235 Key Words: native rangelands, season- land in central North Dakota. Hofmann and fine-silty mixed Typic Haploborolls ality, crested wheatgrass, smooth et al. (1993) also demonstrated that soils. Flat native pastures were located bromegrass, western wheatgrass cool-season grasses seeded on non- on soils comparable to the seeded pas- mined land and grazed at the same sea- tures, but rolling pastures were located Seeded cool-season grasses are usual- son-long (133 days) stocking rate as on loam, silt loam, and silty clay Entic ly recommended for spring or fall graz- native range produced comparable year- and Typic Haploborolls soils. ing in the Northern Great Plains because ling steer weight gains. In that study The 73 ha of land to be seeded was it is widely believed that their nutritive both cool-season and native pastures divided into 6.1 ha plots and randomly -1 quality in mid-summer is too low to were stocked at 1.5 AUM ha , which seeded to 1 of the 3 cool-season grasses support adequate livestock weight gains. was 15% greater than the USDA-SCS by no-till seeding into spring wheat This pattern of use was first suggested (1984) recommended stocking rate for stubble (Triticum aestivum L.) in the fall by Sarvis (1941) and Williams and Post native range in this area. However, for- of 1985 (Hofmann et al. 1993). Four, (1945), when they suggested that crested age utilization rates were only 30Ð40%, 6.1 ha plots were fenced on both flat and wheatgrass and smooth bromegrass indicating that a higher stocking rate rolling native rangelands (Hofmann et were best used in the spring when they might be warranted. The objective of the al. 1993). In the current study there were provided nutritious forage for grazing study reported here was to determine the 5 pasture treatments, 2 stocking rates 2Ð3 weeks earlier than native grasses. effect of moderate (1.6) and heavy (2.4 and 2 replications of each pasture treat- -1 They also indicated these grasses should AUM ha ) season-long (140 days) ment by stocking rate combination for a be completely utilized by about 1 July. stocking rates on forage utilization and total of twenty pastures. Stocking rates Rogler et al. (1962) reported that "most the performance of yearling steers graz- were established by adjusting pasture cool-season grasses lose quality rapidly ing crested wheatgrass, western wheat- size to 5.7 and 3.8 ha for moderate and after heading in late June". grass, smooth bromegrass, and flat and heavy stocking rates, respectively. Nevertheless, both Sarvis (1941) and rolling native range. Pastures were stocked with 3 Hereford Williams and Post (1945) demonstrated steers (Bos taurus) weighing approxi- that crested wheatgrass and smooth mately 294 kg at the beginning of each bromegrass could be grazed season-long Materials and Methods grazing season. Due to limited forage and produce animal weight gains com- utilization (30Ð42%) in the previous parable to native range. Heavy stocking Grazing studies were conducted near study (Hofmann et al. 1993) stocking rates and early starting dates were used Mandan, North Dakota for 140 days rates of 1.6 (moderate), and 2.4 AUM -1 by Williams and Post (1945), thus graz- each summer from 1992Ð1994. Grazing ha (heavy) were established. None of ing periods ranged from 79 to 180 days periods were from 14 May to 30 the pastures were fertilized for at least for crested wheatgrass and 69-160 days September 1992; 20 May to 6 October 10 years before the current study was for smooth bromegrass depending on 1993; and 19 May to 5 October 1994. initiated in 1992, and pastures received yearly precipitation patterns. Because Seeded species were ‘Nordan’ crested no fertilizer during this study. producers needed a dependable summer- wheatgrass (Agropyron desertorum Rolling and flat native pastures dif- long forage supply, the practice of fully [Fisch. Ex Link] Schult.), ‘Rodan’ west- fered in species composition as well as utilizing cool-season grasses in spring, ern wheatgrass (Pascopyrum smithii topography. Flat native pastures con- followed by grazing native rangeland in [Rydb.] Löve), and ‘Lincoln’ smooth tained blue grama (Bouteloua gracilis summer was adopted. Later Frisch- bromegrass (Bromus inermis Leyss.). [H.B.K.] Griffiths), green needlegrass knecht et al. (1953) suggested that crest- Native pastures were located on adja- (S. viridula Trin.), needleandthread ed wheatgrass could also be used for fall cent class II and III, 2Ð6% slope (flat (Stipa comata Trin. and Rupr.), western grazing, if regrowth was adequate. In native) and class IV and VI, 9Ð25% wheatgrass, sedges (Carex spp.), three- the 1970's extensive coal mining in the slope (rolling native) land. Seeded and awn (Aristida spp.) and Kentucky blue- Northern Great Plains (Ries et al. 1977) native pastures included 2 replications at grass (Poa pratensis L.). Rolling native brought the related need to stabilize both moderate and heavy stocking rates. pastures contained the same species as reclaimed land with perennial grasses. Seeded pastures were established on flat native pastures plus little bluestem Before mining, about 51% of this land fine-silty, mixed Pachic Haploborolls (Schizachyrium scoparium [Michaux] was in native rangelands and used for grazing. Reseeding mined land to native Table 1. Precipitation (mm) and mean monthly ambient air temperatures (C) between April and grass mixtures was expensive, slow, and October for 1992Ð1994 and the 80-year average. difficult, while introduced cool-season 1992 1993 1994 80-yr avg grasses could be more readily estab- Month Precip. Temp. Precip. Temp. Precip. Temp. Precip. Temp. lished at lower cost and grazed more quickly after seeding. Hofmann and April 8 6 36 6 31 6 38 6 May 38 15 69 13 18 16 58 12 Ries (1989) using moderate stocking June 117 17 114 16 71 19 89 18 rates based on recommendations result- July 69 17 343 18 48 19 56 21 ing from lessons learned in the 1930's August 41 17 48 19 5 19 51 21 drought (Dyksterhuis 1949), found that Sept 23 13 5 12 69 16 41 14 cool-season grasses could be grazed sea- Oct 5 7 0 6 147 9 20 8 son-long (126 days) on reclaimed mined Total 301 615 389 353

236 Journal of Range Management (52)3, May 1999 Table 2. Pasture means averaged over stocking rate and years for end of season standing crop term. Data were considered significant (standing crop), forage utilization, and yearling steer weight gains for 140-day grazing seasons in at P<0.05 unless otherwise indicated. 1992Ð1994)1.

Standing Crop Utilization Steer performance Results and Discussion Pasture Ungrazed Grazed Gain steer-1 ADG Gain ha-1 ------(kg ha-1)------(%) ------(kg)------Smooth bromegrass 2,898d 1,946b 31a 143a 1.02a 93a During this 3-year study, western Western wheatgrass 5,018a 3,476a 28a 141a 1.01a 91a wheatgrass had the highest average end Crested wheatgrass 4,367b 3,164a 27a 136a 0.97a 89a of season standing crop dry matter per Flat native 3,885bc 3,134a 19a 131a 0.94a 86a hectare and the greatest amount of Rolling native 3,371cd 2,470b 25a 115b 0.82b 75b grazed residue at the end of the grazing SE 202 175 3 4.5 0.03 3.2 season, while smooth bromegrass had 1Means within a column with different letters differ (P<0.05). the lowest end of season standing crop dry matter and the least amount of grazed residue (Table 2). There were Nash) and patches of western snowberry were observed every 20 minutes from significant (P<0.05) interactions (Symphoricarpos occidentalis Hook), dawn to dusk for 9 days in August and between pasture treatment and year for and buffaloberry (Shepherdia argentea September. All treatments were both ungrazed and grazed end of season [Pursh] Nutt.). observed on the same days. Time spent standing crop and utilization rates, indi- End of the grazing season standing grazing, lying, standing, walking, play- cating that forage production and uti- crop was estimated by hand clipping ing, scratching, salting and drinking lization differences among pasture treat- forage at a 5 mm stubble height from 36 were recorded, and the percent of the ments were not entirely consistent over by 36-cm plots, randomly located inside total daily observation time spent in the 3 years. In a previous study on this 0.9 by 4.3-m cages used to prevent graz- these activities was calculated. Walking, site (Hofmann et al. 1993), western ing. Similar procedures were used to playing, scratching, salting, and drink- wheatgrass was also the most productive collect forage samples at random loca- ing were minor activities which occu- and had the most grazed residue at the tions outside of each cage to estimate pied little time and thus were grouped end of the season, but hilly (rolling) the amount of residual forage where and analyzed as all other activities. native pastures produced the least end of grazing occurred. Samples were clipped Daily precipitation and ambient mini- season standing crop and had the least far enough from cages to eliminate any mum and maximum air temperatures remaining grazed residue. Despite animal-cage effects. One sample was were recorded at a weather station locat- increased stocking rates, forage utiliza- clipped from both inside and outside of ed about 3.2 km north of the study site. tion over both moderate and heavy treat- each cage and cages were relocated each Monthly precipitation totals and mean ments in the current study was lower year. Three cages were randomly locat- monthly temperatures (minimum and than utilization rates for these pastures ed in each seeded pasture, 4 cages in maximum) were calculated and com- reported by Hofmann et al. (1993). each flat native, and 5 cages in each pared to historic values (Table 1). Utilization in the current study ranged rolling native pasture. More cages were Data were analyzed by analysis of from 19% for flat native to 31% for located in native pastures because of the variance according to a completely ran- smooth bromegrass pastures, but differ- diversity of vegetation. Forage samples domized design. Pasture treatment, ences among pastures were not signifi- were dried at 60¡C and used to calculate stocking rate and pasture treatment x cant. Utilization rates were probably end of season standing crop dry matter stocking rate were tested using repli- lower in the current study because pre- (DM), and grazed residue per hectare. cate(pasture treatment x stocking rate) cipitation levels were generally above Forage utilization was calculated as fol- as the test term. Year, pasture treatment normal (Table 1), while the earlier study lows: Utilization (%)=[(end of season x year, stocking rate x year, and pasture included 2 drought years. standing crop DM (within cage sam- treatment x stocking rate x year were Although smooth bromegrass pastures ples)-grazed residue DM (outside of tested using year x replication (pasture produced the least forage, steers gained cage samples))/end of season standing treatment x stocking rate) as the test similarly on smooth bromegrass and crop DM (within cage samples)] x 100. Steers were weighed following an overnight stand without feed or water at Table 3. Stocking rate means for end of season standing crop (standing crop), forage utilization, and yearling steer weight gains averaged over pasture treatments and years for 140-day grazing the beginning and end of each study and 1,2 at 21-day intervals during the studies. seasons in 1992Ð1994 . Trace mineralized salt containing 96- 98.5% salt, 0.35% zinc, 0.34% iron, Standing Crop Utilization Steer performance 0.20% manganese, 0.033% copper, Stocking rate Ungrazed Grazed Gain steer-1 ADG Gain ha-1 0.007% iodine, and 0.005% cobalt ------(kg ha-1)------(%) ------(kg)------(Akzo Salt, Inc., Clarks Summit, Penn.) Moderate 4,223a 3,189a 22b 138a 0.99a 73b was available at all times. Heavy 3,582b 2,480b 30a 128b 0.92b 101a In 1994 steer activities on smooth SE 129 112 2 2.9 0.02 2 1Moderate and heavy stocking rates were 1.6 and 2.4 AUM ha-1, respectively. bromegrass, western wheatgrass, crested 2 wheatgrass, and flat native pastures Means within a column with different letters differ (P<0.05).

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 237 Fig. 1. Three year (1992Ð1994) average cumulative yearling steer weight gains by weighing period (± SE) for seeded cool-season and flat and rolling native pastures. western wheatgrass (Table 2). There grasses changed little after mid-July. grazed pastures, therefore forage disap- were no significant (P>0.20) interac- Although Sarvis (1941) reported season- pearance per hectare through grazing, tions among pasture treatment, stocking long average daily gains on smooth trampling and deterioration was compa- rate, or year with regard to animal per- bromegrass (0.92 kg) which were com- rable between moderate and heavy formance data. Weight gains were not parable to weight gains in this study stocking rates at 1,034 and 1,102 kg ha-1, different among seeded cool-season pas- (Table 2), they sometimes reported large respectively. tures and the flat native treatment, but weight losses in August and September Average daily gains were significantly weight gains were lower for the rolling which were not encountered in the cur- higher (8%) at the moderate compared to native treatment. Hofmann et al. (1993) rent study (Fig. 1). the heavy stocking rate, but weight gains reported that weight gains for the level End of season standing crop averaged per hectare were 39% higher for the (flat) native pastures were significantly over all cool-season and native pastures heavy stocking rate (Table 3). higher than gains from the seeded cool- was greater (P<0.05) at the moderate Cumulative weight gains for moderate season pastures. Differences between compared to the heavy stocking rate, and heavy stocking rate treatments aver- the 2 studies suggest that when precipi- which suggests the heavier stocking rate aged over all 3 years did not vary signifi- tation is average or above, cool-season may have adversely affected forage pro- cantly (P<0.11) until mid-July (Fig. 2). grasses support good animal perfor- duction (Table 3). Grazed residue was From early August until October, steers mance, but with drought conditions also greater (P<0.05) on moderately grazing at the moderate stocking rate had weight gains from the well established native grasses may be greater. Table 4. Mean end of year standing crop (standing crop), forage utilization, and yearling steer weight gains averaged over pasture treatment and stocking rates for 140-day grazing seasons in Three-year-average cumulative weight 1992, 1993, and 19941. gains show that beginning with the second weighing period in late June, steers Standing Crop Utilization Steer performance grazing smooth bromegrass tended to Year Ungrazed Grazed Gain steer-1 ADG Gain ha-1 have the highest weight gains while ------(kg ha-1)------(%) ------(kg)------steers grazing rolling native pastures 1992 3,483c 2,544c 27a 135a 0.96a 88a had the lowest (Fig. 1). These weight 1993 3,928b 3,136a 20b 139a 0.99a 91a gain differences continued to increase 1994 4,296a 2,818b 32a 125b 0.89b 82b until the end of the season, but cumula- SE 120 81 2 2 0.01 1.2 tive gain differences among the other 1Means within a column with different letters differ (P<0.05).

238 Journal of Range Management (52)3, May 1999 Fig. 2. Three year (1992Ð1994) average cumulative yearling steer weight gains by weighing period (± SE) for moderate and heavy stocking rates. progressively higher weight gains than tures spent significantly more time graz- Steers spent significantly less time steers grazing at the heavy stocking rate. ing than steers on western wheatgrass, grazing and more time standing on mod- Ungrazed end of season standing crop crested wheatgrass, or flat native pas- erately stocked than on heavily stocked increased significantly from 1992Ð1994, tures (Table 5). Time utilized in other pastures (Table 6). Steers on heavily but the grazed end of season standing activities was similar among pastures, stocked pastures may have spent more crop was lower in 1994 than 1993 but steers grazing smooth bromegrass time grazing because lower forage pro- (Table 4). Higher ungrazed end of sea- tended to spend less time lying, stand- duction (Table 3) necessitated increased son standing crop each year may have ing, and in all other activities compared grazing time for them to reach satiety. reflected a gradual recovery from the to steers on the other pastures. Smooth drought of 1988 and 1989 and carry bromegrass pastures produced the low- over water from 1993. Lauenroth and est dry matter yield, yet steer weight Conclusions Sala (1992) also reported a lag time gains were comparable to the other pastures, therefore grazing time may have between increased precipitation and a Results of this study show that cool- been greater because steers were exer- response in forage production. Forage season seeded grasses can be grazed cising a high degree of diet selectivity or utilization was highest in 1994, and continuously over a 140-day season in because less available forage necessitat- lowest in the extremely wet year of the Northern Great Plains, where precip- 1993. End of season grazed and ed increased grazing time. Hardison et al. (1954), Kirby and Stuth (1982), and itation and environmental conditions are ungrazed standing crop differences many others have reported that cattle similar to central North Dakota, and between 1993 and 1994 could have been graze selectively and that they prefer support yearling steer weight gains due to less growing season precipitation leaves to stems. comparable to good quality native in 1994, which would have resulted in less forage regrowth and lower grazed Table 5. Percentage of observation time steers were engaged in various activities during dawn to end of season standing crop. The high dusk observation for 9 days in August and September 19941. ungrazed end of season standing crop in 1994 probably occurred early in the ACTIVITY summer as a result of stored moisture Pasture Grazing Lying Standing All Other from 1993. It is unlikely that ungrazed ------(%)------end of season standing crop levels were Smooth bromegrass 64a 26a 7a 3a Western wheatgrass 57b 31a 9a 4a substantially affected by a build up of b a a a the previous year’s forage growth, Crested wheatgrass 59 30 7 4 Flat native 58b 28a 10a 4a because cages were relocated each year. SE 1.1 1.1 0.7 0.3 Steers on smooth bromegrass pas- 1Means within a column with different letters differ (P<0.05).

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 239 Table 6. Effect of stocking rate on the percentage of time steers were engaged in various activities Hofmann, L. and R. E. Ries. 1989. Animal during dawn to dusk observations over 9 days in August and September 19941,2. performance and plant production from continuously grazed cool-season reclaimed and ACTIVITY native pastures. J. Range Manage. Stocking rate Grazing Lying Standing All Other 42:248Ð251. Hofmann, L., R. E. Ries, J. F. Karn, and A. ------(%)------B. Frank. 1993. Comparison of seeded and Moderate 58b 30a 9a 4a native pastures grazed from mid-May Heavy 61a 28a 7b 4a through September. J. Range Manage. SE 0.8 0.8 0.5 0.2 46:251Ð254. 1Moderate and heavy stocking rates were 1.6 and 2.4 AUM ha-1, respectively. 2 Kirby, D. R. and J. W. Stuth. 1982. Seasonal Means within a column with different letters differ (P<0.05). diurnal variation in composition of cow diets. J. Range Manage. 35:7Ð8. rangelands and better than more typical easier to establish, and they can be Lauenroth, W. K. and O. E. Sala. 1992. Long- rolling rangelands. These data agree grazed sooner after seeding compared to term forage production of North American with results reported by Hofmann et al. native grass mixtures. Planting intro- shortgrass steppe. Ecol. Appl 2:397Ð403. Ries, R. E., F. M. Sandoval, and J. F. Power. (1993), but do not support the popular duced cool-season grasses also gives 1977. Reclamation of disturbed lands in the management philosophy that cool-sea- producers the additional flexibility of lignite area of the Northern Plains, p. son grasses should only be used for alternating between small grain and 309Ð327. In: G. H. Gronhovd and W. R. spring and fall grazing. Data from this perennial grass production. Kube (eds.), Proc., 1977 Symposium on study further indicate that it may be pos- technology and use of lignite. GFERC/IC- sible to graze cool-season grasses sea- 77/1. Energy Res. and Dev. Admin.-Univ. of Literature Cited North Dakota. Grand Forks, N.D. son-long at stocking rates as much as Rogler, G. A., R. J. Lorenz, and H. M. 80% higher than USDA-SCS (1984) Schaaf. 1962. Progress with grass. North recommendations for native range. Dyksterhuis, E. J. 1949. Condition and man- Dakota Agr. Exp. Sta. Bull. 439. Acceptable stocking rates will vary agement of range land based on quantitative Sarvis, J. T. 1941. Grazing investigations on according to soil type and precipitation ecology. J. Range Manage. 2:104Ð115. the Northern Great Plains. North Dakota Agr. Frischknecht, N. C., L. E. Harris, and H. K. Exp. Sta. Bull. 308. amounts. Good quality native rangelands Woodward. 1953. Cattle gains and vegetal USDA-SCS. 1984. Guide to range sites, condi- should not be plowed and reseeded to cool- changes as influenced by grazing treatments tion class and initial stocking rate. USDA- season grasses, but disturbed lands or on crested wheatgrass. J. Range Manage. Soil Conservation Service Field Office marginal croplands can be seeded to 6:151Ð158. Technical Guide, Section 2, Rangeland cool-season grasses for season-long Hardison, W. A., J. T. Reid, C. M. Martin, Interpretations, Missouri Slope. p. 4. and P. G. Woolfolk. 1954. Degree of grazing. Introduced cool-season grass Williams, R. M. and A. H. Post. 1945. Dry herbage selection by grazing cattle. J. Dairy land pasture experiments. Montana Agr. Exp. seed is less expensive, the grasses are Science. 37:89Ð102. Sta. Bull. 431.

240 Journal of Range Management (52)3, May 1999 J. Range Manage. 52:241Ð248 May 1999 Caatinga vegetation dynamics under various grazing intensities by steers in the semi-arid Northeast, Brazil

SEVERINO G. DE ALBUQUERQUE

Author is Caatinga research scientist, Embrapa-CPATSA/Caixa Postal 23. 56300-000 Petrolina, PE Brasil.

Abstract Resumo

The effects of cattle grazing were evaluated on range dynamics Estudaram-se 3 taxas de lotação anual (TL) (pesada, 1 boi 6,7 of the Caatinga which is a deciduous dry woodland, covering most ha-1; média, 1 boi 10,0 ha-1 leve, 1 boi 13,3 ha-1), além de uma of the semi-arid Brazilian Northeast. Three stocking rates (SR) exclusão livre de pastejo (zero). Na 1a etapa (1978Ð81), cada TL -1 -1 were studied (heavy, 1 steer 6.7 ha ; moderate 1 steer 10 ha ; foi estudada sob 2 sistemas de pastejo, contínuo e rotacional- -1 light, 1 steer 13.3 ha ), in addition to an ungrazed exclosure (zero mente deferido com 3 subdivisões. Na 2a etapa (1981Ð84), as sub- stocking). In the first phase (1978Ð81) each stocking rate was test- divisões foram eliminadas, passando a ser repetições do pastejo ed under continuous and deferred grazing. In the second phase contínuo. Usaram-se 6 bovinos machos por tratamento. Não (1981Ð84), deferred grazing was eliminated, so that pastures houve influência nem da TL e nem do curto periodo de pastejo became replications of continous grazing. Six steers per pasture diferido na freqüência (FQ) das espécies (spp) herbáceas. A were used, and pasture size was used to vary stocking rate. There grande variação na precipitação anual se refletiu na FQ delas, as was no effect of stocking rate or grazing system period on the fre- quais foram divididas em 3 grupos, sendo o principal deles for- quency of the herbaceous species. They were, however, influ- mado por 16 spp muito influenciadas, para as quais houve corre- enced by rainfall in the period, and could be divided into 3 lação (P<0,05) entre os 2 parametros. Não houve efeito da precip- groups. Sixteen species increased with increasing rainfall during itação na FQ de duas spp importantes (Herissantia crispa (L.) the last months of the rainy season, and reached the highest fre- Briz. e. Selaginella convoluta Spring.). A mortalidade de 5 arbus- quency in 1984. Eleven species also increased with increasing tos etiquetados (Lippia microphylla Cham., Croton rhamnifolius rainfall but reached the highest frequency in 1983. Rainfall had (Kunth em.) Mull. Arg., Calliandra depauperata Benth, Cordia no effect on the frequency of 2 important species, Herissantia leucocephala Moric. e. Bauhinia cheilantha (Bong.) Steud.) crispa (L.) Briz. and Selaginella convoluta Spring. Death rate of 5 diminuiu com o decréscimo na TL, sendo 11, 7, 9, 3, 7, 7 e 4,5% shrubs (Lippia microphylla Cham., Croton rhamnifolius (Kunth respectivamente para Pesada Média, Leve e Zero, havendo difer- em.) Mull. Arg. Calliandra depauperata Benth, Cordia leuco- ença significativa entre pastejo e exclusão. Houve diferença cephala Moric., and Bauhinia cheilantha (Bong.) Steud.) (P<0,05) entre arbustos, sendo maior naqueles mais fáceis de decreased with decreasing stocking rate, 11.7, 9.3, 7.7, and 4.5%, serem quebrados como L. microphylla e C. leucocephala. A TL respectively on heavy, moderate, light, and zero stocking. Death também influenciou significativamente no crescimento em altura, rates were higher in easily broken shrub species, L. microphylla sendo respectivamente de Ð2,7 e 9,8% para Pesada e Zero. A den- and C. leucocephala. Stocking rate also influenced the height sidade de arbustos e árvores, determinada pelo Método do Ponto growth rate of the tagged shrubs, being respectively Ð2.7 and Quadrante, foi respectivamente de 21.109 e 447 plantas ha-1 em 9.8% for heavy and zero stocking. Mean density of shrubs and 1982, e de 13.230 e 401 plantas ha-1 em 1984, sendo a grande mor- trees, determined by the Point-Centered Quarter Method, was talidade de arbustos (37,3%) no período 1982–84 provavelmente -1 respectively 21,109, and 447 plants ha in 1982, and 13,230 and devido a grande seca de 1982, nao havendo influência da TL. 401 plants in 1984; the main cause of the high shrub death Considerando-se as 7 áreas experimentais isoladamente, não (37.3%) was probably the 1982 drought. Density was not affected houve regressão entre as densidades de arbustos de 1982 e 1984, by stocking rate. Considering the 7 experimental areas separate- havendo contudo regressão entre a densidade de arbustos em ly, there was no regression between 1982 and 1984 shrub densi- 1982 e a diferença entre as densidades de 1982 e 1984. ties. There was, however, regression between 1982 density and the difference between 1982 and 1984 densities.

The Caatinga, a thorny, deciduous, dry woodland that cov- Key Words: tropical woodlands, stocking rate, native pastures, ers most of the Brazilian Northeast, is dominated by woody grazing system, shrubs and trees, herbaceous stratum, frequency. plants, and may be a range with the highest density of shrubs and trees in the world. Most of the Caatinga types are shrub dominated, although as quoted by Sampaio (1995), some The author wishes to thank Drs. José Givaldo G. Soares and Clovis Guimarães Filho, for their help in designing and setting up the study; Drs. Antônio de Pádua authors have claimed that most of the Caatinga area was M. Fernandes, Mário de A. Lira and Dárdano de Andrade-Lima (in memoriam), orginally covered with trees. Most woody species are decidu- for their support in the approval and setting up the research: Dr. João Ambrósio de ous, and leaf litter represents an important source of forage in Araújo Filho, for assistance in methodological procedures; Dr. Everardo V.S.B. Sampaio for reviewing this manuscript. the dry season (Kirmse et al. 1987). In spite of being less sen- Published with the approval of CPATSA Publishing Committee. sitive to overgrazing than pastures dominated by herbaceous Manuscript accepted 15 Aug. 1998.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 241 vegetation, the Caatinga has also been folius (Kunth em.) Mull. Arg., the 5 dominant ones already cited plus degradated (Vasconcelos Sobrinho 1949, Calliandra depauperata Benth., Cordia Lantana camara L. and Croton sonderi- Andrade-Lima 1981). Degradation was leucocephala Moric, and Bauhinia anus Mull. Arg. Measurements were probably caused by shifting cultivation cheilantha (Bong.) Steud. The last 3 are taken biennially, by monitoring up to 10 and fuel wood harvest because the endemic to the Caatinga (Prado 1991). plants of each shrub species in each Caatinga-covered region is densely pop- The distinction between shrubs and trees macro plot, starting in 1980. ulated. The paucity of grasses is one of follows criteria of Walker (1976). A third aspect was evaluated in the its characteristics (Cole 1960), but there The study was conducted from August experimental units in 1982 and 1984: is evidence that the Caatinga invaded 1978 to August 1984, being divided into the density of trees and shrubs (height > some formerly more open areas covered 2 phases. In the first one (August 0.5 m) as determined by the Point- by grasses (Smith 1974), a situation 1978ÐAugust 1981), 3 stocking rates Centered Quarter Method (PQM) analogous to those in USA and were tested: heavy, H, 1 steer 6.7 ha-1 (Cottam and Curtis 1956). Our purpose Australia, in which overgrazing resulted (40 ha); moderate, M, 1 steer 10 ha-1 (60 was to measure effect of stocking rate in establishment of woody plants (Herbel ha); light, L, 1 steer 13.3 ha-1 (80 ha). on the woody stratum as a whole. In the 1985, Harrington and Hodgkinson These were combined with 2 grazing areas of 40 ha (1 steer 6.7 ha-1 and 1986). systems: continuous (C), and rotational- exclusion), 60 ha (1 steer 10.0 ha-1) and Annual stocking rate (SR) is the most ly deferred, with 3 sub-divisions (D). A 80 ha (1 steer 13.3 ha-1), 100, 150, and important factor in maintaining the sta- fourth treatment was an ungrazed exclo- 200 points were placed respectively. bility of a native pasture. In Caatinga, a sure of zero stocking (40 ha). None of The sampling points were distributed in stocking rate of 15 ha head-1 of cattle is the 7 experimental units was replicated. lines, and in each one, the minimum and usually recommended (Banco do In the second phase (August maximum distances between sampling Nordeste do Brasil 1971, Rodrigues and 1981ÐAugust 1984), the pastures used points were 21 and 26, m respectively, Borges 1979). On most of the proper- for deferred grazing were eliminated, any distance in this interval being deter- ties, however, cattle are associated with becoming replications of the continuous mining by sorting. From each quarter of goats and sheep. Adjustments in stock- grazing treatments. Six steers per exper- each sampling point, I measured the dis- ing rate may be followed by adjustment imental unit were utilized, resulting in a tance to the nearest shrub and to the in grazing system for range rest (Hickey total area of 400 ha, including the exclo- nearest tree. Shrubs were classified into 1977). Rotationally deferred grazing is sure of 40 ha. 3 heights (H1 > 0.5 Ð 1 m; H2 > 1Ð2; designed to help degraded ranges recov- To evaluate the effect of grazing H3 > 2m). er (Sampson 1913). intensity on vegetation, macro plots of Data were interpreted, taking into This research was undertaken to study 20 x 5 m were located systematically account the effect of stocking rate and the effect of stocking rate and rotational- within each experimental unit, with 6 in year on the frequency of herbaceous ly deferred grazing by cattle on the continuous grazing and in the exclosure, species, the effect of stocking rate and Caatinga of the semi-arid region of and 12 in deferred grazing (4 in each biennium on the performance of the Pernambuco State, in the Brazilian sub-division). In each macro plot, the tagged shrubs, and the effect of the year Northeast. following 2 aspects were evaluated: (1) on the density of shrubs and trees. Data frequency of herbaceous species and on mortality and changes in height and density of seedlings (height <0.5) of canopy cover of the tagged shrubs were Materials and Methods woody plants, determined annually, in transformed from percentage to arc sin May (in 1979, 1981, 1983, and 1984) or and evaluated with analysis of variance The research was conducted at the in April (in 1980 and 1982), by using (Snedecor and Cochran 1976). As treat- Caatinga Experimental Station (CEC) five 1-m2 (2 x 0.5 m) quadrats, placed at ments were not replicated, the interac- (9¡21' S Lat; 370 m altitude) of the random; and (2) mortality and growth in tions involving stocking rate, biennia, Brazilian Agricultural Research height and in canopy cover of 7 alu- and plant species were used as residual Corporation (EMBRAPA)-Agricultural minum tagged shrub species, including variance. Research Center for Semi-Arid Tropics (CPATSA), in Petrolina municipality. Table 1. Rainfall from October 1978—September 1984, and historical mean, October 1963— September 1997. The area has flat topography and red- yellow podzolic soils (Burgos and Precipitation Cavalcanti 1991) with the following Month 1978Ð79 1979Ð80 1980Ð81 1981Ð82 1982Ð83 1983Ð84 Historical characteristics: pH = 5.8; Ca2+ + Mg2+ = mean 3+ 3.3 meq/100 g;/ Al = 0.7 meq/100 g; P ------(mm) ------3 ppm. Annual potential evaporation is Oct.—Nov. 42.2 52.1 56.0 14.6 0.0 62.8 58.7 2,630 mm, with a mean annual precipi- Dec. 12.2 54.8 34.0 90.7 82.8 7.2 75.1 tation of 567 mm (Table 1). The vegeta- Jan. 118.1 186.0 20.3 10.4 60.0 20.5 72.2 Feb. 96.4 201.3 4.8 20.6 166.4 3.9 83.7 tion is an arboreous-shrubby Caatinga, Mar. 28.3 44.7 340.3 79.1 205.2 314.2 140.9 with the tree stratum dominated by Apr. 118.4 10.6 20.5 97.4 0.8 122.9 86.4 Mimosa tenuiflora (Wiild.) Poir., and May 18.4 1.2 0.5 1.4 0.0 44.6 20.2 the shrub stratum dominated by Lippia Jun.ÐSep. 24.0 1.3 8.4 50.2 37.6 29.2 30.3 microphylla Cham., Croton rhamni- Total 458.0 552.0 484.8 364.4 552.8 605.3 567.5

242 Journal of Range Management (52)3, May 1999 Results and Discussion tem effect on the tagged shrubs was not have been obtained, confirming that in evaluated because measurements were semi-arid climates, increases in rainfall started in 1980 and the deferred system result in linear increases in primary pro- There were 29 herbaceous species was discontinued in 1981. ductivity (Whittaker 1975). In the sec- which occurred with average frquency > Through the years, there was a ond group (11 species), the frequencies 4.33%, considering the means calculated remarkable variation in frequency of also varied strongly through the years, over 6 years and 4 treatments. Plant most of the herbaceous species, proba- but their highest value was in 1983. types were distributed as follows: pteri- bly caused by variation in the precipita- There was a change of positions dophyte, 1 species; monocotyledonous, tion. Based on this aspect, the species between the first 2 groups in 1983 and 6 species (5 grasses and 1 sedge); and were divided into 3 groups (Table 2). 1984, i.e. the first group had the highest dicotyledonous, 22 species. The lowest The frequencies of the first 16 species mean frquency in 1984, while the sec- limit of 4.33% was chosen arbitrarily. were strongly infuenced by rainfall, and ond group did so in 1983. The third Grazing system had no effect on either an association (P<0.01) was detected group, made up of 2 very important herbaceous species frequencies or on between rainfall in the period Caatinga forbs, H. crispa and S. convo- woody seedling densities. This was an FebruaryÐApril (or Feb.ÐMar.) and their luta, was little influenced by rainfall. S. expected result because each deferred mean frequency (Fig. 1). In another convoluta is a reviviscent pteridophyte sub-division was kept free of grazing Caatinga, Araújo Filho (1985) detected that passes the dry season completely only once during the rainy season in the that herbaceous plants reacted more to desiccated, i.e., in air dried stage with a first phase. In the review conducted by fluctuation in rainfall than to livestock saturation deficit of up to 72Ð74% Gammon (1978), long duration of use. If phytomass had been measured, (Morello 1954). On a small ranch near deferred grazing was necessary to get similar associations would probably the experimental site, these species pro- improvement in pastures. Grazing sys-

Table 2. Frequency of herbaceous species in 3 groups (Group 1 = very influenced by rainfall but with highest frequency in 1984; Group 2 = Same as Group 1, but with highest frequency in 1983; Group 3 = little influenced by rainfall), in the period 1979Ð84.

Frequency Group/Species Plant 1979 1980 1981 1982 1983 1984 class ------(%) ------Group 1 Paspalum scutatum Nees ex Trin. Grass 2.14 9.29 7.38 0.71 2.38 4.05 Hyptis suaveolens (L.) Poit. Forb 0.00 8.81 11.19 0.00 5.24 10.95 Eragrostis ciliaris (L.) R. Br. Grass 4.05 3.10 8.33 0.00 13.33 14.52 Diodia teres Walt. Forb 1.19 5.24 10.00 3.57 11.43 15.71 Phyllanthus ninuri (L. em.) Mull Arg. Forb 0.00 3.81 25.71 0.00 8.10 12.62 Marsypianthes chamaedrys (Vahl.) Kuntze Forb 0.00 11.67 4.76 0.00 1.67 32.38 Borreria ocymoides DC. Forb 0.00 0.00 4.29 0.00 16.67 37.62 Cuphea sp. Forb 0.00 3.81 12.14 0.71 10.71 31.19 Polygala brizoides St. Hil. Forb 0.95 3.81 7.14 1.90 19.52 33.33 Mitracarpus frigidus K. Schum. Forb 1.19 8.81 32.14 8.57 17.14 19.05 Turnera pumilea L. Forb 0.00 3.81 19.29 12.38 35.48 39.29 Schwenkia americana Roy ex L. Forb 6.67 24.52 38.33 0.00 41.67 61.67 Cyperus uncinulatus Schrad. ex Nees Sedge 0.48 17.62 40.48 34.05 39.52 48.57 Gymnopogon rupestris Ridley Grass 29.52 45.78 35.95 17.86 37.38 51.67 Panicum trichoides Sw. Grass 18.10 49.05 55.48 2.38 58.09 75.71 Centratherum punctatum Cass. Forb 25.48 54.29 59.05 1.67 54.29 73.81 Mean 5.61 15.84 23.22 5.24 23.29 35.13 Group Macroptilium martii Benth. Forb 0.00 2.86 1.43 0.71 16.67 6.19 Triumfetta sp. Forb 0.00 0.00 2.38 1.90 20.00 10.48 Croton glandulosus (L. em) Forb 1.43 8.10 5.00 2.38 22.86 2.62 Mull. Arg Pavonia cancellata (L.F.) Cav. Forb 1.67 5.95 2.86 1.90 21.19 14.52 Portulaca oleracea L. Forb 0.00 3.33 7.14 0.00 33.33 16.67 Bernardia sinoides Mull Arg. Forb 0.00 5.00 9.29 13.10 24.76 16.43 Brachiaria molis (Sw.) L. Parodi Grass 6.67 8.57 0.48 0.95 45.00 9.29 Microtea scabrida Urb. Forb 0.48 0.48 15.48 1.43 38.81 24.52 Althernanthera brasiliana (L.) Kuntze Forb 7.38 26.90 24.76 1.67 38.81 35.71 Hybanthus calceolaria (L.) Forb 2.86 19.52 29.29 9.05 51.43 24.52 Schulze Corchorus argutus H.B.K. Forb 6.90 29.05 21.90 3.10 48.33 37.86 Mean 2.49 9.98 10.91 3.29 32.83 18.07 Group 3 Selaginella convoluta Spring Pteridophyte 29.76 26.90 30.00 34.05 30.00 35.48 Herissantia crispa (L.) Briz. Forb 48.81 39.05 42.86 37.86 59.05 58.33 Mean 39.28 32.97 36.43 35.95 44.52 46.90

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 243 but to a denser woody vegetation. This is probably the reason that most of our herbaceous species were less abundant in the exclosure. The density of woody species seedlings was not related to stocking rate (Table 3). Friedel (1986) detected influence from sheep and rabbits on woody seedlings, but no effect from cattle. An important point is the trend for the density of woody seedlings to decrease in the exclosure, following the same trend as most of the herbaceous species. The woody canopy of Caatinga in rest trends to become closed, curtailing the space suitable for the herbaceous stratum. Seedling densi- Fig. 1. Relationship between rainfall of last months of rainy season and mean frequency of ty also decreased under deferred heavy 16 herbaceous species. and moderate stockings, but the decrease in the exclosure was more evi- vided 70% of the herbaceous phytomass both the herbaceous and woody strata dent. Seedlings of Tabebuia spongiosa at the end of the dry season (Leal 1996). (Albuquerque and Bandeira 1995), steer Rizzini were not included in Table 3 There were so few plants of L. camara diets were made up of herbs and shrub because in 1983 a great germination of and C. sonderianus tagged in 1980 that leaves, decreasing the grazing pressure seeds of this tree occurred (mean of 16.4 neither the effect of biennium nor of on the lower stratum. In addition, the plants m-2), being completely different stocking rate could be estimated. L. high density of Neoglaziovia variegata from other woody species. microphylla suffered high mortality, so (Arr. Cam.) Mez and other thorny Mortality of the tagged shrubs that parameters other than death rate species in the lower stratum could form increased with increasing stocking could not be evaluated. There was a sig- micro-sites, protecting the herbs from intensity, and grazing treatments were nificant influence of biennia on mortali- over-use, and providing them the oppor- significantly different from the zero ty rate, which was highest in the second tunity to set seed. In South Africa, medi- grazing (Table 4). This indicates that biennium. The influence of the 1981Ð82 um sized stone rubble was considered cattle could cause overgrazing in the drought could only be detected in 1984. important in protecting grass seedlings shrub stratum as detected by Toutain Rainfall in the crop year 1980Ð81 was from overgrazing by large herbivores (1986), and damage can also be caused not low, but was highly concentrated in (Van der Walt 1980). by trampling (Chesterfield and Parsons March. Most of the herbaceous species 1985). On the other hand, Kelly and There was no apparent effect of stock- occurred less frequently in the exclosure Walker (1976) did not detect any harm- ing rate on frequency of herbaceous than in the grazed areas, except A. ful influence of cattle and goats on the species, and differences among treat- brasiliana and P. ninuri, which occurred woody stratum. Mortality was higher for ments were attributed to heterogeneity most frequently in the exclosure. Most C. leucocephala and L. microphylla, of the area. Frequency data might not research on grazing in range pastures shrubs that can be broken easily. have been sensitive enough to measure have been done in the USA, where Stocking rate had a significant effect on effect of stocking rate as in the work of ranges are dominated by grasses and the height of the 4 shrubs (Fig. 2). Hacker (1984) in a semi-arid range in forbs. Dyksterhuis (1949) based his Shrubs under heavy stocking always Australia. On the other hand, in a dense work on such pastures, and recognized decreased in height in both biennia, vegetation with a forage phytomass of that rest in woodlands does not lead to although in one of the pastures, heavy ca. 1,000 kg ha-1, distributed equally in range dominated by herbaceous plants, stocking under deferred grazing until August 1981 (HD), all shrubs except C.

Table 3. Density of woody species seedlings (height <0.5 m) in the period 1979Ð84, under 4 stocking intensities [Heavy (H) = 1 steer 6.7 ha-1; Moderate (M) = 1 steer 10 ha-1; Light (L) = 1 steer 13.3 ha-1; Zero = No grazing] and 2 grazing systems (C = Continuous grazing; D = Deferred grazing until 1981).

Stocking intensity/Grazing system Year HC HD MC MD LC LD Zero Mean ------(seedlings m-2)------1979 3.03 5.72 2.50 7.68 2.33 0.87 8.67 4.40 ± 3.0 1980 2.47 7.28 2.20 5.77 2.17 2.37 8.63 4.41 ± 2.8 1981 4.03 5.74 2.36 4.65 1.80 1.79 6.50 3.84 ± 1.9 1982 1.40 5.10 1.53 4.35 2.27 1.45 3.60 2.81±1.5 1983 1.83 4.08 5.10 3.95 2.80 1.03 3.57 3.20 ± 1.4 1984 2.23 3.37 2.53 3.75 2.53 4.14 3.77 3.19 ± 0.7 Mean 2.50 ± 0.9 5.21 ± 1.4 2.71 ±1.2 5.03±1.5 2.32 ± 0.3 1.94 ± 1.2 5.79± 2.5 3.64

244 Journal of Range Management (52)3, May 1999 Table 4. Mortality of 5 tagged shrubs, submitted to 4stocking intensities, in the period 1980Ð84 Mean density of shrubs was 37.3% (heavy = 1 steer 6.7 ha-1; Moderate = 1 steer 10 ha-1; Light = 1 steer 13.3 ha-1; Zero = No grazing). lower in 1984 than in 1982, the difference being considered very high (Table Stocking intensity Shrub species Heavy Moderate Light Zero Mean 5). Apparently, this was not caused by stocking rate, because in HD as well as ------(%) ------1st biennium (1980Ð82) in zero stocking, there were increases in C. leucocephala 10.90 11.38 7.38 0.00 7.41 shrub densities (Fig. 4). One point that L. microphylla 22.92 21.31 13.16 0.00 14.35 shows the difference very clearly is the B. cheilantha 0.00 8.79 6.09 0.00 3.72 C. depauperata 0.90 0.00 1.39 0.00 0.57 correlation between the mortality of the C. rhamnifolius 1.55 0.65 0.00 0.00 0.55 7 tagged shrubs in 1982Ð84 and their Mean 7.25 8.43 5.60 0.00 5.32 A “mortality” detected by point-centered 2nd biennium (1982Ð84) quatrum method (PQM) (Table 6). The C. leucocephala 32.25 31.08 17.70 11.43 23.86 difference in the columns of this table L. microphylla 21.62 4.17 9.09 30.00 16.22 B. cheilantha 18.18 7.23 12.04 2.08 9.88 might be attributed to the plant size. C. depauperata 2.73 4.17 7.04 0.00 3.48 Shrubs marked in 1980 had a mean C. rhamnifolius 3.15 4.55 3.61 1.96 3.32 height of 1 m, while shrubs 0.5 m high Mean 16.19 10.24 9.90 9.09 11.35 B were included in the PQM measure- biennia mean ments, and younger plants of lower C. leucocephala 23.07 21.23 12.54 5.71 15.64 a L. microphylla 22.27 12.74 11.12 15.00 15.28 a stature are more likely to die. Shrubs B. cheilantha 9.09 8.01 9.06 1.04 6.80 b 0.5Ð1 m high made up 67.3 and 47.7% C. depauperata 1.81 2.08 4.21 0.00 2.03 b of the total in 1982 and 1984, respec- C. rhamnifolius 2.35 2.60 1.80 0.98 1.93 b Mean 11.72 a 9.33 a 7.75 a 4.55 b 8.34 tively, while shrubs >2 m high made up 9.3 and 21.4% in 1982 and 1984, *Means with same lower case letters in the same line, in the same column, and with same capital letters in the same column, are not statistically different (Duncan, P>0.05). In spite of the difference between Heavy and Light being higher respectively. The loss of shrubs in the than between Light and Zero, there is statistical difference between Light and Zero, because of angular transformation. period 1982Ð84, that was detected by PQM, might have been caused by the leucocephala increased in height in the It could be concluded that cattle might 1982 density, i.e. a high density caused period 1982Ð84. This increase was prob- cause degradation of the shrub stratum, high competition for water in 1982. ably caused by the good rains in 1983 but even in the heaviest stocking rate in There was no regression (P>0.05) and 1984 and lack of competition this study, the degradation was lighter between 1982 and 1984 densities, there because HD had the lowest shrub densi- than the alteration caused by the being however regression (P<0.01) ty among the 7 experimental units in 1981Ð82 drought. Stocking rates are between 1982 density and the difference 1982. Moderate stocking had the oppo- more likely to be determined by animal between 1982 and 1984. In 2 of the 3 site effect compared to HD, plants gen- performance, rather than vegetation shrub species in which there were no erally grew until 1982, then lost height regression between 1982 density and the following biennium and became response. number of plants that disappeared in the shorter than the initial height. Under light and zero grazing, there was a Table 5. Shrub densities in 1982 and 1984, and percent of shrubs with height between 0.5 and 1 m. steady increase in growth of all species in the 2 biennia, except for C. leuco- Botanical name Density Height (0.5Ð1 m) cephala, which is easily broken and 1982 1984 1982 1984 probably reacted to browsing even with ------(plants ha-1 ± SD) ------(%)------the lightest use. C. leucocephala 4,739 ± 2,585 2,231 ± 864** 75.6 52.6 There was no significant influence of C. depauperata 2,909 ± 1,755 2,254 ± 1,415 ns 84.2 68.5 stocking rate on mean canopy cover of C. rhamnifolius 2,675 ± 912 2,417 ± 860 ns 39.2 26.7 the 4 shrubs, even though cover tended B. cheilantha 1,545 ± 1,433 1,241 ± 810* 77.6 56.6 to increase under light to zero stocking, L. microphylla 1,463 ± 1,142 770 ± 469** 47.4 72.3 and decrease under heavier grazing (Fig. L. camara 1,268 ± 1,063 349 ± 262** 89.5 72.3 3). Shrub height was generally measured C. sonderianus 953 ± 1,085 942 ± 1,153 ns 21.9 14.5 on the central twig in the canopy, which C. microphylla1,2 540 ± 520 428 ± 208** 97.8 42.9 1,2 was most likely to be damaged by M. arenosa 520 ± 821 233 ± 325** 98.3 65.1 M. tenuiflora1 258 ± 28 214 ± 91* 87.3 40.2 browsing. Canopy cover was deter- 1 ns mined by taking 2 perpendicular mea- Other tree spp 688 ± 282 516 ± 543 94.8 64.3 Other shrub spp 3,550 ± 3,213 1,633 ± 976* 68.4 49.8 sures of canopy diameters, and there 3 3 Total 21,109 ± 6,825 13,229 ± 3,601** 67.3 47.7 was always the chance that all 4 twigs **,*, and ns—Refers to regression coefficient (r2) between shrub densities in 1982 and difference between shrub densi- were not browsed, mainly when they are ties in 1982 and 1984, in the 7 experimental areas. tangled with other shrubs. This is proba- 1Tree species in shrub state. 2 bly the reason for the lack of statistically Complete botanical names of species appearing for the first time: Caesalpinia microphylla Mart.; Mimosa arenosa (Willd.) Poir. significant effect of stocking rate on 3Weighted mean canopy cover.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 245 -1 ha -1 ha -1 ha

-1 ha -1 ha -1 ha

Fig. 2. Effect of stocking rate on height of C. depauperata, C. leucocephala, B. cheilantha, C. rhamnifolius, and the mean of these 4 shrubs. period 1982Ð84, C. rhamnifolius and C. and this is evidence that some plants folius, because a number of plants of sonderianus, the difference between passed from lower height to height > 0.5 these 2 species passed from seedling 1982 and 1984 densities were very low, m. Under heavy stocking, C. rhamni- stage to the height >0.5 m stage, and resembling the performance of trees. folius increased in height, while C. leu- compensated for the number of plants There were increases in shrub density cocephala decreased. Although C. killed by browsing. In addition, the few from 1982 to 1984 in heavy deferred depauperata and B. cheilantha plants of L. microphylla that escaped and zero grazing, the areas with lowest decreased in mean height, these 2 shrubs mortality increased in height, although densities, and the origin of these addi- had small increases in height in heavy this parameter was not evaluated statisti- tional plants can be questioned. It is sup- deferred from 1982 to 1984. C. depau- cally. posed that competition and browsing perata, in spite of having a small Therefore, the high shrub disappear- determines which plants will die, which increase in height in this paddock, ance in the period 1982Ð84 might be ones will stay below 0.5 m height, and decreased in density, together with C. attributed mainly to the drought, and the which ones will grow. In the zero grazed leucocephala. Most of the increase in amount was determined by the density exclosure, the 4 main shrubs always density in heavy deferred in 1982Ð84 present in 1982, not only by grazing increased in height from 1982 to 1984, came from B. cheilantha and C. rhamni- intensity. In the Sahel, the drought of

246 Journal of Range Management (52)3, May 1999 Table 7. Tree densities in 1982 and 1984.

Tree density Botanical name 1982 1984 - - - - (plants ha-1 ± SD) - - - - M. tenuiflora 174.2 ± 94.90 148.8 ± 89.20** C. microphylla 55.6 ± 41.61 49.1 ± 38.76** T. spongiosa 54.2 ± 26.66 55.9 ± 21.08* M. arenosa 41.7 ± 59.56 34.1 ± 40.84** C. phyllacanthus1 30.5 ± 18.75 22.3 ± 11.65** M. pseudoglaziovii1 24.6 ± 12.35 25.7 ± 18.90* B. leptophloeos1 19.1 ± 11.52 16.8 ± 9.05** C. vitifolius1 14.5 ± 12.53 15.7 ± 10.60** A. piauhiensis1 10.0 ± 13.04 11.7 ± 13.40** P. obliqua1 3.3 ± 2.37 3.4 ± 1.98* S. tuberosa 1.8 ± 1.87 2.8 ± 2.80ns Fig. 3. Effect of stocking rate on mean canopy cover of 4 shrub species cited in Fig. 2. Other spp 17.3 ± 7.23 14.7 ± 5.20ns Total 446.8 ± 94.21 401.0 ± 91.93* **,*, and ns—Refers to regression coefficient (r2) between tree densities in 1982 and 1984, in the 7 experimental areas. 1Complete botanical names of species appearing for the first time: Cnidoscollus phyllacanthus (Pohl) Mull. Arg.; Manihot pseudoglaziovii Pax & K. Hoffm.; Bursera leptophloeos (Mart.) Engl.; Cnidoscollus vitifolius (Mull. Arg.) Pohl; Acacia piauhiensis Benth; Piptadenia obliqua L. (Pers.).

Among the trees present in the study area, neither seedlings (height<0.5 m) nor shrubs of Spondias tuberosa Arr. Cam. were found. The causes are unknown. It is one of the most important Caatinga fruit trees, providing employ- ment for many people during the 2 Fig. 4. Relationship between shrub density in 1982 and difference between shrub density in 1982 and 1984 [H = heavy grazing (1 steer 6.7 ha-1); M = moderate grazing (1 steer 10.0 ha- month fruiting season. 1; L = light grazing (1 steer 13.3 ha-1); Zero = no grazing; C = continuous grazing always; D = rotationally deferred grazing until 1981]. Conclusion 1972 caused death of 50% of shrubs ance, and it is difficult to distinguish (Bille 1978), even though the density between temporary and permanent Neither stocking rate nor the short was much lower in comparison to our changes (Dodd 1994). Data from both period of rotationally deferred grazing study. Extreme weather fluctuations in shrub and herb strata prove that system affected frequency of herbaceous almost all semi-arid environments Caatinga and African ranges similarly species. Most of them tended to be less caused changes in ecosystem appear- react more to climate than to grazing abundant under zero grazing, which pressure (Ellis and Swift 1988). might have been caused by the trend of Table 6. Mortality of all tagged shrubs indepen- For density of trees, the difference woody plants in the Caatinga to become dently of treatments from 1982 to 1984, and from 1982Ð84 was only 10.3% there denser when in rest. There was a difference between 1982 and 1984 densities.. being an association (r2 = 0.71; P<0.05) remarkable increase in the frequency of between both densities (Table 7). The herbaceous species with an increase in Shrub Mortality Difference between rainfall in the last months of the rainy when tagged 1982 and 1984 difference might be attributed to normal tree mortality, being aggravated a little season. Low precipitation in 1982 prob- C. rhamnifolius 3.611 9.642 by drought. Bille (1978) found tree ably caused a high mortality of shrubs. C. depauperata 3.67 22.52 Increasing stocking rate significantly C. sonderianus 7.95 1.15 morality of 20%. This normal rate of tree mortality is illustrated by the fact increased mortality and decreased B. cheilantha 10.20 19.74 growth in height of most shrubs. Cattle L. microphylla 13.77 47.37 that in 1982 there was an association (r2 C. leucocephala 27.36 52.92 can cause degradation in the shrub stra- = 0.67; P<0.05) between density of trees tum, but the 1982 drought caused more L. camara 46.76 72.48 and density of tree species in shrub Mean 16.19 32.33 damage than the heaviest stocking rate stage. In 1984, there was no association -1 **Correlation coefficient (r) between columns = 0.89 of 1 steer 6.7 ha . (P<0.01). (P>0.05) probably due to the high mor- 1Data based on whole number of tagged plants. 2 tality of 30.6% among trees in shrub From Table 5. stage in the period 1982Ð84.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 247 Literature Cited Congress. Cambridge Univ. Press, (eds.), Seasonally dry tropical forests. Cambridge, U.K. Cambridge Univ. Press. Cambridge, U.K. Gammon, D.M. 1978. A review of experi- Sampson, A.W. 1913. Range improvement Albuquerque, S.G. de, and G.R.L. ments comparing systems of grazing man- by deferred and rotation grazing. USDA Bandeira. 1995. Effect of thinning and agement on natural pastures. Proc. Bull. 34. Washington, D.C. slashing on forage phytomass from a Grassld. Soc. So. Africa. 13:75Ð82. Smith, E.L. 1974. O papel do manejo de Caatinga of Petrolina, Pernambuco, Brazil. Hacker, R.B. 1984. Vegetation dynamics in pastagens nativas no Brasil. (In por- Pesquisa Agropecuária Brasileira a grazed mulga shrubland community. II. tuguese). [The role of native pastures man- 30:885Ð891. The ground storey. Australian J. Botany agement in Brazil]. Paper presented at XI Andrade-Lima, D. de. 1981. The Caatingas 2:251Ð261. Annual Meeting of Brazilian Soc. Anim. dominium. Revista Bras. de Botânica Harrington, G.N. and K.C. Hodgkinson. Sci., Fortaleza, Brazil. 4:149Ð163. 1986. Shrub-grass dynamics in Mulga Snedecor, G.W. and W.G. Cochran. 1976. Araújo Filho. J.A. de. 1985. Pastoreio 0 communities of Eastern Australia, p. Statistical methods. 7 ed. The Iowa State múltiplo, p. 209–233. In: Anais do 7 26Ð28. In: P.J. Joss, P.W. Lynch, and O.B. Univ. Press. Ames, Iowa. 593 p. Simpósio sobre manejo de pastagens. (in Williams (eds.), Rangelands: a resource Toutain, B. 1986. Recent vegetation portuguese). [Multiple grazing. In: Proc. nd th under siege, Proc. 2 Internat. Rangeland changes and degradation in some Sahelian 7 Symposium on pasture management]. Congress. Cambridge Univ. Press, pastoral ecosystems of Western Africa, p. FEALQ. Piracicaba, Brazil. Cambridge, U.K. 73Ð74. In P.J. Joss, P.W. Lynch, and O.B. Banco do Nordeste Do Brasil. Herbel, C.H. 1985. Vegetation changes on Williams (eds), Rangelands: a resource Departamento de Estudos Econômicos arid rangelands of the Southwest. under siege, Proc. 2nd Internat. Rangeland do Nordeste. 1971. Perspectivas de desen- Rangelands 7:19Ð21. Congress. Cambridge Univ. Press, volvimento do Nordeste até 1980; perspec- Hickey Jr., W.C. 1977. A discussion of Cambridge, U.K. tivas da agricultura oferta agricola. (In grazing management systems and some Van der Walt, P.T. 1980. A phytosociologi- portuguese). [Perspectives for the develop- pertinent literature (abstracts & excerpts) cal reconnaissance of the Moutain Zebra ment of the Brazilian Northeast until 1980; 1895Ð1966. USDA-Forest Serv. Albu- National Park. Koedoe 23:1Ð32. perspectives for agriculture agricultural querque, N.M. Vasconcelos Sobrinho, J. de. 1949. O supply]. v. 3, T. 2. Fortaleza, Brazil. Kelle, R.D. and B.H. Walker. 1976. The grande deserto central brasileiro, p. Bille, J.C. 1978. Woody Forage Species in effects of different forms of land use on 151Ð160. In: As regiões naturais do the Sahel: Their Biology and Use, p. st the ecology of a semi-arid region in South- Nordeste, o meio e a civilização. (In por- 392Ð395. In: D.N. Hyder (ed.), Proc. 1 eastern Rhodesia. J. of Ecol. 64:553Ð576. tuguese). [The big central brazilian desert, Int. Rangeland Congress. Soc. Range Kirmse, R.D., F.D. Provenza, and J.C. In: The natural regions, the environment Manage. Denver, Colo. Malechek. 1987. Clearcutting Brazilian and the civilization of the Brazilian Burgos, N. and A.C. Cavalcanti. 1991. Semiarid Tropics: Observation on its Northeast]. CONDEPE. Recife, Brazil. Levantamento detalhado dos solos da área effect on small ruminant nutrition during Walker, B.H. 1976. An approach to the de sequeiro do CPATSA, Petrolina-PE. (In the dry season. J. Range Manage. monitoring of changes in the composition portuguese). [Detailed soil survey of 40:428Ð432. and utilization of woodland and savanna CPATSA dry farming area, Petrolina-PE]. Leal, T.M. 1996. Efeito da complementação vegetation. So. African J. Wildl. Research Embrapa-SNLCS, Boletim de Pesquisa 38. alimentar no pós-parto sobre o desempen- 6:1Ð32. Rio de Janeiro, Brazil. ho reprodutivo de cabras sem raça definida Whittaker, R.H. 1975. Communities and Chesterfield, C.J. and R.F. Parsons. 1985. (SRD) e o desenvolvimento das crias, cri- Ecosystems. 2. ed. Macmillan Publ. Co., Regeneration of three species in arid adas na Caatinga não cercada. (In por- New York, N.Y. Southeastern Australia. Australian. J. Bot. tuguese). [Effect of post-partum dietary 33:715Ð732. supplementation on the reproductive per- Cole, M.M. 1960. Cerrado, Caatinga and formance of undefined breed does and kid Pantanal: the distribution and origin of growth, raised in a fencless Caatinga area]. savanna vegetation of Brazil, Geograph. J Master Thesis, Univ. Federal Rural de 136 (part 2):168Ð179. Pernambuco. Recife, Brazil. Cottam, G. and J.T. Curtis. 1956. The use Morello, J. 1954. Ecologia de una planta of distance measures in phytosociological reviviscente de la Caatinga. (In spanish). sampling. Ecol. 37:451Ð460. [Ecology of a reviviscent plant from Dodd, J.L. 1994. Desertification and degra- Caatinga]. Revista Brasileira de Biologia dation of Sub-Saharan Africa—The role of 14:83Ð108. livestock. BioSci. 44:28Ð34. Prado, D.E. 1991. A Critical Evaluation of Dyksterhuis, E.J. 1949. Condition and man- the Floristic Links between Chaco and agement of range land based on quantita- Caatingas Vegetation in South America. tive ecology. J. Range Manage. PhD. Thesis, Univ. of Saint Andrews. 2:104Ð115. Saint Andrews, U.K. Ellis, J.E. and D.M. Swift. 1988. Stability Rodrigues, A. and J.F. Borges. 1979. of African pastoral ecosystems: alternate Pesquisas com pastagem em área seca: paradigms and implications for develop- relatório 1977/78. (In portuguese). ment. J. Range Manage. 41:450Ð459. [Researches on pastures in dry areas: Friedel, M.M. 1986. The interaction with report 1977/78]. Secretaria da Agricultura climate, soil and land use of Central e Abastecimento da Paraíba. João Pessoa, Australian tree and shrub population, pp. Brazil. 45Ð46. In: P.J.Joss, P.W. Lynch, and O.B. Sampaio. E.V.S.B. 1995. Overview of the Williams (eds), Rangelands: a resource nd Brazilian Caatinga, p. 35Ð63. In: S.H. under siege, Proc. 2 Internat. Rangeland Bulloc, H.A. Mooney and E. Medina

248 Journal of Range Management (52)3, May 1999 J. Range Manage. 52:249Ð257 May 1999 Rangeland cover component quantification using broad (TM) and narrow-band (1.4 NM) spectrometry

EDWARD W. BORK, NEIL E. WEST, KEVIN P. PRICE, AND JOHN W. WALKER

Authors are assistant professor, Department of Agricultural, Food, and Nutritional Science, University of Alberta, Edmonton, Alberta, Canada T6G 2P5; professor, Department of Rangeland Resources, Utah State University, Logan, Utah, USA 84322-5230 U.S.A.; associate professor, Kansas Applied Remote Sensing Center, University of Kansas, Lawrence, Kans. USA 66045-2969; and resident director of research, Texas Agricultural Experiment Station, 7887 North Highway 87, San Angelo, Tex., USA 76901. At the time of research, Bork was a graduate research assistant in the Department of Rangeland Resources at Utah State University, and Walker was range scientist, USDA, ARS, U.S. Sheep Experiment Station, Dubois, Ida..

Abstract Resumen

Calibrated predictive relationships obtained from simple Se evaluaron relaciones predictivas calibradas obtenidas a and multiple regression of thematic mapper or broad-band través de regresiones simples y múltiples de datos de mapas (BB) and 1.4 nm interval or narrow-band (NB) spectral data temáticos de banda amplia (BA) y banda estrecha (BE) de 1.4 were evaluated for quantifying 11 rangeland components nm de intervalo. El objetivo fue evaluarlas para cuantificar (including total vegetation, forb, grass, shrub, litter, and bare 11 componentes del pastizal (incluyendo vegetación total, soil) and distinguishing among 6 long-term grazing treat- hierbas, zacates, arbustos, mantillo y suelo desnudo) y distin- ments of sagebrush steppe. In general, all 4 data types pre- guir entre 6 tratamientos de apacentamiento de largo plazo dicted similar values for each rangeland cover component. en una estepa de "sagebrush". En general los 4 tipos de datos Multiple regression models usually had little advantage over predijeron valores similares para cada uno de los compo- simple regression models for predicting cover, particularly nentes de cobertura del pastizal. Los modelos de regresión for abundant cover components, although this trend was múltiple usualmente tuvieron poca ventaja sobre los modelos inconsistent among components. Consequently, simple pre- de regresión simple para predecir cobertura, especialmente dictive models are recommended for quantifying rangeland para los componentes de cobertura abundante, aunque esta indicator components using remotely-sensed data. The use of tendencia fue inconsistente entre componentes. Con- NB spectral data resulted in lower standard errors of predic- secuentemente, para cuantificar los componentes indicadores tion (SEP), although these reductions were inconsistent del pastizal mediante el uso de datos de sensores remotos se among rangeland components. Although both data types dis- recomienda modelos predictivos simples. El uso de datos de tinguished among grazing treatments with major plant com- espectro de BE produjeron errores estandard de prediccion positional differences (P < 0.00) using a multivariate analysis mas bajos (ESP), aunque estas reducciones fueron inconsis- of variance (MANOVA), only the NB data distinguished tentes entre componentes del pastizal. A pesar de que ambos between grazing treatments with minor ecological differences tipos de datos y utilizando el análisis multivariado (MANO- (P < 0.01). These results suggest that in a practical context, VA) distinguieron los tratamientos de apacentamiento con NB data are advantageous for quantifying rangeland cover mayores diferencias de composición de plantas (P<0.001) solo components and distinguishing among grazing treatments los datos de BE distinguieron los tratamientos de apacen- under the condition of our study. tamiento con diferencias ecológicas menores (P<0.01). Estos resultados sugieren que en un contexto practico y bajo las condiciones de nuestro estudio, los datos de BE son venta- Key Words: indicators, long-term grazing, predictability, josos para cuantificar los componentes de cobertura del pas- sagebrush steppe tizal y distinguir entre tratamientos de apacentamiento.

Remote sensing of western U.S. rangelands has increased the speed and efficiency of gathering information on this livestock grazing) on vegetation composition, productivity, extensive resource (Tueller 1989). In addition to collecting and soil degradation (e.g., Bastin et al. 1993, Pickup et al. baseline inventory data, this technology can be used as a mon- 1993). In part, remote sensing is viewed as a favorable alter- itoring tool for evaluating the influence of disturbances (e.g., native to the traditional, ground reconnaisance monitoring methods because the latter are slow, laborious, limited to localized areas, and subject to great variation (West and Smith This work was primarily funded by the Utah Agricultural Experiment Station, 1997). of which this is Journal Paper No. 6042. The authors thank the U.S. Sheep Experiment Station for hosting this research, Previous remote sensing of rangelands has been particularly particularly Scott McKoy for aiding with data collection. successful at monitoring simple biological attributes such as Manuscript accepted 28 Aug. 1998.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 249 total vegetational cover (e.g., Curran Another problem with mixture models is component relative to point-based sam- 1980, Gamon et al. 1993). The defini- that non-linear spectral mixing can pling methods, and (2) determine tion and conceptualization of rangeland occur (Roberts et al. 1993, Borel and whether the NB spectral variables "health" or condition, however, has Gerstl 1994, Ray and Murray 1996), improved differentiation among grazing changed to include the amount of bare particularly in arid environments. treatments compared with BB spectral soil, litter, microphytes (algae, lichen, A possible alternative to the spectral variables. and moss), standing dead vegetation, unmixing approach for quantifying and various living plant growth forms rangeland cover components (e.g., plant (e.g., grasses, forbs, and shrubs) as well growth forms) is alternative spectral Methods (National Research Council, 1994). variables obtained from the increased Soil-based information is necessary to spectral resolution of NB remotely- Study Site protect the long-term sustainability of sensed data. These data are collected This research was conducted over 2 the ecosystem, such as ensuring that the over many more, but narrower, spectral years at the U.S. Sheep Experiment minimum threshold of vegetational bandwidths within each ground resolu- Station, 10 km north of Dubois, Ida. cover on the soil (e.g., Packer 1951) is tion element (or pixel) than BB data. (44¡14'44" N. Latitude, 112¡12'47" W. maintained. Quantitative information on Because the accuracy of quantifying Longitude). This rangeland is situated at the relative abundances of plant growth individual cover components depends about 1,650 m elevation in the northeast- forms are particularly important for directly on the potential for establishing ern portion of the sagebrush steppe managing livestock because they are not relationships between the abundance of ecosystem type (West 1983). The cli- only indicative of previous grazing each component and the spectral infor- mate is semiarid with cold winters and activities (e.g., Mueggler 1950, Laycock mation within individual pixels, an warm summers. Average annual precipa- 1967, Bork et al. 1998a), but also differ increased number of spectral bands may tion was 324.6 mm over the past 64 in palatability and nutritional value to improve the ability to quantify individ- years, including 70 cm of snow, with an livestock. ual surface components. Price et al. average annual temperature of 6.1¡C Rangeland assessment using remote (1992) used this technology to establish (NOAA 1993). About 44% of the annual sensing has typically been done using relationships between NB reflectance precipitation falls from May to August, small spatial-scale (i.e., coarse-grained, and total plant cover and leaf area index with peak precipitation in May and June. large pixel size) data collected from in a tall grass prairie. Testing is needed, In the year of this study, precipitation broad-band (BB) sensors such as however, within more structurally com- was 30 mm below average, with April Landsat TM and MSS. These data have plex vegetation types such as the sage- and May slightly above average and been effective in classifying and quanti- brush steppe, where plant cover is sub- June through August slightly below. fying total vegetational cover and phy- stantially lower. In addition to having Data for this study were collected tomass (e.g., Mouat et al. 1981, Curran less living cover, vegetational features from 6 long-term seasonal grazing treat- and Williamson 1985, Ustin et al. 1986, are discontinuous and mixed with vary- ments, in fenced paddocks varying in Lloyd 1990, Paruelo and Lauenroth ing proportions of litter, rock, and bare size from 4 to 12.5 ha, located near the 1995) and the cover of growth forms ground (Tueller 1987), creating a het- station headquarters. Vegetation in the such as shrub (Boyd 1986) and grass erogeneous environment that compli- area is dominated by three-tip sagebrush (Paruelo and Golluscio 1994). More cates the use of remotely-sensed data. (Artemisia tripartita Rydb.), bluebunch detailed plant community information This study evaluated the accuracy and wheatgrass (Pseudoroegnaria spicata has also been obtained from coarse spa- reliability of broad-band (BB) and nar- [Pursh] A Löve), and arrowleaf balsam- tial resolution, remotely sensed data row-band (NB) spectra for quantifying root (Balsamorhiza saggitata [Pursh] using spectral unmixing models, which indicator cover components, particularly Nutt.) (Laycock 1963; with currently interpolate the amount of contributing those required to evaluate rangeland preferred latin names of Kartesz 1994). surficial sub-components from either "health" (NRC 1994) and biological All of the grazing treatments were estab- BB spectra (Huete 1986, Pech et al. integrity (EPA 1994, West et al. 1994). lished between 1923 and 1950, resulting 1986a, 1986b, Ustin et al. 1986) or nar- A previous investigation indicated that in a minimum of 46 consecutive years row-band (NB) spectra (Gamon et al. calibrated predictive relationships of treatment. Of the 6 treatments cur- 1993, Mustard 1993, Roberts et al. obtained from NB spectral data may rently in place, one has been annually 1993, Wessman et al. 1997). These improve the quantification of detailed fall-grazed and one annually spring- models, however, are not without prob- cover components (i.e., increase R2) rel- grazed since 1923. Two others have lems. For example, they are often creat- ative to BB spectral data (Bork 1997). been fall-grazed and spring-grazed since ed under simplified, contrived condi- The general objective of this study was 1950, respectively, but were previously tions where only one (or a few) primary to examine whether these apparent grazed during the opposite season vegetation component(s) vary for inves- improvements, when applied to a series between 1923 and 1950. These 4 treat- tigational purposes (e.g., Elvidge and of historical grazing treatments, translat- ments will hereafter be referred to as the Chen 1995). Because rangelands are ed into more accurate quantification of old fall, old spring, new fall, and new complex environments with a great rangeland cover components. spring, respectively. The final 2 treat- diversity of cover components (and thus, The specific objectives of this investi- ments were exclosures ungrazed by live- component combinations), the accuracy gation were to, (1) determine whether stock, created in 1940 and 1950, respec- of these models under practical field the leading BB and NB spectral vari- tively. Prior to establishment, the newer conditions is often limited (Price 1994). ables predict similar cover values per exclosure was spring-grazed, while the

250 Journal of Range Management (52)3, May 1999 Table 1. Abundance of cover components within each grazing treatment based on 1995 and 1996 plot at the end of the boom using a cover data (from Bork et al. 1998a). water-level and plumb-bob. This produced a circular instantaneous field-of- Cover (by Grazing Treatment1) view (IFOV) with a diameter of 1.51 m Component O-Fall N-Fall O-Spring N-Spring O-Excl. N-Excl. and ground surface area of approximate- 2 Vegetational: ------(%) ------ly 1.75 m . The observers and frame Live Forb 14.6 11.8 6.4 8.0 13.3 7.1 were consistently positioned on the Live Grass 16.3 17.8 20.0 11.5 13.3 16.0 north side of the plot to avoid shadow- Total Herb 30.9 29.7 26.4 19.5 26.6 23.0 ing and disturbing the vegetation. Live Shrub 17.4 20.5 25.6 28.8 20.5 22.2 Spectal sampling was preceded by cal- Total Veg. 48.3 50.2 52.0 48.3 47.0 45.3 ibration of the spectrometer using a stan- Other: dardized white SPECTRALONTM1 panel Dead Shrub 3.6 6.6 9.2 10.1 7.4 8.5 Lichen 7.1 4.6 3.3 3.2 3.9 3.9 and dark reading adjustment. Integration Litter 16.6 15.0 18.3 18.5 17.9 17.0 time per reading was 175 milliseconds. Moss 1.5 1.1 2.4 1.1 3.3 5.2 Calibration was redone at least every 30 Rock 1.1 1.5 1.3 0.8 0.7 0.8 min. during each sampling session. Each Soil (bare) 21.7 20.9 13.5 17.9 19.6 19.1 spectral file was recorded as the average 1 Treatments are as follows: O-Fall, annually fall-grazed since 1923; N-Fall, annually fall-grazed since 1950 but spring- of 5 readings over a 2 second period to grazed from 1923Ð1950; O-Spring, annually spring-grazed since 1923; N-Spring, annually spring-grazed since 1950 but minimize the impact of plant movement fall-grazed from 1923 to 1950; O-Excl, no sheep grazing since 1940 with fall-grazing from 1923 to 1940; N-Excl., no sheep grazing since 1950 with spring-grazing from 1923Ð1950. by wind or other factors (e.g., insects) on the readings. Spectral readings were recorded as the older, fall-grazed. All livestock grazing alluvium overlying basalt bedrock . All 6 proportion (%) of incident spectral ener- within the treatments since the start of soil types within the study area are gy reflected, from 380 to 1075 nm the long-term study has been by sheep. Mollisols, with soil characteristics het- wavelength in consecutive 1.4 nm wide For a complete history of the timing and erogeneous across the landscape because bandwidths (i.e., 500 bands). The short- stocking rates on each grazing treat- of the variable thickness of unconsolidat- est wavelength measured was 400 nm to ment, see Bork et al. (1998a). ed parent material (Natural Resources eliminate the ultraviolet region. The Based on a previous study, average Conservation Service or NRCS, 1995). upper range was set at 960 nm to cover values within each grazing treat- The mix of soils within pastures, howev- include the minor water absorption band ment are summarized in Table 1 for er, is similar (Bork et al. 1998b). Litter, at this upper limit (Hoffer 1978), but each of the 11 cover components. In lichen, moss, bare soil, and exposed rock eliminate the noisy bands beyond this general, repeated spring-grazing by are relatively common. range. In addition to the narrow-band sheep has removed the perennial forbs (NB) data, the software used with the such as balsamroot, resulting in a heavy Spectral Sampling spectrometer provided simulated broad- cover of sagebrush with abundant annu- Repeated, close-range multispectral band (BB) values (i.e., thematic mapper) al herbs such as cheatgrass. Annual fall- measurements were made on cloud free that were used as a benchmark to which grazing by sheep has maintained the days at 30 randomly-positioned circular the NB measurements could be com- native herbs and removed shrubs, pro- plots in each of the 6 grazing treatments, pared. Simulated thematic mapper data ducing a more balanced mixture of during mid June, July, and August of consisted of the relative percent perennial grasses, forbs, and shrubs 1996. All plots were clearly marked to reflectance over the same waveband (Mueggler 1950, Laycock 1967, Bork et facilitate re-sampling. The sampling intervals as on the Landsat thematic al. 1998a). The exclosures are interme- dates were selected to maximize the mapper sensor, but as measured by the diate in composition between the fall- phenological separation documented PS-II, over the intervals 450Ð520, and spring-grazed areas. Additional among the major vegetational compo- 520Ð600, 630Ð690, and 760Ð900 nm, residual effects are evident within the 3 nents within the study area (Blaisdell respectively (ASD 1991). These data switchover treatments (i.e., the new 1958), and coincided with peak green- were not adjusted for differences in exclosure, fall, and spring paddocks). In ness of all growth forms (early to mid gains, etc., between instruments, nor to particular, the new exclosure and new June), moderate herb senescence (mid top of atmosphere equivalent. fall-grazed paddock remain greater in July), and advanced herb senescence shrub cover than their older counterparts (late August), respectively. A Personal Cover Component Sampling (Laycock 1967, Bork et al. 1998a). The Spectrometer II (Analytical Spectral Cover data were collected from each significant historical precedent and Devices [ASD] Inc., Boulder, Colorado, of the 180 plots within 48 hours follow- information available from this grazing 1991), mounted on a light-weight, ing June spectral sampling. Cover data trial make it an ideal location to test the portable aluminum boom, was used to were measured using a variation of the effectiveness of BB and NB spectral measure each plot between the peak point sampling method (Floyd and data for distinguishing among range- sunshine hours of 1100 hours and 1500 Anderson 1982), with the minimum lands (i.e., treatments) with distinctly hours MDT (i.e., within 2 hours of solar different vegetational compositions. noon). The spectral receptor had a 25¡ Soils in the area have been derived field of view. It was oriented vertically 1Registered trademark of Labsphere Inc., North from wind-blown loess, residuum, or 3.4 m directly over the center of each Sutton, N.H. 03260-0070.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 251 number of points determined from a well as composite variables involving original fall, spring, and exclosed areas pilot study (Bork 1997). Each plot was more than one reflectance (i.e., complex were compared to the new fall, new sampled with a point-frame containing variables). Examples of the latter includ- spring, and new exclosed treatments, 164 equally-spaced points, 10.5 cm ed ratio indices such as the Difference respectively. Note that our objective was apart. Two layers of cross wires 20 cm Vegetation Index (near infrared [NIR]- not to document the comprehensive dif- apart were used to facilitate vertical red), Normalized Difference Vegetation ferences in cover components among cross-sighting and reduce parallax. Index ([NIR-red]/[NIR+red]) (Tucker the treatments as that has been done The surface feature immediately 1979), and the Soil Adjusted Vegetation elsewhere (Bork et al. 1998a), but rather below each point was recorded, with the Index ([NIR-red]*[1+L]/[NIR+red+L]; to evaluate how the BB and NB spectral data compiled into average cover values where L is an adjustment for bare soil data may differentiate among treatments per component per plot. All 11 surface effects) (Huete 1988), as well as deriva- relative to the 2 primary management cover components (Table 1, left column) tive (i.e., slope-based) indices for the NB questions. examined were rangeland soil and vege- data. Multi-date measurements were To test the regressions for distinguish- tation indicators that had either been used to determine the change in each ing among treatments and contrast the previously linked directly to rangeland spectral variable between June and July, calibrations from each data type, spec- condition within the study area July and August, and June and August. tral data from the plots not used in the (Mueggler 1950, Laycock 1967, Bork et From these analyses, 4 regressions were initial calibration procedure (N = 119; al. 1998a), or were important on the established for each cover component, 2 20 in each treatment except the new basis of current rangeland condition the- based on the leading simple regression spring paddock, where the marker for 1 ory (NRC 1994). using single spectral variables (one for plot was lost due to wind) were used to each data type) and 2 multiple regression predict the cover of each component Data Analysis models (one for each data type). These using all 4 regressions (i.e., of the vari- Previous work (Bork 1997) had ana- regressions are provided in Table 2. ables in Table 2). This was done by lyzed one-third of the study plots, 10 In the present study, the unique histo- inserting the required spectral from each grazing treatment (N = 60), ry of each grazing treatment allowed variable(s) into each regression and cal- with simple and stepwise multiple evaluation of how well the calibrated culating the predicted cover in each plot. regression, to determine, within each of spectral data could address the 2 prima- Predicted cover from each regression the 11 cover components, the potential ry management questions that had been (BB simple and multiple, and NB sim- improvement in predictability from identified within the study area (Bork et ple and multiple) within each plot was using narrow-band (NB) as compared al. 1998a). The first of these questions subsequently compared to measured with broad-band (BB) spectral data. examined the impact of seasonal sheep cover to evaluate the accuracy of the Data analysis evaluated a wide range of grazing since 1950 (i.e., the general dif- calibrations within each cover compo- spectral variables from each sampling ferences among the fall, spring, and nent. Overall accuracy was determined date in each type of spectral data (BB exclosed areas). The second question for each component by computing the and NB), including isolated spectral examined the residual impact of the sea- adjusted standard error of predicted reflectances (i.e., simple variables), as sonality of sheep grazing prior to 1950 cover across all 119 validation plots (i.e., rangeland resilience), wherein the using the formula:

Table 2. Summary of the leading spectral variables used to predict each cover component with broad-band (BB) and narrow-band (NB) data, using simple and multiple regression.

Component: Broad-Band1 Narrow-Band1 Simple Multiple Simple Multiple Vegetational: Forb Jn-Aug DVI2 Jn Green; Jn NIR Jn-Aug SL32 Jn B769; Jn B784; Aug B755; Aug B897 Herb Jn-Jy SAVI Aug Blue; Aug Red; Jn Blue; Jn NIR Jn-Aug SL16 Jn B769; Aug B599; AugB684 Grass Aug NIR Aug NIR Jn-Jy SL8 Aug B954 Shrub Aug NDVI Aug Red; Aug NIR Aug SAVI 698/670 Aug B670; Aug B698; Aug B883 Total Veg. Jn NDVI* Jn Blue; Jn NIR Jn SAVI 698/670* Jn B542; Jn B613; Jn B698 Other: Dead Shrub Aug Red* Aug Blue; Aug Red; Jn Green Aug SLPA* Jn B585; Aug B514; Aug B670 Lichen Jn Blue Jn Blue Jn B499 Jn B443; Jn B457; Jn 528; Jn B542; Jn B883 Litter Jn Red Jn Red Jn ARred Jn B641 Moss Jy G/B NDVI Jy G/B NDVI Jy SAVI 570/400 Jy B755 Rock Jy Blue Jy Blue Jy SAVI 698/670* Jy B400 Soil (bare) AugNDVI* Aug Blue Aug NDVI Aug B514; Aug B570; Aug B954 Max NIR/MinRed* 1Simple regressions marked with a '*' are 2nd-order, curvilinear functions. 2Spectral variables derived from multi-temporal sampling dates are represented by differences between data on those dates.

252 Journal of Range Management (52)3, May 1999 2 2 {[ Sumi=1..N(PCi - MCi) ] - [ (Sumi=1..N(PCi - MCi)) /N ] } tions) of interest. Roy's greatest root, a St. Error Prediction Sqrt = (1) test statistic derived from comparing the Sqrt [ N - 1 ] among-grazing treatment variation (of where: dom contrasts were used within the spectral variables) to within-treatment PC = predicted cover, model to address the management-based variation (of spectral variables), was MC = measured cover, questions. Three contrasts examined the used to evaluate results of the MANO- N = total number of plots, and data for main grazing treatment differ- VA (Scheiner 1993). i = the ith plot. ences (i.e., fall vs spring, fall vs To directly assess whether any of the To determine whether the spectral exclosed, and spring vs exclosed) and 3 4 regressions of spectral data for each data could distinguish among the 6 graz- assessed the residual effects of manage- cover component had a practical impact ing treatments, a Multivariate Analysis ment prior to 1950 (i.e., the original fall, on distinguishing among grazing treat- of Variance (MANOVA) using SAS spring, and exclosure vs newer fall, ments, the same 6 contrasts were done proc GLM (SAS Institute Inc. 1988) spring, and exclosure, respectively). As (P<0.10) within each cover component was performed on the set of leading BB a result, the MANOVA tested the ability on the BB and NB predicted values spectral variables isolated from the sim- of each spectral data type (BB and NB) from both the simple and multiple ple regressions (Bork 1997). A similar to distinguish significant differences regressions. Contrasts among treatments procedure was done on the NB data among treatments, both overall and of the measured cover values (P<0.10) (Bork 1997). Six single degree-of-free- within the specific contrasts (i.e., ques- provided the benchmark against which

Fig. 1. Mean narrow-band spectral response curves for June (solid line), July (dashed line), and August (dotted line) 1996, for each grazing treatment (n = 30).

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 253 Table 3. Mean measured and predicted cover (standard error of prediction [SEP] in parentheses was under-estimated using the leading for predicted cover) using the simple and multiple regression calibrations of the broad-band (BB) single NB spectral variable while herb and narrow-band (NB) data within each of the 11 cover components examined. cover was under-estimated by the leading single broad-band (BB) spectral Cover Component Actual Cover Predicted Cover (N = 119) Simple Regression Multiple Regression variable. In contrast, lichen was over- BB Data NB Data BB Data NB Data estimated using multiple regression of ------(%) ------the NB data. In general, the standard error of pre- Vegetational: diction calculated for each of the 4 Forb 8.8 8 4.7 8 8.3 regression types did not appear to differ (5.62) (5.80) (4.99) (5.84) within individual cover components Grass 15.6 15.5 15.3 15.6 15.4 (7.59) (7.17) (7.59) (7.75) (Table 3). In only one case did multiple Herb 24.2 17.9 23.4 23.6 23.1 regression reduce the standard error of (7.62) (7.64) (7.16) (7.14) prediction by at least 10% relative to Live Shrub 21.7 22.7 21.3 22.8 23.1 simple regression (BB forb, from 5.62 to (7.87) (7.33) (7.86) (7.14) 4.99). Other decreases in the standard Total Vegetation 45.9 46.2 47.2 47.4 47.2 error of prediction from using multiple (6.52) (6.27) (7.14) (6.78) regression occurred for bare soil (NB), Other / Soil-Based: Dead Shrub 7.2 7.4 7.4 7.6 7.6 live shrub (NB), and herb (BB and NB) (4.88) (4.81) (4.92) (4.92) (Table 3). In several situations, the stan- Lichen 4.7 4.1 4.1 4.1 6.4 dard error of prediction was greater as a (9.02) (9.04) (9.02) (9.03) result of using multiple regression. For Litter 19.2 19.6 19.6 19.5 19.6 example, the error for grass using NB (5.91) (5.93) (5.92) (5.89) data went from 7.17 to 7.75. Other Moss 2.7 3.1 2.9 n/a1 2.9 increases in the standard error were evi- (2.91) (2.93) (2.92) dent for bare soil (BB) and total live Rock 1.4 0.8 0.8 0.7 0.7 vegetation (both BB and NB) (Table 3). (2.73) (2.64) (2.74) (2.72) Little difference was observed in the Soil (bare) 18.9 18.8 19.1 18.5 19.2 (7.09) (7.47) (7.54) (7.20) standard error of prediction from using 1 NB data (Table 3). Of the differences No BB spectral variables met the minimum significance level for entry using multiple regression. that did occur, results were mixed regarding the type of spectral data. For to check for any practical differences ferences are apparent, both between example, when NB data were used, the within the 4 calibrated regressions. This main treatment types and within treat- standard error for forb was greater using procedure evaluated the calibrated ment replicates (i.e., old vs new). either simple or multiple regression, but regressions at a practical level, and Among all 6 treatments, the 2 fall- lower for live shrub and total vegetation facilitated testing of the NB data for grazed areas had sine wave-like spectral using each regression strategy (Table 3). improved predictability of cover compo- response curves in June, which flattened While the herb component showed vir- nents among treatments. For each cover by July (mid-summer). While the tually no difference in standard error, component, we determined, (1) the response curve of the old fall treatment grass had a mixed response with only number of contrasts per calibrated flattened out from a drop in the near simple regression of NB data resulting regression that correctly represented infrared and increase in red reflectance, in a lower SEP. Among the non-living measured cover differences, (2) the the new fall curve changed almost and soil-based cover components, dif- number of contrasts significant in the exclusively in visible reflectance (i.e., ferences in the SEP from using NB data measured data that were missed (i.e., near infrared was stable). This latter pat- were minimal, with the only exception found not significant) within the con- tern of temporal change was also appar- being bare soil (Table 3). trasts of predicted data, and (3) the num- ent in the spectral response curves of Results of the MANOVA showed sig- ber of non-significant contrasts within both spring-grazed treatments. nificant differences between the BB and the measured data that were found sig- Interestingly, both spring treatments had NB subsets of simple spectral variables nificant in the predicted data. These val- increased in reflectance in July and among the 6 grazing treatments (Table ues represented correct classifications, declined in August, with the old spring 4). Furthermore, these significant differ- omission errors, and commission errors, showing greater changes over time. ences (P<0.001) were apparent in all the respectively. Both exclosures had spectral response main effect contrasts using both data curves similar to the old fall treatment, types (Table 4). This trend, however, although the magnitude of change was did not continue into the 3 contrasts Results less than that found in the latter. addressing the differences among graz- Average predicted cover values within ing treatments grazed similarly since The mean narrow-band (NB) spectral each cover component were similar to 1950, but differently before then (i.e., response curves from June, July, and measured cover regardless of regression examining residual effects or rangeland August for each grazing treatment are type (Table 3). Only 3 noticeable excep- resilience). Although all 3 of these con- depicted in Fig. 1. Several notable dif- tions were found. Predicted forb cover trasts were significant using the NB

254 Journal of Range Management (52)3, May 1999 Table 4. Results of the overall and contrast-based MANOVA for the broad-band (BB) and narrow- spring-, new spring-, and new fall- band (NB) data. grazed treatments was somewhat sur- prising. There are, however, several pos- Test: BB Spectra (N = 10) NB Spectra (N = 11) 1 1 sible explanations for this observation. F - Value P F - Value P 2 These paddocks had the greatest annual Overall Model 10.98 p < 0.001 13.72 p < 0.001 herb cover (both forb and grass) (Bork Main Effect Contrasts: et al. 1998a). Rapid senescence of annu- Fall vs Spring 5.83 p < 0.001 9.11 p < 0.001 als may have increased the relative Fall vs Excl. 5.44 p < 0.001 6.03 p < 0.001 reflectance from soil-based components Spring vs Excl. 4.68 p < 0.001 4.69 p < 0.001 over time, thereby increasing near Residual Effect Contrasts: infrared reflectance. In contrast, the Old Fall vs New Fall 6.88 p < 0.001 7.3 p < 0.001 long-term fall-grazed treatment and 2 Old Spring vs New Spring 1.44 p = 0.18 2.45 p < 0.01 Old Excl. vs New Excl. 1.27 p = 0.26 5.38 p < 0.001 exclosures, which were dominated by deep-rooted perennial herbs, may have 1F - values are based on Roy's greatest root. 2 The overall model tests collective spectral differences among all 6 grazing treatments. remained green longer into the summer and produced abundant litter in the shrub interspaces that persisted (Comanor and data, when the BB spectral data were Discussion Staffeldt 1979), thus concealing more of used, only the comparison between the the soil surface. Litter from perennial old and new fall-grazed treatment was This study is unique in that it simulta- species may also have contributed to a significant (Table 4). greater organic matter content at the soil Statistical contrasts among the grazing neously evaluated both narrow-band surface although this notion has not treatments based on the measured and (NB) and broad-band (BB) spectral data predicted cover from the 4 regressions for quantifying rangeland cover compo- been independently verified. Finally, were also compared to determine the nents (e.g., growth forms) through the surface soil moisture may have played a number of significant contrasts within use of localized plots. The results indi- role, particularly if spectral observations each type of predicted cover data that cated not only how well these data pre- were obtained before the soil surface were correct, commission errors, and dict various cover components, but fully dried following precipitation omission errors, relative to the contrasts facilitated the subsequent separation of events during the summer. of measured data (data not shown). In grazing treatments. Although previous Within individual cover components, general, the number of correctly predict- remote sensing work has addressed the mean predicted cover was similar to ed significant contrasts were highly task of quantifying plant growth forms measured cover, regardless of the variable among the cover components, such as shrubs (e.g., Boyd 1986) and regression type used. Inspection of the regardless of data type and regression grasses (e.g., Paruelo and Golluscio residual differences between predicted strategy. Living cover components were 1995) on rangelands, these studies have and measured cover data at the plot particularly poor, with the number of used coarse resolution satellite data and level, however, showed that deviations contrasts correctly predicted ranging ocular estimates of component cover, (i.e., predicted-measured) were highly from 25% (total vegetation) to 75% making them more difficult to interpret. variable among sampling plots. These (forb and shrub). Non-living and soil- The prominent changes in spectral differences continue to reflect the inher- based components were not much better, response curves within each of the graz- ent "white-noise" of the data within the ranging from 50% (litter) to 83% (dead ing treatments early in the summer (from study area. These results indicate that shrub). As might be expected given the June to July) are consistent with the rela- while the average predictability of com- low proportion of correct predictions, tively rapid progression of phenologies ponents among many plots may be commission and omission errors were among most range plants common in the maintained under various regression also common among nearly all predicted area (Blaisdell 1958). The fall-grazed strategies, caution should be used when contrasts for all cover components, with treatments, particularly the old fall, had trying to extrapolate to the individual relatively greater errors associated with the greatest herb cover (Bork et al. plot level. For those cover components the vegetational components. 1998a), resulting in these treatments where predicted cover and measured Comparison of the detailed contrasts exhibiting the largest degree of temporal cover showed greater discrepancies among grazing treatments using predict- change in reflectance within the spectral (e.g., forb, herb, and lichen), the indeed and measured cover values showed response curve. Although the exclosures that the use of NB data instead of BB were also high in herb cover, these areas pendent spectral variables used to pre- data, resulted in an increase in the num- had abundant shrub cover as well. The dict cover may be less reliable. In the ber of correct contrasts (from 63 to 79% higher shrub:herb ratio likely produced case of lichen, 5 NBs were selected into and from 52 to 63% for simple and mul- more shadow and overtopping of the the multiple regression model indicating tiple regression, respectively). A corre- herbaceous understory, reducing green the poor predictability may be due to sponding reduction was evident in the vegetation reflectance (Wilson and regression overfitting. number of commission, and in particu- Tueller 1987). As a result, the spectral Overall, the standard error of predic- lar, omission errors, from using NB data response curves of these treatments var- tions within cover components varied (ommission errors changed from 38 to ied less throughout the growing season. little among the 4 regression types eval- 21%, and 42 to 38%, for simple and The lack of temporal change in the uated. The smallest differences occurred multiple regression, respectively). near infrared spectral region of the old within the less abundant cover compo-

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 255 nents (i.e., rock, moss, lichen, and dead Interestingly, of the 3 contrasts evaluat- Boyd 1986, Paruelo and Golluscio shrub) in Table 3. Where differences did ing residual effects, the old and new 1995). In contrast, collecting localized occur, the use of multiple regression and fall-grazed treatments had the most dis- data in well-defined plots and correlat- NB data did not consistently reduce the similar cover component compositions ing it directly with broad-band (BB) and standard error, indicating that the ideal (Bork et al. 1998a) due to the slow narrow-band (NB) spectral data circum- type of regression strategy for predicting recovery following spring-grazing prior vents the logistical problems associated cover varies among components. to 1950 in the latter (Laycock 1967). with coarse-resolution data, such as geo- Furthermore, changes in standard error Unlike the BB data, the MANOVA with rectification and ground-truthing. As a appeared to depend on (i.e., interact NB data showed all 3 contrasts for result, this study was able to evaluate with) both the mode of regression (sim- residual effects were significant. This is the potential of NB remotely sensed data ple and multiple) and data type (BB and despite the 2 spring-grazed treatments for assessing rangeland condition. NB). Multiple regression frequently pro- both being near-monocultures of shrub This study indicates that NB spectral duced lower standard errors for the more and the 2 exclosures being similar data offer an advantage over BB data for detailed components, particularly plant stands of spatially mixed shrub and qualitatively distinguishing among sage- growth forms (i.e., forb, herb, and perennial herb. Thus, it appears that brush steppe rangelands in different shrub). Interestingly, multiple regression while BB and NB spectral data are both vegetational states, particularly when within the more general cover compo- useful for qualitatively distinguishing ecologically-induced differences from nents [e.g., total vegetation and bare soil among grazing treatments with promi- grazing are subtle. The advantage of NB (BB only)] often failed to reduce the nent compositional differences, the data for quantitatively predicting range- standard error of prediction. Hence, it increased spectral resolution associated land cover components and distinguish- appears that complex predictive models with NB data may improve the qualita- ing among unique grazing treatments, containing multiple spectral variables tive separability of grazing treatments however, remains questionable due to may only be beneficial for rangeland that are dissimilar with respect to more highly inconsistent results among data monitoring when more detailed cover subtle, but managerially important, eco- types within cover components. components are of biological interest. logical characteristics. Consequently, at present, it is doubtful The use of NB data relative to BB The increased number of correct con- that NB data can fill the needs of range- data more often than not reduced the trasts and decreased number of omission land managers to accurately quantify standard error of prediction. Unlike the errors as a result of using NB data sug- specific cover components (i.e., growth trend for simple and multiple regression, gests that despite the lack of differences forms that serve as indicators) (NRC however, NB data were more advanta- between BB and NB standard errors of 1994) any better than BB spectral data. geous for quantifying total living vege- prediction, the NB data may still confer Despite this conclusion, further research tation and other relatively abundant liv- an advantage within practical remote and testing is clearly needed using both ing growth forms. Components that did sensing applications designed to quantita- BB and NB data at various spatial scales not have a lower standard error using tively distinguish among grazing treat- in order to identify strategies through NB data included those with relatively ments. The extent to which NB data may which remote sensing technology may little cover, such as forb. be beneficial, however, will depend on offer a practical advantage for assessing Examination of the spectral data using both the a priori degree of inherent differ- rangeland condition. a MANOVA provided a qualitative ences among treatments and the required method to directly examine the grazing extent of separability during application. treatments for the ecological differences The lack of consistent improvements also Literature Cited of concern to rangeland managers. suggests that to determine whether NB Significant differences showed that data will be advantageous, a direct com- Analytical Spectral Devices Inc. 1991. when either the BB or NB spectral vari- parison between BB and NB data rele- Personal spectrometer II user’s reference. ables from the simple regressions were vant to the problem will be necessary Boulder, Colo. collectively examined, all of the main within the area of interest (i.e., on a case Bastin, G.N., A.D. Sparrow, and G. effect contrasts among grazing treat- study basis). Pearce. 1993. Grazing gradients in central ments were significant. These results are Australian rangelands: ground verification of remote sensing-based approaches. supportive of the fact that the fall, Conclusion Austr. Rangel. J. 15:217Ð233. spring, and exclosed areas have the Blaisdell, J.P. 1958. Seasonal development and greatest vegetation compositional differ- yield of native plants on the Upper Snake ences (Bork et al. 1998a) and hence, are This study utilized a "bottom-up" River Plains and their relation to certain cli- easily separable. However, significant framework as an alternative to the tradi- matic factors. USSA Tech. Bull. 1190. differences among the old and new pad- tional "top-down" scientific approach Borel, C.C. and S.A.W. Gerstl. 1994. Non- docks within each main treatment were for problem solving in natural resource linear spectra mixing models for vegeta- not consistent among the BB and NB management (Shrader-Frechette and tive and soil surfaces. Remote Sens. data. Using BB data, only the 2 fall- McCoy 1993). Top-down monitoring of Environ. 47:403-416. grazed treatments differed. The lack of Bork, E.W. 1997. Rangeland condition rangelands using remote sensing has assessment using high spatial and spectral differences within the 2 exclosed and 2 typically used coarse resolution spectral resolution remote sensing. Ph.D. Diss, spring-grazed areas indicated that the data and poorly ground-truthed informa- Dept. of Rangeland Res, Utah State BB data were unable to detect the resid- tion on ecosystem characteristics (e.g., University, Logan, Utah. 299 pp. ual effects of management prior to 1950.

256 Journal of Range Management (52)3, May 1999 Bork, E.W., N.E. West, and J.W. Walker. shortwave vegetation index imagery. Int. Ray, T.W. and B.C. Murray. 1996. Non- (1998a). Three-tip sagebrush steppe J. Remote Sens. 11:2269Ð2279. linear mixing in desert vegetation. Remote response to long-term seasonal sheep graz- Mouat, D.A., J.B. Bale, K.E. Foster, and Sens. Environ. 55:59Ð64. ing. J. Range Manage. 51(3): 293Ð300. B.D. Treadwell. 1981. The use of remote Roberts, D.A., Smith, M.O., and Adams, Bork, E.W., N.E. West, J.A. Doolittle, and sensing for an integrated inventory of a J.B. 1993. Green vegetation, nonphoto- J.L. Boettinger. (1998b). Soil depth semi-arid area. J. Arid Environ. synthetic vegetation, and soils in AVIRIS assessment of sagebrush grazing treat- 4:169Ð179. data. Remote Sens. Environ. 44:255Ð269. ments using electromagnetic induction. J. Mueggler, W.F. 1950. Effects of spring and SAS Institute Inc. 1988. SAS/STATTM Range Manage. 51(4):469Ð474. fall grazing by sheep on vegetation of the user's guide, Release 6.03 Edition, Cary, Boyd, W.E. 1986. Correlation of rangeland Upper Snake River Plains. J. Range N.C. brush canopy cover with Landsat MSS Manage. 3:308Ð315. Scheiner, S.M. 1993. MANOVA: multiple data. J. Range Manage. 39:268Ð271. Mustard, J.F. 1993. Relationships of soil, response variables and multispecies inter- Comanor, P.L. and E.E. Staffeldt. 1979. grass, and bedrock over the Kameah actions. p. 94-112, In: Scheiner, S.M. and Decomposition of plant litter in two west- Serpentinite Melange through spectral J. Gurevitch (eds.). Design and analysis of ern North American deserts. p. 31-49, In: mixture analysis of AVIRIS data. Remote ecological experiments, Chapell & Hall West, N.E. and J.J. Skujins (eds.). Sens. Environ. 44:293Ð308. Inc., New York. Nitrogen in desert ecoystems. Dowden, National Research Council. 1994. Shrader-Frechette, K.S. and E.D. McCoy. Hutchinson and Ross, Stroudsberg, Penn. Rangeland health: new methods to classify, 1993. Method in ecology: strategies for Curran, P. 1980. Multispectral remote sens- inventory, and monitor rangelands. conservation. Cambridge University Press, ing of vegetation amount. Prog. Phys. National Academy Press, Washington, D.C. New York. Geogr. 4:315Ð340. Natural Resources Conservation Service. Tucker, C.J. 1979. Red and photographic Curran, P.J. and H.D. Williamson. 1985. 1995. Soil investigation of ARS United infrared linear combinations for monitor- The accuracy of ground data used in States Sheep Experimental Station heading vegetation. Remote Sens. Environ. remote sensing investigations. Int. J. quarters range. USDA Nat. Res. Cons. 8:127Ð150. Remote Sens. 6:1637Ð1651. Service, Rexburg, Ida. Tueller, P.T. 1987. Remote sensing science Elvidge, C.D. and Z. Chen. 1995. NOAA. 1993. Climatological Data Annual applications in arid environments. Remote Comparison of broad-band and narrow- Summary—Idaho. NOAA Report, Vol. Sens. Environ. 23:14Ð154. band red and near-infrared vegetation 96, #13. Tueller, P.T. 1989. Remote sensing technol- indices. Remote Sens. Environ. 54:38Ð48. Packer, P.E. 1951. An approach to water- ogy for rangeland management applica- Environmental Protection Agency. 1994. shed protection criteria. J. Forestry, tions. J. Range Manage. 42:442-453. Landscape monitoring and assessment 49:639Ð644. Ustin, S.L., J.B. Adams, C.D. Elvidge, M. research plan. US EPA 620/R-94/009. Paruelo, J.M. and R.A. Golluscio. 1995. Rejmanek, B.N. Rock, M.O. Smith, Floyd, D.A. and J.E. Anderson. 1982. A Range assessment using remote sensing in R.W. Thomas, and R.A. Woodward. new point interception frame for estimat- northwest Patagonia (Argentina). J. Range 1986. Thematic mapper studies of semi- ing cover of vegetation. Vegetatio Manage. 47:498-502. arid shrub communities. BioSci. 50:185Ð186. Paruelo, J.M. and W.K. Lauenroth. 1995. 36:446Ð452. Gamon, J.A., C.B. Field, D.A. Roberts, Regional patterns of normalized difference Wessman, C.A., C.A. Bateson, and T.L. S.L. Ustin, and R. Valentini. 1993. vegetation index in North American shrub- Benning. 1997. Detecting fire and grazing Functional patterns in an annual grassland lands and grasslands. Ecol. 76:1888Ð1898. patterns in tallgrass prairie using spectral during an AVIRIS overflight. Remote Pech, R.P, R.D. Graetz, and A.W. Davis. mixture analysis. Ecol. Appl.7:493Ð511. Sens. Environ. 44:239Ð253. 1986b. Reflectance modelling and the West, N.E. 1983. Western intermountain Hoffer, R.M. 1978. Biological and physical derivation of vegetation indices for an sagebrush steppe. p. 351Ð397, In: N.E. considerations in applying computer-aided Australian semi-arid shrubland. Int. J. West (ed.). Ecosystems of the World 5: analysis techniques to remote sensor data. Remote Sens. 7:389Ð403. temperate deserts and semi-deserts. p. 227Ð289, In: Swain, P.H, and S.M. Davis Pech, R.P. A.W. Davis, R.R. Lamacraft, Elsevier Scientific Publ. Co., New York. (eds.). Remote sensing: the quantitative and R.D. Graetz. 1986a. Calibration of West, N.E. and E.L. Smith. 1997. approach. McGraw-Hill Inc., New York. Landsat data for sparsely vegetated semi- Improving the monitoring of rangelands. Huete, A.R. 1986. Separation of soil-plant arid rangelands. Int. J. Remote Sens. Rangelands 19(6):9Ð14. spectral mixtures by factor analysis. 7:1729Ð1750. West, N.E, K. McDaniel, E.L. Smith, P.T. Remote Sens. Environ. 19:237Ð251. Pickup, G., V.H. Chewings, and D.J. Tueller, and S. Leonard. 1994. Huete, A.R. 1988. A soil adjusted vegeta- Nelson. 1993. Estimating changes in vege- Monitoring and interpreting ecological tion index (SAVI). Remote Sens. Environ. tation cover over time in arid rangelands integrity on arid and semi-arid lands of the 25:295Ð309. using Landsat MSS data. Remote Sens. western United States. Report from the Kartesz, J.T. 1994. A synonymized check- Environ. 43:243Ð263. Western Regional Research Coordinating list of vascular flora of the United States, Price, J.C. 1994. How unique are spectral Committtee-40 on Rangeland Research, Canada, and Greenland: Vol. 1 - checklist. signatures? Remote Sens. Environ. New Mexico Range Improvement Task 2nd Ed. Timber Press, Portland, Ore. 49:181Ð186. Force, Rep. No. 37. Laycock, W.A. 1963. An annotated bibliog- Price, K.P., V.C. Varner, E.A. Martinko, Wilson, R.O. and P.T. Tueller. 1987. Aerial raphy of range management publications and D.C. Rundquist. 1992. Analysis of and ground spectral characteristics of from the U.S. Sheep Experimental Station. multispectral narrow-band spectrora- rangeland plant communities in Nevada. Wyoming Range Management, Issue No. diometer measurements from six prairie Remote Sens. Environ. 23:177Ð191. treatments in Kansas. unpubl. paper pre- 175. sented at the ACSM/ASPRS '92 Annual Laycock, W.A. 1967. How heavy grazing Convention and Exposition, Feb. 29— and protection affect sagebrush-grass March 5, Albuquerque, N. M. ranges. J. Range Manage. 20:206Ð213. Lloyd, D. 1990. A phenological classification of terrestrial vegetation cover using

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 257 J. Range Manage. 52:258Ð262 May 1999 Defoliation time and intensity of wall barley in the Mediterranean rangeland

MOH’D KHAIR J. EL-SHATNAWI, HANI Z. GHOSHEH, HAIL K. SHANNAG, AND KHALIL I. EREIFEJ

Authors are assistant professor and associate professor, Faculty of Agriculture, Jordan University of Science and Technology, P.O. Box 3030, Irbid, Jordan.

Abstract Resumen "Wall barley" (Hordeum murinum L.) es la especie domi- Wall barley (Hordeum murinum L.) is the dominant species nante de los pastizales del noreste de Jordania y la cual in northeastern rangeland of Jordan that decreases under decrece bajo el apacentamiento. En un pastizal semiárido de grazing. We investigated the responses of wall barley to clip- Jordania investigamos durante 2 estaciones de crecimiento, ping time and height during 2 growing seasons in the semi- 1994/1995 y 1995/ 1996, la respuesta del "wall barley" a la arid rangeland of Jordan. A natural stand was utilized to con- época e intensidad de defoliación. Se utilizó una población duct the experiments that were arranged in a randomized natural para conducir los experimentos, los cuales se complete block design during 1994/95 and 1995/1996 growing establecieron bajo un diseño de bloques completos al azar. seasons. Treatments were combinations of clipping heights (5 Los tratamientos fueron combinaciones de alturas de corte (5 or 10 cm above soil surface) and plant growth stages (tiller- o 10 cm arriba de la superficie del suelo) y etapas de crec- ing, jointing, or booting), in addition to unclipped check. imiento de la planta (ahijamiento, encañe o embuche), y se Results showed that clipping to 5 and 10 cm stubble height at tuvo un testigo si cortar. Los resultados mostraron que el tillering produced 1,167 and 1,349 kg ha-1 dry matter, respec- -1 corte a los 5 o 10 cm efectuados en el ahijamiento produjeron tively, compared to 1,122 kg ha for unclipped check. 1,167 y 1,349 kg ha-1 de materia seca respectivamente, com- Clipping to 5 and 10 cm stubble height reduced shoot weight parado con 1122 kg ha-1 producidos por el testigo sin defo- by 28 and 21% at jointing stage and 52 and 38% at booting liar. Cortar a 5 o 10 cm de altura del rastrojo remanente stage. Defoliation during tillering stage did not impact plant redujo el peso de los tallos en 28 y 21 % cuando los cortes se height of regrowth nor seed yield. Weed biomass were higher realizaron en la etapa de encañe y en 52 y 38% cuando se when plant defoliation was delayed to the jointing and boot- efectuaron en la etapa de embuche. La defoliación durante la ing stages. Therefore, it is recommended to defoliate wall bar- etapa de ahijamiento no impactó la altura del rebrote ni la ley early at tillering stage but before plants reach jointing or producción de semilla. La biomasa de la maleza fue mayor reproductive stages. cuando la defoliación fue retrasada hasta los estados de encañe o embuche. Por lo tanto, es recomendable defoliar a "wall barley" en la etapa de ahijamiento, pero antes de que Key Words: Jordan, Hordeum murinum L. clipping, plant las plantas alcancen el estado de encañe o la etapa reproduc- height, shoot, seed, annual grass, weeds tiva.

Jordanian grassland is located within the 200 to 350 mm rainfall zone of the eastern Mediterranean region. This semi- to defoliation allows manager to choose the proper timing for arid rangeland occupies an important part of the country and plant use and rest which permits better management of range needs to be well understood to achieve optimum management. plant communities (Jameson and Huss 1959). Plant response Vegetation dynamics and seasonal growth patterns need spe- to green material removal varies greatly among seasons and cial attention as many plant communities are exposed to long- defoliation intensities. Defoliation timing and frequency term overgrazing. Wall barley (Hordeum murincum L.) is an affects range grass development (Mullahey et al. 1990). Miller annual native grass that is the key species for many local plant and Donart (1979) reported that forage production and crown communities in semiarid rangelands. Wall barley withstands diameters of 2 grass species were affected by forage removal the harsh climatic conditions of dry areas by its ability to season and quantity. The most critical period was during the reproduce readily from seed and to recover quickly from reproductive stage. Plants become more sensitive to defolia- drought events. tion as reproductive stage is approached (Tarassoum 1982, Deciding when and to what extent plants should be defoliat- Moser and Perry 1983). Dual purpose barley (H. spontaneum ed are considered the most critical and difficult tasks that a L) yield was reduced by late clipping (AL-Rawi et al. 1995). range manager must undertake. Understanding plant responses The ability of grasses to regrow after defoliation depends on plant genotype and the amount of green leaf tissue remaining in the stubble (Davies 1974). Plant injuries decreased with Manuscript accepted 19 Jul. 1998.

258 Journal of Range Management (52)3, May 1999 increasing stubble heights after clipping Agropyron spicatum (Sullivan and Sprague 1953, Mclean and Wikeen 1985). Dual purpose barley is becoming more popular in Iraq because it tolerates grazing at the pre-stem elongation stage (AL-Rawi et al. 1995). Grazing barley at pre-stem elongation stage can recover and produce grain equivalent to ungrazed stand (Morey 1961). However, severe defoliation of many semiarid range species was detrimental when defoliated plants relied upon nutrient reserves for regrowth (Trlica and Cook 1971). Therefore, knowledge of defoliation extent and timing is crucial for successful management of annual species such as wall barley range. Jordanian rangeland is subjected to continuous livestock overgrazing, and thorough understanding of vegetation dynamics is not available. The responses of wall barley to defoliation has not been reported. The objective was to Fig. 1. Rainfall distribution for 1994/95, 1995/96 growing seasons and long-term precipita- determine the responses of wall barley tion average (LTPA) at JUSTES. to clipping time and height in the semi- arid rangelands of Jordan. intensive agricultural use. The long-term 3 times and arranged in a randomized precipitation average (LTPA) for the complete block design. All plots were site is 230 mm, whereas annual rainfall harvested to 5 cm stubble height at seed Materials and Methods for the 1994/95 and 1995/96 growing maturity. seasons were 275 and 182 mm, respec- Data collected each year included; Study Area tively (Fig. 1). plant height (average of 3 vertical mea- Field studies were conducted on the surements from the soil surface to the Jordan University of Science and Treatment and Statistical Analysis highest point in the plant and soil sur- Technology Experimental Station Field plots of 6 x 5 m were established face recorded at physiological seed (JUSTES) located 22 km east of Irbid on a native stand of wall barley. maturity); total oven-dry forage weight (32¡ 34' N, 36¡ 0' E) in the 1994/95 and Treatments were combinations of clip- determined by adding the weight of tops 1995/96 growing seasons. The site has ping height (5 or 10 cm above soil sur- removed at the time of defoliation to the an altitude of 520 m and is characterized face) and timing, that corresponded to total forage weight (vegetative shoot by a flat to gently rolling topography plant growth stages of tillering, jointing, part weight) at time of harvest. Oven dry with less than 8% slope. The soil is and booting. Tillering is the growth weight was obtained by heating shoot at brown weakly cracked, calcareous, deep stage at which 50% of the plant tillers 70¡C for 48 hours. Seed yield was deter- silty clay with a pH of 6.8. Natural veg- have 4 leaves per tiller; jointing is the mined by threshing and cleaning dry etation at JUSTES is a typical Mediter- stage at which the second node appear heads. Weed biomass was also deter- ranean semiarid grassland with the dom- in 50% of the tillers; and booting is the mined at the end of the season by inating species being Hordeum spp., stage that immediately precedes the weighing aboveground weeds after dry- Avena fatua L. and Trigonella spp. This emergence of the inflorescence from the ing at 70¡C for 48 hours. All weight 2 site was subjected to long term over- sheath of the flag leaf in approximately measurements were collected for 1 m grazing and cultivation until 1986, after 30% of tillers. Treatment combinations quadrats. Analysis of variance (Table 1.) that it was partially protected from and an unclipped check were replicated for data combined over the 2 years were performed as outlined for a randomized

Table 1. Mean square and source of variation table showing the effect of defoliation time and height of Hordeum murinum L.

Source of Mean Square Variation df Plant height Forage yield First cut forage Seed yield Weed biomass Year 1 0.46 7817.36 9102.26 11270.10 586.90 Error (a) 4 4.12 1798.43 229.17 1074.70 725.00 Treatment 6 788.00 ** 436388.80 ** 638564.07 ** 266279.30 ** 53457.30 ** Treatment*Year 6 11.00 11221.47 3476.45 ** 699.90 246.20 * Error (b) 24 5.13 4943.71 3088.52 1182.00 707.04

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 259 complete block design using MSTATC computer program (Michigan State University). Means were separated by Duncan’s Multiple Range Test (DMRT) at P<0.01 (Steel and Torrie 1980).

Results and Discussion

Rainfall Variation in rainfall total and distribution were high between the 2 seasons which is expected for semiarid lands (Noy-Meir and Seligman 1979). In 1994/95, more than 50 % of rain occurred during November, whereas March and April were dry months (Fig. 1). This poor distribution masked the Fig. 2. Effect of defoliation time and height on plant height of Hordeum murinum L. Columns effect of high precipitation on plant with different letters differ significantly according to DMRT at P ≤ 0.01. growth. Although the precipitation for 1995/96 was lower than LTPA, the distribution was better especially during the period from January to April. This allowed the plant to produce forage yields equivalent to that produced in 1994/95.

Plant Height Clipping plants during the tillering stage did not affect plant height when measured at physiological maturity (Fig. 2). However, wall barley height was significantly reduced when clipping was performed at the jointing (27 and 29 cm) and boot stage (13 and 18 cm). Clipping height effects were only observed when cutting was practiced at boot. Wall barley height was greatly reduced when plants were clipped 5 cm above soil surface at the boot stage.

Forage Yield Forage production from the first cut increased with decreasing clipping height (Fig. 3). The greatest forage yield (1349 kg ha-1) was obtained when wall barley plants were clipped to 10 cm above soil surface at tillering stage (Fig. 3). However, clipping wall barley plants to 5 cm above soil surface at tillering lower ed forage yield (1,167 kg ha-1) but was equivalent to unclipped checks. This indicates that clipping heights of 5 and 10 cm at tillering stage did not impede regrowth following clipping in wall barley. Barley herbage at tillering stage has high protein content equivalent to that of forage legumes (Droushiotis and Wilman Fig. 3. Effect of defoliation time and height on first cut (top) and total forage yield (bottom) of Hordeum mirinum L. Columns with different litters differ significantly according to 1987) which may also improve animal DMRT at P ≤ 0.01. feed quality. Similar results were

260 Journal of Range Management (52)3, May 1999 Tiller production was inhibited when the wall barley plants were defoliated at the booting stage therefore seeds produced at the end of the season were from late maturing tillers, due to hot and dry condition during April. Large reductions in H. murinum L. seed production might eliminate this grass from range communities since it depends on soil seed bank for continuity and persistence. Seed production was not adversely affected when clipping occurred at tillering stage regardless of clipping height (594 and 583 kg ha-1 for 10 and 5 cm clipping height, respectively). This indicate that seed reserve will not be depleted if defoliation occurred early in the season. Mowing range grasses early in spring did not affect seed head production, whereas, late spring mowing progressively reduced the density of seed heads (Sims et al. 1971). Day time temperature Fig. 4. Effect of defoliation time and height on seed yield of Hordeum murinum L. Columns with different letters differ significantly according to DMRT at P ≤ 0.01. (Franke et al. 1992), moisture supply (Thakur and Shands 1954), and proper grazing management (Morey 1961), obtained where clipping heights of 4 and (Agropyron spicatum) produced higher contribute to high forage and seed yields 10 cm did not affect forage yield of shoot biomass due to better opportunity of small grain crops under simulated meadow foxtail stand (Smith et al. 1973). to regrow through the growing season. However, a 10 cm stubble height clip- grazing. ping improved plant forage yield but this Seed Yield yield was reduced when plants were Seed yield was low when plants were Weed Biomass The dominating weed species on the clipped down to 5 cm stubble height clipped at jointing or booting stages site were Cardaria draba (L.) Desv., (Leyshon and Campbell 1992). Previous (Fig. 4). Clipping at booting stage Sinapis arrensis L., and Diplotaxis eru- observations indicated that plant respons- resulted in a severe reduction in seed coides (L.) D.C. Weed interference as es to defoliation were highly variable production. Seed yields at booting stage detected by weed biomass significantly among time and intensity of defoliation were 98 and 169 kg ha-1 for clipping (Miller and Donart 1979), and the most height of 5 and 10 cm, respectively. increased as defoliation was delayed and critical defoliation time was either during the reproductive stage or under unsuit- able environmental conditions (Mulla- hey et al. 1990). In addition, clipping height generally has less influence on forage yield than clipping time (Dovel 1996). This agrees with our result where plant defoliation during jointing or booting stages reduced forage yield. These reductions were more severe when clipping was practiced at lower stubble heights (Fig. 3). Clipping during booting stage coincided with high temperature and low precipitation at the end of the rainy season. This lead to reduction in plant regrowth at booting stage. Italian ryegrass (Lolium multiflorum) regrowth rate and tillering capacity were minimized at high temperature, whereas primary growth rate was at maximum (Hill and Pearson 1985). These results agree with data reported by Kennett et al. (1992) where early season defolia- Fig. 5. Effect of Hordeum murinum L. defoliation time and height on weed biomass. Columns tion of bluebunch wheatgrass with different letters differ significantly according to DMRT at P ≤ 0.01.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 261 clipping intensity (5 vs 10) increased Davies, A. 1974. Leaf tissue remaining after Sims, P.L., L. J. Ayuko, and D. N. Hyder. (Fig. 5). Weed biomass was 201 and cutting and regrowth in perennial ryegrass. 1971. Developmental morphology of 223 kg ha-1 in stands subjected to 10 and J. Agi. Sci. 82:165Ð172. switchgrass and sideoats grama. J. Range 5 cm clipping heights at tillering stage, Dovel, R.L. 1996. Cutting height effects on Manage. 24:357Ð360. respectively. Clipping at booting stage wetland meadow forage yield and quality. Smith, D., A.V.A. Jacques, and J.A. J. Range Manage. 49:151Ð156. Balasko. 1973. Persistence of several tem- reduced wall barley stand and led to Droushiotis, D.N. and D. Wilman. 1987. perate grasses grown with alfalfa and har- increased weed infestation (412 to 426 Effects of harvesting program and sowing vested two, three, and four times annually -1 kg ha ). Clipping during tillering did not date on the forage yield, digestibility, at two stubble heights. Crop Sci. reduce wall barley competing ability nitrogen concentration, tiller and crop 13:553Ð556. and this prevented weed invasion. This fractions of barley in Cyprus. J. Agr. Sci. Steel, R.G.D. and J.H. Torrie. 1980. negative impact of clipping on competi- (Camb.) 109:95Ð106. Principles and Procedures of Statistics A tive ability was stronger when range Frank, A.B., A. Bauer, and A.L. Black. biometrical Approach, 2nd ed. McGraw- plants were clipped at later growth 1992. Effects of air temperature and nitro- Hill Co., New York, N.Y. stages. Kennett et al. (1992) also report- gen fertilizers on spike development in Sullivan, J.T. and V.G. Sprague. 1953. ed that spotted knapweed (Centaurea spring barley. Crop Sci. 32:793Ð797. Reserve carbohydrates in orchardgrass cut Hill, M.J. and C.J. Pearson. 1985. Primary for hay. Plant Physiol. 49:304Ð313. maculosa Lam.) growth and density growth and regrowth responses of temper- Tarassoum, T. 1982. Some effects of clip- were influenced by bluebunch wheat- ate grasses to different temperature and ping on the developmental morphology of grass defoliation treatments. Weed inva- cutting frequencies. Aust. J. Agr. Res. mesa dropseed. M.S. Thesis. New Mexico sion to any range site is associated with 36:25Ð34. State Univ., Las Cruces, N.M. reduction in desirable species range Jameson, D.A. and D.L. Huss. 1959. The Thakur, C. and H. L. Shands. 1954. Spring cover and normally reduces the grazing effect of clipping leaves and stems on small grain agronomic response to plant value of that site. Defoliation alters number of tiller, herbage weights, root clipping when seeded at two rates and fer- plant competitive ability and influences weights, and food reserves of little tilized at two levels of nitrogen. Agron. J. its role in the community (Maschinski bluestem. J. Range Manage. 12:122Ð126. 46:15Ð16 and Whitham 1989). Kennett, G.A., J.R. Lacey, C.A. Butt, K.M. Trlica, M.J. and C.W. Cook. 1971. Rutz, and M.R. Haferkamp. 1992. Defoliation effects on carbohydrate Effects of defoliation, shading and compe- reserve of desert species. J. Range tition on spotted knapweed and bluebunch Manage. 24:418Ð425. Conclusion wheatgrass. J. Range Manage. 45:363Ð369. Leyshon, A.J. and C.A. Campbell. 1992. Maximum forage production was Effect of timing and intensity of first defoli- achieved when wall barley was defoliat- ation on subsequent production of 4 pasture ed at either at the early tillering stage or species. J. Range Manage. 45:379Ð384. at plant maturity. Adequate soil moisture Maschinski, J. and T. C. Whitham. 1989. and temperature will encourage optimal The continuum of plant responses to herbivory: the influence of plant association, plant regrowth. Defoliation during joint- nutrient availability, and timing. Amer. ing or booting stages of wall barley Natur. 134:1Ð19. reduces forage and seed yield and Mclean, A., and S. Wikeen. 1985. Influence increases weed interference. Severe of season and intensity of defoliation on reduction in seed yield occurs when bluebunch wheatgrass survival and vigor plants are defoliated at the booting stage in southern British Columbia. J. Range which threatens a possible elimination of Manage. 38:21Ð26. the species from the rangeland as a result Miller, R.F. and G.B. Donart. 1979. of heavy unplanned grazing practices. Response of Bouteloua eripoda (Torr.) Future studies are needed to determine Torr. and Sporobolus flexuosus (Thurb.) the optimum number of seeds in the Rydb. to season of defoliation. J. Range seed bank that are required to maintain a Manage. 38:21Ð26. healthy natural wall barley population. Morey, D.D. 1961. Forage production of small grains under maximum favorable conditions. Agron. J. 53:57Ð59. Literature Cited Moser, L.E. and L.J. Perry. 1983. Yield, vigor, and persistence of sand lovegrass (Eragrosits trichodes) following clipping AL-Rawi, B.A., A.M. AL-Shamma and treatments. J. Range Manage. 36:236Ð238 J.A. Mohammed. 1995. Effect of delay of Mullahey, J.J., S.S. Waeler, and L.E. the second round clipping on the perfor- Moser. 1990. Defoliation effects on produc- mance of some dual-purpose barley lines. tion and morphological development of lit- In: N. Haddad, R. Tutwiler and E. tle bluestem. J. Range Manage. 36:236Ð238. Thomson (ed), Improvement of Crop- Noy-Meir, I. and N.G. Seligman. 1979. Livestock Integration Systems in West Management of semi-arid ecosystems in Asia and North Africa. Savvy Press, Israel, p.111Ð160. In: B.H. Walker (ed.), Syria. Management of semi-arid ecosystems. Elsevier Sci. Publ. Co.. Amsterdam.

262 Journal of Range Management (52)3, May 1999 J. Range Manage. 52:263Ð270 May 1999 Plant responses to defoliation and resource supplementation in the Pryor Mountains

JACE T. FAHNESTOCK AND JAMES K. DETLING

Authors are research scientist, Department of Renewable Resources, University of Wyoming, Laramie, Wyo. 82071-3354; and professor, Department of Biology, Colorado State University, Fort Collins, Colo. 80523-1878. At the time of the research, the senior author was graduate research assistant, Department of Biology, Colorado State University, Fort Collins, Colo.

Abstract Resumen

Field studies were conducted in 2 types of grasslands in the Se condujeron estudios de campo en 2 tipos de pastizal en el Pryor Mountain Wild Horse Range of northern Wyoming pastizal Pryor Mountain Wild Horse ubicado en el noreste de and southern Montana to examine plant biomass production Wyoming y sureste de Montana. El objetivo de estos estudios and nitrogen responses to the separate and combined effects fue examinar los efectos individuales y combinados de la of graminoid defoliation and increased environmental defoliación de gramíneas y el incremento de recursos ambien- resource (water or nutrients) supply. Short-term plant tales (agua o nutrientes). Las respuestas de la planta a corto responses were monitored over 2 years which differed sub- plazo fueron monitoreadas durante 2 años, los cuales stantially in growing season precipitation. In the arid, low ele- difirieron substancialmente en precipitación durante la vation grassland, total grass biomass was significantly lower estación de crecimiento. En la parte árida, pastizal de baja in the dry year than the wet year in all treatments. elevación, la producción total de biomasa de gramíneas en Defoliation of the grasses did not reduce their aboveground todos los tratamientos fue significativamente menor en el año biomass production in the wet year, but severely reduced it in seco que en el año húmedo. En el año húmedo, la defoliación the dry year, primarily because of a decrease in tiller density. de los zacates no redujó su producción de biomasa aérea, sin Mass of remaining individual tillers increased with clipping embargo, fue severamente reducida en el año seco, debido in the dry year, and nitrogen concentrations of the grasses principalmente a la disminución de la densidad de hijuelos. increased with clipping in both years. Irrigation alone Durante el año seco, la biomasa remanente de los hijuelos increased total belowground biomass compared to the other individuales se incrementó con la defoliación. En ambos años, treatments, but did not increase the aboveground biomass las concentraciones de nitrógeno de los zacates se incrementó production of any plant functional group. Clipping plus irri- con la defoliación. Comparado con los otros tratamientos, la gation resulted in greater total aboveground biomass produc- irrigación sola incrementó la biomasa total subterránea, pero tion and higher nitrogen concentrations of the grasses than no incrementó la producción de biomasa aérea en ninguno de control or irrigated treatments. Clipping graminoids in the los grupos funcionales de plantas. La defoliación mas irri- more mesic montane grassland did not decrease their biomass gación produjó mayores cantidades de biomasa total aérea y production in either year, but did increase their nitrogen con- concentraciones de nitrógeno que los tratamientos control o centrations and increase the collective aboveground biomass de irrigación. La defoliación de gramíneas en el pastizal mon- production of the other plant functional groups. Fertilization tano más mésico no disminuyó su producción de biomasa en and fertilization plus clipping significantly increased total ninguno de los años de estudio, pero si incrementó su concen- aboveground biomass production in both years, and total tración de nitrógeno e incrementó la producción colectiva de belowground biomass was greatest in fertilized plots. biomasa aérea de los otros grupos funcionales de plantas. En ambos años de estudio, la fertilización y la fertilización mas defoliación incrementaron significativamente la producción Key Words: wild horses, Pryor Mountain Wild Horse Range, total de biomasa aérea y la producción total de biomasa sub- primary production, nitrogen, Pseudoroegneria spicata terránea fue superior en las parcelas fertilizadas.

Many large ungulate herbivores preferentially consume grasses relative to their proportion in the plant community as plant species or functional groups are differentially affected (Schwartz and Ellis 1981, Vinton et al. 1993). This not only by the direct and indirect effects of selective herbivory. For changes the relative ability of grasses to acquire resources, but example, selective grazing of dominant grasses by bison in also alters the competitive interactions within the community tallgrass prairie can increase photosynthesis and growth of neighboring ungrazed plants of the same and other species Research was funded in part by the Bureau of Land Management and the (Fahnestock and Knapp 1993, 1994). National Park Service. The ability of grasses to compensate for biomass consumed Manuscript accepted 26 Aug. 1998.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 263 by large herbivores is controlled in part resource limitations (i.e., water or nutri- Löve (bluebunch wheatgrass) is the by the availability of environmental ent limitations) to growth following most abundant plant in these lowland resources that can readily be acquired by defoliation. Because the relative compet- communities, accounting for nearly 50% the plants (Chapin and McNaughton itive ability of other plants may be dif- of the total herbaceous plant cover and 1989). When resources such as water, ferentially affected by selective grass over 90% of the total grass cover light, and mineral nutrients are abundant herbivory, we also investigated whether (Fahnestock 1998). Since water presum- and readily available to plants, increased selective defoliation of the grasses would ably limits plant growth in these low- aboveground productivity of grasses in result in increased biomass production or lands, a grass defoliation and water sup- response to moderate grazing levels is nitrogen content of other plant functional plementation experiment was carried out possible (Archer and Tieszen 1980, groups, particularly forbs. We hypothe- in this community. McNaughton 1985, Seastedt 1985, Hik sized that biomass production and nitro- The second experiment was conducted and Jeffries 1990, Pandey and Singh gen concentrations would increase in in a more mesic, montane meadow at 1992, Frank and McNaughton 1993). plants with additional resource supply, 2,400 m above sea level that receives Conversely, when environmental and that the potential negative effects of over 500 mm of precipitation annually. resources are limited, the ability of defoliation on grass growth would be This site is located approximately 4 km grasses to replace tissue lost to grazers ameliorated with increased resource sup- north-west of the lowland site and pre- is reduced. The persistence of plants in ply. cipitation data for this community were grassland ecosystems, therefore, largely obtained from a weather station located depends on the ability of individuals to about 1 km north of this site. Vegetation cope with complex and dynamic interac- Materials and Methods in this community is more abundant tions between herbivory and limitations than the lowlands, averaging about 177 of environmental resources. Site description g m-2, and consists primarily of grasses The environmental resources that The experiments took place during the and sedges (44% of total plant cover) most frequently limit the growth of summers of 1993, a year with above and forbs (51% of total plant cover). plants and their ability to recover from average growing season precipitation, Much of the precipitation at this site grazing and other disturbances often dif- and 1994, a relatively dry year, in the comes from frequent summer showers, fer with the type of grassland being Pryor Mountain Wild Horse Range, an and thus the availability of nutrients, studied. In arid and semi-arid grass- 18,000 ha refuge located 80 km south of rather than water, presumably limits lands, water availability most often lim- Billings, Mont. Elevation ranges from plant growth here. We carried out a its plant growth (Lauenroth 1979, Sala 1,190 m to 2,440 m in the Pryor graminoid (grasses and sedges) defolia- et al. 1988), while in more humid grass- Mountains, and annual precipitation tion and nutrient supplementation exper- lands, or in wet years, light and nutri- varies from about 130 mm in some low- iment at this site. ents, especially nitrogen availability, are land areas to over 500 mm at upper ele- more likely to limit plant growth (Knapp vations. Our research was carried out in Experimental design and treatments and Seastedt 1986, Burke et al. 1991). 2 types of grasslands in the Pryor A completely randomized 2 X 2 facto- The response of plants to grazing may Mountains—a low elevation arid grass- rial design with 20 replicate plots was also differ between and within species, land and a more mesic montane grass- utilized for each experiment. Each depending on the nutrient and water land—in which plant growth was pre- experiment was established in a repre- stress tolerance of the species present, sumed to be limited, respectively, by sentative area of the upland or lowland and on the type of herbivory and its water and nutrient availability. To con- grassland community at that site, and intensity and frequency (Crawley 1983, trol for grazing, both experiments were plots were selected that were relatively Wallace et al. 1984, Coughenour et al. established in long-term (>20 years) uniform in terms of plant cover and 1985a, 1985b, Polley and Detling 1988). fenced areas that precluded wild horse species composition to control for dif- In the Pryor Mountain Wild Horse use but were similar in botanical compo- ferences within and between plots. Each Range in southern Montana and northern sition as those currently grazed by wild 3.2 m X 3.2 m plot was subdivided into Wyoming, grasses comprise over 70% of horses (Fahnestock 1998). 4 equal subplots to which a particular the annual diet of wild horses (Kissell The first experiment was conducted in treatment was randomly assigned. In the 1996). Little is known, however, about an arid lowland community at 1,300 m arid lowlands the treatments were grass the ability of the grasses in this system to above sea level that receives only ca. defoliation (D), irrigation (W), grass maintain production in response to this 230 mm of precipitation annually. defoliation plus irrigation (DW), or potentially substantial grazing pressure, Precipitation data (1982Ð1994) for this untreated control (C). In the more mesic or about the interaction of herbivory community were obtained from a weath- uplands the treatments were graminoid with water or nutrient availability in the er station located about 1 km south of defoliation (D), fertilization (F), Pryor Mountains. The principal objective our study site. Aboveground biomass in graminoid defoliation plus fertilization of this research, therefore, was to deter- this community averages about 128 g m-2 (DF), or control (C). mine if the dominant grasses in the Pryor (Fahnestock 1998) and is dominated by Vegetation was sampled in each sub- Mountain Wild Horse Range could perennial grasses (51% of total plant plot with 0.25 m2 circular quadrats that acquire the resources necessary to main- cover) and forbs (26% of total plant were randomly placed in one of 4 possi- tain biomass and nitrogen production in cover), with lesser quantities of cushion ble locations in each subplot. No vegeta- response to selective grass defoliation, or plants, dwarf shrubs and succulents. tion sample was collected within 20 cm whether there were environmental Pseudoroegneria spicata (Pursh) A.

264 Journal of Range Management (52)3, May 1999 of the edge of each subplot to avoid pled combustion/reduction and gas chro- K) was surface-broadcast applied to the edge effects. To simulate recurrent, matography (CHN-1000, LECO appropriate randomly selected subplots selective ungulate herbivory, all Corporation, St. Joseph, Mich.). once near the beginning (late-May) of the graminoid biomass in entire defoliated Aboveground N yield was calculated by 1993 and 1994 growing seasons. Soils and defoliated plus irrigated or defoliat- multiplying the N concentration by the were moist at the time of fertilization and ed plus fertilized subplots of each exper- aboveground biomass of each functional rain fell on the site within 10 days of fer- iment was clipped to a height of 2 cm group. Root biomass was estimated at tilization. All fertilizer was dissolved into once a month in May, June, July, and the time of final harvest in 1994 by the soil within 4 weeks of application. August of 1993 and 1994, and clipped excavating 5 alternate 0.25 m2 quadrats biomass within the 0.25 m2 quadrats of to the depth of 30 cm for each treatment. Statistical analyses each treatment was retained. This level Soils were air-dried and roots and rhi- In each experiment, total biomass (i.e., of defoliation was equivalent to 60Ð70% zomes were separated from soil by sieving. MayÐAugust clipped biomass plus final utilization of the graminoids over the In the lowland experiment, irrigated August harvest biomass) and nitrogen course of the growing season. subplots were watered with hand-held responses to each treatment were com- Additionally, at the lowland site, tillers sprinklers with water obtained from a pared using the General Linear Models of P. spicata were counted in the 0.25 nearby well. A 26.6 mm rainfall event procedure of the Statistical Analysis 2 m quadrats of each plot, and individual was simulated in May, June, and July of System (SAS 1989). Models included tiller mass was quantified by weighing each year, approximating a small the main effects of treatment by func- 100 tillers from each treatment in increase in the number of large rainfall tional group and the interaction of treat- August (i.e., 5 randomly selected tillers events this lowland area receives during ments. Type III sums-of-squares were from each of the 20 plots). the growing season. A total of 80 mm of used for significance (P < 0.05) testing, All aboveground biomass in the 0.25 water was added to irrigated plots during and least squares methods were used to 2 m quadrats of each treatment was har- each growing season, resulting in a near examine associations between treat- vested to ground level in August of both doubling of the long-term average pre- ments by functional group. We used t- years and oven-dried at 60¡C for 48 cipitation received at this site in May test procedures to compare biomass and hours. Because all aboveground biomass ÐJuly. Each watering event was applied nitrogen responses of each functional was harvested to ground level in 1993, over a 30-hour span so that there was no group to treatments between years. alternate 0.25 m2 quadrats within each significant puddling or runoff, and water subplot were sampled in 1994. These appeared to uniformly infiltrate the soil alternate plots had received the same to a depth of at least 25 cm (pers. obs. Results treatments except for the ground level based on excavations). This irrigation harvest in 1993. Biomass and total off- scheme appeared to increase the avail- Biomass responses take values for each treatment were sort- ability of soil water to irrigated plants by Growing season (ca. MarchÐJuly) pre- ed by functional group (e.g., live and nearly 3 weeks during the growing sea- cipitation in the lowland site was 202 dead graminoids, forbs, etc.) and son (i.e., irrigated soils were appreciably mm in 1993, well above the long-term weighed. Pooled biomass samples (n=5) more moist at 20 cm than non-irrigated average of 143 mm, while in 1994 it of each functional group were ground soils for about 1 week following each was only 101 mm. Live grass biomass through an 850 µm (20-mesh) screen in watering event). In the upland experi- and live plus dead grass biomass were -2 a Wiley mill, subsampled, and nitrogen ment, a moderate level (39 g m ) of significantly lower in all lowland treat- concentrations were determined by cou- slow-release fertilizer (20-10-5 of N-P-

Table 1. Mean aboveground biomass (g m-2) ±1 SE of grasses, forbs, and total biomass in control plots and defoliated, irrigated, and defoliated plus irrigated treatments from arid lowland sites (n = 20 for each functional group x treatment). Values shown are for final harvest in August 1993 and 1994 and include clipped grass biomass from May through August (see text for additional details).

Control Defoliated Irrigated Defoliated and Irrigated ------(g m-2) ------1993 Grasses 31.8 ± 1.0a*‡ 33.3 ± 1.1a* 33.4 ± 1.0a* 27.5 ± 0.8a* Live 10.5 ± 1.1a* 13.2 ± 1.4a* 12.5 ± 1.4a* 11.8 ± 1.2a* Dead 21.3 ± 0.8a 20.1 ± 3.3a* 20.9 ± 2.8a* 15.7 ± 2.0a Forbs 25.8 ± 2.9a 33.3 ± 3.4a 30.0 ± 2.6a 27.2 ± 3.2a Total aboveground biomass 137.2 ± 4.5a 135.6 ± 4.7a 135.4 ± 4.5a 126.8 ± 5.2aa 1994 Grasses 22.7 ± 1.0a* 8.6 ± 0.4b* 19.0 ± 0.6a* 14.2 ± 0.7ab* Live 6.2 ± 1.1a* 3.1 ± 0.3b* 5.6 ± 0.7ab* 4.4 ± 0.6ab* Dead 16.6 ± 3.0a 5.5 ± 1.2b* 13.4 ± 1.9a 9.8 ± 2.3ab Forbs 21.0 ± 2.8a 24.4 ± 3.7a 30.3 ± 4.5a 29.8 ± 3.4a Total aboveground biomass 119.2 ± 7.2a 123.3 ± 12.2a 134.6 ± 10.2ab 149.5 ± 8.5b ‡Columns with different letters within a functional group indicate significant differences (P < 0.05) between treatment means. Asterisks indicate significant differences (P < 0.05) between 1993 and 1994 values.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 265 tiller density in control plots was about 270 tillers m-2 in both years. In defoliated treatments, density averaged 260 tillers m-2 in 1993, but only 140 tillers m-2 in 1994 (Fig. 1b). Irrigation did not increase aboveground biomass production of any plant functional group in either year (Table 1). Total belowground biomass, however, was greatest in irrigated plots by the end of the experiment (Fig. 2a). Total belowground biomass in irrigated plots was 158 g m-2, while that in control, defoliated, and defoliated plus irrigated plots averaged 98 g m-2 (Fig. 2a). Irrigation apparently also enabled the grasses to compensate, at least in part, for tissue lost to clipping during the dry year; that is, in the dry year defoliation alone reduced live and dead aboveground grass biomass, but biomass of grasses in defoliated plus irrigated plots was not different from that of control plots (Table 1). The combined effects of defoliation and irrigation did not significantly alter aboveground biomass of any individual plant functional groups, but did result in greater total aboveground biomass production than control and defoliated treatments by the end of the -1 -2 Fig. 1. (a) Individual grass tiller mass (mg tiller ) and (b) tiller density (no. m ) in 1993, a experiment in 1994. wet year, and 1994, a dry year, in the lowland experimental site of the PMWHR. In the upland study site, growing sea- Measurements were made on Pseudoroegneria spicata, the dominant grass in these lowlands. Different letters above bars indicate significant (P < 0.05) differences in means son (ca. AprilÐAugust) precipitation was between treatments and years. Treatments are control (C), grasses defoliated (D), irrigated also lower in 1994 (193 mm) than in (W), and defoliated plus irrigated (DW). 1993 (399 mm). Nevertheless, we found no interannual differences in above- ments in 1994 than 1993 (Table 1). reduced, but defoliation did reduce tiller ground biomass production of any plant However, forb and total aboveground density in the dry year (Fig. 1b). Grass functional group in control plots (Table biomass were not significantly lower in 1994 than 1993 in any treatment. In the Table 2. Mean aboveground biomass (g m-2) ±1 SE of grasses, forbs, and total biomass in control lowland experiment, there was not a sig- plots and defoliated, fertilized, and defoliated plus fertilized treatments from montane sites (n = nificant effect of any of the treatments 20 for each functional group x treatment). Values shown are for final harvest in August 1993 (defoliated, irrigated, or defoliated plus and 1994 and include clipped graminoid biomass from May through August (see text for addi- irrigated) on biomass production of any tional details). plant functional group in 1993, the wet Control Defoliated Fertilized Defoliated and year (Table 1). This was not the case, Fertilized however, in 1994, where grass biomass -2 production was much lower in defoliat------(g m ) ------2 1993 ed plots (9 g m ) than in control plots a a a a (23 g m-2; Table 1). Graminoids 48.5 ± 6.4 ‡ 54.2 ± 5.0 61.7 ± 7.2 * 65.4 ± 6.6 Live 25.4 ± 2.9a 30.9 ± 2.6a 42.0 ± 4.8b* 45.2 ± 4.6b Much of the lower live grass biomass Dead 23.0 ± 4.2a 23.2 ± 2.7a 19.7 ± 4.1a* 20.1 ± 3.1a production in 1994, the dry year, was Forbs 82.0 ± 10.7a 84.0 ± 7.0a 93.6 ± 11.3a* 110.3 ± 17.9a the result of lower mass of individual Total aboveground biomass 175.6 ± 11.1a 193.1 ± 13.1ab 234.9 ± 17.4bc* 234.5 ± 15.2bc* live grass tillers. In control plots, mass -1 1994 was 32 mg tiller in 1993 but only 14 a a b a -1 Graminoids 49.7 ± 6.0 41.2 ± 6.2 105.6 ± 8.1 * 59.6 ± 6.5 mg tiller in 1994 (Fig. 1a). A similar Live 27.8 ± 3.1ab 21.8 ± 2.4a 57.5 ± 4.9c* 36.1 ± 4.4b reduction in tiller mass was seen in Dead 21.9 ± 3.2a 19.4 ± 3.5a 48.1 ± 6.3b* 23.5 ± 2.5a grasses that were irrigated. In contrast, Forbs 90.1 ± 12.0a 101.2 ± 12.8ab 146.8 ± 16.8c* 137.4 ± 13.7bc individual mass of grass tillers that were Total aboveground biomass 177.7 ± 18.2a 211.7 ± 22.5b 338.3 ± 19.8c* 282.8 ± 17.4c* defoliated (both defoliated and defoliat- ‡Columns with different letters within a functional group indicate significant differences (P < 0.05) between treatment ed plus irrigated treatments) were not means. Asterisks indicate significant differences (P < 0.05) between 1993 and 1994 values.

266 Journal of Range Management (52)3, May 1999 2). Selective defoliation of the Table 3. Nitrogen concentrations (%) of live and dead grass and forbs in 1993, a wet year, and graminoids also did not significantly 1994, a dry year, from the lowland experimental site (n = 5 for each functional group x treat- reduce their biomass production in ment). either year. However, defoliated plots did have greater total aboveground bio- Control Defoliated Irrigated Defoliated and mass than control plots by the end of the Irrigated experiment (Table 2). ------(%) ------Fertilization increased live graminoid 1993 biomass in both years, and total live and Live grass 0.81 ± 0.01a*‡ 1.26 ± 0.05b* 0.85 ± 0.04a* 1.25 ± 0.03b* Dead grass 0.45 ± 0.01a* 0.62 ± 0.02b* 0.46 ± 0.02a* 0.61 ± 0.03b* dead graminoid biomass was highest in a a a a fertilized plots by the end of the experi- Forbs 0.78 ± 0.03 0.84 ± 0.03 0.82 ± 0.02 0.95 ± 0.04 ment in 1994. At that time, live (58 g m-2) 1994 and dead (48 g m-2) graminoid biomass in Live grass 0.66 ± 0.02a* 0.80 ± 0.02b* 0.62 ± 0.02a* 0.78 ± 0.02b* Dead grass 0.53 ± 0.03a* 0.76 ± 0.01b* 0.55 ± 0.04a* 0.75 ± 0.01b* fertilized plots was more than twice that a a a a in control or defoliated plots (Table 2). Forbs 0.80 ± 0.03 0.95 ± 0.19 0.79 ± 0.04 0.81 ± 0.03 Aboveground biomass production of all ‡Columns with different letters within a functional group indicate significant differences (P < 0.05) between treatment means. Asterisks indicate significant differences (P < 0.05) between 1993 and 1994 values. plant functional groups was higher in fertilized plots in 1994 than in 1993 (Table 2). Total belowground biomass in ed plots were higher in both years than were higher in all treatments in 1993 fertilized plots (620 g m-2) was greater those in either control or irrigated plots than in 1994, and N concentrations of than that of control (391 g m-2) or defoli- (Table 3). The N concentration of live the forbs were not affected by any treat- ated (306 g m-2) plots by the end of the grasses was significantly lower in all ment in either year. experiment, but was not significantly treatments in 1994, the dry year, than in Total aboveground nitrogen yield (g N different from defoliated plus fertilized 1993, the wet year (Table 3). In contrast, m-2) of the lowland and upland plots (453 g m-2 Fig. 2b). The defoliation the N concentration of dead grasses was graminoids was increased by defoliation plus fertilization treatment increased live higher in all treatments in 1994 than in only in 1993, the wetter year (Fig. 3). graminoid biomass in 1993 compared to 1993. The N concentrations of the forbs Aboveground N yield of all lowland control plots, but total grass biomass was were not significantly affected by selec- plants combined was increased in the not significantly changed by this treat- tive grass defoliation in either year. defoliated plus irrigated treatment by the ment (Table 2). Additionally, total In the more mesic montane grassland, end of the experiment in 1994, and in the aboveground biomass was significantly all treatments (defoliated, fertilized, and upland experiment, fertilization and higher in defoliated plus fertilized plots defoliated plus fertilized) resulted in defoliation and fertilization increased the than in control plots in both years. increased N concentrations in the live aboveground N yield of the graminoids graminoids, and defoliated and defoliat- and of all plants combined in both years. Nitrogen responses ed plus fertilized increased the N concentration of the dead graminoids, in In the arid lowland grassland, N con- both years (Table 4). As at the lowland Discussion and Conclusions centrations of both live and dead grass site, N concentrations of live graminoids in defoliated and defoliated plus irrigat- In the wet year, 1993, selective defoliation of the lowland grasses, primarily P. spicata, had no effect on their aboveground biomass production; that is, the grasses fully compensated for the shoot biomass removed. In the dry year, however, defoliation in the absence of water supplementation severely reduced grass biomass production. Irrigation in that year enabled P. spicata to compensate for shoot biomass lost to clipping. Consistent with our hypothesis, these data suggest that in the arid lowlands of the Pryor Mountain Wild Horse Range, regrowth following grazing is closely linked to water availability. Similar reductions in grass biomass

-2 production in response to the simultane- Fig. 2. Total belowground biomass (g m ) in the (a) lowland and (b) upland experimental ous pressures of clipping and water sites at the conclusion of the experiment in August, 1994. Columns headed by different letters indicate significant (P < 0.05) differences between treatments. Treatments are control stress have also been found in other (C), graminoids defoliated (D), irrigation (W) or fertilization (F), and graminoids defoliat- studies (e.g., Toft et al. 1987, ed plus irrigation (DW) or graminoids defoliated plus fertilization (DF). Note different Georgiadis et al. 1989, Simoes and scales for the 2 sites. Baruch 1991, Busso and Richards

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 267 either the dry or wet year. Why P. spicata failed to increase its aboveground biomass production in response to artifi- cially increased water supply is not known. Although bare ground evaporation was undoubtedly high, soil excavations showed that soil moisture was appreciably higher in irrigated compared to non-irrigated plots for at least 1 week following each watering event. It is possible that the grasses in these arid lowlands are rather insensitive to large puls- es of increased water availability, such as those simulated in this study, and that frequent, smaller rainfall events may be more important to their growth (e.g., Sala and Lauenroth 1982, but see Weaver 1985). Additionally, the overall environment in these lowlands was still arid which may have negated the effects of our irrigation treatment. Nevertheless, total root biomass was significantly higher in the lowland irrigated plots than in the other treatments by the end of the experiment, indicating that plant growth responses to our levels of irrigation may Fig. 3. Aboveground nitrogen yield (g N m-2; n = 5 for each vertical bar) for graminoids, have been belowground only. forbs, and total N yield in the lowland and upland experiments in August, 1993 and In the lowland site, the reduction in August, 1994. Treatments as in Fig. 2. Asterisks denote significantly higher N yields (calcu- grass biomass with clipping that was lated as N concentration multiplied by aboveground biomass) in treatment than control. measured in the dry year was the result Note different scales for each graph. of a decrease in grass tiller density and not individual tiller mass, which was 1995). This response, however, is in Richards (1995) similarly found that much higher in clipped than control contrast to the prediction of Hilbert et mass of individual unclipped tillers of P. plots that year. Bluebunch wheatgrass, al. (1981) that plants which are stressed spicata was severely reduced under the dominant grass in these lowlands, in some way, and consequently growing drought conditions compared to more has been shown to be a decreaser under slowly compared to their potential rates, average precipitation levels. They also increasing grazing pressure (Williams are most likely to increase production found that plants that were irrigated 1963). A decrease in tiller number result- following grazing. The response of the showed lower tiller height, leaf area, and ing from grazing has frequently been lowland grasses in the Pryor Mountains mass compared to plants under more reported (Branson 1956, Ellison 1960, was also not consistent with those of average precipitation conditions. Our Caldwell et al. 1981, Carman and Briske McNaughton et al. (1983) and levels of irrigation did not increase indi- 1985, Polley and Detling 1989), although Coughenour et al. (1985 a, 1985b, vidual grass tiller mass of P. spicata in an increase in tiller production (Richards 1990); in their studies, the effects of et al. 1988, Zhang and Romo 1995) and water stress and defoliation were found to often act singly and 1 factor could Table 4. Nitrogen concentrations (%) of live and dead graminoids and forbs in 1993, a wet year, reduce or ameliorate the negative effect and 1994, a dry year, from the montane experimental site (n = 5 for each functional group x of the other. In our study, additional treatment). water supply ameliorated the negative Control Defoliated Fertilized Defoliated and effects of defoliation in the dry year. In Fertilized the semiarid shortgrass steppe, ------(%) ------Milchunas et al. (1994) have found that 1993 long-term primary production across 50 Live graminoids 1.37 ± 0.04a* 1.50 ± 0.02b* 1.72 ± 0.02c* 1.76 ± 0.06c* years of cattle grazing treatments was Dead graminoids 0.90 ± 0.03a* 1.08 ± 0.05b 1.04 ± 0.05ab* 1.17 ± 0.05b most sensitive to precipitation and least Forbs 1.47 ± 0.05a 1.56 ± 0.03a 1.58 ± 0.07a 1.59 ± 0.06a sensitive to grazing intensity. 1994 In the dry year, decreased live grass Live graminoids 1.13 ± 0.02a*‡ 1.27 ± 0.05bc* 1.24 ± 0.02b* 1.37 ± 0.06c* biomass in the lowland control plots Dead graminoids 0.70 ± 0.07a* 0.94 ± 0.08b 0.55 ± 0.01a* 1.00 ± 0.10b was the result of lower individual tiller Forbs 1.27 ± 0.03a 1.33 ± 0.08a 1.35 ± 0.05a 1.25 ± 0.06a mass, not tiller density. Busso and ‡Columns with different letters within a functional group indicate significant differences (P < 0.05) between treatment means. Asterisks indicate significant differences (P < 0.05) between 1993 and 1994 values.

268 Journal of Range Management (52)3, May 1999 increased regrowth of surviving tillers plants (Chapin 1980, Jaramillo and other plant functional groups in the (Branson 1956, Caldwell et al. 1981, Detling 1992). Fertilization also resulted Pryor Mountains, indicating that the rel- McNaughton et al. 1983, Carman and in defoliated graminoids producing more ative competitive abilities of plants in Briske 1985), has also been observed. In live, but not dead, biomass in 1993, than the upland and lowland communities are this system there appears to be an inverse unclipped, unfertilized plants. Therefore, not greatly, or at least rapidly, altered by relationship between tiller size and densi- the effect of increased nutrient supply on selective graminoid defoliation. ty, as is generally the case (Risser 1969, defoliated graminoids in the uplands may Gorham 1979, Christiansen and Svejcar be to decrease their rate of senescence, 1988). The simultaneous pressures of and to increase their belowground bio- Literature Cited clipping and water stress reduced tiller mass production (see above). density of P. spicata, but increased the Our results suggest that the dominant Archer, S. and L.L. Tieszen. 1990. Growth mass of the surviving tillers relative to graminoids in the Pryor Mountains are and physiological responses of tundra those of unclipped plants. Additions of able to withstand fairly heavy levels of plants to defoliation. Arct. Alp. Res. water to clipped plots in the dry year did defoliation through compensatory 12:531Ð552. not prevent a reduction in tiller density growth. In the uplands, the graminoids Branson, F.A. 1956. Quantitative effects of clipping treatments on five range grasses. following clipping, providing further are able to compensate for tissue lost to J. Range Manage. 9:86-88. evidence to suggest that P. spicata is not grazers without additional resource sup- Burke, I.C., T.G.F. Kittel, W.K. very responsive to the large, infrequent ply. In the lowlands, however, their abil- Lauenroth, P. Snook, C.M. Yonker, and rainfall events simulated in this short- ity to fully regrow following grazing is W.J. Parton. 1991. Regional analysis of term study. only possible when water availability is the central Great Plains. BioSci. In the upland defoliation-fertilization not too low. Increased water availability 41:685Ð692. experiment, the decrease in growing sea- in the lowlands of the Pryor Mountains Busso, C.A. and J.H. Richards. 1995. son precipitation in 1994 compared to will most likely stimulate belowground Drought and clipping effects on tiller demography and growth of two tussock 1993 did not result in a decrease in above- growth in these plant communities, and grasses in Utah. J. Arid Environ. ground biomass production of any plant will increase the nitrogen concentrations 29:239Ð251. functional group, suggesting that water of all plants except perhaps the forbs. Caldwell, M.M., J.H. Richards, D.A. did not limit plant growth at this site dur- In both the lowland and upland com- Johnson, R.S. Nowak, and R.S. Dzurec. ing the study. This was not true at all munities, selective graminoid defolia- 1981. Coping with herbivory: photosyn- upland sites in the Pryor Mountain Wild tion increased the N concentrations of thetic capacity and resource allocation in Horse Range, however, since decreases in graminoids in both years, and increased two semiarid Agropyron bunchgrasses. Oecologia 50:14-24. plant cover and biomass were measured total aboveground N yield of graminoids Carman, J.G. and D.D. Briske. 1985. at some sites in the dry year (Peterson et in the wet year. This may have resulted Morphologic and allozymic variation al. 1997, Fahnestock 1998). At this site, from increased N uptake by defoliated between long-term grazed and non-grazed our finding that defoliation did not reduce grasses, increased allocation of N to populations of the bunchgrass Schizachy- live or total graminoid biomass produc- aboveground plant tissue, or both. rium scoparium var. frequens. Oecologia tion in either year indicates that the Nitrogen uptake, N concentration, and 66:332Ð337. graminoids are apparently able to com- physiological activity are often higher in Chapin, F.S., III. 1980. The mineral nutrition of wild plants. Annu. Rev. Ecol. Syst. pensate for biomass lost to grazers with- plants that have been grazed or other- 11:233Ð260. out additional resource supply. wise defoliated than in plants from Chapin, F.S., III and S.J. McNaughton. Fertilization increased the growth of ungrazed areas (e.g., Jameson 1963, 1989. Lack of compensatory growth under most plant functional groups such that Detling et al. 1979, McNaughton 1979, phosphorus deficiency in grazing-adapted total above- and belowground biomass Coppock et al. 1983, McNaughton et al. grasses from the Serengeti plains. production were higher in fertilized only 1983, Ruess et al. 1983, Detling and Oecologia 79:551-557. plots than unfertilized plots by the end of Painter 1983, Polley and Detling 1988, Chapin, F.S., III, P.M. Vitousek, and K. Van Cleve. 1986. The nature of nutrient the experiment. In individual plants, Jaramillo and Detling 1988). Increased limitation in plant communities. Amer. nutrient limitation is recognized by an N concentrations in grazed graminoids Nat. 127:48Ð58. increase in growth in response to an addi- results in higher quality forage available Christiansen, S. and T. Svejcar. 1988. tion of the limiting nutrient. The analo- to herbivores. This may have important Grazing effects on shoot and root dynam- gous response at the community level is consequences for subsequent food pref- ics and above- and below-ground non- an increase in total community produc- erence by herbivores in the Pryor structural carbohydrates in Caucasian tion in response to fertilization (Chapin et Mountain Wild Horse Range. In addi- bluestem. Grass Forage Sci. 43:111Ð119. Coppock, D.L., J.K. Detling, J.E. Ellis, al. 1986). In this experiment only, the tion, the higher N-yield of defoliated and M.I. Dyer. 1983. Plant-herbivore upland graminoids increased their above- graminoids suggests that herbivores interactions in a North American mixed- ground biomass production in response may, via their grazing activities, grass prairie. I. Effects of black-tailed to fertilization in the first year, but by the increase not only the quality of their for- prairie dogs on intraseasonal aboveground second year all plant functional groups age but also the quantity of crude pro- plant biomass and nutrient dynamics and had increased biomass production. These tein subsequently available to them plant species diversity. Oecologia 56:1Ð9. results suggest that species or functional when they regraze the same patch. Coughenour, M.B., S.J. McNaughton, and L.L. Wallace. 1985a. Responses of an groups differed in their ability to respond Defoliation of the graminoids did not African tall-grass (Hyparrhenia filipendu- to increased nutrient supply or that nutri- change the aboveground biomass pro- la staph.) to defoliation and limitation of ent limitations were not the same for all duction or nitrogen concentrations of the water and nitrogen. Oecologia 68:80Ð86.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 269 Coughenour, M.B., S.J. McNaughton, and Jaramillo, V.J. and J.K. Detling. 1988. Sala, O.E. and W.K. Lauenroth. 1982. L.L. Wallace. 1985b. Responses of an Grazing history, defoliation, and competi- Small rainfall events: an ecological role in African graminoid (Themeda triandra tion: effects on shortgrass production and semiarid regions. Oecologia 53:301Ð304. Forsk.) to frequent defoliation, nitrogen, nitrogen accumulation. Ecol. 69:1599Ð1608. Sala, O.E., W.J. Parton, L.A. Joyce, and and water: a limit of adaptation to her- Jaramillo, V.J. and J.K. Detling. 1992. W.K. Lauenroth. 1988. Primary produc- bivory. Oecologia 68:105Ð110. Small scale heterogeneity in a semi-arid tion of the central grassland region of the Coughenour, M.B., J.K. Detling, I.E. North American grassland. I. Tillering, N United States. Ecol. 69:40Ð45. Bamberg, and M.M. Mugambi. 1990. uptake and retranslocation in simulated SAS Institute. 1989. SAS/STAT User's Production and nitrogen responses of the urine patches. J. Appl. Ecol. 29:1Ð8. Guide, Fourth Edition. SAS Institute Inc., African dwarf shrub Indigofera spinosa to Kissell, R.J., Jr. 1996. Competitive interac- Cary, N.C. defoliation and water limitation. Oecologia tions among bighorn sheep, feral horses, Schwartz, C.C. and J.E. Ellis. 1981. 83:546Ð552. and mule deer in Bighorn Canyon Feeding ecology and niche separation in Crawley, M.J. 1983. Herbivory: the dynam- National Recreation Area and Pryor some native and domestic ungulates on the ics of animal-plant interactions. University Mountain Wild Horse Range. Ph.D. Diss. shortgrass prairie. J. Appl. Ecol. of California Press, Berkeley, Calif. Montana State Univ., Bozeman, Mont. 18:343Ð353. Detling, J.K. and E.L. Painter. 1983. Knapp, A.K., and T.R. Seastedt. 1986. Seastedt, T.R. 1985. Maximization of pri- Defoliation responses of western wheat- Detritus accumulation limits productivity mary and secondary productivity by graz- grass populations with diverse histories of of tallgrass prairie. BioSci. 36:662Ð668. ers. Amer. Nat. 126:559Ð564. prairie dog grazing. Oecologia 57:65Ð71. Lauenroth, W.K. 1979. Grassland primary Simoes, M. and Z. Baruch. 1991. Detling, J.K., M.I. Dyer, and D.T. Winn. production: North American grasslands in Responses to simulated herbivory and 1979. Net photosynthesis, root respiration, perspective. p. 3Ð24. In: N.R. French (ed.), water stress in two tropical C4 grasses. and regrowth of Bouteloua gracilis follow- Perspectives in grassland ecology. Oecologia 88:173Ð180. ing simulated grazing. Oecologia 41:127- Springer-Verlag, New York, N.Y. Toft, N.L., S.J. McNaughton, and N.J. 134. McNaughton, S.J. 1979. Grazing as an opti- Georgiadis. 1987. Effects of water stress Ellison, L. 1960. Influence of grazing on mization process: grass-ungulate relation- and simulated grazing on leaf elongation plant succession of rangelands. Bot. Rev. ships in the Serengeti. Amer. Nat. and water relations of an East African 26:1Ð78. 113:691Ð703. grass, Eustachys paspaloides. Australian J. Fahnestock, J.T. 1998. Vegetation respons- McNaughton, S.J. 1985. Ecology of a graz- Plant Phys. 14:211Ð226. es to herbivory and resource supplementa- ing ecosystem: the Serengeti. Ecol. Mono. Vinton, M.A., D.C. Hartnett, E.J. Finck, tion in the Pryor Mountain Wild Horse 55:259Ð294. and J.M. Briggs. 1993. Interactive effects Range. Ph.D. Diss. Colorado State Univ., McNaughton, S.J., L.L. Wallace, and of fire, bison (Bison bison) grazing and Fort Collins, Colo. M.B. Coughenour. 1983. Plant adaptation plant community production in tallgrass Fahnestock, J.T. and A.K. Knapp. 1993. in an ecosystem context: effects of defolia- prairie. Amer. Midl. Nat. 129:10-18. Water relations and growth of tallgrass tion, nitrogen, and water on growth of an Wallace, L.L., S.J. McNaughton, and prairie forbs in response to selective grass African C sedge. Ecol. 64:307Ð318. M.B. Coughenour. 1984. Compensatory herbivory by bison. Int. J. Plant Sci. 4 Milchunas, D.G., J.R. Forwood, and W.K. photosynthetic responses of three African 154:432-440. Lauenroth. 1994. Forage production graminoids to different fertilization, water- Fahnestock, J.T. and A.K. Knapp. 1994. across fifty years of grazing intensity treat- ing and clipping regimes. Bot. Gaz. Plant responses to selective grazing by ments in the shortgrass steppe. J. Range 145:151Ð156. bison: interactions between light, her- Manage. 47:133Ð139. Weaver, T. 1985. Summer showers: their bivory and water stress. Vegetatio Pandey, C.B. and J.S. Singh. 1992. Rainfall sizes and interception by surface soils. 115:123Ð131. and grazing effects on net primary produc- Amer. Midl. Nat. 114:409Ð413. Frank, D.A. and S.J. McNaughton. 1993. tivity in a tropical savanna, India. Ecol. Williams, C.S. 1963. Ecology of bluebunch Evidence for the promotion of above- 73:2007Ð2021. wheatgrass in northwestern Wyoming. ground grassland production by native Peterson, J., J.T. Fahnestock, and J.K. Ph.D. Diss., Univ. of Wyoming, Laramie, large herbivores in Yellowstone National Detling. 1997. Ungulate/vegetation inter- Wyo. Park. Oecologia 96:157Ð161. actions at the Pryor Mountain Wild Horse Zhang, J. and J.T. Romo. 1995. Impacts of Georgiadis, N.J., R.W. Ruess, S.J. Range. Final report to the National Park defoliation on tiller production and sur- McNaughton, and D. Western. 1989. Service. vival of northern wheatgrass. J. Range Ecological conditions that determine when Polley, H.W. and J.K. Detling. 1988. Manage. 48:115Ð120. grazing stimulates grass production. Herbivory tolerance of Agropyron smithii Oecologia 81:316Ð322. populations with different grazing histo- Gorham, E. 1979. Shoot height, weight, and ries. Oecologia 77:261Ð267. standing crop in relation to density of Richards, J.H., R.J. Mueller, and J.J. monospecific stands. Nature 279:148Ð150. Mott. 1988. Tillering in tussock grasses in Hik, D.S. and R.L. Jeffries. 1990. Increases relation to defoliation and apical bud in the net above-ground primary produc- removal. Ann. Bot. 62:173Ð179. tion of salt-marsh forage grass: a test of Risser, P.G. 1969. Competitive relationships the predictions of the herbivore-optimiza- among herbaceous grassland plants. Bot. tion model. J. Ecol. 78:180Ð195. Rev. 35:251Ð284. Hilbert, D.W., D.M. Swift, J.K. Detling, Ruess, R.W., S.J. McNaughton, and M.B. and M.I. Dyer. 1981. Relative growth Coughenour. 1983. The effects of clip- rates and the grazing optimization hypoth- ping, nitrogen source and nitrogen concen- esis. Oecologia 51:14Ð18. tration on the growth responses and nitro- Jameson, D.A. 1963. Responses of individual gen uptake of an East African sedge. plants to harvesting. Bot. Rev. 29:532-594. Oecologia 59:253Ð261.

270 Journal of Range Management (52)3, May 1999 J. Range Manage. 52:271Ð274 May 1999 Imbibition temperature affects winterfat (Eurotia lanata (Pursh) Moq.) seed hydration and cold-hardiness response

YUGUANG BAI, D. TERRANCE BOOTH, AND J.T. ROMO

Authors are range ecologist, Agriculture and Agri-Food Canada, Kamloops Range Res. Unit Kamloops, BC, V2B 8A9, Canada, rangeland scientist, USDA- ARS, Cheyenne, Wyo. 82009, USA, and professor, Dept. of Plant Sciences, Univ. of Saskatchewan, Saskatoon, SK, S7N 5A8, Canada. At the time of the research Bai was research scientist, Dept. of Plant, Soil and Insect Sciences, Univ. of Wyoming, Laramie, Wyo. 82071, USA.

Abstract Resumen

Winterfat (Eurotia lanata (Pursh) Moq.) diaspores harvest- Semillas de "Winterfat" (Eurotia lanata (Pursh) Moq. ) ed from 1 Canadian and 2 USA sites were imbibed at 0, 5, 10, cosechadas de un sitio canadiense y dos de U.S.A fueron and 20¡C. It was hypothesized that imbibition temperature imbibidas a 0, 5, 10, y 20°C. Se hipotetizó que la temperatura affects seed hydration which is related to cold-hardiness of de imbibición afecta la hidratación de la semilla la cual esta winterfat. Weight gain was measured at 8-hour intervals until relacionada con la resistencia al frío del "winterfat". La full hydration, and embryo water content was determined. ganancia de peso se midió a intervalos de 8 horas hasta la Water content of fully hydrated seeds differed among collec- hidratación total y se determinó el contenido de agua del tions and lower imbibition temperatures were always associ- embrión. El contenido de agua de semillas completamente ated with greater seed water content. Differences in water hidratadas difirió entre colecciones y las bajas temperaturas content of seeds imbibed at different temperatures was relat- de imbibición siempre estuvieron asociadas con un mayor ed to cold-hardiness. When water content of embryos was contenido de agua de la semilla. La diferencia del contenido measured, differences among imbibition temperatures exist- de agua de las semillas imbibidas a diferentes temperaturas ed, but were reduced. Differences in seed water content fue relacionada a la resistencia al frío. Cuando se midió el among imbibition temperatures were mainly due to contenido de agua de los embriones existían diferencias entre endosperm other than the embryo because the embryo las temperaturas de imbibición, pero eran mínimas. Las hydrated faster than other seed parts. Suggestions were made diferencias en el contenido de agua de las semillas entre tem- for modeling seed water relations based on this study. peraturas de imbibición fueron debidas principalmente al endospermo mas que el embrión, esto porque el embrión se hidrató más rápido que otras partes de la semilla. Basado en Key Words: Krascheninnikovia, Ceratoides, embryo, thresh- este estudio se hicieron sugerencias para modelar las rela- old water content. ciones hídricas de la semilla.

Environmental stress affects plants at all life stages, though dry seeds are most tolerant. As seed water content increases, physiological activities of seeds are increasingly influenced by measured by seed germination percentage and rate depended temperature-water interactions. Many seed studies have dealt on imbibition temperature (Bai et al. 1998). Since seed water with species that are injured by imbibing cold water (Hobbs relations are also related to cold-hardiness, a closer look at and Obendorf 1972, Ashworth and Obendorf 1980, Herner hydration processes of winterfat seeds may provide additional 1986). Few have examined basic themes using seeds with information on its cold-hardiness mechanism. unusual tolerance to extremes, such as winterfat (Eurotia Seeds must reach a threshold water content before germina- lanata (Pursh) Moq.), which may be unaffected or even bene- tion starts. Water content at the onset of germination was the fited by cool conditions (Booth 1992). same among osmotic conditions (Bradford 1986, Gray et al. Winterfat seeds can germinate over a wide range of temper- 1990) or certain temperatures (Gummerson 1986). On the atures (Dettori et al. 1984), including 0 or near 0¡C (Hilton other hand, if different seed parts imbibe water at different 1941, Woodmansee and Potter 1971, Dettori et al. 1984, rates, then using the whole seed water content as a pre-condi- Booth 1987). A recent study indicated that hydrated winterfat tion for seed germination may be questionable. In the study seeds were tolerant to cooling as low as Ð30¡C even after the reported here, seed hydration of winterfat was tested at tem- occurrence of the low temperature exotherm; cold-hardiness peratures ranging from 0 to 20¡C. Objectives of this study were to determine effects of imbibition temperature on seed Funding was provided in part by Saskatchewan Agriculture Development water in relation to cold-hardiness and to incipient germina- Fund, Ducks Unlimited Canada and Canada-Saskatchewan Agricultural tion in winterfat. Green Plan to J.T.R. The authors thank Drs. K.J. Bradford, J.Q. Hou, P.H. Li, and C. Stushnoff for reviewing an earlier version of this manuscript. Manuscript accepted 22 Aug. 1998.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 271 Materials and Methods

Seed sources and post-harvest handling Winterfat diaspores (seed-containing dispersal units) were hand-collected in October 1994 from Matador, Saskatchewan, Sterling, Colo., and Pine Bluffs, Wyo. (Table 1). The first 2 sites are located in Mixed Prairie while the third is located in Shortgrass Prairie (based on a map by Dodd; in Tetlyanova et al. 1990). Harvested diaspores were stored in paper bags at room temperature until February 1995. Dry weight of threshed seeds from Matador and Sterling were 38 and 28% heavier than from Pine Bluffs.

Seed water content as affected by imbibition temperature We placed diaspores of the 3 collections on moistened germination paper (Anchor Paper Corp., St. Paul, Minn.) over plastic slant-boards, and covered them with 1 layer of cellulose tissue (Jones and Cobb 1963). Slant-boards were then placed in closed germination boxes (25 x 40 x 20 cm), which were filled with distilled water to 3-cm depth. These boxes were placed in incubators at 0, 5, 10, and 20¡C in darkness. The experimental design was a randomized complete block with 3 replicates arranged over time. For each experimental unit (slant board), we retrieved 10 diaspores from incubators at 8-hour intervals until full Fig. 1. Water content (mean±SE) of winterfat seeds collected from Matador, Sask., Sterling, Colo., and Pine Bluffs, Wyo. after imbibition at 0, 5, 10, or 20¡C. See Table 2 for regression equations. hydration (1 to 8 days, depending on imbibition temperature). Bracts were removed and surface water of seeds was each collection. At the end of imbibi- Results blotted away with tissue paper. Seeds tion, embryos of half of the seeds were were then sealed in 0.25 ml tin capsules separated from other seed parts and the Seed water content as affected by (Leco Corp., St. Joseph, Mich.) and water contents of both seeds and imbibition temperature embryos were determined as described weighed (Booth and Bai 1998) before Initial seed water content was similar above. being oven dried at 80¡C for 24 hours among seed collections, averaging 4.8% and dry weight determined. Weighing (Fig. 1). However, the rate of water was done with a micro-balance to 0.001 Data analysis uptake during imbibition, time required mg and seed water content was Data were first analyzed with to reach full hydration, and seed water expressed on a dry weight basis (%DW). ANOVA or GLM in a randomized com- content when fully hydrated depended We defined full hydration as the stage 8 plete block design (Snedecor and on seed collection and imbibition tem- hours before germination began. Cochran 1980) over the 3 seed collec- perature. Water content of seeds at the tions, and then analyzed within each time of germination decreased with Comparison of water contents of collection because of the interaction increasing imbibition temperatures. As seeds and embryos at full hydra- between seed collection and imbibition expected, the rate of seed water uptake temperature. Data were further analyzed increased, and time required to reach tion separately for each imbibition tempera- The above imbibition procedures were full hydration decreased with increasing ture. Statistical significance was imbibition temperature. repeated in June 1995 (4 months after ≤ assumed at P 0.05 and means were sep- Seeds from Matador required more the above study), with 4 replicates for arated using LSD.

272 Journal of Range Management (52)3, May 1999 Table 1. Descriptions of sources and habitats for winterfat seeds used in the study.

Site Geographical location Vegetation and Seed weight dominant species (g 100 seeds-1) Matador Saskatchewan, Canada Mixed Prairie 0.25 50¡42'N, 107¡43'W, elev. 685 m Agropyron dasystachyum (Hook.) Scribn Agropyron smithii Rydb. Koeleria cristata Pers. Stipa viridula Trin. Eurotia lanata (Pursh) Moq. Sterling Colorado, USA Shortgrass Prairie 0.23 40¡37'N, 103¡13'W, elev. 1181 m Bouteloua gracilis (H.B.K.) Lag. Buchloe dactyloides (Nutt.) Engelm. Pine Bluffs Wyoming, USA Mixed Prairie 0.18 1¡10'N, 104¡09'W, elev. 1554 m Stipa comata Tri. & Rupr. Agropyron smithii Rydb. Bouteloua gracilis (H.B.K.) Lag. time to reach full hydration, particularly of Phase II during germination, or 24 in the embryo than in the endosperm at 0 and 5¡C, than those from Sterling or hours before germination started. Others was found in several species such as Pine Bluffs (Fig. 1). When seeds were imbibed seeds for given durations corn (Zea mays L.) (Styles 1948, imbibed at 0¡C, seed water content at regardless of the physiological stage, Vertucci 1989). The seed coat of winter- full hydration was greater for the and simply called them "imbibed" or fat is membranous and highly permeable Matador collection than the Sterling or "hydrated" seeds (Ishikawa and Sakai (Booth and McDonald 1994), enabling Pine Bluffs collections. At 5, 10, and 1982, Gray and Steckel 1983, Cremer seeds to absorb water quickly and store 20¡C, seed water content was highest and Mucha 1985). In the present study it in the endosperm. Nevertheless, seed for the Matador collection and lowest we defined full hydration as 8 hours hydration generally increases with for the Sterling collection. before germination started, or the end of increasing temperature (Shull 1920, The relationship between seed water Phase II. Seed water content at full Allerup 1958) and at warmer tempera- content and time after imbibition were hydration varied by seed collection, but tures the embryo may hydrate faster described in linear, quadratic and cubic was always greater with lower imbibi- than the endosperm. Variation in hydra- equations (Fig. 1, Table 2). Phases I and tion temperatures. The greater water tion rate explain the lower whole-seed II during imbibition (as defined by Ching content at 0¡C explains the warmer low- water content at higher temperatures 1972, 1973) were more distinct for seeds temperature exotherm for seeds imbibed than at lower temperatures. Therefore, imbibed at lower temperatures than those at lower temperatures than those water content of the whole seed may not imbibed at higher temperatures. imbibed at higher temperatures (Bai et accurately reflect the true water status al. 1998). related to seed germination. Instead, the Comparison of water contents of The fact that embryo water content water content of an embryo is a better measurement of germination readiness seeds and embryos at full hydra- was much lower than whole seed water content indicates that the rate of hydra- than that of the whole seed, at least for tion tion, and the water holding capacity, seeds similar to winterfat. Water content of seeds and embryos at vary among seed parts. Faster hydration full hydration was similar among collections (P = 0.26 and 0.17 for seeds and Table 2. Regression equations describing relationships between water content (Y, %DW) and time embryos, respectively) and data were after imbibition (X, hour) of seeds collected from Matador, Sask., Sterling, Colo., and Pine pooled (Fig. 2). Seed water content Bluffs, Wyo. Curves are presented in Fig. 1. again decreased with increasing imbibi- 2 tion temperatures, ranging from 169 to Seed Imbibition Regression equation R P-value Collection Temperature 231%. Water content of embryo decreased between 0 and 10¡C imbibi- Matador 0 Y=8.58+3.49X-0.0286X2+0.000085X3 1.00 0.000 5 Y=2.96+5.39X-0.0722X2+0.000364X3 0.99 0.000 tion temperatures, but increased at 20¡C, 2 with the range from 94 to 114%. 10 Y=5.94+5.13X-0.0477X 0.99 0.000 20 Y=2.85+8.35X-0.1600X2 0.99 0.000 Sterling 0 Y=6.66+2.31X-0.0096X2 0.99 0.000 Discussion 5 Y=3.75+3.03X-0.0201X2 0.99 0.000 10 Y=7.46+2.56X 1.00 0.001 20 Y=2.60+5.37X 1.00 0.001 The definition of full hydration of Pine Bluffs 0 Y=5.89+2.36X-0.0105X2 0.99 0.000 seeds has largely depended on 5 Y=2.45+3.30X-0.0198X2 0.99 0.000 researchers and the intervals used for 10 Y=4.92+4.38X-0.0367X2 0.99 0.000 seed water content determination. Keefe 20 Y=2.14+6.29X-0.0958X2 0.99 0.001 and Moore (1981) used the middle point

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 273 Dahal, P., N. Kim, and K.J. Bradford. 1993. Respiration and germination rates of tomato seeds at suboptimal temperatures and reduced water potentials. J. Exp. Bot. 47:941Ð947. Dettori, M.L., H.J. Ballitette, J.A. Young, and R.A. Evans 1984. Temperature profiles for germination of two species of winterfat. J. Range Manage. 37:218Ð222. Gray D. and J.R.A. Steckel. 1983. Freezing injury during seed germination: Its influence on seedling emergence in the onion (Allium cepa). Seed Sci. Tech. 11:317Ð322. Gray D., J.R.A. Steckel, and L.J. Hands. 1990. Responses of vegetable seeds to controlled hydration. Ann. Bot. 66:227Ð235. Gummerson, R.J. 1986. The effect of constant temperature and osmotic potentials on the germination of sugar beet. J. Exp. Bot. 37:729Ð741. Herner, R.C. 1986. Germination under cold soil Fig. 2. Water content of seeds and embryos at full hydration at 0, 5, 10, or 20¡C imbibition temper- conditions. HortSci. 21:1118Ð1122. ature for winterfat seeds collected from Matador, Sask., Sterling, Colo., and Pine Bluffs, Wyo. Hilton, J.W. 1941. Effects of certain micro-eco- Data were pooled for collections and means with the same letter within seed or embryo were not logical factors on the germinability and early significantly different at P≤ 0.05. development of Eurotia lanata. Northwest Sci. 15:86Ð92. Hobbs, P.R. and R.L. Obendorf. 1972. The dependence of seed water content Literature Cited Interaction of initial seed moisture and imbibi- at full hydration on imbibition tempera- tional temperature on germination and productivity of soybean. Crop Sci. 12:664Ð667. ture has been observed in seeds of other Allerup, S. 1958. Effect of temperature on species (C.W. Walters, personal com- uptake of water in seeds. Plant Physiol. Hunter, J.R. and A.E. Erickson. 1952. Relation munication), but has not, to our knowl- 11:99Ð105. of seed germination to soil moisture tension. Ashworth, E.N. and R.L. Obendorf 1980. Agron. J. 44:107Ð109. edge, been reported. The finding Ishikawa, M. and A. Sakai 1982. implies, the critical moisture content Imbibitional chilling injury in soybean axes: Relationship to stelar lesions and seasonal Characteristics of freezing voidance in com- concept (Hunter and Erickson 1952) environments. Agron. J. 72:923Ð928. parison with freezing tolerance: A demonstra- and the hydrothermal time model Bai, Y., D.T. Booth, and J.T. Romo. 1998. tion of extraorgan freezing. p. 325-340. In: (Gummerson 1986, Dahal et al. 1993) of Winterfat (Eurotia lanata (Pursh) Moq.) P.H. Li and A. Sakai (eds.). Plant cold hardi- seed germination must be modified seedbed ecology: Low temperature exotherms ness and freezing stress (Volume 2). Academic and cold-hardiness in hydrated seeds as influ- Press, New York, N.Y. based on embryo water content, and Jones, L.G. and R.D. Cobb 1963. A technique expanded to a wider temperature range, enced by imbibition temperature. Ann. Bot. 81:598Ð602. for increasing the speed of laboratory germina- particularly for temperatures near 0¡C . Booth, D.T. 1987. Contributions to the reproduction testing. Proc. Assoc. Official Seed What remains unanswered is the exis- tive autecology of winterfat (Eurotia lanata Analysts 53:144Ð160. tence of differences in embryo water (Pursh) Moq.) with notes on direct seeding Keefe, P.D. and K.G. Moore 1981. Freeze des- content among imbibition temperatures. methods. Ph.D. Diss., Univ. Wyo., Laramie, iccation: A second mechanism for the survival One possibility is that cold cell walls Wyo. of hydrated lettuce (Lactuca sativa L.) seed at Booth, D.T. 1992. Seedbed ecology of winterfat: sub-zero temperatures.Ann. Bot. 47:635Ð645. with lower energy level provide greater Imbibition temperature affects postgermination Shull, C.A. 1920. Temperature and rate of mois- resistance to cell extension and elonga- growth. J. Range Manage. 45:159Ð164. ture intake in seeds. Bot. Gazette 69:361Ð390. tion. Therefore, greater turgor pressure Booth, D.T. and Y. Bai 1998. Measuring mois- Snedecor, G.W. and W.C. Cochran. 1980. (higher water content) is required for ture content of small seeds. J. Range Manage. Statistical methods (7th ed.). Iowa State Univ. germination to proceed. 51:179Ð180. Press, Ames, Iowa. In summary, differences in seed water Booth, D.T. and M.B. McDonald. 1994. Cation Styles, I.E. 1948. Relation of water to the germi- concentration in post-imbibed winterfat seeds nation of corn and cotton seeds. Plant Physiol. contents were related to cold-hardiness as influenced by imbibition temperature. J. 23:201Ð222. of winterfat. Higher seed water content Range Manage. 47:485Ð488. Tetlyanova, A.A., R.I. Zlotin, and N.R.A. at full hydration was associated with Bradford, K.J. 1986. Manipulation of seed French. 1990. Changes in structure and func- lower imbibition temperatures, appar- water relations via osmotic priming to improve tion of temperature-zone grasslands under the ently a result of different hydration rates germination under stressed conditions. influence of man. p. 301-334. In: A. HortSci. 21:1105Ð1112. between embryo and endosperm. Breymeyer (ed.). Managed grasslands. Ching, T.M. 1972. Metabolism of germinating Elsevier, Amsterdam. Further studies using embryos instead of seeds. p. 103Ð218. In: T.T. Kozlowski (ed.). Vertucci, C.W. 1989. Relationship between ther- the whole seeds over a wider range of Seed biology (Volume 2). Academic Press, mal transitions and freezing injury in pea and temperatures are needed for modeling New York, N.Y. soybean seeds. Plant Physiol. 90:1121Ð1128. water relations for seed germination. Ching, T.M. 1973. Biochemical aspects of seed Woodmansee, R.G. and L.D. Potter. 1971. vigor. Seed Sci. Tech. 1:73Ð88. Natural reproduction of winterfat (Eurotia Cremer, K.W. and S.B. Mucha. 1985. Effects lanata). J. Range Manage. 24:24-30. of freezing temperatures on mortality of air- dry, imbibed and germinating seeds of euca- lyptus and radiata pine. Australian Forest Res. 15:243Ð251.

274 Journal of Range Management (52)3, May 1999 J. Range Manage. 52:275Ð282 May 1999 Grazing steer fecal output dynamics on south Texas shrubland

JERRY W. STUTH AND ROBERT K. LYONS

Authors are professor, Department of Rangeland Ecology and Management, Texas A&M University, College Station, Tex.. 77843-2126 and assistant professor and extension range specialist, Texas A&M Research and Extension Center, Uvalde, Tex. 77802-1849.

Abstract Resumen

Combined with other information, fecal output appears to Combinada con otra información, La producción fecal have potential use in models to predict forage intake. Better parece tener un uso potencial en los modelos para predecir el understanding of fecal output dynamics relative to forage consumo de forraje. Un mejor entendimiento de la dinámica availability could improve model estimates of animal perfor- de producción fecal relativa a la disponibilidad de forraje mance. Field trials were conducted during 4 different periods podría mejorar las estimaciones de los modelos acerca del to investigate the relationship between 1) declining forage comportamiento productivo animal. Se condujeron estudios mass or forage component availability and beef steer fecal de campo durante 4 diferentes períodos con el objetivo de output and between 2) browse consumption and available for- investigar la relación entre 1) la disminución de la masa de age mass. Fecal output was estimated using the rare-earth forraje o la disponibilidad del componente de forraje y el marker ytterbium. Initial fecal output as a percentage of rendimiento fecal de novillos para carne y 2) el consumo de body weight was greatest in March (1.24%) and least in forraje de arbustos y la masa disponible de forraje. La pro- August (0.96%). Regression slopes were negatively correlated ducción fecal fue estimada usando como marcador el Iterbio. (Ð0.73) with initial forage mass. As indicated by regression La mayor producción fecal inicial, expresada como porcenta- slopes, fecal output declined most rapidly in March (slope = je del peso corporal, se obtuvo en marzo (1.24%) y la menor 0.57) and slowest in August (slope = 0.13). Expression of en agosto (0.96%). Las pendientes de regresión fueron negati- available forage mass as either daily grass allowance or daily vamente correlacionadas (-0.73) con la masa inicial de forra- grass leaf allowance, both as g dry matter/kg live weight, pro- je. Como indicaron las pendientes de regresión, la producción duced similar regression equation statistics. Development of fecal disminuyó más rápidamente en marzo (pendiente = regressions for individual pastures within trials did, however, 0.57) y más despacio en agosto ( pendiente = 0.13). La expre- improve equation statistics in all trials except August. Browse sión de la masa de forraje disponible ya sea como la asi- consumption was <10% until daily grass allowance fell below gnación diaria de zacate o la asignación diaria del material 50 g/kg live weight then increased to between 53 and 64% foliar de zacate, ambas expresadas como g de materia seca/ below 25 g/kg live weight, but was not adequate to maintain kg de peso vivo, produjeron ecuaciones de regresión simi- fecal output. Apparent seasonal differences in fecal output lares. El desarrollo de regresiones para potreros individuales suggest lower forage intake (29%) in August compared to dentro de los ensayos mejoró la ecuación estadística en todos March. Fecal output was not affected by daily grass los ensayos excepto en agosto. El consumo de forraje de allowance above 100 g. Fecal output declined to below 0.6% arbustos fue <10% hasta que la asignación diaria de zacate of body weight below 100 g daily grass allowance. Data are cayó abajo de 50 g/kg de peso vivo, entonces, abajo de los 25 interpreted to suggest that different fecal output curves g/kg de peso vivo, se incrementó entre 53 a 64% pero no fue and/or adjustment factors may be needed to account for sea- adecuada para mantener la producción fecal. Las aparente son and initial forage mass. diferencias estacionales en la producción fecal sugieren un bajo consumo de forraje (29%) en agosto comparado con marzo. El rendimiento fecal no fue afectado cuando la asi- Key Words: herbage allowance, beef cattle, browse consump- gnación de forraje fue arriba de los 100 g. Cuando la asi- tion gnación diaria de zacate fue menos de 100 g, la producción fecal disminuyó a menos del 0.6% del peso corporal. Los Grazing animal forage intake is routinely estimated using datos son interpretados para sugerir que diferentes curvas de fecal output-forage indigestibility ratios (Langlands 1975, producción fecal y/o factores de ajuste pueden ser necesitados cuando para reportar por estación y masa de forraje inicial. Cordova et al. 1978). Fecal output can be relatively stable within a physiological stage (McCollum and Gaylean 1985, Ellis et al. 1988) or across a wide range of digestibility (Ellis et al. 1992), gut capacity or forage type (McCollum and et al. 1988) but can change with physiological stage (Sprinkle Gaylean 1985), and forage availability (McCollum and Gaylean 1985, McKown et al. 1991). Although fecal output Manuscript accepted 8 Aug. 1998. and body weight have been used to predict feed or roughage

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 275 intake (Conrad et al. 1964, Forbes used during this study. One set was used min at the beginning of each sample col- 1983), fecal output dynamics with in the March and May trials, a second lection in each trial. Ten total regard to forage availability need further set in August, and a third set in January. esophageal fistulated steers were avail- clarification for use in grazing animal Within each trial, each group of steers able for sampling. Sample collections models. Rayburn (1986) reported rapid- was 9 to 14 months old and weighed were staggered by one day for the 2 pas- ly declining intake for cattle below 240 to 340 kg. tures used within each trial. Between Dominant grasses in pastures included each collection event, collector steers 2,250 kg/ha forage availability; howev- purple threeawn (Aristida purpurea were grazed in pastures adjacent to and er, this threshold may vary from <3,000 Nutt.), plains bristlegrass [Setaria leu- with the same vegetation as trial pas- kg/ha (Allden and Whittaker 1970) to copila (Scribn. & Merr.) K.Schum.] and tures. Collector steers and their mothers <135 kg/ha (Handl and Rittenhouse Texas bristlegrass (Setaria texana were experienced with the vegetation in 1972). Vegetation characteristics, e.g., W.H.P. Emery), Texas tridens [Tridens the region before weaning. Collected leaf and stem proportions, may be more texanus (S. Wats.) Nash], hooded wind- extrusa was dried 48 hours at 60¡C and highly correlated to forage intake and millgrass (Chloris cucullata Bisch), buf- subjected to macrohistological analysis fecal output than gross forage mass. In felgrass (Cenchrus ciliaris L.), fall to determine steer diet composition by shrublands, these relationships may be witchgrass [Leptoloma cognatum plant group and plant part (Araujo affected by a large, diverse supply of (Schult.) Chase var. cognatum], Kleberg 1985). Diet samples were analyzed for browse. bluestem (Dichanthium annulatum crude protein (CP) content on a dry mat- Stapf) and Hall panicum (Panicum hallii ter basis by micro-Kjeldahl procedure This research was conducted 1) to Vasey var. hallii). Common woody (AOAC 1960). Digestible organic mat- examine relationships between available plants occurring in pastures were ter (DOM) was determined by in vitro forage mass and fecal output and between mesquite (Prosopis glandulosa Torr.), procedures using a 48-hour fermentation forage components and fecal output and spiny hackberry (Celtis pallida), lime (Tilley and Terry 1963) followed by 2) to determine how browse consumption pricklyash (Zanthoxylum fagara (L.) neutral detergent fiber procedure (Van relates to available forage mass. Sarg.), coyotillo [Karwinskia humbold- Soest and Wine 1967). tiana (R.& S.) Zucc], desert yaupon Grass standing crop components (live (Schaefferia cuneifolia Gray), shrubby leaf, dead leaf, live stem, and dead stem) Methods blue sage (Salvia ballotaeflora Benth.), were estimated using extrusa grass com- and whitebrush [Aloysia gratissima ponent composition, grass standing crop This study was conducted on the La (Gill. & Hook.) Troncoso]. Plant growth disappearance, and initial grass standing Copita Research Area (27¡ 40'N, 98¡ initiates in late March with a bimodal crop mass. Grass component mass was 12'W) in northeastern Tamaulipan peak of grass standing crop in June and estimated by multiplying percent grass Province, about 80 km W of Corpus October depending on rainfall patterns. component in extrusa obtained at each Christi, Tex. To establish contrasting Browse standing crop peaks in July and sample collection by the amount of forage conditions, four 0.4-ha pastures remains stable for most species through grass disappearance occurring during were established on a gray sandy loam early November. the interval between that sample collec- site that had been chained 7 years earli- Five sample collections were conduct- tion and the next and summing interval er. Two of the 4 pastures (A and B) ed during each trial except in January values for each component over the were chained only while the other 2 (C when only 4 were possible. Sample col- entire trial. These totals were used to and D) were also sprayed 3 years before lections occurred at the beginning and calculate initial mass and initial percent- the study with picloram at a rate of 1.0 end of each trial and at 5-day intervals age of each grass component and grass kg active ingredient/ha to create grazing during the trial. Twelve, 1 x 30 m belt component composition of the residual conditions with high grass production transects were established in each pad- grass throughout the trial. Herbage dis- and low available browse. dock to characterize available forage appearance between clipping dates was Four grazing trials were conducted throughout each trial. Available browse assumed to be equal to consumption. during seasons representing different standing crop was measured in each belt Little or no growth occurred during the phenological stages. Trials were con- transect using the browse volume- 5 days intervals between sampling dates ducted in 1) March,1985; 2) May, 1985; weight method (Lopes and Stuth 1984). within each trial. Because available 3) August, 1985; and 4) January, 1986. Two, 0.5 x 1 m quadrats were randomly grass was so low during the last interval Two pastures (1 chained only and 1 located in each of the 12 belt transects of each trial, composition of the grass chained and sprayed) were grazed by 6 and herbage composition by weight standing crop was assumed to be pro- Beefmaster steers each for 21 days or visually estimated and all herbaceous portional to extrusa sample grass com- until utilization of the grass standing aboveground biomass subsequently ponent composition at the beginning of crop was judged to be 90% during each clipped to ground level. Standing crop this interval. trial. Pastures A (chained only) and C (kg/ha) was determined by species by To estimate fecal output, non-cannu- (chained and sprayed) were grazed in multiplying estimated species composi- lated steers used to graze pastures were March and August while pastures B tion by total clipped herbage for each dosed daily with the rare-earth element (chained only) and D (chained and sample collection within each trial. ytterbium in powdered acetate form via sprayed) were grazed in May and Four, esophageal fistulated steers (270 gelatin capsules. Steers were dosed daily January, allowing 150 to 210 days kg, 9 months of age) per pasture were beginning 9 days before each trial. between trials. Three sets of steers were allowed to graze for approximately 30 During this pre-trial period, these steers

276 Journal of Range Management (52)3, May 1999 Table 1. Mean and standard error for initial and end of trial forage crude protein (CP, dry matter basis) and digestible organic matter (DOM) for March, May, August, and January trials by pasture.

CP, DOM, % Pasture A or B1 Pasture C or D2 Pasture A or B Pasture C or D Trial Initial Ending Initial Ending Initial Ending Initial Ending ------(%) ------(%) ------Mar. 16.6±0.2 12.1±0.5 17.0±0.3 10.5±0.2 65.2±1.2 34.0±1.3 67.5±1.2 36.0±3.0 May 12.1±0.2 12.4±0.4 11.5±0.2 9.9±0.3 74.8±0.5 46.0±1.2 71.8±0.5 45.0±2.5 Aug. 9.8±0.2 9.2±0.4 9.3±0.2 7.6±0.3 74.8±0.5 44.0±1.2 71.4±1.1 46.0±2.5 Jan. 11.6±0.2 10.2±0.8 11.5±0.3 8.9±0.1 70.6±2.5 38.0±5.0 69.7±0.5 34.0±2.0 1Pastures A & B were chained 7 years before this study. 2Pastures C & D were chained and then sprayed with picloram at 7 and 3 years, respectively, before this study. n=4 were grazed in pastures adjacent to and Regression and correlation analyses als. Generally, DOM levels declined with the same vegetation as trial pastures. (SAS 1988) were performed across and more drastically than CP levels. These Because vegetation was the same in these within trials to determine the relation- dietary trends reflect a shift toward adjacent pre-trial and trial pastures, fecal ship between available forage, diet com- browse as available grass declined. output estimates should not have been position, and dietary browse consump- affected. Each trial pasture was grazed by tion and fecal output on a dry matter Daily Grass or Grass Leaf 6 steers. Fecal samples were collected basis as a percentage of body weight. Allowance and Fecal Output from each steer 24 hours after each diet Non-log functions were tested first in Initial fecal output was associated collection and analyzed for ytterbium regressions. Log functions were then with a range of ungrazed grass standing concentration using atomic absorption used in regressions because of the crop of 1500 to 3000 kg/ha with lowest spectroscopy (Ellis et al. 1982). apparent visual trends in the data. herbage mass in March and highest in Galyeanet et al. (1986) reported that over August (Table 2). Initial fecal output as a variety of experimental conditions, Results and Discussion a percent of body weight differed among average continuous-dose marker esti- trials (P<0.10) from a high in March mates of fecal output using ytterbium (1.24%), to a low in August (0.96%), were 104% of total fecal collection with Diet Quality with May (1.09%) and January (1.08%) a range of 86 to 144%. Musimba et al. In general both CP and DOM declined between these extremes. Initial fecal (1987) suggested that this technique is during each trial as quantity of forage output in May, January, and August was most useful for relative comparisons. decreased. Highest CP levels occurred 88, 87, and 77% of initial March fecal Because marker recovery was not deter- in March and lowest levels in August output, respectively. mined, results are valuable because they (Table 1). Initial CP levels were similar Using stepwise regression, the log may indicate general relationships and between pastures within trials. Ending function of grass standing crop areas for future investigation. Steers were CP levels tended to be higher in pastures explained 74 to 94% of the variation in weighed at the beginning of each trial at A and B (chained only) with more avail- fecal output for individual trials (Table 0600 hours after penning at 2000 hours able brush. Initial and ending DOM lev- 3). Except for January, the log of grass the previous evening to determine fecal els were relatively similar between pas- standing crop was the only variable output on a percent body weight basis. tures within trials, but varied among tri-

Table 2. Initial standing crop (kg/ha ± standard error) of grass, forbs, and browse and initial percent grass standing crop component by trial and pastures.

Initial Standing Crop Trial Pasture Grass Forbs Browse Pasture Grass Forbs Browse ------(kg/ha-1) ------Mar. A1 1040±87 163±22 1976±28 C2 1850±114 101±15 632±15 May B1 1512±183 700±107 2200±36 D2 2788±395 164±20 1305±29 Aug. A 1256±113 664±131 2986±46 C 3005±413 356±149 1808±38 Jan. B 765±100 239±43 839±16 D 1578±179 71±27 769±19 Initial Grass Standing Crop Component Live leaf Dead leaf Live stem Dead stem A/B C/D A/B C/D A/B C/D A/B C/D ------(%) ------

Mar. 68 34 13 15 7 8 12 43 May 43 53 15 9 16 18 26 20 Aug. 31 28 24 27 25 9 20 36 Jan. 9 10 47 26 6 26 38 38 1Pastures A & B were chained 7 years before this study. 2Pastures C & D were chained and then sprayed with picloram at 7 and 3 years, respectively, before this study.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 277 Fig. 1. Influence of daily grass allowance, g dry matter (DM)/kg live weight (LW) on marker-estimated fecal output of steers. Predictive equations are based on the log of daily grass allowance with pastures A and C or B and D combined for March, May, August, and January. Pastures A & B were chained 7 years before this study. Pastures C & D were chained and then sprayed with picloram at 7 and 3 years, respectively, before this study. Endpoints of lines for each trial indicate the grass allowance range. selected. The strongest regression rela- ences in marker recovery, diet quality crop and daily grass allowance than pas- tionships between grass standing crop differences or an unwillingness of ani- tures A and B (chained only) in all trials and fecal output occurred in March and mals to shift from highly desirable but (Table 2, Fig. 1). In all trials except August (Table 3). minimally occurring and rapidly declin- August, analysis of daily grass The relationship between grass standing forage components, e.g., live leaf in allowance-fecal output relationships by ing crop and fecal output does not inte- March. August live:dead leaf ratio was individual pastures improved regression grate pasture size, animal number, and about 1:1 compared to >2:1 in March equation statistics for one or both pas- animal size; therefore, the relationship and May, and < 1:2 in January. Mnene tures (Table 4). Initial daily grass between fecal output and daily grass et al. (1996) noted reduced fecal output allowance was negatively correlated allowance, g dry matter/kg live weight, when cattle diets predominated by dead (Ð0.73) with regression equation slopes. was examined for each trial using the leaf shifted to diets predominated by In all trials except March, regression log function of grass allowance. Across live leaf available at low levels. equation slopes for pastures A and B trials, initial fecal output corresponded Pastures C and D (chaining plus piclo- were greater than those for pastures C to daily grass allowances of 100 to 200 ram) had a higher initial grass standing and D indicating a more rapid decline in g/kg live weight (Fig. 1) with lower levels occurring in January and accompa- Table 3. Stepwise regression analysis of log functions of kg grass (LGKG), forb (LFKG), and nied by a fecal output decline in pasture browse (LBKG) standing crops and fecal output as a percent of body weight.

B (chained only) almost twice that in 2 2 pasture D (chained and sprayed, Table Trial Variable Partial R Model R Probability 4). As indicated by slopes of the regres- Mar. LGKG 0.94 0.94 0.0001 sion functions (Fig. 1), daily grass May LGKG 0.75 0.75 0.0055 allowance had the least impact on fecal Aug. LGKG 0.84 0.84 0.0002 output in August and the greatest impact Jan. LGKG 0.74 0.74 0.0060 in March. Perhaps this difference in LFKG 0.16 0.90 0.0391 regression slopes indicates either differ- LBKG 0.06 0.96 0.0640

278 Journal of Range Management (52)3, May 1999 Fig. 2. Effect of daily grass allowance, g dry matter (DM)/kg live weight (LW) on dietary browse content. Predictive equations are based on the log of daily grass allowance with pastures A and C or B and D combined. Pastures A & B were chained 7 years before this study. Pastures C & D were chained and then sprayed with picloram at 7 and 3 years, respectively, before this study. fecal output in pastures A and B with unpublished data). Redmon et al. (1995) mental temperatures. Apparent lower lower initial grass allowances. These suggested similar plateau levels for initial fecal output in May could also be lower initial grass allowances probably daily forage allowance on wheat. related to high environmental tempera- resulted in an inability to maintain Regression of the log of daily grass leaf tures. Because animals were in the ther- rumen fill. Although not consistent allowance, g dry matter/kg live weight moneutral zone during the January trial, across trials, McCollum and Galyean and fecal output by trial and by pasture fecal output values were affected by fac- (1985) reported greater fecal output in within trials resulted in little or no tors other than temperature. conjunction with increased undigested improvement in equation statistics over Differential preferences among for- fill, which was attributed to either grass allowance equations (Table 4). ages with contrasting phenologies may expanded gut capacity or the forb com- Initial fecal output was associated with also account for the apparent lower ini- ponent of the diet. daily grass leaf allowances of 30 to 120 g. tial fecal output levels in May, August, In the present study, the greatest One potential explanation for lower and January. If steers consumed less for- decline in fecal output appears to have initial fecal output in August is the age and less indigestible fill, fecal out- occurred below 100 g of daily grass influence of high environmental temper- put could be depressed. Estimated dead allowance. Ellis et al. (1984) reported atures. Average maximum temperature leaf in the initial forage mass increased that a progressive reduction in daily was 26¡C in March, 32¡C in May, 38¡C almost 2 to 3 fold in August and herbage allowance of ryegrass from 700 in August, and 22¡C in January. January, respectively, compared to to 100 g dry matter/kg body weight had Depressions in forage intake have been March and May (Table 2). no significant effect on daily fecal out- associated with high environmental tem- Diet quality, specifically with regard put. The NRC (1987) intake function peratures (NRC 1987). It seems logical to protein, is also a potential explanation suggests that 100% forage intake is that, like depressed fecal output associ- for some of the apparent differences in achieved above 200 g daily dry matter ated with low forage availability fecal output observed in this study. forage allowance/kg body weight. On (McCollum and Galyean 1985, Milford and Minson (1965) reported wheat pastures, steers have exhibited McKown et al. 1991), depressed fecal depressed intake in relation to CP levels increased gain with increasing daily for- output could be expected if forage below 7%. In the present study, no CP age allowance up to 250 g (Pinchak intake were depressed by high environ- levels were below this threshold (Table

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 279 Table 4. Regression comparisons for fecal output as a percent of body weight and daily grass were minor components of the standing allowance (GA, g dry matter/kg live weight), daily grass leaf allowance (GL, g dry mattter/kg live crop. Although these observations may weight) and most highly correlated single grass species across pastures and by pastures. not have general application, we believe they indicate that partitioning the total Across Pastures GA GL Species Trial GL A/B1 C/D2 A/B C/D A/B C/D grass standing crop can provide greater understanding of fecal output and intake Mar. r2 0.71 0.84 0.99 0.87 0.96 0.97 0.99 dynamics. The tradeoff between diet SEy 0.13 0.10 0.03 0.09 0.05 0.05 0.004 selection and acquisition of dry matter Slope 0.46 0.54 0.54 0.61 0.46 0.74 0.60 can be seen from these analyses. Int3 0.32 0.006 0.11 0.004 0.41 0.19 0.42 RT4 log log log log log log log May Dietary Components and Fecal r2 0.76 0.65 0.98 0.71 0.98 0.91 0.99 Output SEy 0.07 0.10 0.007 0.09 0.007 0.05 0.0005 Slope 0.27 0.35 0.13 0.28 0.09 0.33 0.01 Correlation analysis of fecal output Int 0.52 0.29 0.75 0.50 0.85 0.69 0.98 and dietary components indicated differ- RT log log log log log log linear ences among trials (Table 5). During Aug. March, fecal output was positively cor- 2 r 0.86 0.88 0.87 0.89 0.85 0.86 0.90 related (0.67) with live grass stem and SEy 0.04 0.05 0.03 0.05 0.03 0.06 0.03 negatively correlated (Ð0.60) with live Slope 0.15 0.21 0.13 0.18 0.12 0.20 0.12 Int 0.63 0.50 0.62 0.61 0.66 0.67 0.75 browse stem. May fecal output was pos- RT log log log log log log log itively correlated with live grass leaf Jan. (0.87) and negatively correlated with r2 0.80 0.99 0.91 0.99 0.80 0.99 0.99 live forb leaf (Ð0.76), live forb stem SEy 0.08 0.02 0.05 0.02 0.07 0.03 0.002 (Ð0.91), and live browse leaf (Ð0.63). Slope 0.25 0.54 0.30 0.35 0.17 0.06 0.40 Int 0.61 0.12 0.32 0.59 0.67 0.58 0.95 August fecal output was correlated RT log log log log log linear log (0.78) with live grass leaf. In January, 1Pastures A & B were chained 7 years before this study. dead grass leaf was the only dietary 2Pastures C & D were chained and then sprayed with picloram at 7 and 3 years, respectively, before this study. component correlated (0.89) with fecal 3Int = intercept output. Penning et al. (1994) reported 4RT = regression type that bite mass was more highly correlated (0.82) with green leaf mass than any 1). McCollum (1995) suggested that for- Individual Grass Species and other sward measurement. Only in the ages below 10% CP may be deficient in Fecal Output March trial was grass leaf, live or dead, ruminally degradable protein. Improved Within each trial and pasture, correla- not correlated with fecal output. In this intake has been reported in relation to tions were calculated between grass trial, dietary live leaf content tended to ruminally degradable protein (Hannah et standing crop of each species and fecal remain high despite declining forage al. 1991, Lintzenich et al. 1995, Köster output. Standing crop of the single grass availability. This relatively high level of et al. 1996). Initial dietary CP was species most highly correlated with live leaf is probably an indication of ani- below 10% in the August trial. It is con- fecal output was used to develop log or mal drive to consume leaf over stem and ceivable that a lack of ruminally degrad- linear regression equations. Except for live over dead tissue relative to quantity able protein may have contributed to a August, where no improvement was of leaf available per tiller. depressed intake and the apparent observed, single species equation stan- For the 3 trials in which grass leaf, depression in initial fecal output in dard error of estimate was reduced by live or dead, was correlated with fecal output, maximum observed fecal output August. Escape protein (Donaldson et half to an order of magnitude compared was achieved when leaf made up 50 to al. 1991) and amino acid profile (Hill to equations using grass allowance or 70% or more of the diet. However, and Ellis 1991) have also been reported grass leaf allowance (Table 4). Only in to function in intake control. The poten- when grass leaf fell below 10% in the the March chained pasture did the single diet, fecal output was depressed by 25 to tial influence of these factors should not species used in these regressions rank be overlooked. Although ending dietary 40%. In all trials, maximum observed first in standing crop. In most cases, fecal output was attained only when CP was above 10% in all trials except these species ranked second or third in total dietary grass content (leaf and August, these values were influenced by standing crop and in 2 instances they stem) was near 100%. dietary browse (Fig. 2). South Texas browse species CP levels are reported Table 5. Correlation of diet composition (%) of total grass (TG); live leaf (GLL), dead leaf (GDL) in the 10 to 30% range (Taylor et al. and live stem (GLS); forb live leaf (FLL) and live stem (FLS); and browse live leaf (BLL) and 1997). However, recent evidence sug- live stem (BLS) to fecal output as a percent of body weight. gests (Barnes et al. 1991) that crude protein values overestimate the nutritional Trial TG GLL GDL GLS FLL FLS BLL BLS value of these species. Therefore, it is Mar. 0.68 0.10 0.50 0.67 0.06 0.25 -0.45 -0.60 possible that crude protein, ruminally May 0.94 0.87 0.48 0.24 -0.76 -0.91 -0.63 -0.58 degradable protein, escape protein, Aug. 0.64 0.78 0.34 0.16 -0.36 -0.21 -0.22 -0.22 and/or amino acids were deficient Jan. 0.74 0.17 0.89 0.48 -0.26 -0.54 -0.55 -0.27 toward the end of the trials.

280 Journal of Range Management (52)3, May 1999 Table 6. Correlation of grass standing crop Conclusions Diss., Texas A&M Univ., College Station, (GSC; kg/ha) and fecal output as a percent of Tex. body weight (FO) or dietary browse content Barnes, T.G., L.H. Blankenship, L.W. (DBC; %) by trial. Whether due to environmental condi- Varner, and J.F. Gallagher. 1991. tions, differential forage quality, or dif- Digestibility of guajillo for white-tailed Correlation (r) by Trial ferential forage preferences, apparent deer. J. Range Manage. 44:606Ð610. Relationship Mar. May Aug. Jan. differences in fecal output among sea- Conrad, H.R., A.D. Pratt, and J.W. Hibbs. GSC vs FO 0.91 0.81 0.79 0.84 sons found in this study could reflect 1964. Regulation of feed intake in dairy GSC vs DBC -0.71 -0.76 -0.50 -0.51 important differences in forage intake. cattle: I. Change in importance of physical FO vs DBC -0.70 -0.92 -0.75 -0.43 Assuming equal dry matter digestibility and physiological factors with increasing digestibility. J. Dairy Sci. 47:54Ð62. and using March and August fecal out- Cordova, F.J., J.D. Wallace, and R.D. Dietary Browse, Fecal Output, and puts, forage intake for a 300 kg steer Pieper. 1978. Forage intake by grazing would be 29%, 29%, and 16% lower in livestock: a review. J. Range Manage. Daily Grass Allowance August than March at 1) initial fecal 31:430Ð438. Both grass standing crop and fecal out- output levels, 2) 100 g, and 3) 50 g daily Donaldson, R.S., M.A. McCann, H.E. put were negatively correlated with grass allowance/kg live weight, respec- Amos, and C.S. Hoveland. 1991. Protein dietary browse content (Table 6). Daily tively. In reality, forage quality would and fiber digestion by steers grazing win- grass allowance levels at which dietary be expected to decline with rapidly ter annuals and supplemented with ruminal browse levels increased corresponded to declining forage availability, further escape protein. J. Anim. Sci. where declining fecal output was 69:3067Ð3071. diminishing forage intake and animal Ellis, W.C., M.J. Wylie, and J.H. Matis. observed (Fig. 1 and Fig. 2). These rela- nutritional status. Although potentially tionships indicate that 1) cattle selected 1988. Dietary-digestive interactions deter- important, the rapid declines in fecal mining the feeding value of forages and browse in significant amounts only when output that occurred below 100 g daily roughages. 1988. p. 177Ð229. In: ¯rskov, grass standing crop became limiting and grass allowance would most likely only E.R. (ed.): Feed Science. Elsevier Sci. 2) browse consumption was not ade- be observed under extremely high stock Publ. New York, N.Y. quate to maintain fecal output. Seasonal densities. Rapid declines in fecal output Ellis, W.C., C. Lascano, R. Teeter, and differences in the correlations between associated with lower initial daily grass F.N. Owens. 1982. Solute and particulate grass standing crop and fecal output sug- allowance emphasize the need for lower flow markers. pp. 37Ð56 In: Protein gest that forage characteristics other than Requirements for Cattle. Oklahoma State stock densities under these conditions. It Univ. Misc. Pub. 109. Stillwater, Okla. simply grass standing crop warrant study appears that, if stock density is main- to gain a clearer understanding of factors Ellis, W.C., J.P. Telford, H. Lippke, and tained so that daily grass allowances are M.E. Riewe. 1984. Forage and grazing influencing fecal output. above 100 g dry matter/kg live weight, effects on intake and utilization of annual To better understand these relation- forage availability will have minimal ryegrass by cattle, p. 223Ð34. In: G.W. ships in shrubland situations, daily grass effect on fecal output and on forage Horn (ed.), National Wheat Pasture allowance was compared to dietary intake. Using the dry matter digestibility Symposium Proceedings. Oklahoma Agr. browse content (Fig. 2) using the log of assumptions above, estimated forage Exp. Sta. Pub. Misc. Pub. 115. daily grass allowance. Browse com- intake depressions at about 100 g daily Forbes, J.M. 1983. Models for the predic- prised less than 10% of steer diets until tion of food intake and energy balance in grass allowance were 3 to 8% below dairy cows. Livestock Prod. Sci. daily grass allowance fell below about that calculated using initial fecal output. 50 g/kg live weight. Prediction equa- 10:149Ð157. We interpret the data to indicate that dif- Galyean, M.L., L.J. Krysl, and R.E. Estell. tions showed an exponential increase in ferent fecal output curves or adjustment 1986. Marker-based approaches for esti- browse consumption rising to nearly factors may be needed to characterize mation of fecal output and digestibility in 60% of steer diets as daily grass fecal output relative to 1) seasonal dif- ruminants, p. 96Ð113. In: F.N. Owens allowance dropped below 25 g/kg live ferences and 2) different initial grass (ed.): Feed intake of beef cattle symposium. Oklahoma State Univ. Misc. Pub. weight. Highest actual browse consump- standing crops and grazing pressure. tion was 64% in January, followed by 121. Stillwater, Okla. Handl, W.P. and L.R. Rittenhouse. 1972. 58, 57, and 53% in March, May, and Herbage yield and intake of steers. Proc. August, respectively. These browse lev- Literature Cited West. Sec. Amer. Soc. Anim. Sci. els occurred at actual daily grass 23:197Ð200. allowance levels less than 25 g/kg live Allden, W.G. and I.A.McD. Whittaker. Hannah, S.M., R.C. Cochran, E.S. weight and mostly in pastures A and B 1970. The determinants of herbage intake Vanzant, and D.L. Harmon. 1991. (chained only). When contrasted by by grazing sheep: the interrelationship of Influence of protein supplementation on trial, browse consumption had the factors influencing herbage intake and site and extent of digestion, forage intake, availability. Australian. J. Agr. Res. and nutrient flow characteristics in steers strongest relationship with daily grass consuming dormant bluestem-range for- allowance in May followed by March, 21:755Ð766. AOAC. 1960. Micorchemical methods: age. J. Anim. Sci. 69:2624Ð2633. January and August in that order (Fig. Hill, T.M. and W.C. Ellis. 1991. Effect of 2). The shift to browse occurred earliest Nitrogen. pp. 643Ð644. In: Official methods of analysis. 9th ed. Assoc. Official fish meal or feather meal supplementation in May and latest in August, probably Agr. Chem. Washington, D.C. on intake, digestion and weight gains of because May browse offered a reasonable Araujo, M.R. 1985. Dietary selection by calves grazing bermudagrass. p. 35-36. In: alternative food, but August offered less cattle as influenced by stocking rates in a Beef cattle research in Texas. Texas A&M alternative forage because of leaf drop. short duration grazing system. Ph. D. Univ., College Station, Tex.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 281 Köster H.H., R.C. Cochran, E.C. McCollum, F.T. and M.L. Galyean. 1985. Redmon, L.A., F.T. McCollum III., G.W. Titgemeyer, E.S. Vanzant, I. Cattle grazing blue grama rangeland II. Horn, M.D. Cravey, S.A. Gunter, P.A. Abdelgadir, and G. St. Jean. 1996. Seasonal forage intake and digesta kinet- Beck, J.M. Mieres, and R. San Julian. Effect of increasing degradable intake pro- ics. J. Range Manage. 38:543Ð546. 1995. Forage intake by beef steer grazing tein on intake and digestion of low-quality, McKown, C.D., J.W. Walker, J.W. Stuth, winter wheat with varied herbage tallgrass-prairie forage by beef cows. J. and R.K. Heitschmidt. 1991. Nutrient allowances. J. Range Manage. Anim. Sci. 74:2473Ð2481. intake of cattle on rotational and continu- 48:198Ð201. Langlands, J.P. 1975. Techniques for esti- ous grazing treatments. J. Range Manage. Sprinkle, J.E., D.D. Kress, D.E. Dornbos, mating nutrient intake and its utilization 44:596Ð601. D.C. Anderson, M.W. Tess, R.P. by the grazing ruminant. p. 320Ð332. In: Mnene, W.N., J.W. Stuth, and R.K. Z.W. McDonald and A.C.I. Warner (eds.), Ansotegui, B.E. Olson, and N.J. Roth. Lyons. 1996. Effects of herbage and bush Digestion and metabolism in the ruminant. 1992. Fecal output of different biological Univ. of New England Pub. Unit. Sydney, level on diet selection and nutrient intake type of beef cattle on native range Australia. of cattle in a Commiphora savanna. throughout a production year. Proc., Lintzenich, B.A., E.S. Vanzant, R.C. Tropical Grassl. 30:378Ð388. Western Section, Amer. Society of Anim. Cochran, J.L. Beaty, R.T. Brandt, Jr., Musimba, N.K.R., M.L. Galyean, J.L. Sci. 43:31Ð34. and G. St. Jean. 1995. Influence of pro- Holechek, and R.D. Piper. 1987. SAS. 1988. Users Guide: Statistics. SAS cessing supplemental alfalfa on intake and Ytterbium-labeled forage as a marker for Inst., Inc. Cary, N.C. digestion of dormant bluestem-range for- estimation of cattle fecal output. J. Range Taylor, R. B., J. Rutledge, and J.G. age by steers. J. Anim. Sci. 73:1187-1195 Manage. 40:418Ð421. Herrea. 1997. A field guide to common Lopes E.A., and J.W. Stuth. 1984. Dietary NRC. 1987. Predicting feed intake of food- South Texas shrubs. Texas Parks and selection and nutrition of spanish goats as producing animals. National Academy Wildlife Press. Austin, Tex. influenced by brush management. J. Range Press. Washington, D.C. Tilley, J.M.A., and R.A. Terry. 1963. A Manage.37:554Ð560. Penning P.D., A.J. Parsons, R.J. Orr, and two-stage technique for the in vitro diges- Milford, R. and Minson, D.J. 1965. Intake G.E. Hooper. 1994. Intake and behaviour tion of forage crops. J. Brit. Grassl. Soc. of tropical pasture species. Proc. 9th Int. responses by sheep to changes in sward 18:104Ð111. Grassl. Cong. 1:815Ð822. characteristics under rotational grazing. Van Soest, P.J., and R.H. Wine. 1967. Use McCollum, F.T. 1995. Ruminal degradabili- Grass and Forage Sci. 49:476Ð486. ty of supplemental protein for grazing beef of detergents in the analysis of fibrous Rayburn, E.B. 1986. Quantitative aspects of feeds. IV. Determinations of plant cell- cattle, p. 177Ð189. Proc. Southern Pasture pasture management. Seneca Trail RC&D Forage Crop Improvement Conf. May 23- wall constituents. J. Assoc. Official Agr. Technical Manual. Seneca Trail RC&D. 25, 1994, Knoxville, Tenn. USDA-ARS, Chem. 50:50Ð55. New Orleans, La. Franklinville, N.Y.

282 Journal of Range Management (52)3, May 1999 J. Range Manage. 52:283Ð289 May 1999 Cattle use affects forage quality in a montane riparian ecosystem

REBECCA L. PHILLIPS, M.J. TRLICA, WAYNE C. LEININGER, AND WARREN P. CLARY

Authors are research assistant, Graduate Degree Program in Ecology, Colorado State University; professor and associate professor, Rangeland Ecosystem Science Department, Colorado State University, Fort Collins , Colo. 80523; and project leader, Intermountain Research Station, Boise, Idaho 83702.

Abstract Resumen

Forage nitrogen (N) and phosphorous (P) concentrations Se midió la concentración de nitrógeno (N) y fósforo (P) del and in-vitro dry-matter digestibility (IVDMD) were mea- forraje y la digestibilidad in vitro de la materia seca sured in 2 important riparian species the year following (DIVMS) de 2 importantes especies ribereñas. Las short-term, high-intensity cattle grazing treatments in a mon- mediciones fueron hechas al año siguiente de aplicar con tane riparian ecosystem in northcentral Colorado. Current ganado los tratamientos de apacentamiento de corta duración year’s growth of water sedge (Carex aquatilus Wahlenb.) and y alta intensidad en un ecosistema ribereño de montaña de la planeleaf willow (Salix planifolia Pursh.) was collected month- región norcentral de Colorado. El crecimiento de "water ly from May to September 1996. The effects of grazing and sedge" (Carex aquatilus Wahlenb.) y "planeleaf willow" season of grazing in 1995 on forage quality the following (Salix planifolia Pursh.)se registró mensualmente de mayo a growing season was determined. Season of grazing (i.e., late- septiembre de 1996. Se determinaron los efectos que el spring, early-summer, late-summer, and fall) the previous apacentamiento y la época de apacentamiento de 1995 year did not differentially affect forage quality in either tuvieron en la calidad del forraje de la siguiente estación de species. However, grazing by cattle the previous year did crecimiento. La época de apacentamiento (por ejemplo, fin de increase forage quality of water sedge as compared with primavera, inicio de verano, fin de verano y otoño) del año plants that were not previously grazed. Grazed water sedge anterior no afecto la calidad de forraje de ninguna de las plants had higher concentrations of N and P and greater especies. Sin embargo, el apacentamiento con ganado en el IVDMD than ungrazed controls. Nitrogen and P concentra- año anterior incrementó la calidad del forraje de "water tions of browsed planeleaf willow were not different from sedge" ya que fue superior a la de plantas que no fueron controls, but current year’s growth collected in the fall from apacentadas anteriormente. Las plantas apacentadas de previously browsed plants was 11% more digestible than cur- "water sedge" tuvieron mayores concentraciones de rent year’s growth from non-browsed willow. The 2 species nitrógeno y fósforo y mayor DIVMS que las plantas control responded uniquely to cattle use, which suggested that these 2 sin apacentamiento. La concentración de N y P de plantas life forms differ in response to herbivory. This study support- ramoneadas de "planeleaf willow" fueron similares a las de ed the hypothesis that grazing by cattle would improve forage las plantas control, pero el forraje colectado en otoño de quality in a riparian ecosystem, although results varied with plantas anteriormente ramoneadas fue 11% más digestible en life form. comparación del forraje producido en la misma época por plantas intactas de "planeleaf willow". Las 2 especies Key Words: Water sedge, Carex aquatilus, planeleaf willow, respondieron en forma única al uso por el ganado, lo cual sugiere que estas dos formas de vida responden en forma Salix planifolia, nitrogen, phosphorous, in-vitro dry-matter diferente a la herbívora. Este estudio soporta la hipótesis de digestibility que el apacentamiento por ganado podría mejorar la calidad del forraje de los ecosistemas ribereños, aunque los resulta- Previous studies have shown that large herbivores utilize dos varían con la forma de vida. riparian areas disproportionately heavy relative to upland areas (Roath and Krueger 1982, Platts and Nelson 1985). Heavy grazing might change plant species composition, production, stand density, vigor, and seed production (Ryder 1980), and several investigators have observed changes in riparian vegetation biomass, height, composition, and cover in Research was funded by the U.S. Forest Service Research Joint Venture response to specific grazing regimes (Kauffman et al. 1983, Agreement INT-96080-RJVA and the Colorado State University Agricultural Schulz and Leininger 1990, Popolizio et al. 1994, Clary Experiment Station. 1995). Nutritive quality of forage species following grazing The authors would like to thank Dr. James Detling for his critical review of the has not been determined for montane riparian ecosystems, and manuscript. The U.S. Forest Service is recognized for granting permission to use land and other resources for this study. We also thank Barbara Oskroba for her data that indicate how livestock grazing affects forage quanti- assistance and use of facilities for conducting the laboratory analyses. ty in riparian ecosystems are needed for better management of Manuscript accepted15 Aug. 1998. riparian grazing (Platts 1986).

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 283 Two common montane riparian Center 1948Ð1990). For the study peri- in the 2 contrasting species was deter- species, water sedge (Carex aquatilus od, average growing season precipita- mined the year following grazing. Wahlenb.) and planeleaf willow (Salix tion at the Red Feather Lakes weather Random samples were taken for both planifolia Pursh.), were evaluated to station was 285 mm, and temperatures species within each paddock 4 times determine how previous cattle use ranged from 4¡to 16¡C (NOAA 1995- during the 1996 growing season. Initial affected forage N and P concentrations 1996). forage samples were gathered from all and IVDMD. Plants were evaluated at 4 The Naz soil series dominates the paddocks 1 June, representing early- phenological stages of plant develop- Sheep Creek region. These are deep, spring growth. Early-summer, late-sum- ment following 1 year of short-term, well drained soils formed from granitic mer, and fall samples were taken at the high intensity grazing treatments in an parent material. Soils in the riparian beginning of each successive month area that had been excluded from cattle study areas are primarily Naz 70, with 1 thereafter. These collection times corre- grazing for 40 years. It was hypothe- to 3% slopes. Texture of the soil is a sponded to the periods of the 1995 graz- sized that water sedge and planeleaf wil- clay loam and is classified as coarse ing treatments: late-spring, early-sum- low would have increased N concentra- loamy pachic cryoborol. The A horizon mer, late-summer, and fall. Grab sam- tion in current year’s growth after graz- has high organic matter (4Ð17%) and is ples of water sedge leaves and culms ing, along with concomitant levels of from 20 to 80 cm thick (USDA 1980). were clipped, while leaves and stem tips increased P and digestibility. Other Overstory vegetation at the study site of current year’s growth were removed researchers have reported elevated N was dominated by planeleaf willow, by hand from planeleaf willow through- concentrations for upland graminoid Geyer willow (S. geyeriana Anderss.), out each paddock. Forage samples for species from 4 weeks to 6 months after and yellow willow (S. lutea Nutt.). willows were taken only to a height of 2 defoliation (Jaramillo and Detling 1988, Understory herbaceous vegetation con- m to insure that samples represented Polley and Detling 1988, Rhodes and sisted of Kentucky bluegrass (Poa previously-browsed regrowth. Leaf and Sharrow 1990). This study, however, pratensis L.), fowl bluegrass (P. palus- stem tissue of water sedge was clipped was conducted 8Ð12 months following tris L.), water sedge, Nebraska sedge (C. at ground level within each paddock and cattle use treatments, and riparian rather nebraskensis Dewey), beaked sedge (C. bagged as a sample. A total of 60 sam- than upland species were evaluated. rostrata Stokes), tufted hairgrass ples for each species, (5 grazing treat- (Deschampsia caespitosa L.), bluejoint ments x 4 dates of collection x 3 pad- reedgrass (Calamagrostis canadensis docks) were collected. Materials and Methods Michx.), and dandelion (Taxaxacum Forage samples were placed in paper officinale Wiggers) (Schulz and bags, oven-dried at 50¡ C, weighed, and Study Site Leininger 1991, Popolizio et al. 1994). ground through a 1-mm mesh screen. The Sheep Creek Allotment is located Each homogenized sample represented a in north central Colorado, 80 km north- Methods species within a paddock at a particular west of Fort Collins, within the Small (0.25 ha) paddocks utilized in time for a grazing treatment. Carbon and Roosevelt National Forest at an eleva- this study were randomly located within nitrogen contents were determined using tion of approximately 2,500 m. The the exclosures. Short-duration, high- a LECO CHN-1000 instrument (LECO Allotment consists of 5,340 ha with intensity, seasonal grazing treatments Corp. 1993). An acid digest was per- 1,050 ha classified as grazeable range. with steers were applied in these pad- formed after samples were ashed in a The riparian area was heavily grazed docks in 1995. A set of 3 replicated pad- muffle furnace at 500¡ C. The digest was from the 1890's to the mid-1950's. Three docks that represented each of 5 grazing analyzed for P content with an inductive- exclosures comprising a total of 40 ha treatments; late-spring, early-summer, ly coupled plasma atomic emissions and 2.5 km of stream and adjacent ripar- late-summer, fall, and control (not spectrometer (Baker et al. 1964). The ian meadows were constructed in 1956 grazed) were assigned at random to the IVDMD was determined following the to exclude cattle use (Schulz and paddocks. Five steers were placed in the procedure of Tilley and Terry (1963), as Leininger 1990, Popolizio et al. 1994). designated paddocks at the beginning of modified by Pearson (1970). Each sam- The study plots (paddocks), located each season and allowed to graze until ple was inoculated with ruminal fluid within these exclosures, had been pro- herbaceous utilization reached approxi- obtained from a fistulated steer on an tected for 40 years. mately 65% (Pelster 1998). The average alfalfa hay diet and allowed to digest for The nearest weather station to Sheep time to reach this level of utilization was 48 hours at 39¡ C. This was followed by Creek is located at Red Feather Lakes, approximately 4 days. According to the an acid pepsin digest for an additional 48 15 km southeast of the study site at an stubble-height measurement technique hours to simulate digestion of material elevation of 2,542 m. Average annual (Kinney and Clary 1994), 65% percent leaving the rumen. precipitation at Red Feather Lakes is of the herbaceous biomass was utilized, All data were analyzed using analysis 406 mm, while average precipitation for and utilization of individual plants with- of variance techniques for a completely the growing season (MayÐSeptember) is in a paddock was fairly consistent. All randomized block design with a factori- 236 mm. Average daily temperatures willow plants within each paddock were al arrangement of treatments. Data for range from Ð11¡C in January to 25¡C in browsed to a height of 2 m, although the each species were analyzed separately July, and average daily temperatures proportion of willow in cattle diets grad- and collectively using the SAS (1996) during the growing season range from ually increased from late-spring to fall general linear models procedure. A 0¡to 25¡C (National Climatic Data 1995 (Pelster 1998). The forage quality repeated measures procedure was used to determine significant effects (p<0.10)

284 Journal of Range Management (52)3, May 1999 of collection time, species, grazing treatment, and interactions. Differences between grazed plants (all seasonal treatments combined) versus those that were not grazed (control) were compared using a linear contrast statement (SAS 1996). Five models tested the main factors of interest: a) a multivariate repeated measures analysis of variance that included grazing treatments and both species to determine species-level differences and three-way interactions, b) 2 multivariate repeated measures models, 1 for each species, where the variance among all treatments at each collection time was determined, and c) 2 univariate models, 1 for each species, that contrasted the grazed (all 4 seasonal treatments combined) plants with those that were not used by cattle (control).

Results and Discussion

A multivariate repeated measures analysis of variance revealed that collection time was a significant factor for Fig. 1. Nitrogen concentration in current year’s growth for grazed and ungrazed plants of 2 both species (p<0.01) for levels of N, P, riparian species, water sedge (Carex aquatilus) and planeleaf willow (Salix planifolia). Data points represent means (± standard error) for the dates that samples were collected. and IVDMD (F3,61 = 303.4, 255.8, Collection times correspond with the following seasons: late-spring (1 June), early-sum- 40.97, respectively). Nitrogen and P mer (July 1), late-summer (1 Aug.), and fall (1 Sept.). concentrations and IVDMD, including all grazing treatments plus controls, decline found for N and P over the sum- The comparable changes in N and P decreased in both water sedge and mer. Nitrogen and P were not good pre- concentrations and IVDMD in water planeleaf willow as phenological devel- dictors of digestibility, and differences sedge between grazed and control plants opment continued from young leaves in the seasonal patterns of N and P as with season of cattle use showed no sig- and shoots to senescence (Fig. 1, 2, 3). compared with IVDMD indicated that nificant collection time by grazing treat- other factors, such as fiber or secondary ment interactions. Therefore, a simple Water Sedge chemicals, influenced digestibility. effect of grazing was determined with a Average N concentration in water Water sedge nutritive characteristics linear contrast statement (SAS 1996). sedge of all treatments and controls was varied seasonally, and comparative This analysis combined the 4 seasonal 2.5% on 1 June. Average N declined research that includes riparian forage grazing treatments and compared the approximately 20% each successive quality data throughout an entire grow- forage quality with ungrazed plants. month, to 1.3% N on 1 September (Fig. ing season is lacking. Coppock et al. Results showed that grazed water sedge 1). Phosphorous decreased each month (1983) reported N concentrations of 1.0 N and P concentrations and IVDMD from a high of 0.34% on 1 June to a low to 1.7% and IVDMD values from 52 to were greater (p<0.10) compared with of 0.11% on 1 September (Fig. 2). The 63% over the growing season from May plants that were not previously grazed, to October for Carex spp. in a mixed- IVDMD for water sedge, however, did (F3,30 = 3.21, 6.14, 17.97, respectively). not steadily decline over the growing grass prairie. At the Central Plains Nitrogen in grazed water sedge was 5 to season (Fig. 3). Average IVDMD for Experimental Range, N concentration 10% higher than in comparable water sedge ranged from 68% in early- and IVDMD for Bouteloua gracilis ungrazed plants (Fig. 1). Grazed water spring to 54% in the fall. Sharp declines (H.B.K.) at peak standing crop sedge plants remained higher in N than in IVDMD were found between late- (September 1) contained about 1.8% N ungrazed plants throughout the entire spring and early-summer samples (7%) and was 60Ð65% digestible (Milchunas growing season. Phosphorous concen- and between late-summer and fall sam- et al. 1995). A native sedge of Serengeti trations were also higher in grazed water ples (12%). The decline in water sedge National Park in Tanzania, Kyllinga ner- sedge (Fig. 2). At the beginning of the IVDMD between early-summer and vosa (Steud.), had N concentrations of growing season, grazed plants were 14% late-summer, however, was only 1.6%. 2-3% and P concentrations of 0.2% dur- higher in P than ungrazed plants. This ing the growing season (McNaughton Digestibility values between early- and differential declined from about 12% in and Chapin 1985). All these data are late-summer remained about 63%, July and August to <1% in September. within the range reported for water which contrasts with the progressive Thus, the effect of grazing on water sedge in the present study.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 285 for N and P (Fig. 1 and 2). However, there was a significant (p<0.01) collection time by browsing treatment interaction for IVDMD of planeleaf willow (F12,22 = 7.53). Browsed willow had greater digestibility during early growth and senescence than during the middle of the growing season (Fig. 3). Digestibility of browsed willow remained the same between late-summer and fall, while IVDMD of unbrowsed willow dropped substantially over the same period of time. The disparate effects of time describe this interaction and illustrate how seasonal nutritive trends may change as a result of cattle use the previous year. The simple effect of treatment varied between browsed and unbrowsed willow IVDMD in the fall. Browsed willow was higher (p<0.10) in IVDMD than unbrowsed willow at this time (F4,10 = 4.6). Data for planeleaf willow nutritive characteristics with time in the literature are lacking. Nitrogen concentration and digestibility among browse species have been reported under natural and simulated browsing, but these data represent Fig. 2. Phosphorous concentration in current year’s growth for grazed and ungrazed plants individual points in time with browsing of 2 riparian species, water sedge (Carex aquatilus) and planeleaf willow (Salix planifolia). treatments that varied considerably from Data points represent means (± standard error) for the dates that samples were collected. Collection times correspond with the following seasons: late-spring (1 June), early-summer those in this present study. Nonetheless, (July 1), late-sumer (1 Aug.), and fall (1 Sept.). previous research does provide a framework for comparison. For example, sedge leaf and stem P concentration represented current year’s growth that nitrogen concentrations in forage for diminished by the end of the growing was not grazed the year of collection. Betula pubescens (Ehrh.) of 1.6 to 2.4% season. Digestibility of grazed water Additionally, there was significantly have been reported, along with sedge was also greater than in plants that higher production of sedges in previous- pepsin/cellulase digestibility values of were not previously grazed (Fig. 3). ly grazed paddocks as compared with 48-49% (Danell and Huss-Dannell Differences in IVDMD from 2.9 to 6.3% ungrazed controls (Schenck 1996, pers. 1985). Forage quality variables have also were found between grazed and ungrazed comm.), thereby repudiating the proba- been determined for Salix spp. in water sedge, and grazed plants had con- bility that nutrients in the ungrazed Yellowstone National Park (Singer et al. sistently higher digestibility at each col- standing-crop were diluted by greater 1994). They found nitrogen concentra- lection time than did ungrazed plants. aboveground biomass. Other comparable tions of 1Ð2% and dry-matter digestibili- The higher forage quality found in results are not available, so we suggest ty values from 45Ð53% for several wil- grazed water sedge as compared with that these protracted responses may be low species. These IVDMD and N data ungrazed control plants is in agreement attributed to water sedge nutrient storage were collected in August and are compa- with several greenhouse and field exper- strategies, accelerated mineralization in rable to the IVDMD and N data found in iments where other defoliated gram- soils of grazed paddocks and increased August for planeleaf willow in our study. inoids have been studied (Ruess and nutrient availability, or altered riparian Data for planeleaf willow N and P McNaughton 1984, McNaughton and nutrient-cycling dynamics in grazed pad- concentrations were reported for leaf Chapin 1985, Jaramillo and Detling docks as compared with controls. samples taken at Sheep Creek from 1988, Polley and Detling 1988). In these JulyÐSeptember 1994 and 1995 by Dernburg (1997). The P concentrations studies the samples were collected with- Planeleaf Willow in the same season of treatment and from leaf expansion to senescence were Data analysis for N and P concentra- 0.32Ð0.27% and N concentrations over imply that intensive grazing or clipping tions in planeleaf willow indicated that removes older growth and facilitates the the same time period ranged from 2.4 to these nutrients were not affected 1.8%. These values are slightly lower subsequent replacement by younger tis- (p>0.10) by the season of cattle use 1 sue with lower C:N ratios (Jameson than those determined in this study, but year after plants were browsed. Also, year-to-year variation is expected. 1963). In this study samples were gath- there were no significant collection ered the year following cattle use, and A univariate analysis that contrasted times by grazing treatment interactions browsed planeleaf willow plants with

286 Journal of Range Management (52)3, May 1999 those that were not browsed showed that among woody species. Again, these seasonal decline that was noted in water N and P concentrations were not differ- responses were measured during the sedge P concentration did not parallel ent (p>0.10) over all collection times the same growing season as when plants the steep decline that was found in year after grazing treatments were were browsed or were under continuous planeleaf willow P concentration (Fig. applied (Fig. 1 and 2). The IVDMD time use. These increases, then, may not be 2). A grazing treatment x species x time by treatment interaction precluded the meaningful in comparison with results of sampling interaction was found use of this univariate model, but the pres- from this study when plant responses (p<0.01) for phosphorous (F12,61 = ence of this interaction indicated differ- were measured in the growing season 4.77) and IVDMD (F14,81 = 4.77) in the ences between browsed and unbrowsed after browsing. full model that included both species willow through time. Browsed willow and treatments. This may be explained collected in the fall was 11% more Species Comparisons by differences in seasonal dynamics digestible than unbrowsed willow col- Water sedge and planeleaf willow N, between the 2 species that caused the lected at the same time (Fig. 3). P, and IVDMD were different (p<0.01) interaction. The decline in P concentra- Forage quality differences have been tion in willow plants between early- from one another (F1,11 = 566.6, 93.3, reported to differ for some woody browse 246.5, respectively). Averaged across all summer and late-summer was greater species as affected by defoliation intensi- grazing treatments and controls, plane- than that found for water sedge. Also, ties (Bryant 1981, Danell et al. 1985, leaf willow contained 43% and 52% differences in P between browsed and Danell and Huss-Danell 1985, Singer et more forage N and P than did water unbrowsed willow plants during the al. 1994). Moderate to high levels of sedge, while digestibility of water sedge summer were greater, as compared with browsing have resulted in increased leaf was approximately 25% higher than that grazed and ungrazed water sedge plants. N and dry-matter digestibility (Danell of planeleaf willow. Water sedge N and Seasonal differences also contributed and Huss-Danell 1985), lower IVDMD P concentrations were highly correlated similarly to the IVDMD grazing treat- and tannins (Singer et al. 1994), and (r = 0.91), but this was not the case for ment x species x time interaction; for increased palatability (Danell et al. 1985) planeleaf willow (r = 0.10). The gradual willow IVDMD varied more through time than the gradual changes found in water sedge (Fig. 3). These data reflect distinctive species-level responses among seasons that contributed to the significant 3-way interactions. Species-level differences were also found as a result of varying effects that previous cattle use had on forage quality. Concentrations of N and P increased along with IVDMD in forage of water sedge as a result of previous grazing, while only fall IVDMD increased in planeleaf willow foliage the year following cattle use. The greatest increase in water sedge digestibility for grazed plants, compared with controls, was found in early-summer, while the greatest difference in willow digestibility was found in the fall. Grazed water sedge N and P concentrations were consistently higher than ungrazed control plants each season, while planeleaf willow N and P concentrations did not change from late- spring to fall as a result of browsing the previous year. The contrasting responses between this riparian sedge and shrub are evidence of variability associated with plant life forms that may represent adap- tive strategies unique to each life form. Disparate strategies were demonstrated further by significant differences in N Fig. 3. In-vitro dry-matter digestibility of current year’s growth for grazed and ungrazed and P concentrations and IVDMD plants of 2 riparian species, water sedge (Carex aquatilus) and planeleaf willow (Salix plan- between the 2 species throughout the ifolia). Data points represent means (± standard error) for the dates that samples were col- growing season. Furthermore, although lected. collection times correspond with the following seasons: late-spring (1 June), early- this willow and sedge are commonly summer (1 July), late-summer (1 Aug.), and fall (1 Sept.). associated with western riparian ecosys-

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 287 tems (Youngblood et al. 1985), their nity-level spatial scale in a montane Kinney, J.W. and W.P. Clary. 1994. A nutritive responses to previous cattle use riparian ecosystem. Clarification of how photographic utilization guide for key were quite different. compensatory nutrient uptake and allo- riparian graminoids. USDA Forest Ser. cation patterns operate requires species- Gen. Tech. Rep. INT-GTR-308. level, ecophysiological approaches be LECO Corporation. 1993. Instruction Conclusions Manual CHN-1000 Elemental Analyzer. used that explore mechanisms that affect Form No. 200-516-014. Leco Corporation, these response variables in the context of St. Joseph, Mich. Forage quality of water sedge and highly interactive soil-plant ecology. McNaughton, S.J. and F.S. Chapin, III. planeleaf willow was affected the year Evidence of higher forage quality in 1985. Effects of phosphorous nutrition and following short-term, high-intensity cat- grazed water sedge in this study suggests defoliation on C4 graminoids from the tle grazing in a montane riparian ecosys- that grazing induced a change in avail- Serengeti Plains. Ecol. 66:1617Ð1629. tem, and the disparate responses were able nutrients and nutrient uptake in Meyers, L.H. 1989. Grazing and riparian water sedge. A quantitative understand- management in southwestern Montana. pp. indicative of ecophysiological mecha- 117Ð120. In: Practical Approaches to nisms unique to each life form. ing of the underlying, indirect mechanisms will require greater study of the Riparian Resource Management- An Increased N, P, and IVDMD in water Education Workshop. 8Ð11 May 1989; sedge as a result of grazing the previous riparian soil-plant nutrient exchange Billings, Mont. BLM-MT-PT-89-001- year may be evidence of increased complex. 4351. uptake kinetics, greater nutrient avail- Milchunas, D.G., A.S. Varnamkhasti, ability, or reallocation of reserves. W.K. Lauenroth, and H. Goetz. 1995. Greater IVDMD in browsed planeleaf Literature Cited Forage quality in relation to long-term willow in the fall may suggest lowered grazing history, current-year defoliation, levels of fiber or secondary compounds, Baker D.E., G.W. Gorsline, C.G. Smith, and water resource. Oecologia 101:366Ð374. and may help explain greater cattle pref- W.I. Thomas, W.E. Grube and J.L. Ragland. 1964. Techniques for rapid National Climatic Data Center. erence for the willow during later 1948Ð1990. NCDC Summary of the Day, growth stages (Meyers 1989, Kinch analysis of corn leaves for eleven elements. Agron. J. 56:133Ð136. West 1. Earth Info. Inc., Boulder, Colo. 1989, Pelster 1998). Bryant, J.P. 1981. Phytochemical deter- NOAA (National Oceanic Atmospheric Season of use did not affect forage rence of snowshoe hare browsing in Administration). 1995Ð1996. Climatological quality the following year, although sea- adventitious shoots of four Alaskan trees. Data. Red Feather Lakes, Colorado. U.S. son of use can affect other variables Science 213:889Ð890. Dept. of Commerce, Ashville, N.C. such as plant cover, production, and Clary, W.P. 1995. Vegetation and soil Pearson, H.A. 1970. Digestibility trials: in streambank erosion (Kauffman et al. responses to grazing simulation on riparian vitro techniques. pp. 85Ð92. In: Range and 1983). These data demonstrated how meadows. J. Range Manage. 48:18Ð25. Wildlife Habitat Evaluation-A Research Coppock, D.L., J.K. Detling, J.E. Ellis, Symposium. May, 1968, Flagstaff, grazing might interact with aboveground Ariz.USDA Forest Ser. Misc. Publ. No. biomass production the following year and M.I. Dyer. 1983. Plant-herbivore interactions in a North American mixed- 1147. in a riparian area on sites that have been grass prairie. Oecologia 56:1Ð9. Pelster, A.J. 1998. Diet selection and live- excluded from grazing for an extended Danell, K. and K. Huss-Danell. 1985. stock management in a montane riparian time, thereby improving graminoid for- Feeding by insects and hares on birches zone. M.S. Thesis. Colorado State Univ., age quality in the year following cattle affected by moose browsing. Oikos 44: Fort Collins, Colo. use. Conclusions are based on these one 75Ð81. Platts, W.S. 1986. Managing riparian stream time, short-term, high-intensity cattle Danell, K., K. Huss-Danell, and R. habitats. pp. 81Ð86. In: Proceeding of the use treatments. Other cattle management Bergstrom. 1985. Interactions between Symposium, American Fisheries Society. regimes and other sites that are regularly browsing moose and two species of birch March 5Ð6, 1986; Fort Collins, Colo. USDA Forest Ser. Intermountain Res. Sta., grazed may not yield similar results. in Sweden. Ecol. 66:1867Ð1878. Dernburg, A.R. 1997. Cattle preferences for Boise, Ida. This study did demonstrate, however, willows (Salicaceae) in a montane riparian Platts, W.S. and R.L. Nelson. 1985. that previous cattle use can increase for- site. Ph.D. Diss.. Colorado State Univ., Streamside and upland vegetation use by age quality of a montane riparian com- Fort Collins, Colo. cattle. Rangelands 7: 5Ð7. munity, although substantiation with Jameson, D.A. 1963. Responses of individ- Polley, H.W. and J.K. Detling. 1988. studies at different spatial and temporal ual plants to harvesting. Bot. Rev. Defoliation, nitrogen, and competition: scales and in other riparian areas is rec- 29:532Ð594. effects on plant growth and nitrogen nutri- ommended. Jaramillo, V.J. and J.K. Detling. 1988. tion. Ecol. 70:721Ð727. Further field studies are needed to Grazing history, defoliation, and competi- Popolizio, C.A., H. Goetz, and P.L. determine spatial and temporal variation: effects on shortgrass production and Chapman. 1994. Short-term response of nitrogen accumulation. Ecol. 69:1599Ð1608. riparian vegetation to 4 grazing treatments. tions in riparian nutrient dynamics and Kauffman, J.B., W.C. Krueger, and M. J. Range Manage. 47:48Ð53. the extent of graminoid responses to her- Vavra. 1983. Effects of late season cattle Rhodes, B.D. and S.H. Sharrow. 1990. bivory. Ideally, a long-term study would grazing on riparian plant communities. J. Effect of grazing by sheep on the quantity include data collection prior to defolia- Range Manage. 36: 685Ð691. and quality of forage available to big game tion, during regrowth that season, and in Kinch, G. 1989. Riparian area management: in Oregon's Coast Range. J. Range new growth the following year. This grazing management in riparian areas. Manage. 43:235Ð237. experiment represents data for 1 growing USDI Bur. of Land Manage. Tech. Rep. Roath, L.R. and W.C. Krueger. 1982. season that encompassed only a commu- 1737-4. U.S. Government Printing Office, Cattle influence on a mountain riparian Washington D.C. zone. J. Range Manage. 35:100Ð104.

288 Journal of Range Management (52)3, May 1999 Ruess, R.W. and S.J. McNaughton. 1984. Schulz, T.T. and W.C. Leininger. 1990. Tilley, J.M. and R.A. Terry. 1963. A two- Urea as a promotive coupler of plant-herbi- Differences in riparian vegetation structure stage technique for the in vitro digestion of vore interactions. Oecologia 63:331Ð337. between grazed areas and exclosures. J. forage crops. J. Brit. Grassl. Soc. 18:104Ð111. Ryder, R.A. 1980. Effects of grazing on bird Range Manage. 43:295Ð299. USDA, Soil Conservation Service and habitats. pp. 51Ð66. In: Management of Schulz, T.T. and W.C. Leininger. 1991. Forest Service. 1980. Soil survey report. Western Forests and Grasslands for Impacts of livestock grazing on nongame Larimer County area, Colorado. Naz 70 Nongame Birds. USDA Forest Ser. Gen. wildlife populations in a montane riparian soil series. U.S. Government Printing Tech. Rep. INT-86. area. Great Basin Natur. 51:286Ð292. Office 239-812/48. Washington, D.C. SAS Institute. 1996. SAS user's guide. Singer, F.J., L.C. Mark, and R.C. Cates. Youngblood, A.P., W.G. Padgett, and A.H. Raleigh, N.C. 1994. Ungulate herbivory of willows on Wilson. 1985. Riparian type classification Schenck, S. 1996. Personal communication. Yellowstone’s northern range. J. Range of eastern Idaho-western Wyoming. USDA Rangeland Ecosystem Science Dept., Manage. 47:435Ð443. Forest Ser. Region 4, R4-Ecol-8501. Colorado State Univ., Fort Collins, Colo.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 289 J. Range Manage. 52:290Ð295 Evidence of cell deterioration in winterfat seeds during refrigerated storage

D. TERRANCE BOOTH, ROCHMAH AGUSTRINA, AND ROLLIN H. ABERNETHY

Authors are rangeland scientist, USDA, Agricultural Research Service (ARS), High Plains Grassland Research Station, Cheyenne, Wyo, 82009; former graduate student, and professor, Plant, Soil, and Insect Sciences Dept., University of Wyoming, Laramie, Wyo. 82071.

Abstract Resumen

Effective storage of wildland seeds helps alleviate supply El almacenamiento efectivo de semillas silvestres ayuda shortages and mitigates variable production associated with aliviar la falta de suministro y mitigar la producción variable annual weather patterns. The storage environment is critical asociada con los patrones anuales de clima. El ambiente de for seeds like winterfat [Eurotia lanata (Pursh) Moq.) that almacenamiento es critico para semillas como las de "winter rapidly lose viability under ambient conditions. Defining seed fat [Eurotia lanata (Pursh) Moq.]que rápidamente pierden response to storage conditions is basic to effective seed stor- viabilidad bajo las condiciones ambientales. El definir la age programs. We used electron micrographs of freshly col- respuesta de la semilla a las condiciones de almacenaje es lected, and of stored winterfat seeds, with vigor tests to com- básico para implementar programas efectivos de almace- pare seedling vigor and to relate seed performance to seed namiento de semilla. Utilizamos micrográficas del microsco- cell biology as influence by; (a) seed age under known storage pio electrónico de semillas recién cosechadas y semillas de conditions, and (b) imbibition temperatures. We found that "winter fat" almacenadas. Se utilizaron las micrográficas con imbibition temperatures had little influence on the vigor of pruebas de vigor para comparar el vigor de la plántula y fresh seeds but significantly influenced aged seeds. relacionar el comportamiento de la semilla con la biología Mitochondrial deterioration was evident in winterfat seeds celular influenciado por: (a) la edad de la semilla bajo condi- stored 5-6 years at 5¡C, and in fresh, but incompletely ciones de almacenamiento conocidas y (b) las temperaturas hydrated seeds held at 20¡C. We recommend seeds be held at de imbibición. Encontramos que la temperatura de imbibi- Ð18¡C or colder for long-term storage and that field seedings ción tiene poca influencia en el vigor de semillas recién be done during the cold season to reduce the chance that cosechadas, pero influencio significativamente el de las semil- incompletely hydrated seeds will be exposed to warm temper- las envejecidas. La deterioro mitocondrial fue evidente en atures. semillas de "winter fat" almacenadas durante 5–6 años a 5°C y en semillas frescas pero hidratadas incompletamente a 20¡C. Para el almacenamiento por largos periodos, recomen- damos que las semillas deben ser conservadas a Ð18¡C o tem- Key Words: Seed aging, mitochondria, vigor, imbibition, peraturas mas frías y que las siembras de campo sean hechas Eurotia, Ceratoides, Krascheninnikovia durante la época fría para reducir la probabilidad de que semillas hidratadas incompletamente sean expuestas a tem- Only in laboratory studies has it been possible precisely to peraturas calientes. separate effects on germination from effects on subsequent survival.—John L. Harper (1977) in his book, 'Population Biology of Plants'. (1974b) recommended refrigeration for long-term storage and Winterfat [Eurotia lanata (Pursh) Moq.]1 and closely related reported little germination from collections held at ambient species are important forage plants on the cold deserts of temperatures for 8 years compared to 68% germination for North America and Asia, but seeds of these species rapidly seeds held at 5¡C. Although 5¡C storage extends seed viabili- lose viability when stored at ambient conditions (Wilson ty, its effect on seedling vigor is unknown. 1931, Hilton 1941, Springfield 1968, 1974b). Springfield Our objective was to determine if seed aging under 5¡C- storage reduced winterfat seedling vigor. Our hypothesis was developed from studies by Booth (1992), Booth and Funding was provided by Partners for International Education and Training, Washington, D.C.; USDA-ARS; and the Wyoming Agricultural Experiment McDonald (1994), and Agustrina (1995). Booth (1992) report- Station. Authors thank Carol Hearne, Wyoming State Veterinary Laboratory for ed seedling vigor was reduced by warm imbibition tempera- helping us to obtain micrographs, and Dr. Patricia Berjak, University of Natal, tures (Fig. 1). Booth and McDonald (1994) found that rapid Durban, South Africa for her invaluable assistance in interpreting micrographs of cell structure. We thank L.W. Griffith and H. Jansen for technical assistance and 1 Drs R.L. Anderson, B.N. Smith, and A.G. Taylor for reviewing the manuscript. Krascheninnikovia Gueldenstaedt. is the current synonymic favorite, replacing Mention of trade or company names is for information only and does not imply an Ceratoides J.T. Howell. To call attention to the need to stabilize scientific nomencla- endorsement. ture except where new evidence clarifies the phylogeny, we have retained Eurotia Manuscript accepted 24 Jul. 98. (Adanson 1763) which was used by most publications between 1840 and 1971.

290 Journal of Range Management (52)3, May 1999 Germination and Growth of 1986 and 1993 Seeds Booth (1992) and Agustrina (1995) report tests for differences in seedling axial length (a measure of seedling vigor) due to imbibition temperature. We used these reports to compare seedling vigor from the 1986 and 1993 seeds by calculating the 95% confidence interval for mean axial lengths of seedlings from the 1986 seeds. We used the MSE from an ANOVA comparing seedling axial length by imbibition temperature for day 10 and for day 12 of incubation (n = 75) (Booth, unpublished 1992 data). Confidence intervals were not calculated for the 1993 data because interactions among the several treatments prevented pooling of experimental units (Agustrina 1995). In their procedures Booth (1992) and Agustrina (1995) divided seeds into groups of 20, weighed the groups to 0.01 g, and humidified the seeds for 3 days at 2¡C prior to imbibition to reduce imbibitional injury (Vertucci and Leopold 1984, Vertucci 1989). The seeds were then incubated on slant boards (Jones and Cobb 1963). The slant boards containing humidified seeds were placed in plastic boxes with tight fitting lids; the boxes filled to a depth of 5 cm with distilled water, then placed in incubators for 4 days at treatment temperatures that included 5 and 20¡C. All treatments Figure 1. The effect of imbibition temperature on winterfat seedling vigor from seeds collect- were then incubated in the dark at 20¡C. ed in Nevada and Wyoming, and from the released variety 'Hatch', as reported by Booth The axial lengths of germinants were (1992). measured after 5, 7, 10, and 12 days of incubation using a digitizing tablet imbibition of winterfat seeds at warm Methods and Materials (Booth and Griffith 1994). temperatures did not damage seed membranes. Agustrina (1995) found freshly Plant Material harvested winterfat did not exhibit the Mitochondrial Ultrastructure Winterfat diaspores (seed-containing Study imbibition-temperature response evident dispersal units) were collected in Winterfat diaspores of the 1986 and in the stored seeds used by Booth October, 1986 (Booth 1992); and from 1993 Cheyenne collections were used (1992). Because increases in respiratory the same stand on the High Plains for an electron microscope study of activity during germination of other Grassland Research Station, Cheyenne, mitochondrial structure. The diaspores species have been associated with rapid Wyo., (elevation of 1,909 m) in 1993 were threshed, the seeds weighed, and development of mitochondrial ultra- (Agustrina 1995). The diaspores of both placed in a humidity chamber at 5 or structure, including rehydration of its collections were stored at room tempera- 20¡C until the seed weight reached 1.2 membrane system (Nawa and Asahi ture for approximately 2 months, then or 1.5 times the dry weight. After 1973, Solomos et al. 1972), we hypothe- stored at 5¡C (Springfield 1974a). Using humidification, 1 mm was cut from the sized (1) that winterfat seeds stored the methods described by Booth and tip of the embryonic radicle and fixed in more than 2 years at 5¡C contained Griffith (1984), a portion of the diaspores 5% glutaraldehyde at room temperature mitochondrial-ultrastructure changes were threshed and seeds stored at 5¡C. for 4 to 5 hours. For comparison, dry that reduced seedling vigor and, (2) that The 1986 seeds were tested by Booth in seeds were submerged in 1% glutaralde- imbibition temperatures affected winter- 1989 and 1990. Seeds from diaspores hyde in 0.1 M sodium cacodylate (pH > fat seedling vigor by influencing mito- collected in Cheyenne in 1993 were test- 4) for 48 hours, then fixed in 5% glu- chondrial hydration, metabolism and ed about 6 months after collection taraldehyde. Standard methods were development in seeds stored more than 2 (Agustrina 1995). years at 5¡C. used for specimen dehydration and

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 291 mounting (Hayat 1986). The specimens Table 1. Seedling axial-length comparison MA/CA were trimmed using an LKB ultratome between Cheyenne seeds collected in 1986 Neither imbibition temperature, col- and tested between 1989Ð1990 vs. Cheyenne (LKB Instruments, Inc., Rockville, Md.) seeds collected in 1993 and tested in 1994. lection year, moisture, nor their interac- For light microscopy, the 1.0 micron Seeds were imbibed at the temperature indi- tions significantly affected numbers of sections were cut using a Reichert Jung cated for 4 days, then incubated at 20¡C. The mitochondria. P values ranged from Ultracut (Reichert-Jung Optische Werke 95% confidence interval is indicated for the 0.14 for the temperature x year interac- Ag, Hernalser Hauptstr, Wien, Austria), 1986 seeds. tion (Table 2) to 0.84 for moisture x stained with 1% toluidine blue, and then year. However, the MA/CA ratio was Mean Length observed under the light microscope. by Collection affected by an imbibition-temperature x Using the light microscope we selected Temperature Days collection-year interaction (P=0.03). a representative area from the winterfat Measured 1986 1993 The greatest MA/CA ratio was observed radicle tip for study by transmission --¡C------(mm)------from 1993 seeds imbibed at 20¡C (Table electron microscope (TEM). A section 5 10 48 + 4.8 80 2). This was significantly greater (P < approximately 79 nanometers thick was 12 48 + 4.7 78 0.05) than for 1986 seeds imbibed at cut from the chosen area using the 20 10 25 + 4.8 77 20¡C. The 2 collections did not differ Reichert Jung Ultracut, stained with 12 25 + 4.7 76 when imbibed at 5¡C. uranyl acetate and lead citrate, and then The seedling axial length means for 1986 seeds were observed by TEM. A second group of obtained from earlier data (Booth, data on file). Means for 1993 are for seeds, at treatment temperature, without Discussion representative cells were chosen from fungicide (Agustrina 1995). these TEM observations. Micrographs were obtained using an internally ference in the seedling vigor between Mitochondrial Degradation mounted camera on the TEM, the mag- the 2 seedlots. Early metabolic activity, which begins nification was 50,700 x. Samples were near 20% seed moisture, consists of cut from 3 different seeds in each treat- membrane reorganization, glycolysis, ment and 2Ð6 micrographs were made Mitochondrial Ultrastructure repair of organelles, and activation of from each sample. Dry Seeds. Radicle-tip cells of dry enzymes in preparation for an increasing seeds from 1986 and 1993 collections respiration rate and subsequent radicle Ratio of Mitochondrial Area to contained mitochondria with few cristae emergence (Bewley and Black 1985). and numerous translucent areas in the Loss of seedling vigor due to seed aging Cell Surface Area (MA/CA) mitochondria matrix. Hydration recon- Slides of TEM photos were used to has been associated with a marked stituted radicle mitochondria for both decline in soluble carbohydrate determine the mitochondria number and a 1986 and 1993 seeds. ratio of mitochondrial area to cell area (Bermal-Lugo and Leopold 1992), Hydration at 5¡C. When hydrated at increases in chromosomal aberration (MA/CA). Relative area was measured by 5¡C to the 20% moisture level, the mito- tracing projected images on tracing paper. (Dimitrov 1994, Guiterrez et al. 1993), chondria of 1993 seeds contained, (1) changes in ribosomal protein content Traced images were cut out, weighed, and more distinct lipid bilayers of the outer the data used to calculate the ratio. (Zalewski 1985, Hallam et al. 1973), membrane, (2) a greater number of and abnormal mitochondrial structures cristae, and (3) a more uniformly dense (Hallam et al. 1973, Abu-Shakra and Experimental Design and matrix— typical for normal mitochon- Ching 1967). Aging produces an accu- Statistical Analysis dria of hydrated cells—than did cells mulation of a free fatty acid mixture in To compare MA/CA, mitochondrial from 1986 seeds (Fig. 2a versus 2b). number, and size of mitochondria we used When hydrated at 5¡C to 50% mois- Table 2. Data for Cheyenne seeds collected in a random design and the ANOVA proce- ture content, mitochondrial ultrastruc- 1986 and 1993 showing the effects of temper- dure in SAS. Comparison of individual ture of 1986 and 1993 seeds was similar ature and collection year on total surface treatment means used the Least (data not shown). area of mitochondria per surface area of the Hydration at 20¡C. Hydration to 20% cell (MA/CA) and on the number of mito- Significant Difference (LSD) for variables chondria per cell. The temperature x year with a significant F-statistic (SAS 1985). moisture at 20¡C resulted in 1993 seeds interaction was significant for MA/CA with thinner mitochondrial cristae than (P=0.035). were observed in seeds hydrated at 5¡C Results (data not shown); however, the mito- Temperature Year Mean Mean number chondrial matrix was more uniformly MA/CA Mitochondria2 Germination and Growth of 1986 distributed. Hydration to 50% moisture NT (dry)1 86 0.268a2 31.67 and 1993 Seeds at 20¡C produced 1993 seed with the NT (dry) 93 0.310ab 33.67 Seedlings from the 1993 seeds had most deteriorated mitochondrial struc- 5 86 0.305ab 40.50 tures seen among all treatments (Fig. 3). 5 93 0.299ab 37.34 about twice the axial length of seedlings 20 86 0.274a 32.67 from the 1986 seeds (Table 1), and axial The outer membrane and cristae were completely degraded while in the 1986 20 93 0.373b 50.17 lengths from 1993 seeds did not fall 1 seeds the outer membrane was partially Dry and not treated within the axial-length confidence inter- 2Means with the same letter are not different as deter- degraded and the cristae structures could vals of seedlings from 1986 seeds. mined by LSD0.05. There were no differences in mito- still be observed. chondria numbers among treatments (P ranged from 0.14 These findings imply a significant dif- for temperature x year, to 0.85 for moisture x year).

292 Journal of Range Management (52)3, May 1999 membrane, vacuolarization of the matrix, loss of the matrix content, and total disappearance of cristae (Luzikov et al. 1985); in short, a decrease in respiration capability and efficiency. Thus the lack of mitochondrial ultrastructure correlates with the low vigor of the 1986 seeds (Table 1).

Effects of Seed Moisture and Temperature There were few differences between dry or hydrated seeds. This agrees with Baird et al. (1979) who also reported no appreciable differences between mitochondria in dry and imbibed radicle tissues. Seed hydration at low temperature did produce differences in MA/CA between the 1986 and 1993 seeds as evident by the significant temperature x collection- year interaction. The greater MA/CA for 1993 seeds imbibed at 20¡C, compared to that of 1986 seeds imbibed at 20¡C, and the lack of difference when these Fig. 2a seeds were imbibed at 5¡C (Table 2), suggest that early metabolic activity at 5¡C, maintained or repaired mitochondria in the 1986 seeds. Thus, these data provide a physiological explanation for Booth's (1992) finding that winterfat seedling vigor decreased with increasing imbibition tempertures (Fig. 1). The same explanation may apply to other chenopods shown to benefit from imbibition at temperatures suboptimal for germination (Haferkamp et al. 1990, Shaw et al. 1994). Hydration at warm temperatures produced unexpected differences due to moisture content of the 1993 seeds. Why did these seeds show so much mitochondrial degradation at 50% moisture whereas the 1986 seed did not? Bain and Mercer (1966) reported that transportation of hydrolyzed food reserves from cotyledon cells of peas did not occur until the moisture level in the seeds reached more than 93%. This Fig. 2b suggests that translocation of Fig. 2. The organelles of embryonic winterfat radicle cells (50,700 x) of 1986 (a) and 1993 (b) hydrolyzed food reserves from winterfat seeds, hydrated at 5¡C to 20% moisture. The organelles are represented as follows: (m) seed-storage tissue may not have mitochondria; (cw) cell wall; (l) lipid body; (N) nucleus; (pd) plastid; (pb) protein body,(R) occurred at 50% moisture. ribosome, and (er) endoplasmic reticulum. Vartepetian et al. (1976) reported that ATP-regeneration was essential in main- mitochondrial membranes as a result of mitochondrial swelling (Luzikov et al. taining the integrity of mitochondria in activation of phospholipase A2 (Luzikov 1985). The characteristics of early mito- excised seedling tissues. Hodson et al. et al. 1985; Nachbaur et al. 1972). This chondrial degradation are dilution of the (1987) hypothesized that a delay in the accumulation induces uncoupling of matrix, swelling and subsequent development of mitochondrial structure oxidative phosphorylation, reduces straightening of the folds formed by the may be related to the available energy endogenous ATP levels, and leads to inner membrane, rupture of the outer supply in the seeds. The lack of activity

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 293 but commented that "subfreezing temperatures appear slightly superior". However, he subsequently (Springfield 1974a) recommended refrigerated storage. The severe degradation of mitochondria in 6-month-old seeds at 50% moisture and 20¡C, suggests that fresh, partially hydrated seeds are at risk during periods of warm weather. We believe that the combination of warm soil temperatures (> 15¡C) and incomplete seed hydration could account for the failure of late-spring and summer winterfat seedings.

Judging the Evidence Micrographs. The micrographs of the 1986 seed were made 3Ð4 years after the vigor test was conducted, and thus represent a more advanced stage of mitochondrial degradation than was present when the seeds were tested. The fact that the micrographs are of seeds that are older than the vigor test is perhaps fortunate since we may not have detected these gradual structural changes in 2- Fig. 3. The organelles of embryonic winterfat radicle cells (50,700x) in 1993 seed hydrated to year-old seeds. 50% moisture at 20¡C. See Fig. 2 for organelle labeling information. Comparing Seedlots from Different Years. The Coefficients of Variation (Steel and Torrie 1980) for mean month- in the mitochondria of pea seedlings Springfield (1968) compared winterfat ly temperature and precipitation for June after their peak activity, was correlated seeds which had been stored in a ware- through September, 1986 and 1993 at with shrinking of the cotyledons house (13 to 35¡C) with refrigerated the High Plains Grasslands Research (Malhotra et al. 1970). storage (3 to 6¡C), and with freezer stor- Station are smaller than the coefficients So lack of an energy supply may have age (Ð20 to Ð32¡C). He did not directly for the three 1993 seed collections caused mitochondrial degeneration in compare refrigerated and freezer storage (Cheyenne and Pine Bluffs, Wyo; and the 1993 seed at 50% moisture. Would not this also occur in the 1986 seed? Table 3. Climatological data and Coefficients of Variation for Cheyenne and Pine Bluffs, Wyo., and Aging may have reduced active catabol- Sterling Colo.1 ic enzymes in the 1986 seeds. Das and Sen-Mandi (1992) reported a decline in Mean Temperature Mean Precipitation amylase activity of aged seeds. Declines Cheyenne 1986 ------(¡C)------(mm)------June 18.2 62 in enzyme activity or in the cellular abil- July 20.4 26 ity to synthesize proteins could indicate August 19.6 39 reduced catabolic activity. Thus, the Sept 13.1 63 1986 mitochondria could be more Cheyenne 1993 degraded than in the 1993 seed at 20% June 15.0 84 July NA NA moisture, then be less degraded at 50% August 18.4 56 moisture. Sept 12.5 80 Pine Bluffs 1993 June 15.8 66 Implication of Findings and July NA NA Related Information August 19.1 22 The evidence is that storage at 5¡C did Sept 12.5 40 not protect winterfat seeds from age- Sterling 1993 June 19.9 52 related degradation. The seeds remained July 22.8 41 viable, but lost vigor. Other chenopod August 22.5 50 seeds stored at Ð12 to Ð23¡C were Sept 15.4 30 reported to have little loss of vigor after Coefficients of Variation for: Temperature Precipitation 22 years (Pack and Owen 1950). Roos Cheyenne 1986 & 1993 17.7 32.6 (1989) recommended seeds be at Ð18¡C Chey., P. B., & Sterl. 1993 20.9 37.7 1 or lower for long term storage. National Oceanic and Atmospheric Administration (1986, 1993). Data for Sterling, Colorado was courtesy of the High Plains Climate Center, University of Nebraska, Lincoln (personal communication).

294 Journal of Range Management (52)3, May 1999 Sterling, Colo.) tested by Agustrina protect winterfat seed-vigor potential. of utricles and seedling establishment of (1995) (Table 3). Agustrina (1995) For that reason we also recommend win- 'immigrant' forage kochia. J. Range detected no significant differences in terfat be seeded during the cold season. Manage. 43:518Ð522. seedling axil length among the three Hallam, N.D, B.E. Robert, and D.J. 1993 seeds collections. This implies that Osborn. 1973. Embryogenesis and germi- Literature Cited nation in rye. III. Fine structure and bio- differences in the seeds due to produc- chemistry of the non-viable embryo. tion year are small relative to aging dif- (Secale cereale L.). Planta 110:279Ð290. ferences, and are unlikely to confound Abu-Shakra, S.S., and T.M. Ching. 1967. Harper, J.L. 1977. Population biology of our interpretation of the data. Mitochondrial activity in germinating new plants. Academic Press. New York. Extrapolation from 1 Seed Source. We and old soybean seeds. Crop Sci. Hayat, M. A. 1986. Basic technique for detected differences in basic cell biolo- 7:115Ð118. transmission electron microscopy. gy that have been observed in seeds of Agustrina, R. 1995. The influence of imbi- Academic Press Inc. Orlando, Fla.. p. 347. other species. A basic process that bition temperature on germination, mito- Hilton, J. W. 1941. Effects of certain micro- chondrial structures, and proteins of win- occurs across species is not likely to ecological factors on the germinability and terfat (Ceratoides lanata (Pursh) J.T. early development of Eurotia lanata. change due to seed source within a Howell). Ph.D. dissertation. Univ. of Northwest Sci. 15:86Ð92. species. Further, the 3 seed sources test- Wyoming, Laramie, Wyo. Hodson, M.J., L. Di Nola, and A.M. ed by Booth (1992) were all more than 2 Bain, J.M. and F.V. Mercer. 1966. Mayer. 1987. The effect of changing tem- years old and had been stored under Subcellular organization of cotyledons in perature during imbibition on ultrastruc- similar conditions and gave similar germinating seed and seedlings of Pisum ture in germinating pea embryonic radicle. responses to imbibition temperature. sativum L. Aust. J. Biol. Sci. 19:69Ð84. J. Exp. Botany. 38:525Ð534. Finally, though Springfield (1968) did Baird, L.A.M., A.C. Leopold, W.J. Jones, L.G. and R.D. Cobb. 1963. A tech- not grasp the importance of the differ- Bramlage, and B.D. Webster. 1979. nique for increasing the speed of laborato- ences between "refrigerated" and "freez- Ultrasructural modifications associated ry germination testing. Proc. Assc. Offic. with imbibition of the soybean radicle. Seed Analysts. 53:144Ð160. er"-stored seeds, his observations add to Bot. Gaz. 140:371Ð377. Luzikov, V.N., A.V. Galkin, and D.B. the evidence that winterfat seed aging Bermal-Lugo, I. and A.C. Leopold. 1992. Roodyn. 1985. Mitochondrial biogenesis and during 5¡C-storage is an important con- Changes in soluble carbohydrates during breakdown. Consultant Bureau. New York. sideration for all winterfat seeds. seed storage. Plant Physiol. 98:1207Ð1210. Malhotra, Surjit S. and Mary Spencer. Bewley, J.D. and M. Black. 1985. Seeds: 1970. Changes in the respiratory, enzy- physiology of development and germina- matic, and swelling and contraction Conclusion and tion. Plenum Press. New York and properties of mitochondria from cotyle- Recommendations London. dons of Phaseolus vulgaris L. during Booth, D.T. 1992. Seedbed ecology of win- germination. Plant Physiol. 46:40Ð44. terfat: imbibition temperature affects post- Nachbaur, J., A. Colbeau, and P.M. We conclude that storage of winterfat germination growth. J. Range Manage. Vignais. 1972. Distribution of mem- seeds at 5¡C resulted in age-related 45:159Ð164. brane-confined phospholipases A in the degradation of seed mitochondria and Booth, D.T. and L.W. Griffith. 1984. rat hepatocyte. Biochem. Biophys. Acta. other organelles during storage, and in a Evaluation of air threshing for small lots 274:426Ð446. significant loss of seedling vigor. Also, of winterfat fruits. J. Range Manage. National Oceanic and Atmospheric that age-related degradation was the 37:287Ð287. Administration. 1986. Climatological physiological reason for the seedling- Booth, D.T. and L.W. Griffith. 1994. Data—Wyoming. Volume 92, Numbers vigor response to imbibition temperature Technical note: Measuring post-germina- 6, 8, 9. tion growth. J. Range Manage. National Oceanic and Atmospheric found by Booth (1992) for 3 separate Administration. 1993. Climatological winterfat seed sources. We now believe 47:503Ð504. Booth, D.T. and M.B. McDonald. 1994. Data—Wyoming. Volume 102, that warm imbibition temperatures do Cation concentration in post-imbibed win- Numbers 6, 8, 9. not significantly reduce vigor of healthy terfat seeds as influenced by imbibition Nawa, Y. and T. Asahi. 1973. seeds. However, incomplete hydration temperature. J. Range Manage. Relationship between the water content (>20% and <90% seed moisture) at 47:485Ð488. of pea cotyledons and mitochondrial warm temperatures (near 20¡C) will Das, G. and S. Sen-Mandi. 1992. Scutellar development during early stage of ger- result in rapid mitochondria degradation amylase activity in naturally aged and mination. Plant Cell Physiol. and loss of vigor. Also, imbibition at accelerated aged wheat seeds. Ann. Bot. 14:607Ð613. Pack, D.A. and F.V. Owen. 1950. warm temperatures will reduce the vigor 69:497Ð501. Dimitrov, B. D. 1994. Types of chromoso- Viability of sugar beet seed held in cold of seeds needing mitochondrial repair. storage for 22 years. Amer. Soc. Sugar For these seeds, cold imbibition reduces mal aberrations induced by artificial seed ageing in Crepis capillaris. Cytobios. Beet Technol. Proc. 6:127Ð129. inefficient respiration and allows repair 77:107Ð114. Roos, E.E. 1989. Long-term seed storage. to proceed. Gutierrez, G., F. Cruz, J. Moreno, V.A. Plant Breeding Reviews 7:129Ð158. We recommend winterfat seeds be Gozales-Hernandez, and J.M. Vazquez- SAS. 1985. SAS User's Guide: Statistical held < Ð18¡C to protect seed vigor dur- Ramos. 1993. Natural and artificial seed Analysis System Institute, Inc. Cary, N. C. Shaw, N.L., M.R. Haferkamp, and E.G. ing long term storage and, that the seeds ageing in maize: Germination and DNA Hurd. 1994. Germination and seedling be imbibed at 0 to 5¡C as a standard lab- synthesis. Seed. Sci. Res. 3:279Ð285. establishment of spiny hopsage in oratory practice. Cold imbibition does Haferkamp, M.R., D.C. Ganskopp, K.L. response to planting date and seedbed no damage to winterfat seeds and it does Marietta, and B.W. Knapp. 1990. environment. J. Range Manage. Environmental influences on germination 47:165Ð174.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 295 Solomos, T., S.S. Malhotra, S. Prasad, Springfield, H.W. 1974b. Winterfat seeds Vertucci, C.W. and A.C. Leopold. 1984. S.K. Malhotra, and M. Spencer. 1972. viable after 8 years refrigerated storage. J. Bound water and its relation to respiration Biochemical and structural changes in Range Manage. 27:78. and imbibitional damage. Plant Phyiol. mitochondria and other cellular compo- Steel, R.G.D. and J.H. Torrie. 1980. 75:114Ð117. nents of pea cotyledons during germina- Principles and procedures of statistics. tion. Can. J. Biochem. 50:725Ð732. Wilson, C.P. 1931. The artificial reseeding Springfield, H.W. 1968. Cold storage helps McGraw-Hill Book Company. N.Y. of New Mexico ranges. N. Mex. Agr. Exp. winterfat seeds retain viability. J. Range Vartepetian, B.B., I.N. Andreeva, and G.J. Sta. Bull. 189. Manage. 21:401Ð402. Kozlova, 1976. The resistance to anoxia Zalewski, K. 1985. The metabolism of aged Springfield, H.W. 1974a. Eurotia lanata and the mitochondrial fine structure of rice seeds the formation of polyribosomes in (Pursh) Moq. p. 398Ð400 In: C.S. seedling. Phytochemistry 88:215Ð225. the embryos of germinating rye grains of Schopmeyer (Tech-Coord.) Seeds of Vertucci, C.W. 1989. The effects of low different viability. Acta Soc. Bot. Pol. Woody Plants in the United States, Ag. water content on physiological activities of 54(4):417Ð427. Handbook No. 450. Forest Service, seeds. Physiol. Plant. 77:172Ð176. USDA, Washington, D.C.

296 Journal of Range Management (52)3, May 1999 Book Reviews

Landscapes of the Interior. Re-explorations of Nature landscapes of the city. This makes for interesting reading. and the Human Spirit. By Don Gayton. 1996. New Landscapes of the Interior is about nature, the environment Society Publishers, Gabriola Island, British Columbia, and land use. Gayton takes the reader on a unique journey Canada. 176p. US$14.95 paper. Can$17.95 paper. US with his philosophical views. He has a true love of the land ISBN 086571-344-8. Canada ISBN 1-55092-285-8. and that love is felt on every page.—Jan Wiedemann, Texas Section, Society for Range Management, Vernon, Texas. Don Gayton is a scientist and range ecologist by training. His in-depth experience of the natural landscapes of western Fields of Grass. Portraits of the Pastoral Landscape and North America has been filled out by a wide variety of unusual jobs which include his being a community developer in Nomads of the Tibetan Plateau and Himalayas. Text Latin America, a hired man on an Okanagan cattle ranch, an and photographs by Daniel J. Miller. 1998. agricultural extension agent on Saskatchewan Indian reserves, International Centre for Integrated Mountain and a steelyard worker. Development, GPO Box 3226, Kathmandu, Nepal. Combining his scientific knowledge and practical experi- 204 p. US$40.00 paper. ISBN 92-9115-824-0. ence, Gayton has written 17 essays about his journey with the The high plateaus, mountainous vistas, and colorful cultures human connection and nature. His stories reach out in original of the Himalayas are the focus of Fields of Grass by Daniel J. ways, reminding us of how nature can be an intense source of Miller, a recent effort at verbal and photographic portraiture of spirituality and renewal. the region's nomadic pastoralism. Originally a Minnesotan, I enjoyed Gayton’s essay about the American cowboy writer and with considerable experience as a Montana cowboy, Mr. and painter Will James. He tells how James created himself as Miller has, since 1974, been active in range-livestock devel- the main character in his own novels: cowboy, painter, writer, opment and wildlife conservation in the Himalayan region. western myth and finally, victim. Most allow fiction to pene- This book is largely a photographic survey of his experiences trate only as far as their bookshelves, stopping it far short of of over 2 decades in Bhutan, China, Tibet, Nepal, Mongolia personal identity. James became part of the landscape and that and Pakistan. is what interests Gayton. After a brief introduction, Fields of Grass is loosely orga- As a boy, Gayton read Smoky, the Cow Horse by James. The nized into 6 main sections with the titles Pastoral Landscape, story features a lone rebellious figure, bonded to a half-wild Pastoral Production, Livestock, Nomads, Changes, and horse, working with animals and against the elements, dealing Future Challenges. Each of these sections contains pho- rarely with other men and almost never with women. James tographs and text focusing on each topic. The quarter-page to portrayed himself as the quintessential American cowboy. full-page photographs are all in black and white. Captions are Thirty years after first reading Smoky, Gayton stumbles usually limited to a statement of the subject, location, and across a reference to Jame’s beginnings in Canada. He wanted date. The brief sections of text are placed among clusters of to learn something from Jame’s overwhelming obsession with photographs. Like the mountain air, the technical content of a certain kind of landscape, and obsession that caused him to the text is rarefied; the content is mostly cultural and econom- bury one identity and create another. Will James was born in ic rather than ecological. In the sections entitled Changes and 1892 as Ernest Nephtali Dufault, in St-Nazaire, Quebec. He Future Challenges, Miller states the urgent need to preserve spent most of his spare time sketching horses and devouring the colorful cultural values and the often sound, but just as dime westerns. He was consumed by the desire to be a cow- often under-appreciated, ecological practices of the boy. At the age fifteen he bolted, and took a train west to the Himalayan nomadic pastoralists. A generous section of Canadian prairies. acknowledgements, a list of suggested readings, and a map of After spending 4 years on the Canadian prairies, James saw Miller's travels follow the 6 sections of text and photographs. the first real images of the West and cattle ranching. He shift- An appendix on "repeat photography" which ends the book ed his language from French to western English. After study- essentially allows the author to utilize several dozen more ing the life of Will James, Gayton concluded that Jame’s small photographs of landscapes. Lacking either captions or obsession was not with the land itself, but land with the other documentation, these attractive photographs are more human totally and perfectly rooted within it. Landscape with useful aesthetically than analytically. figures. As it does such places as the high plains of West Texas, the Another interesting essay concerns Gayton’s experiences in black-and-white photography suits the harsh landscapes and Saskatoon, Saskatchewan. He tells of his days as a graduate hardy peoples of the Himalayas. Even all the color of the student at the University and of living in the small village of powerful landscapes and regional costumes aren't lost, but Sutherland. He tells of his interest in cities, with their corrosive rather are reduced to images more elemental. These images and monstrous qualities. He felt it necessary to look at the arti- are at least as good in black and white as their portraits would ficial/biophysical enterprise—the city. With his interest in nat- be in color, just as on the West Texas plains, a pink Cadillac ural landscapes he takes a philosophical view of the man made convertible can look better than pink.

JOURNAL OF RANGE MANAGEMENT 52(3), May 1999 297