Functional redundancy, non-additive interactions and supply-side dynamics in experimental pond communities.
Ecological experiments have taught us much about the roles of competition and predation in organizing communities (Hairston 1989). Unfortunately, the results of individual experiments can depend critically on site-specific densities of responding species or on the particular choice of manipulated species. General conclusions about the functional roles of broad groups of species, such as top predators, can be skewed by the small sample of species selected for experimental manipulations (see Paine 1992). These uncertainties emphasize the need for experiments designed to address several questions. First, do different species that occupy similar trophic positions have similar effects on community organization, that is, are they functionally redundant (see Lawton and Brown 1993)? Second, are the effects of different species additive, such that aggregate multispecies effects can be predicted by summing the separately measured effects of each species? Third, do the responses to manipulations of these different species depend critically on the densities of responding species? This paper addresses these questions through an experimental study of four community-level phenomena that arise in food webs of modest complexity. First, I describe the somewhat surprising functional equivalence of two predator species, as determined by measures of the strengths of their interactions with a common assemblage of prey. Second, I test the additivity of the impacts of two predator species on an assemblage of three prey species, to assess whether joint effects of two predator species acting in concert are predictable from separate effects of each predator acting in the absence of the other. Third, I explore whether these results depend on variation in initial conditions, by testing whether the initial density of prey in the community influences the outcome of predator-prey interactions. This potential dependence on initial conditions can be considered an aspect of supply-side ecology (see Roughgarden et al. 1988). Fourth, since my experimental system generated situations where predation on herbivores might also reduce the grazing pressure on producers, via a trophic cascade, I also looked for indirect positive effects of predators on producer abundance. More detailed reasons for considering each of these phenomena are outlined below.
Measurement of interaction strengths can show if predators in similar trophic positions have substantially different impacts on initially similar communities. Relative interaction strengths are seldom measured in most communities (see Paine 1992 for an exception). Interaction strengths between the species in a guild of predators and their prey can be estimated by direct comparisons of per capita impacts of different predator species on the same prey communities. Descriptions of food webs often lump many species into feeding groups (Cohen 1978), implying that different predator species have similar interaction strengths or are functionally equivalent. Such assumptions may be risky (see Paine 1992) and are only warranted if predators can be shown to be functionally interchangeable.
Non-additive per capita effects (higher-order interactions) present other problems for the prediction of community patterns (see Strauss 1991, Worthen and Moore 1991, Wissinger and McGrady 1993, Wootton 1993). When joint impacts of two or more species on others are non-additive, it becomes difficult to predict the structure and function of species-rich communities by simply summing up the effects observed between pairs of species in more depauperate communities (Neill 1974, Strauss 1991). Communities generally contain many predator species, making it important to know whether the effects of individual predator species change dramatically in the presence of other predator species.
In communities structured by an interplay between competition among prey and predation, predation's importance may vary strongly with temporal or spatial variation in prey densities (Gaines and Roughgarden 1985, Fairweather 1988, Underwood and Fairweather 1989). Colonization or recruitment by high densities of prey can create situations where intense competition excludes some prey species. Predation in high-density situations can sufficiently reduce prey abundances to moderate interprey competition and thereby enhance prey species richness. Where potential competitors colonize or recruit at low densities, negligible levels of interspecific competition may result, regardless of the effects of predators. Predation in low-density situations would not be expected to moderate already negligible levels of interprey competition. Instead, predators might simply influence such low-density non-competitive communities by eliminating relatively vulnerable prey species (e.g., Addicott 1974). Spatial variation in prey density may also influence the movements and aggregation of mobile predators (Fairweather 1988), further complicating the intensity of interactions between predators and prey.
In this study, factorial manipulations of the densities of two predator species and the initial density of prey (anuran larvae) in artificial ponds permitted tests of the equivalence of interaction strengths and additivity of predator impacts under different initial conditions of prey density. The two predators (the salamanders Notophthalmus viridescens and Ambystoma opacum) and three herbivorous prey (the larval anurans Pseudacris crucifer (formerly Hyla crucifer), Bufo woodhousii fowleri, and Hyla andersonii) are often sympatric in natural ephemeral ponds in New Jersey and other parts of the eastern USA.
I studied the effects of predation by adult Notophthalmus viridescens, the spotted newt, and larval Ambystoma opacum, the marbled salamander. Adult Notophthalmus and larval Ambystoma often coexist in ponds in early spring. Much is known about the population biology of both species (Gill 1978, 1979, Morin 1983a, b, Morin et al. 1983, Stenhouse 1985, Harris 1987, Petranka 1989, Scott 1990, Smith 1990, Fauth and Resetarits 1991). Both species are generalist predators (Wood and Goodwin 1954, Stewart 1956, Gill 1978) although they differ in many other respects. Notophthalmus can act as a keystone predator (see Fauth and Resetarits 1991 for a review), and A. opacum can strongly affect community organization (Stenhouse et al. 1983, Cortwright and Nelson 1990). Less is known about the relative per capita impacts of Notophthalmus and Ambystoma as predators, or about their direct interspecific interactions.
Adult Notophthalmus and larval Arabystoma opacum are already present in ponds when anuran tadpoles begin to hatch in the spring. Both predators feed readily on small anuran tadpoles. Adult Notophthalmus can also prey on small Ambystoma larvae, although larval Ambystoma rapidly grow too large for Notophthalmus to consume. In fact, two of the Notophthalmus collected for this experiment regurgitated small A. opacum larvae. Large size and toxicity make adult Notophthalmus invulnerable to predation by A. opacum. Adult Notophthalmus can occur in ponds throughout the year, while larval Ambystoma usually metamorphose and leave ponds in late spring, often at about the same time that larval Notophthalmus hatch (see Worthington 1969). This phenological separation makes direct interactions between larval Notophthalmus and larval Ambystoma unlikely, Larval Notophthalmus feed primarily on plankton and small invertebrates, and they do not appear to be important predators on anuran larvae (Brophy 1980). Terrestrial adults of Ambystoma opacum oviposit in the basins of dry ephemeral ponds in autumn. Arabystoma eggs hatch when ponds fill with winter or spring rains, ensuring that larvae enter the pond as early as possible and have maximal opportunities for growth.
The focal prey, larval anurans of the species Pseudacris crucifer (= Hyla crucifer), Bufo woodhousii fowleri, and Hyla andersonii frequently occur together in the Pine Barrens of southern New Jersey. These larval anurans feed primarily on phytoplankton and periphyton (Morin and Johnson 1988, Morin et al. 1990). Pseudacris crucifer is competitively inferior, but it manages to coexist with some predators, including Notophthalmus (Morin 1983b, Fauth and Resetarits 1991). The relative abilities of the other two anuran species to resist predators or compete were unknown at the outset of this study.
METHODS AND MATERIALS
This experiment used eight treatments consisting of all possible combinations of three factors (initial prey density, Notophthalmus abundance, and Ambystoma abundance) manipulated in an array of artificial ponds. The rationale for the experimental design was as follows. Stocking two different initial densities of hatchling anurans (low = [200 Pseudacris crucifer + 200 Bufo woodhousii + 100 Hyla andersonii]/[m.sup.3]; high = [600 Pseudacris crucifer + 600 Bufo woodhousii + 300 Hyla andersonii]/[m.sup.3]) mimicked different levels of recruitment while holding the initial relative abundances of species constant. The increase in initial density provided a way to manipulate the potential intensity of competition among the tadpoles (as in Morin 1986). Repeating each of two initial tadpole densities with or without each species of predatory salamander (6 or 0 adult Notophthalmus viridescens/[m.sup.3]; 6 or 0 larval Ambystoma opacum/[m.sup.3]) permitted comparisons of predation at different prey densities. Adding both predators ([6 adult Notophthalmus viridescens + 6 larval Arabystoma opacum]/[m.sup.3]) to ponds containing different initial densities of tadpoles permitted tests of the additivity of the separately measured effects of each predator species. These treatments, however, could not separate non-additive effects caused by interactions among predator species from non-additive effects generated by nonlinear prey responses to increasing predator density. Initial densities of predators and prey fell within the range observed in natural ponds (P. J. Morin 1983b, personal observation). Three replicates of each of the eight treatment combinations were randomly assigned within an array of 24 small artificial ponds (see Morin 1981, 1989).
The 24 artificial ponds (cylindrical steel cattle-watering tanks, 1.52 m in diameter and 0.61 m deep) are near the lower end of the size range of natural temporary ponds. Similar ponds have been used in previous experiments (Morin 1986) at the W. L. Hutcheson Memorial Forest of Rutgers University (Somerset County, New Jersey, USA). An interior coat of white epoxy paint retarded rust and covered the galvanized surface with an inert barrier. Each tank contained [approximately equal to]1000 L of water pumped from nearby Spooky Brook in March 1989, grassy litter (800 g), commercial trout food (50 g), and macrophytes (30 stems of washed Elodea canadensis). Each tank received a 0.460-L inoculum of infusoria, algae, and microinvertebrates suspended in pond water. Each inoculum was a subsample of pooled, mixed plankton collections from several ponds in Ocean County, New Jersey, USA. Lids of fiberglass window screening attached to hexagonal wooden frames retained metamorphosing amphibians and prevented uncontrolled colonization by insects.
Experimental design and responses
The experiment used a 2 x 2 x 2 complete factorial design with three replicates of each combination of the three factors: initial tadpole density, presence/absence of six adult Notophthalmus per tank and presence/absence of six larval Ambystoma per tank. Additions of amphibians took place on 28 April 1989 (Notophthalmus and Ambystoma), 5 May 1989 (Pseudacris and Bufo), and 13 May 1989 (Hyla). Daily collections of metamorphosed amphibians yielded a complete census of survivors from each pond. Amphibians were collected until no further metamorphs appeared. The last stragglers of Pseudacris, Bufo, and Hyla appeared on 25 July, 19 July, and 29 September, although the vast majority of amphibians completed metamorphosis within 1-2 mo after the start of the experiment. Data collected for each survivor included wet mass at tail resorption (in milligrams) and date of collection. Summary statistics for populations of each species collected from each artificial pond included survival to metamorphosis (percentage of the initial number stocked collected as metamorphs), mean mass at tail resorption (in milligrams), mean larval period (days between stocking and metamorphosis), and a linear approximation of mean growth rate in milligrams per day (mean mass/mean larval period). Final anuran species composition, one focus of the analysis, was defined by the vector of relative abundances of metamorphosed froglets of each anuran species, as in Morin (1983b). Species composition was defined by the vector, [P.sub.j] = ([p.sub.1j], [p.sub.2j], [p.sub.3j]), where [p.sub.ij] is the fraction of all the surviving anurans in tank j belonging to species i (Pseudacris, Bufo, or Hyla).
Analysis of effects of initial anuran density, Notophthalmus, and Ambystoma on anuran species composition
A three-factor MANOVA tested whether the vector of anuran relative abundances outlined above depended on initial anuran density, presence/absence of six adult Notophthalmus, and presence/absence of six larval Ambystoma per tank. The main null hypotheses tested were: (1) no effect of initial anuran density, (2) no effect of Notophthalmus, and (3) no effect of Ambystoma. Significant interactions between initial tadpole density and each predator species would indicate that the impact of each predator species on species composition depended on initial tadpole density. A significant interaction between Notophthalmus and Ambystoma would indicate that the joint impact of both predators on species composition was not simply the sum of the separately measured impacts of each predator species (i.e., non-additivity). A significant three-way interaction would indicate that the impacts of all three factors on anuran species composition were interdependent. A MANOVA provided conservative tests of each hypothesis adjusted for the analysis of all variables. ANOVAs of each variable aided the interpretation of the MANOVA and clarified differences in the responses of the three anuran species to the same predators. The choice of statistical methods has been justified elsewhere (Morin 1983b).
TABLE 1. MANOVA for effects of initial tadpole density (Density), Notophthalmus (Newt), and Ambystoma opacum (Ambystoma) on the composition of the anuran guild.
a) MANOVA test criteria (Wilks' lambda) and F statistics
Wilks' Num. Den. Source of variation lambda F df df P
Density 0.6091 4.81 2 15 0.0243 Newt 0.0126 585.99 2 15 0.0001 Density x Newt 0.6091 4.81 2 15 0.0243 Ambystoma 0.0129 571.32 2 15 0.0001 Density x Ambystoma 0.6654 3.76 2 15 0.0472 Newt x Ambystoma 0.0129 571.32 2 15 0.0001 Density x Newt x Ambystoma 0.6654 3.76 2 15 0.0472
b) ANOVA, Relative abundance of Pseudacris crucifer
Source of variation df Type I ss F P
Density 1 0.00046 0.78 0.3891 Newt 1 0.50088 845.34 0.0001 Density x Newt 1 0.00046 0.78 0.3891 Ambystoma 1 0.45970 775.83 0.0001 Density x Ambystoma 1 0.00015 0.26 0.6202 Newt x Ambystoma 1 0.45970 775.83 0.0001 Density x Newt x Ambystoma 1 0.00015 0.26 0.6202 Error 16 0.00948
c) ANOVA, Relative abundance of Hyla andersonii
Source of variation df Type I ss F P
Density 1 0.0028 7.46 0.0148 Newt 1 0.0235 62.30 0.0001 Density x Newt 1 0.0028 7.46 0.0148 Ambystoma 1 0.0153 40.52 0.0001 Density x Ambystoma 1 0.0019 5.09 0.0385 Newt x Ambystoma 1 0.0153 40.52 0.0001 Density x Newt x Ambystoma 1 0.0019 5.09 0.0385 Error 16 0.0061
Analysis of effects of initial anuran density, Notophthalmus, and Ambystoma on anuran performance
Summary responses (survival, mean mass at metamorphosis, mean larval period, and mean growth rates) of each anuran species were analyzed using a three-factor MANOVA as outlined above. Only Pseudacris produced some survivors under all treatment combinations, while Bufo and Hyla were eliminated from most ponds containing predators. Consequently, the analysis of responses other than survival for Bufo and Hyla is restricted to comparisons between salamander-free ponds containing low and high initial tadpole densities. ANOVAs identify the individual variables that contributed to significant differences among treatments. Two statistical models explored the joint effects of both predators on Pseudacris survival. An additive model, based on untransformed survival data, provided a simple test of whether the joint effects of both predators, measured as decreases in survival, are predicted by the separately measured effects of each predator treatment. Artifactual non-additive effects can result when predicted joint effects yield a survival of [less than]0, which will yield a significant interaction in the ANOVA. In contrast, a multiplicative model, obtained by an analysis of logarithmically transformed survival data, is not subject to the potential artifact of predicting negative values for survival (Wootton 1993) and tests whether the probability of survival in the presence of either predator species alone remains independent when both predators are present together. The additive and multiplicative models were compared only for Pseudacris, which produced some survivors in all replicates. For other anuran species, only additive models were explored.
An analysis of covariance (ANCOVA) further explored whether relations between final intraspecific density and growth rate of Pseudacris were influenced by the presence of predators. Data were pooled into two groups, corresponding to results from ponds with or without predators. A significant negative relation between the covariate (final density of surviving Pseudacris) and mean growth rate would indicate intraspecific density dependence of growth within each group. Differences between groups in the slope or elevation of relations between density and growth rate would indicate shifts in density dependence associated with the presence or absence of predators.
Tritrophic level effects: analysis of phytoplankton and periphyton
I also measured the effects of initial anuran density (herbivore density) and each salamander species (predators on the herbivores) on the standing crop of periphyton and phytoplankton sampled in each of the 24 tanks. Sampled periphyton grew on glass microscope slides (25 x 75 mm; see Morin et al. 1988 for a description of the sampling apparatus) placed in the tanks on 30 April 1989 and harvested without replacement at weekly intervals for the first 4 wk of the experiment. Phytoplankton was filtered from 500-mL samples of water collected on each sampling date. Spectrophotometric determinations of chlorophyll concentration followed the trichromatic method of Strickland and Parsons (1968), yielding estimates of periphyton standing crop in micrograms of chlorophyll a per slide and phytoplankton standing crop in micrograms of chlorophyll a per litre. Tadpoles of all three anuran species grazed on algae in the artificial ponds.
A repeated-measures ANOVA of the abundance of periphyton on four successive sampling dates tested for effects in these 1-[m.sup.3] ponds of (1) initial anuran density, (2) presence/absence of six adult Notophthalmus, (3) presence/absence of six larval Ambystoma, and all interactions among these factors.
Interactions among predators
When larval Ambystoma metamorphosed, they were collected and processed as described above for anurans. Their responses to two factors, initial tadpole density and Notophthalmus presence/absence, were measured by a two-factor MANOVA on mean mass, mean head width, and snout-vent length, Larvae were randomly assigned to treatments at the outset of the experiment, and there were no initial differences in these measures among treatments (mass = 0.410 [+ or -] 0.081 g [mean [+ or -] 1 SD], head width = 6,8 [+ or -] 0.4 mm, and snout-vent length = 20.6 [+ or -] 1.6 mm). Larvae of this size were too large for Notophthalmus to capture. Any subsequent differences among treatments could be attributed to ongoing interactions with predators or prey, rather than initial bias.
TABLE 2. Effects of tadpole density, Notophthalmus (Newt), and Ambystoma opacum (Ambystoma) on the response variables measured for Pseudacris crucifer.
Wilks' Num. Den. Source of variation lambda F df df P
Density 0.1497 18.45 4 13 0.0001 Newt 0.1083 26.74 4 13 0.0001 Density x Newt 0.6356 1.86 4 13 0.1772 Ambystoma 0.1038 28.03 4 13 0.0001 Density x Ambystoma 0.7682 0.98 4 13 0.4516 Newt x Ambystoma 0.1984 13.13 4 13 0.0002 Density x Newt x Ambystoma 0.1745 0.56 4 13 0.6909
b) ANOVA, Survival (untransformed)
Source of variation df Type I ss F P
Density 1 0.00243 0.26 0.6204 Newt 1 0.62135 65.12 0.0001 Density x Newt 1 0.00132 0.14 0.7143 Ambystoma 1 0.81831 85.76 0.0001 Density x Ambystoma 1 0.00354 0.37 0.5508 Newt x Ambystoma 1 0.15493 16.24 0.0010 Density x Newt x Ambystoma 1 0.00575 0.60 0.4487 Error 16 0.15266
c) ANOVA, Survival (log transformed)
Source of variation df Type I ss F P
Density 1 0.2468 6.62 0.0204 Newt 1 2.7088 72.66 0.0001 Density x Newt 1 0.1454 3.90 0.0658 Ambystoma 1 2.9927 80.27 0.0001 Density x Ambystoma 1 0.2347 6.30 0.0232 Newt x Ambystoma 1 0.1453 3.90 0.0658 Density x Newt x Ambystoma 1 0.2057 5.52 0.0320 Error 16 0.5965
d) ANOVA, Mass
Source of variation df Type I ss F P
Density 1 25434.5 28.50 0.0001 Newt 1 103 031.5 115.45 0.0001 Density x Newt 1 3415.3 3.83 0.0681 Ambystoma 1 76 602.7 85.83 0.0001 Density x Ambystoma 1 2092.5 2.34 0.1452 Newt x Ambystoma 1 39 894.2 44.70 0.0001 Density x Newt x Ambystoma 1 505.0 0.57 0.4628 Error 16 14 279.4
e) ANOVA, Growth rate
Source of variation df Type I ss F P
Density 1 17.41 53.42 0.0001 Newt 1 38.67 118.65 0.0001 Density x Newt 1 1.39 4.29 0.0549 Ambystoma 1 32.16 98.70 0.0001 Density x Ambystoma 1 0.57 1.76 0.2037 Newt x Ambystoma 1 17.34 53.22 0.0001 Density x Newt x Ambystoma 1 0.24 0.74 0.4026 Error 16 5.21
f) ANOVA, Larval period
Source of variation df Type I ss F P
Density 1 32.90 18.38 0.0006 Newt 1 16.83 9.40 0.0074 Density x Newt 1 0.05 0.03 0.8688 Ambystoma 1 4.08 2.28 0.1504 Density x Ambystoma 1 0.51 0.29 0.6007 Newt x Ambystoma 1 1.76 0.98 0.3361 Density x Newt x Ambystoma 1 0.15 0.08 0.7756 Error 16 28.64
The adult Notophthalmus added to the ponds were somewhat larger than larval Ambystoma (mass = 2.196 [+ or -] 0.370 g [mean [+ or -] 1 SD], head width = 7.4 [+ or -] 0.4 mm, and snout-vent length = 42.0 [+ or -] 2.4 mm). Newts were added to tanks at a ratio of 2 females to 4 males. Notophthalmus reproduced during the experiment, and their metamorphosed larvae were collected to measure effects of initial tadpole density and Ambystoma on reproduction. Metamorphosed newts continued to emerge from the tanks long after the other species had departed, with the last metamorph being collected on 25 September. Their responses to two factors, initial tadpole density and Ambystoma, were measured by a two-factor MANOVA on abundance, mean mass, snout-vent length, and day of metamorphosis. Adult newts were difficult to sample without disrupting the rest of the community, so their growth and survival in the different treatment combinations remained unmeasured. Adult newts are toxic and are avoided by most vertebrate predators, so predation by Ambystoma on adult Notophthalmus was considered to be unlikely.
Anuran species composition
Initial tadpole density, Notophthalmus presence/absence, and Ambystoma presence/absence all influenced anuran species composition (Table 1). Notophthalmus and Ambystoma had very similar effects on the assemblage of larval anurans; each predator species essentially deleted Bufo and Hyla from the ponds, creating a community dominated by surviving tadpoles of Pseudacris [ILLUSTRATION FOR FIGURE 1 OMITTED]. The effects of initial tadpole density depended on the presence or absence of predatory salamanders (see the density x newt x Ambystoma interaction in Table 1). The pattern of species composition generated by the joint impact of Notophthalmus and Ambystoma was very similar to that generated by either predator alone. The only difference between the impact of the two predators was that a very small number of Hyla tadpoles (1-2 per tank) managed to survive in tanks where Ambystoma was the only predator. No Hyla survived in ponds containing only Notophthalmus or Notophthalmus together with Ambystoma.
Effects of initial tadpole density on final species composition appeared only in ponds without predators (Table 1, [ILLUSTRATION FOR FIGURE 1 OMITTED]). High initial density decreased the relative abundance of Hyla. Hyla appeared to be com-petitively squeezed out of the community at high initial densities of larval anurans. The relative abundance of Pseudacris and Bufo both increased slightly, and the ratio of Bufo metamorphs to Pseudacris metamorphs remained essentially unchanged, at higher initial densities.
TABLE 3. ANCOVA for the effects of predators on density-dependent growth in Pseudacris crucifer. Final density refers to that of Pseudacris in each tank at metamorphosis. Predators refers to whether tanks contained predatory salamanders or not. A significant effect of predators means that the relation between growth and final density depends on the presence or absence of predators. See Fig. 3.
Type I Source of variation df ss F P
Dependent variable: Pseudacris growth rate
Model 3 105.40 92.20 0.0001 Error 20 7.62 Corrected total 23 113.02
Dependent variable: Pseudacris growth
Final density 1 19.47 51.11 0.0001 Predators 1 10.88 28.58 0.0001 Final density x predators 1 1.34 3.54 0.0747 TABLE 4. Effects of tadpole density, Notophthalmus (Newt), and Ambystoma opacum (Ambystoma) on the response variables measured for Hyla andersonii.
Source of variation df Type I ss F P
Density 1 0.0357 4.54 0.0490 Newt 1 0.2548 32.34 0.0001 Density x Newt 1 0.0357 4.54 0.0490 Ambystoma 1 0.2426 30.79 0.0001 Density x Ambystoma 1 0.0342 4.35 0.0535 Newt x Ambystoma 1 0.2426 30.79 0.0001 Density x Newt x Ambystoma 1 0.0342 4.35 0.0535 Error 16 0.1261
Density 1 22330.10 10.07 0.0193 Ambystoma 1 133 868.71 60.35 0.0002 Density x Ambystoma 1 5597.00 2.52 0.1633 Error 6 13 309.67
Density 1 705.89 33.12 0.0012 Ambystoma 1 231.28 10.85 0.0165 Density x Ambystoma 1 200.93 9.43 0.0219 Error 6 127.89
Density 1 31.09 27.44 0.0019 Ambystoma 1 70.30 62.05 0.0002 Density x Ambystoma 1 1.90 1.68 0.2421 Error 6 6.79
Anuran larval performance
Pseudacris crucifer survived best in ponds without predators, less well in ponds containing either Notophthalmus or Ambystoma, and worst in ponds containing both predator species (Table 2, [ILLUSTRATION FOR FIGURE 2 OMITTED]). Separate effects of Notophthalmus and Ambystoma on Pseudacris were nearly equivalent. Survival was unaffected by initial tadpole density under any set of predator conditions. Effects of the two predators on Pseudacris survival were non-additive, because more Pseudacris survived the joint impacts of the two predators than would have been predicted from the sum of their separately measured effects (Table 2). A non-additive impact of both predators on Pseudacris survival persists when these data are logarithmically transformed to test the multiplicative model of survival probabilities suggested by Wilbur and Fauth (1990). This result is somewhat more complex, since it involves a significant three-way interaction among the effects of tadpole density, Notophthalmus, and Ambystoma ([F.sub.1,23] = 5.52, P [less than] 0.0320).
Pseudacris growth and mass increased in ponds with predators, and both measures were inversely related to the number of Pseudacris surviving in each pond [ILLUSTRATION FOR FIGURES 2 AND 3 OMITTED]. Larval periods were slightly longer in ponds with predators, within each level of initial tadpole density (Table 2, [ILLUSTRATION FOR FIGURE 2 OMITTED]). Predators also generated differences in the relation between Pseudacris growth rates and the number of Pseudacris surviving in each pond ([ILLUSTRATION FOR FIGURE 3 OMITTED], Table 3). Similar slopes for the relations between Pseudacris growth rates and the number of Pseudacris surviving in each pond suggest similar effects of intraspecific density in ponds with or without pred-ators, while differences in the elevations of the two lines suggest a more intense competitive regime (lower growth at a given density of conspecifics) in ponds without predators where Bufo and Hyla also survived in numbers ([ILLUSTRATION FOR FIGURE 3 OMITTED], Table 3).
Hyla survived better in low density assemblages without predators than in similar communities stocked with high densities of tadpoles ([ILLUSTRATION FOR FIGURE 4 OMITTED], Table 4). Reduced survival in high density assemblages without predators was associated with reduced mass at metamorphosis, prolonged development, and reduced growth rates ([ILLUSTRATION FOR FIGURE 4 OMITTED], Table 4). The few Hyla that managed to survive in ponds containing Ambystoma were much larger than any of their counterparts metamor-phosing from ponds without predators [ILLUSTRATION FOR FIGURE 4 OMITTED].
Bufo survived only in ponds without salamanders, and survival in those ponds was independent of initial tadpole density ([ILLUSTRATION FOR FIGURE 5 OMITTED], Table 5). High initial density was associated with reduced mass at metamorphosis, prolonged development, and reduced growth rates in ponds without predators.
Periphyton and phytoplankton
None of the experimental manipulations affected the abundance of periphytic algae. In contrast, the standing crop of phytoplankton increased significantly in experimental ponds containing predatory salamanders ([ILLUSTRATION FOR FIGURE 6 OMITTED], Table 6). Cascading effects of top predators on the standing crop of primary producers were most pronounced in the first 2 wk of community development.
TABLE 5. Effects of tadpole density, Notophthalmus, and Ambystoma opacum on the response variables measured for Bufo woodhousii fowleri.
Source of variation df Type I ss F P
Density 1 0.0002 0.04 0.8358 Newt 1 0.8784 200.52 0.0001 Density x Newt 1 0.0002 0.04 0.8358 Ambystoma 1 0.8784 200.52 0.0001 Density x Ambystoma 1 0.0002 0.04 0.8358 Newt x Ambystoma 1 0.8784 200.52 0.0001 Density x Newt x Ambystoma 1 0.0002 0.04 0.8358 Error 16 0.0701
Density 1 20 021.92 92.76 0.0006 Error 4 863.39
Density 1 20.53 8.11 0.0465 Error 4 10.13
Density 1 15.74 152.87 0.0002 Error 4 0.41
Responses of top predators
Ambystoma survived equally well under all treatment combinations, but additions of Notophthalmus reduced their mass, snout-vent length, and head width at metamorphosis ([ILLUSTRATION FOR FIGURE 7 OMITTED], Table 7). A three-fold difference in the initial abundance of anuran prey did not significantly affect the growth and survival of Ambystoma larvae.
The abundance and size of larval Notophthalmus that metamorphosed from each pond was unrelated to manipulations of initial tadpole density or Ambystoma. The first larval newts appeared in the tanks at about the same time that Ambystoma metamorphosed, so there may have been little direct interaction between larval Ambystoma and larval newts. There was a significant negative correlation between mean mass at metamorphosis and the number of larval newts emerging from each pond [ILLUSTRATION FOR FIGURE 8 OMITTED].
The functional equivalence of predators
Lawton and Brown (1993) have suggested that communities contain functionally redundant species that potentially play very similar ecological roles. If communities contain functionally redundant species, key aspects of community and ecosystem processes may remain unchanged by changes in species composition, as long as each broad functional group retains at least one functionally competent species. One somewhat indirect approach to testing this idea involves comparing community-level processes in systems containing different numbers of species (Tilman and Downing 1994). Because Tilman and Downing (1994) found that communities differing in species richness also differed in their responses to disturbance, they questioned the validity of the redundant-species hypothesis. The more direct approach taken here involves comparing the functional equivalence of species that are expected, a priori, to function similarly. Several results point to the functional redundancy of Notophthalmus and Ambystoma in simple pond communities. Functional redundancy with respect to certain processes does not rule out other unique species-specific contributions to other unmeasured patterns. However, both predator species appeared to play virtually identical roles in the organization of these communities.
Adult Notophthalmus and larval Ambystoma had nearly identical effects on the composition of the anuran assemblage. Pseudacris, the only species that managed to persist in numbers with predators present, also responded similarly to the two predators. Effects of both predator species also cascaded similarly through the food web to increase phytoplankton abundance. Thus several community-level responses indicate similar direct or indirect impacts of Notophthalmus and Ambystoma on other subsets of the community. Paine's (1992) analysis of several herbivores in a rocky intertidal community pointed to the existence of two groups of functionally similar consumer species, characterized by strong or weak per capita effects on their prey. Analysis of a greater number of predator taxa in artificial pond experiments may eventually reveal a similar pattern, although both salamanders appeared to interact strongly with the anuran assemblage.
Ambystoma and Notophthalmus initially differed in body size, morphology, and life history stages, but they converged rapidly in body size and in community-level effects. Convergence in the effects of these two generalist predators could be explained by an approximate mass-equivalence of per capita effects, where species of similar body size exert similar effects. Comparison of a smaller subspecies of Notophthalmus and the much larger species Ambystoma tigrinum, which attains approximately four times the mass of N. v. dorsalis (see Morin 1983a, b), shows that A. tigrinum has larger per capita effects on tadpole assemblages that parallel its larger size. To determine whether mass is the primary factor accounting for differences in predator effects would require independent manipulations of the total biomass and number of individuals of different predators feeding on the same prey assemblage. An approximate mass-equivalence of generalized predators with similar metabolic rates may be a realistic expectation, since the energy demands and the net feeding rates required to meet those demands should be roughly proportional to the biomass of predators that the community supports (Peters 1983).
The functional equivalence of Notophthalmus viri-descens and Ambystoma opacum indicates that some trophic interactions might be depicted reasonably by lumping species into larger trophic categories. Although trophic lumping affects certain descriptive statistics used to characterize food web architecture (Sugihara et al. 1989, Martinez 1991, Polis 1991), it emphasizes the functional similarity of certain taxa. This study shows that little is gained by referring to the top predator as Notophthalmus or Ambystoma, instead of as a salamander of a certain size. If community patterns can simply be predicted from the biomass of generalist predators, some aspects of community ecology may be greatly simplified by ignoring the specific taxonomy of species that can be lumped into general functional categories (i.e., generalist top predators). It seems unlikely that such simplicity will arise in communities dominated by specialist predators, since their impacts on prey will reflect the idiosyncratic ways that their different feeding specializations affect different subsets of the prey assemblage.
The non-additivity of functionally equivalent predators
Reports of non-additive effects of predators on prey community structure are increasingly common (Van Buskirk 1988, Martin et al. 1989, Fauth 1990, Hurd and Eisenberg 1990, Wissinger and McGrady 1993). The effects of Notophthalmus and Ambystoma on Pseudacris survival were non-additive, in the sense that more Pseudacris survived in the presence of both predators than predicted by the sum of survival decrements in the presence of either predator without the other (see Table 8). This result is not surprising, since combined effects of the two species would have to reduce survival [less than]0, an impossibility, to be statistically additive. Even so, survival in the combined predator treatments was [greater than]0, indicating that this interaction was not simply a statistical artifact. Analysis of the additivity of log-transformed survival, equivalent to a test of a multiplicative model of survival probabilities, also shows a significant departure from additive logarithmic effects (see Table 8). This departure is small, but statistically significant, and shows that the multiplicative model underestimates the combined effects of both predators at the lower density of tadpoles. This departure means that the effects of each predator on the probability of prey survival were not independent. These analyses imply that both additive and multiplicative models depart significantly from observed patterns, but the departure is not very great in size and may be of minor ecological significance.
TABLE 6. Repeated-measures ANOVA for the effects of initial tadpole density, Notophthalmus, and Ambystoma opacum on the standing crops of algal periphyton and phytoplankton. The variable analyzed is the standing crop averaged over four successive sampling dates.
Source of variation df Type I ss F P
Density 1 0.3215 1.87 0.1915 Newt 1 0.0004 0.00 0.9609 Density x Newt 1 0.0010 0.01 0.9389 Ambystoma 1 0.0571 0.33 0.5727 Density x Ambystoma 1 0.0599 0.35 0.5637 Newt x Ambystoma 1 0.3966 2.31 0.1496 Density x Newt x Ambystoma 1 0.0362 0.21 0.6528 Error 15 2.5784
Density 1 0.0425 0.45 0.5137 Newt 1 0.6483 6.79 0.0191 Density x Newt 1 0.2865 3.00 0.1024 Ambystoma 1 1.0219 10.70 0.0048 Density x Ambystoma 1 0.0284 0.30 0.5925 Newt x Ambystoma 1 0.0272 0.29 0.6006 Density x Newt x Ambystoma 1 0.1514 1.59 0.2260 Error 16 1.5276
Several factors could account for less than additive effects of predators on Pseudacris. The first, which is unlikely, is that one predator species reduced the abundance of the other. Notophthalmus did not significantly reduce the survival of Ambystoma, and Ambystoma do not eat toxic adult Notophthalmus (P. J. Morin, personal observation). The second possibility, which cannot be addressed with the data in hand, is that Pseudacris survival was simply a nonlinear function of total pred-ator density, regardless of predator species composition [ILLUSTRATION FOR FIGURE 9 OMITTED]. If this was the case, the impact of either 12 Notophthalmus or 12 Ambystoma on prey survival would also be less than twice the observed effect of 6 of either predator species [ILLUSTRATION FOR FIGURE 9 OMITTED]. Similarly, the combined effects of 6 Notophthalmus and 6 Ambystoma, even if the species were equivalent predators, would also be [less than]2x the expected effect of 6 of either species. Such nonlinear effects of predator density can occur if behavioral interactions decrease the attack rates of predators at higher predator densities (e.g., Wissinger and McGrady 1993). Competition from one predator species may also change the per capita impact of the other predator species on prey survival, through its effects on predator body size. If predators compete, and if competition slows predator growth, smaller predators may have a reduced per capita effect on prey survival. This idea is consistent with observations of reduced Ambystoma size in ponds where Notophthal-mus also occurred. A plot of average Pseudacris survival against the best estimate of final predator biomass in each tank provides some support for this idea [ILLUSTRATION FOR FIGURE 9 OMITTED]. Links between reduced Ambystoma size and a proposed reduced per capita impact on prey survival remain speculative in this case, but have been demonstrated in other systems (Stenhouse et al. 1983). Results from other experiments show that larvae of a larger species, Ambystoma tigrinum, can slow the growth of adult Notophthalmus and reduce the abundance of larval Notophthalmus (Morin 1983a).
The non-additivity of interspecific interactions remains little-studied (Morin et al. 1988, Strauss 1991, Worthen and Moore 1991, Wootton 1993), despite the potential for such higher-order interactions to gravely complicate the prediction of complex community patterns from studies of pairwise interactions among species. Where such effects occur, they may be related to interactions among species that alter the size or behaviors of competitors and predators. While per capita effects may change with species composition, effects standardized by the biomass or behavior of interacting species might be more robust to changes in species composition. For instance, per capita competitive ability in plants is correlated with competitor biomass (Gaudet and Keddy 1988). This raises the possibility of eliminating non-additive per capita effects by changing the community calculus to focus on the population biomass of interacting species rather than their population sizes or by otherwise weighting per capita effects by body size or some behavioral equivalent.
TABLE 7. Effects of tadpole density and Notophthalmus on Ambystoma opacum.
Source of variation df Type I ss F P
Dependent variable: snout-vent length
Newt 1 37.55 30.60 0.0006 Tadpoles 1 6.82 5.56 0.0461 Newt x Tadpoles 1 3.81 3.11 0.1157 Error 8 9.81
Dependent variable: head width
Newt 1 0.71 28.25 0.0007 Tadpoles 1 0.08 3.18 0.1123 Newt x Tadpoles 1 0.12 4.77 0.0605 Error 8 0.20
Dependent variable: mean mass at metamorphosis
Newt 1 0.6979 21.45 0.0017 Tadpoles 1 0.1105 3.40 0.1025 Newt x Tadpoles 1 0.0830 2.55 0.1489 Error 8 0.2603
Wilks' Num. Den. Source of variation lambda F df df P
Newt 0.135004 12.814 3 6 0.0051 Tadpoles 0.481015 2.157 3 6 0.1942 Newt x Tadpoles 0.558454 1.581 3 6 0.2892
Supply-side consequences of variation in prey initial density
Supply-side approaches to community ecology predict that the potential for strong competition among species is greatest in situations with the greatest densities of recruits or settlers (Gaines and Roughgarden 1985, Underwood and Fairweather 1989). In turn, predators should have particularly striking impacts on species composition where they can modify intense competition, and potential competitive exclusion, among prey (Paine 1966, Lubchenco 1978, Morin 1983b). For keystone predators to enhance prey species richness, prey must first be sufficiently abundant for competitive exclusion to reduce species richness in the absence of predators, and then predators must selectively remove competitively dominant species to prevent competitive exclusion.
Negative effects of initial density on various indirect measures of competition (growth, body size, and for one species, survival) showed that elevated initial prey density increased the intensity of competition among prey in communities without predators. However, predators simply eliminated two anuran species, including one, Hyla, that suffered reduced survival from competition in communities without predators. Predation's net effect was to decrease, rather than increase, prey species richness. This result did not depend on initial differences in prey density, which clearly influenced the intensity of competition among prey in ponds without salamanders.
Previous studies of supply-side phenomena have concentrated on whether the impact of predators depends on the initial density of prey that colonize the community. The effects of initial prey density can also depend on predator abundance. Predators may override effects of initial prey density when they cannot move readily among habitat patches, limiting possibilities for predator emigration or immigration in response to prey density. This outcome is likely where variation in prey density occurs among discrete patchy habitats, like ponds, and where predators, like salamanders or fish, move little among habitat patches.
Alternatively, when predators can rapidly reduce prey abundances by means of an efficient functional response, differences in prey density may not persist long enough to have appreciable effects on subsequent interactions. The ability of Notophthalmus and Ambystoma to reduce prey survival to similar levels despite a threefold difference in initial prey densities suggests that the functional responses of both predators were not saturated by the experimental densities of prey.
TABLE 8. Summary of estimated treatment effects of Ambystoma and Notophthalmus (Newts) on the survival of Pseudacris crucifer at two tadpole densities. Treatment effects are shown as departures from the control means. Corresponding net values of survival are shown in parentheses below each treatment effect.(*)
C + A + N Anuran C A N density Control Ambystoma Newts Predicted Observed
Low 0.7150 -0.5234 -0.4367 -0.9601 -0.7006 (0.7150) (0.1916) (0.2783) (0) (0.0144) High 0.7566 -0.5366 -0.5283 -1.0649 -0.6816 (0.7566) (0.2200) (0.2283) (0) (0.0750)
Multiplicative model: log C + log A + log N Anuran log C log A log N density Control Ambystoma Newts Predicted Observed
Low -0.1564 -0.5632 -0.4867 -1.0499 -1.7316 (0.6976) (0.1907) (0.2275) (0.0891) (0.0129) High -0.1218 -0.5380 -0.5458 -1.0838 -1.0247 (0.7754) (0.2189) (0.2150) (0.0825) (0.0714)
* Predicted joint effects for both predators are determined by the sum of C + A + N. Predicted values of survival [less than]0 are truncated to 0. Minor differences between survival in the additive and multiplicative analyses arise from the back transformation of the geometric means obtained in the analysis of log transformed data.
Variable recruitment of larval amphibians in freshwater ponds (see Pechmann et al. 1991) is directly analogous to the variable recruitment of sessile marine organisms that originally inspired supply-side approaches to community dynamics. Such dynamics may apply to a great diversity of organisms having complex life cycles, where the initial abundance of a life history stage in one habitat often depends on input from another life history stage occupying another kind of habitat. The extent to which initial prey density strongly influences interactions with organisms on higher trophic levels in these systems remains controversial and requires much further study in a greater range of systems.
Effects of predators on prey species composition
Pseudacris appeared better able to coexist with predators than either Bufo or Hyla. Pseudacris is a relatively inactive slowly growing species that changes its microhabitat use and behavioral repertoire in the presence of predators (Morin 1986, Lawler 1989). Bufo and Hyla have more active tadpoles that run a greater risk of attracting the attention of visual predators like salamanders. For predators to enhance prey species richness, competitively inferior species excluded at high initial tadpole densities in the absence of predators must also be more resistant to predators than the competitive dominants. That was not the case in this system.
The observed predominance of Pseudacris in ponds containing salamanders resembles its performance in other tadpole assemblages studied in previous experiments (Morin 1983b, Fauth and Resetarits, 1991). The main difference is that predators significantly enhanced the survival of Pseudacris in the more species rich and competitively rigorous communities described in Morin (1983b), whereas in this study Pseudacris simply tolerated salamander predation better than either Bufo or Hyla. Studies of other systems where consumers enhance the species richness or diversity of the consumed (e.g., Paine 1966, Lubchenco 1978, Worthen 1989) underscore the apparent generality of the positive relations between prey competitive ability and susceptibility to predators that appear to drive patterns of predator-mediated competition.
Recent reviews indicate that aquatic communities provide the overwhelming majority of experimental examples of trophic cascades (Menge 1992, Power 1992, Strong 1992). Strong (1992) suggested that the relatively simple linear food chains in some aquatic communities may contribute to the ability to detect cascading effects, since such effects would dissipate if transmitted through many reticulate trophic pathways in more complex communities. Even in my simple reconstructed communities, trophic cascades selectively affected some producers and not others. These trophic cascades also were temporally complex and appeared most prominent early in community development. Even in relatively simple communities, evidence for trophic cascades apparently depends on both the specific taxa observed and the timing of observations.
One explanation for the selective cascade involving phytoplankton invokes differences in the foraging patterns of herbivores that persisted with or were eliminated by predators. Predators selectively removed one species that consumed phytoplankton, while another periphyton grazer managed to persist with or without predators. Predators selectively eliminated Hyla tadpoles, which feed primarily in midwater (see Lawler 1989) where phytoplankton is the primary food source. Hyla was the anuran species most likely to affect phytoplankton abundance, given its tendency to forage in the open water. In contrast, the anuran that persisted with or without predators, Pseudacris, avoids foraging in midwater and feeds primarily by scraping surfaces (Lawler 1989) and is an inefficient feeder on phytoplankton (Seale and Beckvar 1980). The net result of differentially heavy predation on phytoplankton consumers was that ponds with predators would still contain substantial numbers of periphyton grazers, and grazing pressure on periphyton would remain high, regardless of predator (salamander) abundance. It is also possible that predators selectively removed other herbivores, such as zooplankton, that would have preyed selectively on phytoplankton. Either scenario would explain a selective trophic cascade where predators enhance the abundance of phytoplankton without affecting periphyton. The lesson is that even in relatively simple communities, unexpectedly complex patterns can arise from relatively simple manipulations. Theory suggests that such complexity should be expected, even in relatively simple food webs (Abrams 1993).
This study was supported by the NSF under grant BSR 8704519. Marsha Morin and Sharon Lawler helped collect the frogs used to set up the experiment. Sharon Lawler, Lynn Kurzava, and Charles Bristow helped with various aspects of the setup and maintenance of the artificial ponds. Sharon Lawler and Ellen Pehek processed and analyzed phytoplankton and periphyton samples. Christina Kaunzinger, Lynn Kurzava, Mark Laska, Jill McGrady, Paul McMillan, Elizabeth Obee, Ellen Pehek, David Smith, and two anonymous reviewers made many helpful suggestions on previous drafts of the manuscript.
Abrams, P. A. 1993. Effect of increased productivity on the abundances of trophic levels. American Naturalist 141:351-371.
Addicott, J. F. 1974. Predation and prey community structure: an experimental study of the effect of mosquito larvae on the protozoan communities of pitcher plants. Ecology 55:475-492.
Brophy, T. E. 1980. Food habits of sympatric larval Ambystoma tigrinum and Notophthalmus viridescens. Journal of Herpetology 14:1-6.
Cohen, J. E. 1978. Food webs and niche space. Princeton University Press, Princeton, New Jersey, USA.
Cortwright, S. A., and C. E. Nelson. 1990. An examination of multiple factors affecting community structure in an aquatic amphibian community. Oecologia 83:123-131.
Fairweather, P. G. 1988. Consequences of supply-side ecology: manipulating the recruitment of intertidal barnacles affects the intensity of predation upon them. Biological Bulletin 175:349-354.
Fauth, J. E. 1990. Interactive effects of predators and early larval dynamics of the tree frog Hyla chrysoscelis. Ecology 71:1609-1616.
Fauth, J. E., and W. J. Resetarits, Jr. 1991. Interactions between the salamander Siren intermedia and the keystone predator Notophthalmus viridescens. Ecology 72:827-838.
Gaines, S., and J. Roughgarden. 1985. Larval settlement rate: a leading determinant of structure in an ecological community of the marine intertidal zone. Proceedings of the National Academy of Sciences, USA 82:3707-3711.
Gaudet, C. L., and P. A. Keddy. 1988. A comparative approach to predicting competitive ability from plant traits. Nature 334:242-243.
Gill, D. E. 1978. The metapopulation ecology of the red-spotted newt, Notophthalmus viridescens (Rafinesque). Ecological Monographs 48:145-166.
----- 1979. Density-dependence and homing behavior in the red-spotted newt, Notophthalmus viridescens (Rafinesque). Ecology 60:800-813.
Hairston, N. G., Sr. 1989. Ecological experiments. Cambridge University Press, New York, New York, USA.
Harris, R. N. 1987. An experimental study of population regulation in the salamander Notophthalmus viridescens dorsalis (Urodela: Salamandridae). Oecologia (Berlin) 71:280-285.
Hurd, L. E., and R. M. Eisenberg. 1990. Arthropod community responses to manipulation of a bitrophic predator guild. Ecology 71:2107-2114.
Lawler, S. P. 1989. Behavioural responses to predators and predation risk in four species of larval anurans. Animal Behaviour 38:1-9.
Lawton, J. H., and V. K. Brown. 1993. Redundancy in ecosystems. In E.-D. Schulze and H. A. Mooney, editors. Biodiversity and ecosystem function. Ecological Studies Analysis and Synthesis 99:255-270.
Lubchenco, J. 1978. Plant species diversity in a marine intertidal community: importance of herbivore food preference and algal competitive abilities. American Naturalist 112:23-39.
Martin, T. H., R. A. Wright, and L. B. Crowder. 1989. Non-additive impact of blue crabs and spot on their prey assemblages. Ecology 70:1935-1942.
Martinez, N. D. 1991. Artifacts or attributes? Effects of resolution on the Little Rock Lake food web. Ecological Monographs 61:367-392.
Menge, B. A. 1992. Community regulation: under what conditions are bottom-up factors important on rocky shores? Ecology 73:755-765.
Morin, P. J. 1981. Predatory salamanders reverse the outcome of competition among three species of anuran tadpoles. Science 212:1284-1286.
-----. 1983a. Competitive and predatory interactions in natural and experimental populations of Notophthalmus viridescens dorsalis and Ambystoma tigrinum. Copeia 1983:628-639.
-----. 1983b. Predation, competition, and the composition of larval anuran guilds. Ecological Monographs 53:119-138.
-----. 1986. Interactions between intraspecific competition and predation in an amphibian predator-prey system. Ecology 67:713-720.
-----. 1989. New directions in amphibian community ecology. Herpetologica 45:124-128.
Morin, P. J., and E. A. Johnson. 1988. Experimental studies of asymmetric competition among anurans. Oikos 53:398-407.
Morin, P. J., S. P. Lawler, and E. A. Johnson. 1988. Competition between aquatic insects and vertebrates: interaction strength and higher order interactions. Ecology 71:1590-1598.
Morin, P. J., S. P. Lawler, and E. A. Johnson. 1990. Breeding phenology and the larval ecology of Hyla andersonii: the disadvantages of breeding late. Ecology 71:1590-1598.
Morin, P. J., H. M. Wilbur, and R. N. Harris. 1983. Salamander predation and the structure of experimental communities: the responses of Notophthalmus and microcrustacea. Ecology 64:1430-1436.
Neill, W. E. 1974. The community matrix and the interdependence of the competition coefficients. American Naturalist 108:399-408.
Paine, R. T. 1966. Food web complexity and species diversity. American Naturalist 100:65-85.
-----. 1992. Food-web analysis through field measurement of per capita interaction strength. Nature 355:73-75.
Pechmann, J. H. K., D. E. Scott, R. D. Semlitsch, J. P. Caldwell, L. J. Vitt, and J. Whitfield Gibbons. 1991. Declining amphibian populations: the problem of separating human impacts from natural fluctuations. Science 253:892-895.
Peters, R. H. 1983. The ecological implications of body size. Cambridge University Press, Cambridge, United Kingdom.
Petranka, J. W. 1989. Density-dependent growth and survival of larval Ambystoma: evidence from whole pond manipulations. Ecology 70:1752-1769.
Polis, G. A. 1991. Complex trophic interactions in deserts: an empirical critique of food-web theory. American Naturalist 138:123-155.
Power, M. E. 1992. Top-down and bottom-up forces in food webs: do plants have primacy? Ecology 73:733-746.
Roughgarden, J. R., S. Gaines, and H. Possingham. 1988. Recruitment dynamics and complex life cycles. Science 241:1460-1466.
Scott, D. E. 1990. Effects of larval density in Ambystoma opacum: an experiment in large-scale field enclosures. Ecology 71:296-306.
Seale, D. B., and N. Beckvar. 1980. The comparative ability of anuran larvae (genera: Hyla, Bufo, and Rana) to ingest suspended blue-green algae. Copeia 1980:495-503.
Smith, C. K. 1990. Effects of variation in body size on intraspecific competition among larval salamanders. Ecology 71:1777-1788.
Stenhouse, S. L. 1985. Interdemic variation in predation on salamander larvae. Ecology 66:1706-1717.
Stenhouse, S. L., N. G. Hairston, and E. A. Cobey. 1983. Predation and competition in Ambystoma larvae: field and laboratory experiments. Journal of Herpetology 17:210-220.
Stewart, M. M. 1956. The separate effects of food and temperature differences on development of marbled salamander larvae. Journal of the Elisha Mitchell Scientific Society 72:47-56.
Strauss, S. Y. 1991. Indirect effects in community ecology: their definition, study, and importance. Trends in Ecology and Evolution 6:206-210.
Strickland, J. D. H., and T. R. Parsons. 1968. A practical handbook of seawater analysis. Bulletin of the Fisheries Research Board of Canada 167.
Strong, D. R. 1992. Are trophic cascades all wet? Differentiation and donor-control in speciose ecosystems. Ecology 73:747-754.
Sugihara, G., K. Schoenly, and A. Trombla. 1989. Scale invariance in food web properties. Science 245:48-52.
Tilman, D., and J. A. Downing. 1994. Biodiversity and stability in grasslands. Nature 367:363-365.
Underwood, A. J., and P. G. Fairweather. 1989. Supply-side ecology and benthic marine assemblages. Trends in Ecology and Evolution 4:16-20.
Van Buskirk, J. 1988. Interactive effects of dragonfly predation in experimental pond communities. Ecology 69:857-867.
Wilbur, H. M., and J. E. Fauth. 1990. Experimental aquatic food webs: interactions between two predators and two prey. American Naturalist 135:176-204.
Wissinger, S., and J. McGrady. 1993. Intraguild predation and interference competition among dragonfly larvae: non-additive effects on shared prey. Ecology 74:207-218.
Wood, J. T., and O. K. Goodwin. 1954. Observations on the abundance, food, and feeding behavior of the newt Notophthalmus viridescens viridescens (Rafinesque), in Virginia. Journal of the Elisha Mitchell Scientific Society 70:27-30.
Wootton, J. T. 1993. Indirect effects and habitat use in an intertidal community: interaction chains and interaction modifications. American Naturalist 141:71-89.
Worthen, W. B. 1989. Predator-mediated coexistence in laboratory communities of mycophagous Drosophila. Ecological Entomology 14:117-126.
Worthen, W. B., and J. L. Moore. 1991. Higher-order interactions and indirect effects: a resolution using laboratory Drosophila communities. American Naturalist 138:1092-1104.
Worthington, R. D. 1969. Additional observations on sympatric species of salamander larvae in a Maryland pond. Herpetologica 25:227-229.
|Printer friendly Cite/link Email Feedback|
|Author:||Morin, Peter J.|
|Date:||Jan 1, 1995|
|Previous Article:||Seasonal changes in thermoregulation by the frillneck lizard, Chlamydosaurus kingii, in tropical Australia.|
|Next Article:||A behavioral trade-off and its consequences for the distribution of Pseudacris treefrog larvae.|