Asymmetric density-dependent competition and predation between larval salamanders

1. Body size differences among consumers often lead to asymmetric interactions, with larger individuals typically being stronger competitors and/or predators on small individuals. These types of interaction are particularly exemplified in freshwater pond communities, where substantial size variation exists both within and among species of top consumers. We investigated whether density dependence can modify the outcome of size-structured interactions between larval stages of two pond-breeding salamanders, Ambystoma annulatum and Ambystoma opacum . Size structure exists in populations of these species due to variation in the timing of breeding, which we hypothesised would amplify predation rates and competitive asymmetries from the early-arriving species ( A . annulatum ) on the later-arriving species ( A . opacum ). 2. We manipulated the relative densities of both A. annulatum and A . opacum in outdoor mesocosms. We maintained the experiment through metamorphosis, and analysed size at metamorphosis, larval period length and survival of each species. 3. Ambystoma annulatum imparted a strong density-dependent effect on A . opacum through a combination of predation and competition. Survival of A . opacum was negatively related to the density of A . annulatum . For the A . opacum


| INTRODUC TI ON
Body size often plays a critical role in structuring populations and communities in freshwater ecosystems (Woodward & Warren, 2007), in part because it can influence species interaction strengths.For instance, large individuals are typically stronger competitors than smaller individuals because of their greater motility in obtaining resources and ability to generate stronger inference effects (Persson, 1985;Schoener, 1983), leading to asymmetric intra-and interspecific competition.Additionally, increasing size differences between predators and prey greatly impacts predation rates and survival of prey (Kalinkat et al., 2013).Body size variability can also lead to changes in food web stability (Emmerson & Raffaelli, 2004;Woodward et al., 2005), and trophic structure (e.g.trophic cascades; DeLong et al., 2015).
Two factors that can generate variability in body size among individuals in aquatic communities are differences in ontogeny and population density.Many aquatic organisms exhibit ontogenetic variability in their timing of entry into aquatic communities (Wilbur, 1980).For example, many species exhibit biphasic life cycles with terrestrial adults and aquatic larval stages, with the timing of adult breeding differing between either individuals of the same species (i.e. the degree of synchrony) or different species (e.g.priority effects).Larval stages of early-arriving individuals can then grow to attain larger sizes by the time later-arriving individuals enter a community, resulting in asymmetric interaction strengths, i.e. affecting the persistence or life history (e.g.growth rates) of the relatively smaller offspring of later-arriving individuals (Rasmussen et al., 2014).In addition, population density can affect body size through intra-or interspecific competition, slowing growth rates and increasing size variability.
This effect has been repeatedly demonstrated in numerous freshwater taxa (Wilbur, 1997;Wissinger, 1989).Population density is especially important when both predators and prey are simultaneously growing, as it can limit when species reach size refuges by limiting growth of either the predator or prey (Anders, 2001;Wilbur, 1988).
Ontogeny and population density can also affect species interactions via their joint impacts on body size.For instance, the size advantage gained by early arrivers is dependent on suitable growing conditions (Alford & Wilbur, 1985;Wilbur & Alford, 1985).If there is strong negative density dependence in early arrivers, individual growth opportunities may be limited, reducing size asymmetries.For predatory species, this could ultimately shift interactions from predation to competition when body size differences are minimised (Boone et al., 2002;Yang & Rudolf, 2010).Alternatively, if species breed in synchrony but exhibit different growth rates (Griffiths et al., 1994), then population densities may be more critical.Additionally, high densities of later-arriving species from synchronised breeding efforts may satiate early-arriving predators, increasing prey survival relative to when they are at lower densities (Ims, 1990).Higher individual growth rates for later-arriving species may also reverse predator and prey roles once they surpass size thresholds necessary for consuming the early-arriving species (Rasmussen et al., 2014).
Overall, there is a complex interplay between body size, population density, and ontogeny that can be difficult to discern in situ, requiring manipulation in experimental studies.
The importance of ontogeny and density in size-dependent interactions are particularly well exemplified in pond-breeding amphibians.Differences in arrival times to breeding sites, both within and among species, permits offspring of early-arriving adults to hatch before subsequent cohorts arrive, which allows early larvae to attain body sizes large enough to outcompete or prey upon laterarriving individuals (Alford, 1989;Alford & Wilbur, 1985;Anderson et al., 2017;Anderson & Semlitsch, 2014;Boone et al., 2002;Lawler & Morin, 1993;Segev & Blaustein, 2007;Wilbur & Alford, 1985).
However, because individual growth rates are often densitydependent (Anderson & Whiteman, 2015;Scott, 1990;Semlitsch & Caldwell, 1982) and species-specific (Griffiths et al., 1994), the size disparities between species could be altered depending on the relative densities and identities of each species.For carnivorous taxa (e.g.larval salamanders), such differences are key to determining predation rates (Anderson et al., 2016), and ultimately whether species predominately interact as competitors or predator/prey (Boone et al., 2002;Van Buskirk, 2007;Griffiths et al., 1994).
The goal of this study was to determine how density dependence influenced the outcome of size-structured species interactions.Specifically, we tested whether the effects of early-arriving ringed salamanders (Ambystoma annulatum) on demographic traits of individuals (e.g.body size) and survival of later-arriving marbled salamanders (Ambystoma opacum) were dependent on each species' density.Our focal taxa share many similar life history traits, including a complex life cycle where adults breed in the autumn, larvae overwinter in ponds prior to undergoing metamorphosis in late spring to early summer (Hassinger et al., 1970;Semlitsch et al., 2014), and are often found in sympatry in semi-permanent to permanent ponds in the Ozark and Ouachita Mountains (Anderson et al., 2021;Ousterhout et al., 2015;Shaffer, 2010).Larval densities are highly variable among ponds (Ousterhout et al., 2015), as are body size distributions (Anderson et al., 2016).They also generally have overlapping breeding phenologies, with both species having the potential to breed before the other, although their breeding phenology patterns have not been thoroughly explored when occurring in sympatry.Thus, this system is well-suited for understanding how ontogeny and density impact size structure and ultimately species interactions.
Because A. annulatum was added before A. opacum, we expected them to act as predator and prey, respectively.However, we expected that the density of each species may influence the interaction strength and type (i.e.predation vs. competition).As such, we predicted that when A. annulatum arrives early at low abundances, reduced intraspecific competition would lead to larger body sizes and thus higher predation rates on A. opacum.In contrast, when A. annulatum arrives early but at high abundances, intense intraspecific competition would lead to less disparities in size with A. opacum, shifting the interaction type towards competition, with reduced predatory impacts.This effect would be more pronounced with low densities of A. opacum, as they would have higher growth potential.
Additionally, when A. opacum occurred at high abundances, predator swamping effects may occur (Ims, 1990), especially when A. annulatum were at low abundances.

| Experimental design
We tested for density-dependent effects from larval A. annulatum on A. opacum using a response surface design in outdoor mesocosms (Inouye, 2001).We used 16 different combinations of larval abundances, some of which were replicated while others were not (Figure S1).Starting at an abundance of six larvae per tank for each species, we increased in increments of six larvae for each species so that all combinations of numbers between six and 24 were used.We had three replicates of the corners of the response surface (A. annulatum:A. opacum; 6:6, 6:24, 24:6, 24:24) whereas inner treatments (6:12, 6:18, 12:6, 12:12, 12:18, 12:24, 18:6, 18:12: 18:18, 18:24, 24:12, 24:18) were not replicated.The lack of replication for some treatments does not hinder inference because we relied on regression approaches (Cottingham et al., 2005;Inouye, 2001).We also had two treatments that only contained A. opacum to control for the minimum (n = 12) and maximum (n = 48) overall number of salamanders in each mesocosm.These densities fall within the observed densities in natural ponds for each species (Ousterhout et al., 2015).We randomly assigned treatments to mesocosms within the array.

| Experimental setup
The study was conducted in 28 outdoor mesocosms (1,000-L cattle watering tanks; hereafter, tanks) set up at the University of Missouri from September 2018 to June 2019.We filled tanks with water on 15 September 2018 and let tanks sit for 2 days to allow chlorine to evaporate.We then added 1.5 kg of dry leaf litter (primarily Quercus sp., Carya sp., and Platanoides occidentalis) on 17 September.We added c. 1 L of concentrated pond water to each tank to initiate plankton communities.We drilled drain holes in the top lip of each tank to limit water overflowing the top of the tanks during rain events.Tanks remained uncovered to permit natural colonisation by prey insects (e.g.Diptera).While this could have allowed for predatory invertebrates to also colonise the tanks, we did not observe any in the tanks, probably because the experiment took place primarily over the winter months.

We collected c. 30 partial clutches of A. annulatum on 20
September 2018 from a pond at Fort Leonard Wood, Missouri, U.S.A.We brought the eggs back to the lab and let them hatch in plastic containers, which was fully completed by 26 September.We haphazardly caught hatchlings and placed them in individual containers, and then added them to their assigned tanks on 27 September.
We collected seven nests of A. opacum from several ponds at Fort Leonard Wood on 18 October.We placed the nests on moist soil substrate inside plastic containers.We then flooded the containers with water on 21 October.After the eggs hatched, we again haphazardly caught individuals, added them to individual containers, and then added them to their assigned tanks on 26 October.Thus, there was a 29-day difference in addition dates between A. annulatum and A. opacum, matching the relative timing of breeding in the year we conducted this experiment.The natural phenology that year did not permit a test of adding A. opacum first, or both species at the same time to gauge priority effects.All hatchlings were visually similar in size, although we did not measure them.
We captured larvae from each tank in mid-March 2019 to assess the strength of density dependence midway through the experiment.We dorsally photographed individuals over a ruler to obtain total length (TL), snout-vent length (SVL), and head width measurements, using ImageJ (Rasband, 1997).All three measures were highly correlated (r > 0.95); thus, we used TL which had the fewest missing values.While our goal was to catch at least three individuals of each species from each tank to calculate an average size, we were sometimes unsuccessful due to individuals being inactive or not alive, especially A. opacum (i.e. A. opacum had already been consumed by A. annulatum).Thus, we only compared larval body size differences across density treatments for A. annulatum, as well as larval size across species.
We took zooplankton samples from each mesocosm on 3 November 2018 to determine if A. annulatum had reduced prey availability by the time A. opacum arrived into the community (i.e. the effect of exploitative competition).We took two integrated samples from the water column along the wall of each tank in a random cardinal direction, and filtered the samples through an 80μm net.We preserved zooplankton in ethanol, and then subsampled to count approximately 50% of the volume of each sample.We categorised and counted zooplankton as cladocerans, adult copepods, juvenile copepods, and rotifers under a dissecting scope.
We began checking each tank for metamorphosing animals in early May 2019, which corresponded to the timing of metamorphosis in previous experiments (Anderson et al., 2017(Anderson et al., , 2020;;Anderson & Semlitsch, 2014).We removed individuals that had reabsorbed their gills, and recorded their mass (g) and date of metamorphosis.
We again dorsally photographed individuals over a ruler to obtain SVL and TL measurements using ImageJ (Rasband, 1997).We approximated SVL by measuring to the distal junction of the tail and hind legs.We only report results for SVL as it was highly correlated with TL (r = 0.97).
Beginning the second week of June 2019, we drained all tanks and carefully sifted through the leaf litter to find any remaining individuals; only one larval A. annulatum was captured during the deconstruction process, meaning nearly all surviving individuals had metamorphosed.We determined survival of each species by counting the total number of individuals (both larvae and metamorphs) that had been recovered alive.During the larval sampling (March 2019), we discovered four tanks where larvae (n = 1 in three tanks, and n = 4 in one tank) had died due to stranding on the upper lip of the tank when the drain holes became clogged.Based on relative size, these individuals were likely to be A. opacum, so we subtracted these numbers from the initial number of A. opacum in those tanks when analysing survival.

| Analysis
Because our primary objective was to assess interspecific interactions, we first analysed the density combinations containing both species (i.e.excluding the tanks containing only A. opacum).We used multiple linear regression to analyse four response variables for each species: SVL, mass, larval period length, and survival.The two predictors in each model were the initial abundance of A. annulatum and A. opacum.We also tried including quadratic terms of each species' density to account for non-linear effects, but they were all non-significant so we do not report them here.We averaged SVL, mass, and larval period over all individuals that survived in each tank.In cases where only one individual survived (n = 6 tanks for A. opacum), we used that one data point.For survival, we used different model structures for each species.For A. opacum, we used a generalised linear mixed model with a binomial error distribution and an individual level random effect to correct for overdispersion.We compared this model to a generalised linear model with a quasibinomial errors, but they qualitatively had the same result, so we only report the mixed model.We used a generalised linear model with binomial errors for A. annulatum, as no evidence of overdispersion was present.We assessed significance by generating bootstrap confidence intervals (n = 1,000 samples).Because both predictors have the same units, the regression coefficients can be compared as approximate effect sizes on each response.
To tease apart overall density effects versus species identity effects on A. opacum, we conducted a separate set of ANOVAs comparing their responses in tanks with total densities of 12 ( 6A. annulatum: 6 A. opacum vs. 0:12) and 48 salamanders (24:24 vs. 0:48).We analysed the same response variables of SVL, mass, larval period length, and survival.We analysed differences in larval TL between species using ANOVA.We examined the effects of density dependence on TL in larval A. annulatum using linear regression.Because of the potential for low sample sizes influencing mean values for each tank, we reran the analysis on a subset of data where we caught at least three larval A. annulatum per tank, with the results being qualitatively similar; therefore, we report only the results using all the data.
We used linear regression to test for effects of initial A. annulatum abundance on prey resource (zooplankton) levels at the time A. opacum was added to tanks.We analysed groups separately (adult cladocerans, adult copepods or juvenile copepods) and total zooplankton abundance.We did not analyse rotifers as they occurred too infrequently.We log or log(× + 1) transformed each group to improve the normality of model residuals.All analyses were conducted in R (R Core Team, 2020).

| Larval size
On average, TL in mid-March of larval A. annulatum was 51.9 mm and 35.5 for A. opacum, with the difference being statistically significant (F 1,39 = 101.00,p < 0.001; adj.r 2 = 0.71).The range of sizes for each species were nearly completely non-overlapping, with only a few A. opacum being as large as A. annulatum, primarily from the 0:12 treatment (Figure 1a).Larval body size

| Ambystoma annulatum
Overall, A. annulatum was affected by intraspecific density, and unaffected by A. opacum density (Table 1).Average SVL at metamorphosis was negatively related to A. annulatum abundance (Table 1, adj.r 2 = 0.53) with SVL decreasing 0.23 mm for each increase in A. annulatum abundance (Figure 2a).Similarly, mass at metamorphosis for A. annulatum was negatively affected by conspecific abundance (Table 1; adj.r 2 = 0.43), with mass declining 0.036 g for each increase in A. annulatum abundance (Figure 2b).Larval period length was positively related to conspecific abundance (Table 1, adj.r 2 = 0.62; Figure 2c).Larval period was lengthened by 0.60 days for each increase in conspecific abundance.Survival of A. annulatum was unrelated to either conspecific or A. opacum abundance (Table 1; Figure 2d).

| Ambystoma opacum
Intra-and interspecific effects both contributed to explaining the responses of A. opacum (Table 2).The SVL at metamorphosis for A. opacum decreased with increasing abundances of A. annulatum, and was unrelated to conspecific abundance (Table 2; adj.r 2 = 0.23).
SVL decreased 0.17 mm for each increase in A. annulatum abundance (Figure 3a).Mass at metamorphosis for A. opacum was negatively affected by both conspecifics and A. annulatum abundances, with the effect of A. annulatum being slightly greater (Table 2; adj.r 2 = 0.36).Ambystoma opacum mass decreased approximately 0.026 and 0.019 g for each increase in abundance of A. annulatum and A. opacum, respectively (Figure 3b).Larval period length of A. opacum was positively affected by the abundance of A. annulatum (Table 2; adj.r 2 = 0.19).Larval periods increased by 0.28 days for each increase in A. annulatum abundance (Figure 3c).Across all densities, the larval period length of A. opacum was also on average 26 days faster than A. annulatum, showing that A. opacum overcame the 29-day difference in addition dates to metamorphose at roughly the same time as A. annulatum.Survival of A. opacum was only affected by the number of A. annulatum (Table 2; Figure 3d).
The odds of survival decreased by 55% with each increase in the number of A. annulatum (Table 2).
Comparisons of tanks that controlled for overall low and high abundances (comparisons of 6:6 -0:12 and 24:24-0:48, respectively) showed our results were driven by A. annulatum density (Figure S2).At low overall abundances, A. opacum survival was 100% in two of the three replicates of the 6:6 treatment, and 100% in each of the two replicates of the 0:12 treatment.Survival was 0% in the third replicate of 6:6 treatment, although the reason is unclear; the statistical results are similar whether we include or exclude this data point.At high overall abundances, survival of A. opacum was nearly zero with A. annulatum present, whereas survival was nearly 100% without them.Thus, our observed low survival of A. opacum with high abundances of A. annulatum is not simply due to numeric effects.Body size and larval period length in these same comparisons respond primarily to total abundance and not in response to whether A. annulatum was present (Figure S2).

| Prey resources
Total zooplankton density was negatively related to the abundance of A. annulatum (p = 0.06; adj.r 2 = 0.13; Figure S3), indicating some evidence of exploitative competition.Examining individual prey species revealed this relationship was primarily driven by the relationship between A. annulatum and cladocerans (p = 0.02; adj.r 2 = 0.17; Figure S3).Adult copepods and juvenile copepods were not related the abundance of A. annulatum (Figure S3).

TA B L E 1 Parameter estimates and bootstrapped confidence intervals (CIs) for regression models of Ambystoma annulatum responses
Larger body size often confers advantages in competitive and predatorprey interactions (Kalinkat et al., 2013;Persson, 1985;Schoener, 1983).
Developing a better understanding of the underlying mechanisms, such as variation in population density and ontogenetic timing, that produce size variation is therefore critical.We experimentally tested how population densities influence size advantages gained through ontogenetic differences by manipulating the relative abundances of two larval salamanders that were introduced a month apart into pond communities.As expected, we found strong density-dependent effects of the early arriver, A. annulatum, on the late-arriving species,

TA B L E 2 Parameter estimates and bootstrapped confidence intervals (CIs) for regression models of Ambystoma opacum responses
A. opacum, that were from a combination of competition and predation, depending on the demographic trait.Further, the competitive effects are at least partially attributable to density-dependent resource limitation imposed by A. annulatum.Finally, species interactions were highly asymmetric, where A. annulatum was unaffected by A. opacum and instead was only limited by intraspecific density.As this system operates through a combination of competition and predation, it could be classified as intraguild predation (Polis et al., 1989), although we discuss below when and how these interactions may shift.Overall, we show strong for support the hypothesis that the magnitude of sizedependent species interactions can be determined through asymmetric effects of population densities.
The primary factor driving the results of our study was the density of A. annulatum.This species arrived early into the experimental communities, which we hypothesised would allow it to act as a predator and superior competitor of A. opacum due to the size advantages gained from early arrival.Correspondingly, increasing densities of A. annulatum resulted in higher mortality of A. opacum.
Additionally, for individual A. opacum that did survive, we observed strong density-dependent responses for body size and development, indicative of interspecific competition from A. annulatum: metamorphs emerged smaller and took longer to develop at higher densities of A. annulatum compared to A. opacum.Because these life history traits are correlated with adult fitness metrics (Scott, 1994;Semlitsch et al., 1988), the effects of A. annulatum would be expected to translate into population-level effects on A. opacum.
The density-dependent interactions we observed were highly asymmetric, as prey density (A.opacum) largely did not affect many of the responses of either species.In particular, high prey density did not result in swamping effects on the predator that would have increased their survival (Ims, 1990).It is possible that our prey densities were not large enough to satiate the number of A. annulatum present in each tank, although they did match densities found in natural pond communities (Ousterhout et al., 2015).Additionally, the larger A. annulatum may have consumed enough A. opacum to eliminate any density-dependent effects on either themselves or A. opacum in the mesocosms.Evidence from natural ponds indicates that the density of A. opacum does matter for their interactions, as it was the best predictor of larval A. annulatum size (Anderson et al., 2021), although this information was collected in the absence of any information on breeding phenology patterns.As both species can breed earlier than the other, it could be that density dependence only matters for the earliest-arriving species, which would subsequently affect their individual growth rates and predation rates on later-arriving species.We were not able to manipulate additions given their similarity as predators on other amphibians (Anderson et al., 2016;Stemp, 2017).
We hypothesised that high densities of early-arriving species would limit individual growth rates, minimise their size advantage, and result in a switch in interaction type from predation to competition.While we found a strong density-dependent effect, this did not result in a change of interaction type.Our study used densities that fell within those observed in natural populations (Ousterhout et al., 2015), although it could be that the densities of A. annulatum we used were still not high enough to reduce their ability to prey on A. opacum.Thus, only under extreme density dependence may such a shift in interaction type occur.The duration of the phenological timing differences could also play a role.In years where species breed more synchronously, density dependence may be more likely to cause a switch in interaction type.It should be noted that any density-dependent effects we postulate to have occurred assume that the densities in our response surface remained constant throughout the experiment.This was not true for A. opacum, however, as their survival decreased with increasing A. annulatum densities.Ambystoma annulatum survival was very high on average so their final densities were effectively the same as the initial densities.Therefore, our regression-based approach may have limited our ability to detect any density-dependent effects from A. opacum.Using the number of survivors of each species as a predictor variable resulted in similar conclusions to using the initial densities (results not shown), so this potential limitation probably does not alter our overall conclusions.
In some situations, predators can reduce intraspecific competition between prey individuals, resulting in high individual growth (e.g.thinning effects) (Anderson & Semlitsch, 2014;Van Buskirk & Yurewicz, 1998;Davenport & Chalcraft, 2012).We did not observe that here, probably because of the long temporal duration over which the predator and prey overlapped after prey individuals entered a size refuge, suggesting a long subsequent period of competitive interactions.Larval salamanders are often considered to be gape-limited as predators (Smith & Petranka, 1987) and can grow fast enough in ponds to reach size refuges as prey (Urban, 2008).
Thus, the temporal overlap as a consequence of ontogeny may mitigate or intensify thinning effects of predators.
While predation is the likely mechanism by which A. annulatum limited A. opacum persistence in experimental communities, we did find some evidence for exploitative competition for shared resources.
Zooplankton densities at the time A. opacum entered the communities were negatively related to the abundance of A. annulatum, and are the primary food of early larval stages of salamanders (Taylor et al., 1988).Other studies have similarly found resource preemption by early-arriving species in pond communities (Hernandez & Chalcraft, 2012;Segev & Blaustein, 2007), which can sometimes persist well beyond when the predator that initiated the prey reduction had left the community (i.e.legacy effect through metamorphosis) (Rudolf & Van Allen, 2017).We do not expect that A. opacum starved to death, and instead suspect such resource limitation slowed their growth or development rates, resulting in the lower average larval sizes and slower growth rates we observed midway through the experiment.This probably matches what would happen in natural communities, where limited food resources are unlikely to lead to starvation.Further support for this observation comes in our previous work on A. annulatum (Anderson et al., 2017), where resource depletion by A. annulatum did not result in mortality of their prey (A.maculatum), which was instead linked to predation.Because we did not perform behavioral observations, we cannot determine the role of interference in limiting access to food resources, although other studies have suggested only weak interference between our focal species (Stemp, 2017).Future research can combine behavioural observations of taxa with phenology manipulations to uncover the exact mechanisms by which species may exclude each other.
Understanding the mechanistic basis of ontogeny and variability in body size among species is imperative to determine how such factors contribute to structuring both populations and communities (Woodward et al., 2005;Woodward & Warren, 2007;Yang & Rudolf, 2010).Here we show that species densities can play a mediating role in the strength of interactions between species that vary in body size.At the highest densities, this resulted in a combination of predatory and competitive interactions that negatively impacted the smaller species, which in our case was due to phenological differences.Further studies that tease apart how other factors, such as temperature or food web structure, might also modulate the mechanisms of priority effects (e.g.habitat modification or resource preemption) are warranted.Studies of this sort are especially needed because many species' phenologies are shifting (Cohen et al., 2018;Parmesan, 2007) and phenological shifts are likely to impact how and which size classes interact (Yang & Rudolf, 2010), as well as species coexistence (Rudolf, 2019).Furthermore, not all species' phenol- of A. annulatum was negatively related to intraspecific density (p < 0.001; adj.r 2 = 0.61), with A. annulatum being about 22% (12 mm) longer at the lowest densities (n = 6) compared the highest densities (n = 24; Figure1b).

F
Larval size (total length, in mm) of Ambystoma annulatum and Ambystoma opacum caught in mid-March 2019 across all density combinations (a), and the average total length of larval A. annulatum in relation to the initial number of A. annulatum (b).Samples sizes for larval measurements in (a) were n = 94 for A. annulatum and n = 68 for A. opacum.Based on the average total length for each species (A.annulatum = 51.9; A. opacum = 35.5), A. annulatum had higher average growth rates, 0.31 mm/day, than A. opacum, 0.26 mm/day.Colour scale in (b) represents the sample size per tank over which the average total length was taken

F
Predicted relationships for (a) snout-vent length, (b) mass, (c) larval period length, and (d) percent survival of Ambystoma annulatum.The x-axis shows the effect of A. annulatum (intraspecific effects), and the y-axis shows the effect of Ambystoma opacum (interspecific effects of A. opacum before A. annulatum or concurrent arrival due to the natural phenology in the year of the study.However, we suspect that if A. opacum were to enter ephemeral ponds first, they would impart equivalent density-dependent mortality upon A. annulatum, F I G U R E 3 Predicted relationships for (a) snout-vent length (SVL), (b) mass, (c) larval period length, and (d) percent survival of Ambystoma opacum.Only Ambystoma annulatum abundance was a significant predictor for SVL, larval period length, and survival.Numbers, colours, and isoclines are as in Figure2.Only A. annulatum abundance was a significant predictor for SVL, mass and larval period; survival was unaffected by either species.Mass was affected relatively equally by both A. annulatum and A. opacum ogies are changing in the same way (e.g.different shifts in the shapes of phenological distributions; Carter et al., 2018), which could also impact density dependence and/or body size distributions and subsequent species interactions.Without baseline information on the mediating factors of body size variability and priority effects, it will be difficult to predict how historical contingency may change with phenological shifts or climate change more generally.with Missouri Department of Conservation approval (#17673) and experiments conducted with IACUC approval (Missouri: #9297; Appalachian State: #18-19).
). Numbers indicate predicted values for a given isocline.Vertical isoclines indicate only intraspecific effects, horizontal isoclines only interspecific effects and diagonal (45 degree) lines would indicate intra-and interspecific effects are equal.Darker colour shading indicates smaller body lengths and masses, reduced days to metamorphosis, and lower survival.Legends above each panel show the range of values for each response.Only A. annulatum abundance was a significant predictor for snout-vent length, mass and larval period; survival was unaffected by either species All models are general linear models, except survival which is a generalised linear mixed model with a binomial error distribution.The coefficients for survival are log-odds ratio estimates.Bolded values indicate where the CIs do not overlap zero.