Running the gauntlet of predators consumes critical time and energy resources, as all species are vulnerable to one or, typically, more predators at some life stage. Prey employ a vast array of mechanisms to avoid predation, and predators, likewise, come in a bewildering variety. Thus, defensive adaptations are rarely one size fits all. Considerable work has addressed multi-predator consumptive effects, but we now know that non-consumptive effects of predators can dramatically impact individuals, (meta)populations, and (meta)communities. However, little is known regarding the community-wide dynamics of non-consumptive effects generated by multiple predators. Predator avoidance by choosing a patch that is free of a particular predator or predators can be the most effective strategy if conditions at colonization are a reliable predictor of absence, which is often true for fish in freshwater systems. We experimentally manipulated composition of the predator assemblage in aquatic mesocosms in a substitutive design, with zero, one, two, or three caged predatory fish species (one benthic, one pelagic, and one surface fish) at constant density and biomass, and assayed responses of naturally colonizing aquatic insects. We addressed three related questions; first, how do members of a diverse assemblage of colonizing aquatic insects respond to this variation in species and species combinations, second, do individual species (and higher taxa), respond differently to single vs multiple predator species (species richness), and third how do any responses to fish species and species combinations, and effects on species richness, translate into community-wide changes in the composition of colonists. Prey had varied responses to specific predators or combinations of predators, resulting in distinct community composition across treatments and higher β-diversity with predators. Prey showed emergent multi-predator effects, where certain species only responded to predator species combinations, but not to any individual predator, and stronger effects of multiple predator vs single predator treatments, despite strong responses to individual predators in many taxa. Habitat selection effects can range from the individual to the metacommunity, and the dynamics of habitat selection in response to predators is a complex function of predator identity, density, richness, species composition, and patch spatial context.

Materials and Methods

Our experiment was conducted in a large oldfield at the University of Mississippi Field Station (UMFS), Lafayette County, MS. There is a diverse assemblage of aquatic insects at UMFS, including 132 species of aquatic beetles and over 40 species of aquatic hemipterans (Pintar and Resetarits 2020a, 2020b). We set up five arrays (blocks), each with nine, 1300 L (surface area = 2.54 m²) cylindrical mesocosms (N=45) laid out in isosceles trapezoids (Figs. 1a, Appendix 1: Fig. S1).  We crossed the presence/absence of three species of fish: golden topminnows (Fundulus chrysotus), golden shiners (Notemigonus crysoleucas), and brown madtoms (Noturus phaeus) (Fig. 1b), which are among the most commonly encountered species at UMFS.  Each species represents a different habitat/foraging strategy. Notemigonus crysoleucas is a small, pelagic, omnivorous-planktivorous, gape-limited fish that is widespread and abundant in both lentic and lotic habitats (Lee et al. 1980). Fundulus chrysotus is a small, gape-limited, surface-feeding topminnow and is also widespread and abundant (Lee et al. 1980). Noturus phaeus is a small, benthic foraging catfish typically found in lotic habitats (Lee et al. 1980, Chan and Parsons 2000), though is also common in overflow pools, backwaters, slow moving wetlands, and ponds that flood, including those at UMFS (pers. obs.); it is least gape-limited of the three.

We began filling mesocosms with well-water on 11 May 2017, one block at a time, completing filling on 13 May, at which point 1 kg aliquots of mixed, dried leaf litter (mixed hardwoods) were added to each mesocosm. Mesocosms were very quickly colonized by zooplankton and numerous small dipterans whose adults and/or eggs/larvae can pass through the screens (1.3 x 1.13 mm mesh), providing the fish a resource base. The high overall survival rate and positive growth of all three fish species supports the presence of an adequate food base. On 14 May each patch received 6 fish: 6/species in single predator treatments, 3/3 in two predator treatments, and 2/2/2 in three predator treatments, plus fishless controls. Density is on the lower end of biomass density in previous experiments and natural ponds, but above the threshold eliciting avoidance in many aquatic insects and treefrogs (Rieger et al. 2004; unpubl. data). To equalize biomass among fish treatments within blocks, we created complementary (1 “large,” 1 “small”) pairs within each of the three species for each block (by eye to minimize stress), and randomly assigned the appropriate number of pairs to each fish treatment patch within that block, maintaining the same density, approximate biomass, and size-structure across patches within blocks. On 15 May the experiment was begun by submerging screen lids to allow efficient collection of insects, and to separate fish from colonists, obviating consumptive effects (Appendix 1: Fig. S1). Dead fish (18/210) were replaced until 20 May, after which there was no observed mortality. Insects were collected weekly from 22 May until 20 November, with one final collection on 8 December, and preserved for later identification. Insect colonization is not responsive to either intra- or interspecific density, even when insects are not removed weekly (Pintar and Resetarits 2020c), and predation among adult insects is limited to predation by Notonecta irrorata on certain beetles, which always results in decapitation (Pintar and Resetarits 2021). There was no evidence of predation in this experiment, and N. irrorata occurred in only 77/1260 samples (97 individuals). We have not observed any terrestrial or avian predators at our cattle tanks across dozens of experiments.

Insect identification followed Pintar and Resetarits (and sources therein) (Pintar and Resetarits 2020b, 2020a). Only members of the genus Sigara (primarily S. pectenata at UMFS) and Paracymus (primarily P. subcupreus at UMFS) were not all identified to species. Fish survival from 20 May to 8 December was 91% across all three species: 89% for F. chrysotus, 94% for N. phaeus, and 91% for N. crysoleucas. Distribution of mortality was such that only two patches did not hold their full treatment until the end of the experiment; one F. chrysotus×N. phaeus (FC×NP) patch had no surviving F. chrysotus, and one three predator patch (X3) also had no surviving F. chrysotus. We could not track fish survival during the experiment, so we cannot determine when mortality occurred. Since neither tank was and outlier or otherwise out of the ordinary, our assumption was that these tanks held their respective treatments for much of the experiment. Given the overall low fish mortality, this is not unreasonable. Thus, we included both in the final analyses to maintain the balanced design required for independent estimates of multivariate community location and dispersion using PERMANOVA and PERMDISP (see below, Anderson and Walsh 2013).

See details in excel file.