264 TECHNICAL NOTES

Estimating Percentage Dry has evaluated the accuracy of the tech- species. Species of grasses were iden- Weight in Diets Using a nique for estimating dry-weight per- tified by the occurrence and position* centages of plants in the diets of herbi- of such specialized epidermal cells as Microscopic Technique1 vores. cork cells, silica cells, silica-suberose DONNIE R. SPARKS AND The objective of this study was to couples, and asperities. The size and determine if dry-weight composition of shape of the ,guard and subsidiary cells JOHN C. MALECHEKZ a mixture of grasses and forbs could of the stomata were also reliable diag- Graduate Research Assistants, Range be accurately estimated by a micro- nostic features. Science Department, Colorado State scopic technique. The mixed samples were analyzed University, Fort Collins. by examining five slide mounts of Methods and Materials each mixture under a compound bi- nocular microscope. Twenty locations Highlight Samples that contained known were systematically observed on each amounts of grasses and forbs were Percent composition by dry weight slide. A location was considered as an was accurately estimated for 15 mix- artificially mixed. Four mixtures con- area of the slide delineated by a micro- tures of plants that are found in the tained Arizona fescue (Festuca arizon- scope field using 125-power magnifica- diets of some herbivores. The mix- ica), mountain muhly (Muhlenbergia tion. Only those fragments that were tures were sampled by recording the montana), Pennsylvania cinquefoil (Po- recognized as epidermal tissue (other frequency of occurrence of each spe- tentilla pensylvanica), and fringed sage- cies in 100 microscope fields using 125- than hairs) were recorded as positive wort (Artemisia frigida). Eleven mix- power magnification, converting fre- evidence for the nresence of a slant tures included from two to four of the quency to density, and calculating rel- species at a locatidn on the slide. ‘Each following species: western wheatgrass * ative density as an estimate of percent species present for each location was (Agropyron smithii), prairie sandreed composition by dry weight. Dry weight recorded. Frequency percentages (num- (Calamovilfa Zongifolia), summercypress percentages were predicted directly ber of fields that the species occurred from relative density. The microscopic (Kochia scoparia), and alfalfa (Medi- in out of 100 locations) were tabulated technique reported in this paper would cage sa tiva) . for each species in the mixture. The be an accurate means of determining All plants used in the mixtures were frequency percentages were converted the dry-weight composition of stomach actively growing when collected. The to particle density per field using a samples, esophageal samples, rumen plant material was oven dried and samples, and clipped herbage. table developed by Fracker and ground over a one-mm screen to re- Brischle (1944) and the relative den- duce all plant fragments to a uniform A rapid, reliable method for esti- sity, expressed as a percentage, of each size. The mixtures were compounded species in the mixture was calculated. mating plant species composition in of various combinations of species so The relative density of a species was the diets of herbivorous animals is no two samples were alike. The spe- needed. This need is emphasized by used to estimate the percentage dry cies and dry-weight composition of the the variety of methods used for report- weight of that species in the mixture. mixtures were unknown to the authors ing diets (Norris, 1943; Torell, 1956; Regression equations that express until after the sample estimates were Reppert, 1960; Vaughan, 1967). the relationshin between estimated recorded. Mixtures were washed over percentage dry *weight (X) and actual A problem encountered in the study a 200-mesh screen to insure mixing, to percentage dry weight (Y) were devel- of herbivore diets is the lack of a remove dirt, and to remove very small oped for three categories: grasses, forbs, method that can be checked for accu- plant fragments. A small portion of and grass-forb combinations. racy. The technician needs a method the mixture was spread evenly and that builds confidence in his own abil- mounted on a microscope slide using Results and Discussion ity to estimate the composition of a Hertwig’s Solution (Baumgartner and diet. The procedure should be as sim- Martin, 1939) and Hoyer’s Solution Prediction equations for grasses, ple as possible. Botanical analysis of (Baker and Wharten, 1952). The slides forbs, and grass-forb combinations are the diet should accurately reflect both were oven dried at 60 C. Five slides shown in Fig. 1, 2, and 3. The ratio the plant species eaten and the amount were prepared from each mixture. between estimated dry weight percent- of each that was eaten. Tissues of plants that were used in ages (relative densityj and actual dry A microscopic technique for identi- the mixtures were prepared and weight percentages was approximately fication of plants eaten by herbivores mounted on microscope slides in the 1 : 1 for all three categories. Student’s that thoroughly masticate their food same manner for study as reference t-test showed that there was no signifi- was described by Baumgartner and material. Identification of each spe- cant difference (P < .Ol) between re- Martin (1939) and the technique was cies in the mixtures was based on epi- gression equatio& for grasses and forbs later refined by Dusi (1949). This basic dermal characteristics (Davies, 1959; and that the calculated regression technique has been employed fre- Brusven and Mulkern, 1960; Storr, equations for grasses, forbs, and grass- quently in recent years by numerous 1961). These workers found that epi- forb combinations were not statistically other researchers. However, no one dermal characteristics of grasses and different from the equation Y = X. forbs were variable with different Therefore, the percent composition lPublished with the approval of the stages of maturity. Histological fea- based on dry weight of the mixtures Director of the Colorado Agricultural tures such as size and shape of epider- could be predicted directly from the Experiment Station as Scientific Series ma1 hairs, presence or absence of hairs, relative density. Paper No. 1230. 2 Present address: Range Science De- cell shapes, and crystals included in For 11 of the 15 mixtures, the num- partment, Texas A&M University, epidermal cells provided diagnostic ber of epidermal fragments of each College Station. characteristics for identification of forb species at a location was recorded in 265

Y = 1.20 + 0.962 X J

00 * r’s .98

60 ; 60 i m .(D 3” T i 40 = 40. ” : . 2 a

0 10 40 60 80 100

Rolatave Density Relative Density Rclat,ve Denroty

FIG. 1. Relationship of relative density to FIG.2. Relationship of relative density to FIG. 3. Relationship of relative density to actual percent composition by weight actual percent composition by weight actual percent composition by weight for four grasses. for four forbs. for four grasses and four forbs. addition to recording frequency. An consistent with other species or at DAVIES, I. 1959. The use of epider- estimate of the dry-weight percentages other stages of maturity. However, ma1 characteristics for the identifi- for each species was computed from unless the slant Darts in. the diet be- cation of grasses in the leafy stage. the totals. A paired t-test showed no ing analyze; are &ossly different from Brit. Grassland Sot. J. 14:7-16. significant difference (P < .Ol) be- these in our studies, the added accu- DUSI, J. L. 1949. Methods for the tween the estimates obtained from the racy gained by using a prediction determination of food habits by “particle count” technique and the equation more complicated than Y = plant microtechniques and histology “frequency conversion” technique. Ei- X will probably not be worthwhile. If and their application to cottontail ther can be used with a similar degree the specific gravity of a species or the rabbit food habits. J. Wildl. Man- of accuracy, but it is much easier and phenological stage of a species is sus- age. 13:295-298. faster for the technician to determine pected of being different from that FRACKER, S. B., AND J. A. BRISCHLE. if a species is present or absent than used in this study, the investigator can 1944. Measuring the local distribu- to count all recognizable fragments. easily compound mixtures and develop tion of Ribes. Ecology 25:283-303. The “frequency conversion” technique regression equations for estimating dry- HEADY, H. F., AND G. M. VAN DYNE. reported in this paper is recommended. weight percentages. 1965. Prediction of weight compo- There are two requirements that The microscoDe techniaue renorted sition from point samples on clipped must be met before frequency percent- in the present ‘paper cokd also be herbage. J. Range Manage. 18:144- age can be converted to density (Curtis used to determine the species composi- 148. tion of clipped herbage and the weights and McIntosh, 1950). The plant frag- NORRIS, J. J. 1943. Botanical analy- contributed by each species present. ments must be distributed randomly ses of stomach contents as a method Thus, it should be possible to accu- over the slide, and the density of par- of determining forage consumption rately estimate the yields of individual ticles must be such that the most com- of range sheep. Ecology 24: 244-25 1. mon species does not occur in more components in botanical studies. REPPERT, J. N. 1960. Forage prefer- than 86% of the microscope fields. A LITERATURE CITED ence and grazing habits of cattle at random distribution can be attained the Eastern Colorado Range Station. by reducing the particles to a uniform BAKER, E. W., AND B. W. WHARTEN. J. Range Manage. 13:58-65. size and thoroughly mixing them. A 1952. An introduction to acarology. STORR, G. M. 1961. Microscopic anal- frequency of less than 86% for the MacMillan Co., New York. 465 p. ysis of faeces, a technique for ascer- most common species can be main- BAUMGARTNER, L. L., AND A. C. MAR- taining the diet of herbivorous mam- tained by adjusting the amount of ma- TIN. 1939. Plant histology as an mals. Austral. J. Biol. Sci. 14:157- terial mounted on the slide. This is aid in squirrel food-habit studies. J. 164. done by trial and error until the tech- Wildl. Manage. 3:266-268. TORELL, D. T. 1956. Development nician becomes familiar with the ma- BRUSVEN, M. A., AND G. M. MULKERN. of a point system for plant composi- terial. 1960. The use of epidermal charac- tion analyses of hand-clipped and Storr (1961) and Heady and Van teristics for the identification of Dyne (1965) reported that weight per plants recovered in fragmentary con- fistula samples for nutrition studies, unit area of plant material is not con- dition from the crops of grasshop- p. 2. In West. Regional Res. Proi., sistent at different stages of maturity pers. North Dakota Agr. Exp. Sta. W-34, Annual Prog. Rep. California nor is it consistent from species to spe- Res. Rep. No. 3. 11 p. Agr. Exp. Sta. cies. The 1 : 1 relationship between CURTIS, J. T., AND R. P. MCINTOSH. VAUGHAN, T. A. 1967. Food habits estimated dry-weight percentages and 1950. The interrelations of certain of the northern pocket gopher on actual dry-weight percentages for the analytic and synthetic phytosociolog- shortgrass prairie. Amer. Midland species used in this study may not be ical characters. Ecology 3 1:434-455. Natur. 77: 176-189.