Variability in Predicting Edible Browse from Crown Volume

Abstract

Biomass estimates were made with regression techniques using crown volume and weight relationships. The log-log function yielded the highest coefficient of determination for Vasey shin oak, plateau oak, Texas persimmon, and honey mesquite. A quadratic function was best for woollybucket bumelia, littleleaf sumac, agarito, and pricklyash. Sugar hackberry showed equally high coefficients with either the linear or quadratic. Coefficients of determination for catclaw acacia, elbowbush, and skunkbush sumac generally were low regardless of the type of regression equation used. When sampled at various periods over the year, predictive accuracy declined for Vasey shin oak and plateau oak through fall and winter but rose again in spring and early summer. For both species, the log-log function was best from late summer to winter but during spring and early summer the quadratic function was best. Browse biomass commonly is recognized as one of the most difficult of all vegetation components to measure (Blair 1958). Even so, the information has been valuable to foresters for measuring potential flammability (Brown 1976) and to range scientists for estimating utilization by herbivores (Dalrymple et al. 1965; Ferguson and Marsden 1977), forage available to herbivores (Bryant 1977), productivity of herbivores (Leigh et al. 1970), or the response to brush control (Bentley et al. 1970; Scifres et al. 1974). Thus, numerous methods have been developed for inventory of browse or shrub yield. Some of these different methods have been reported by Hamiss and Murray (1976). Recently, techniques used for predicting browse yield have included linear regressions of crown volume on the weight of plants to predict yield. Crown volume requires at least two measurements of the plant in addition to weight, and is well suited because a combination of measurements is usually better than any single measurement (Cook 1960). Also, volume generally yields a better relationship to weight than surface area to weight (Lyon 1968). These volume-weight relationships have usually included only one species (Lyon 1968; Scifres et al. 1974; Rittenhouse and Sneva 1977). Only Bentley et al. (1970) has dealt with more than one shrub species. Previous studies have not examined the same species at different sampling dates within a 1-year period to determine the impact of defoliation on the precision of the predictive equations. Lyon (1968) has, however, found that the precision of regression equations varied with range site for the single species being studied. The authors are assistant professor, Range and Wildlife Management, Texas Tech University; and associate professor, Department of Range Science, Texas A&M University. At the time of this research, Bryant was research assistant, Range Science, Texas A&M University. Financial support from the Texas Agr. Exp. Sta. at Sonora is gratefully acknowledged. This article is TA 14210 from the Texas Agricultural Experiment Station. Manuscript received April 28, 1978. The objective of this research was to evaluate the use of various regression functions for the prediction of biomass from crown volume for several browse species and for selected species at different sampling dates. Methods and Materials Crown volume and weight relationships were used on several browse species to estimate the availability of edible browse to herbivores. Two of the species were evaluated at different sampling dates over a 1-year period. An 8-ha study site was selected at the Texas Agricultural Experiment Station, Sonora, Texas (Fig. 1). The site had been rootplowed 7 years prior. All available browse was in a young age class and less than 2.2 m tall. Since 1970, the browse had received light used by cattle, sheep, Angora goats, and white-tailed deer (Odocoileus virginianus L.). In August, 1975, the density for each species was estimated using either point-center-quarter (Cottam and Curtis 1956) or the correctedpoint-distance technique (Laycock and Batcheler 1975), depending on the degree of aggregation. Two transects, 322 m long, were used and points were established every 30.5 m. The distance to the nearest individual of each species was measured and its height and diameter at the widest point were recorded. Height and diameter measurements were converted to conical volume using the formula