Effects of individual variation in size on growth and development of larval salamanders

Ecology, July, 1996 by Paul E. Brunkow, James P. Collins

INTRODUCTION

Traditional approaches to modeling biological populations often ignore variation among individuals. Parameters defining the behavior of these models, such as r in the Verhulst-Pearl logistic model, [Gamma] and [Lambda] in Schoener's (1973) energy-partitioning model, and [[Phi].sub.ij] in Stewart and Levins' (1973) model are defined as the average of that value for all individuals in the population. Thus, all individuals are assumed to behave like the "average" individual. While this assumption makes these models more tractable mathematically, ecologists have long speculated on whether inclusion of individual variability would qualitatively alter these models' predictions. Such speculations stem from at least two sources: first, the interest of theoretical ecologists in determining whether model equations are affected by including some measure of the variance of a parameter in a model as well as the parameter itself (e.g., Bjornstad and Hansen 1994); and second, the desire to incorporate processes at the individual level into our understanding of patterns and processes at the population level (Hassell and May 1985, Schoener 1986, Kingsolver and Paine 1991).

Recent population models have examined the effects of phenotypic variability by defining homogeneous subclasses of individuals (Schoener 1973, 1986, Begon and Wall 1987) or by defining each individual explicitly (Lomnicki 1978, 1988, DeAngelis and Gross 1992). Other verbal and empirical models have suggested that exploitation of different niches by variable individuals may reduce intraspecific competition, thereby allowing for greater population density and higher average individual fitness (Van Valen 1965, Den Boer 1968, Roughgarden 1972, Polis 1984). All of these models suggest that variability at the individual level can affect population dynamics; however, this hypothesis has been rarely tested in natural populations. An experimental design to test this hypothesis requires populations that differ in the level of variation between individuals in some ecologically relevant trait, such as body mass, but that have equal mean individual mass (Smith 1990). The effects of this treatment on variables like population growth rate, survivorship, equilibrial density, and population stability can then be examined through a long-term study (long-term relative to the life-span of the organism). Detecting a treatment effect in any or all of these variables would indicate that variability among individuals per se could affect population dynamics of a species.

Effects observed at the population level can, however, be mediated through changes in the behavior and physiology of individuals comprising the population (Schoener 1986). The models discussed above all assume to some degree that physiology and behavior of individuals are independent of the degree of variation between individuals within populations. In many situations, however, this assumption may be invalid (Maynard Smith and Brown 1986). Studies describing evolutionarily stable strategies (Maynard Smith 1982) and ideal-free distributions (Parker and Sutherland 1986) conclude that behavior of one individual may depend strongly on behaviors (phenotypes) of other individuals in the population (Wankowski and Thorpe 1979, Dill 1983, Holbrook and Schmitt 1992). Thus, we can profitably examine the effect of individual variability in body mass on growth and development of a single cohort of individuals at a certain point in their life history (Smith 1990).

We report results of a field experiment that tested joint effects of individual variation in body mass and density on growth and development of cohorts of larval Arizona tiger salamanders (Ambystoma tigrinum nebulosum Hallowell). Amphibian populations are generally most strongly regulated in their larval phase (Wilbur and Collins 1973, Wilbur 1980). Growth rate during this phase has strong effects on larval survival by mediating risk of predation (Wilbur 1972, Travis 1980a, b) and cannibalism (Collins and Holomuzki 1984), as well as by reducing risk of mortality in a rapidly drying pond (Scott 1990). Body size at metamorphosis also affects postmetamorphic survival and reproduction (Wilbur 1980, Semlitsch et al. 1988). We expected that changing the distribution of sizes within a cohort would change the distribution of competitive relationships among individuals because of size-dependent differences in diet and feeding rate. We then predicted that such competitive changes would affect larval growth and development. In this experiment, effects of variation in body mass were examined independently of any effects due to mean body size or total biomass of experimental populations.

METHODS

The study system

Populations of Ambystoma tigrinum nebulosum occur naturally throughout montane regions of Arizona, New Mexico, and Colorado (Collins 1981). Adults deposit embryos singly or in loose clumps in the shallow portions of temporary ponds from early March through April. Upon hatching, larval salamanders feed on crustacean zooplankton and grow rapidly. A small percentage of larvae may develop the "cannibal" morphology (Collins et al. 1993) and begin feeding on other larvae; "typical" larvae retain the standard morphology and continue feeding on zooplankton, aquatic insects and insect larvae, snails, and other molluscs (Holomuzki and Collins 1987). Larval densities can vary widely among ponds and between years in the same pond.