What resources are available to desert granivores: seed rain or soil seed bank?

Ecology, April, 1997 by Mary V. Price, Jamie W. Joyner

INTRODUCTION

Spatial and temporal heterogeneity in the availability of resources molds ecological processes at many levels, from the behavior and morphology of individuals to the dynamics of populations, coexistence of species, and functioning of ecosystems. Despite the central importance of resources, however, we can only rarely measure resource availability precisely or understand thoroughly the processes that determine it. This ignorance constrains our ability to answer many ecological questions.

It is not that ecologists have been lazy; several difficulties confront those who try to measure the availability of resources to consumers in nature. First, some forms of a resource may be unavailable because they are toxic, of poor quality, or difficult for consumers with particular morphologies or sensory abilities to detect or process; we need to account for this and measure only usable resources. Second, what we record, generally with random sampling, is likely to underestimate the resources actually encountered by the consumer. The animal may use a nonrandom search strategy that yields higher encounter rates than does random search, it may respond to cues we do not detect, or it may be knowledgeable about the recent history of exploitation within a particular foraging area and avoid searching in depleted areas (Possingham 1989). Finally, and most importantly, we often characterize resource availability in terms of the standing crop, the abundance at one place and time, rather than in terms of cumulative resource input and consumption. Whereas such a "snapshot" of resources may yield an accurate picture of resource availability at a moment in time, it is unlikely to give an accurate measure of availability over longer time periods because resources are dynamic. Equal standing crops can result from high rates of production balanced by high rates of consumption, or from low production balanced by low consumption, but cumulative resource availability to consumers is higher in the first case. In assessing resource availability over any substantial period of time, therefore, it is insufficient to measure the standing crop; one also should measure turnover (Wiens 1984).

Assemblages of specialized granivores that inhabit North American deserts are well-studied by ecologists, and our knowledge of them illustrates several of the points just raised. On the one hand, we now understand a good deal about properties of seeds and of seed patches that affect the ability of these animals to find, harvest, and ingest resources (Brown et al. 1979, Price and Jenkins 1986, Kotler and Brown 1988, Vander Wall 1991, 1993, Crist and MacMahon 1992, Randall 1993, Reichman and Price 1993). We know that changes in the size of granivore populations closely reflect precipitation-related changes in seed production (e.g., Brown et al. 1979, Brown and Harney 1993). We also know that the structure of granivore communities is correlated with average productivity (Brown et al. 1979, Brown and Harney 1993) and with spatial and temporal heterogeneity in factors that affect the costs and benefits of foraging (Price and Brown 1983, Kotler and Brown 1988, Brown and Harney 1993, Reichman and Price 1993). We know much less about the dynamics of seed input and consumption. With considerable effort, workers have extracted seeds from samples of the top few centimeters of desert soils, under the assumption that this measures at least the standing crop of resources available to granivores (e.g., French et al. 1974, Nelson and Chew 1977, Mehlhop and Scott 1983, Reichman 1984, Hassan and West 1986, Price and Reichman 1987, Kemp 1989, Crist and MacMahon 1992). A handful of studies has gone further to estimate seed production by plants, or input to the soil surface via the "seed rain" falling into seed traps (Tevis 1958, Soholt 1973, French et al. 1974, Pulliam and Brand 1975, Whitford 1978). We are aware of only two studies (French et al. 1974, Pulliam and Brand 1975) that have attempted to compare patterns of standing crop in the soil seed bank with the seed rain and seed consumption, and thus to more completely characterize resource dynamics.

Our purpose in this paper is to add to the sparse knowledge of resource dynamics in desert granivore systems by reporting on patterns of temporal and spatial variation in the soil seed bank and seed rain in the Mojave Desert. From discrepancies between patterns for the standing crop and input, we conclude that granivores at this site may make little use of the soil seed bank, contrary to what previous workers have assumed (ourselves included, e.g., Price and Reichman 1987). Instead, the animals appear to rely mostly on newly produced seeds. This possibility has major implications for our understanding of the community ecology of desert granivores, some of which we discuss.

THE STUDY SITE

We worked at 1250 m elevation in the eastern Mojave Desert, 135 km east of Barstow, California, USA (Flynn 15[minutes] USGS quadrangle, T8N R12E S18; 34 [degrees] 47[minutes] N, 115 [degrees] 39[minutes] W), in Granite Cove, a portion of the University of California's Sweeney Granite Mountains Desert Research Center. Cattle were removed from the Center when it was established in the early 1980s. The study site is on a gently sloping piedmont, with soils of coarse sands derived from weathered granite. The site supports a diverse desert shrubland vegetation with elements from the Sonoran, Mojave, and Great Basin Deserts (Thorne et al. 1981). Common shrubs include Acacia greggii, Yucca schidigera, Larrea tridentata, Hymenoclea salsola, Coleogyne ramosissima, Salazaria mexicana, and Acamptopappus sphaerocephalus. Most precipitation results from winter frontal storms, but summer convective storms bring significant rainfall in some years. Yearly totals from 1986 through 1994 averaged 23.8 cm, with 63% falling in cool months (November-April) and 37% in warm months (May-October).