Journal of Animal Ecology 2013, 82, 598–607

doi: 10.1111/1365-2656.12042

Top-down control of prey increases with drying disturbance in ponds: a consequence of non-consumptive interactions? Hamish S. Greig1*, Scott A. Wissinger1,2 and Angus R. McIntosh1 1

School of Biological Sciences, University of Canterbury, Private Bag 4800, Christchurch, New Zealand; and 2Biology Department, Allegheny College, Meadville, PA 16335, USA

Summary 1. Biotic interactions are often expected to decrease in intensity as abiotic conditions become more stressful to organisms. However, in many cases, food-web and habitat complexity also change with abiotic stress or disturbance, potentially altering patterns of species interactions across environmental gradients. 2. We used a combination of field assays and mesocosm experiments to investigate how disturbance from desiccation moderates top-down control of prey by predators across a gradient of pond duration in New Zealand. 3. Field manipulations of predator abundance in ponds led to an unexpected decrease in the top-down control of prey biomass by predatory invertebrates as pond duration increased (decreasing abiotic stress). Predatory fish, which are restricted to permanent ponds, had negligible effects on prey biomass. Mesocosm experiments further indicated the consumptive effects of fish are weak; a result that cannot be explained by an increase in physical habitat refugia in relatively more permanent ponds. 4. Manipulations of invertebrate predator diversity in mesocosms (both substitutive and additive treatments), and the addition of olfactory fish cues, revealed that strong non-consumptive effects of fish reduced predation by predatory invertebrates, and these effects overwhelmed the positive influence of invertebrate predator diversity on prey consumption. 5. These results suggest that decreases in top-down control with increasing pond permanence are likely a result of non-consumptive effects of fish weakening predation by invertebrate predators in the more complex food webs of permanent ponds. Therefore, considering nonconsumptive effects of predators in complex food webs will likely improve the understanding of biotic interactions across environmental gradients. Key-words: Food-web complexity, interaction modification, intraspecific competition, multiple predator effects, pond permanence, risk reduction, trait-mediated interactions

Introduction Understanding how the strength of species interactions changes with abiotic context is essential for predicting the dynamics of ecological communities (Agrawal et al. 2007), especially in response to changes along environmental gradients. Several traditional models of community organization predict decreases in the importance and intensity of biotic interactions as abiotic conditions become more

*Correspondence author. E-mail: [email protected]

stressful to organisms (Connell 1975; Grime 1977; Peckarsky 1983; Menge & Sutherland 1987). For example, abiotic stress can mediate competitive interactions by reducing densities of interacting species below resource carrying capacities (Lubchenco 1980; Crain et al. 2004; Gerhardt & Collinge 2007), and can weaken predation by disproportionately influencing the abundance or effectiveness of predators (Menge & Farrell 1989; Wellborn, Skelly & Werner 1996). Several authors have argued, however, that the effects of abiotic stress on species interactions should vary depending on environmental and community context

© 2013 The Authors. Journal of Animal Ecology © 2013 British Ecological Society

Disturbance increases top-down control (Chesson & Huntly 1997; Crain 2008). For example, physically benign habitats often have higher taxa diversity within and across trophic levels and wider body size variation than disturbed habitats, and hence food webs with more complex connections (Winemiller 1990; Jenkins, Kitching & Pimm 1992; Walters & Post 2008). Omnivory, intraguild predation (IGP) and non-consumptive effects characteristic of these complex food webs weaken pairwise species interactions (Emmerson & Yearsley 2004; Finke & Denno 2004); whereas the low diversity and chain-like resource pathways of simple food webs characteristic of disturbed or stressful habitats can strengthen interactions (Power, Parker & Wootton 1996; Thompson & Townsend 1999). Furthermore, organisms inhabiting physically stressful environments often face life-history constraints that force individuals to trade-off resistance to biotic interactions with rapid development or efficient resource use (Power, Parker & Wootton 1996; Wissinger et al. 2006; Edwards & Stachowicz 2010). These trade-offs may strengthen top-down control in disturbed habitats. Finally, changes in habitat morphology along environmental gradients such as habitat size, complexity and refugia further complicate the community-wide outcome of biotic interactions across environmental gradients. Studies that incorporate these parallel and potentially confounding community and habitat gradients are likely to reveal more realistic effects of abiotic stress on species interactions. Here, we use a combination of field and mesocosm experiments to investigate how shifts in food-web complexity and habitat structure affect aquatic predator–prey interactions across a gradient of pond drying disturbance. The duration and frequency of drying and refilling vary among individual ponds in a landscape, resulting in a well-known gradient of abiotic stress across habitat patches (Wellborn, Skelly & Werner 1996; Williams 1996). Predator body size and diversity increases with pond duration and fish are generally restricted to perennial ponds (Schneider & Frost 1996; Bilton, Foggo & Rundle 2001). Subsequent increases in predation risk with pond duration are often implicated in species turnover across the pond permanence gradient (reviewed in Wellborn, Skelly & Werner 1996; Wissinger 1999). However, pond area, the abundance of macrophyte refugia (Urban 2004; McAbendroth et al. 2005) and food-web complexity (Wissinger et al. 1999) all increase with pond duration, which may instead weaken community-wide predator– prey interactions in more permanent ponds. Therefore, if some predators are able to exploit temporary ponds through rapid colonization or senescence, patterns of topdown control of prey biomass across the habitat permanence gradient may be reversed to be most intense in short duration ponds (Brendonck et al. 2002). We experimentally manipulated predator biomass in natural ponds spanning a gradient of permanence to investigate the influence of drying stress on the strength of top-down control of prey biomass. We then conducted

599

two mesocosm experiments to distinguish the mechanisms driving observed patterns of top-down control across the natural pond permanence gradient. We first manipulated the presence of macrophytes and predatory fish in mesocosms to examine the potential role of refugia in mediating predation on benthic invertebrates in permanent ponds. A second experiment then used additive and substitutive manipulations of predator diversity (within and between trophic levels) to determine the influence of increasing food-web complexity on the strength of top-down control.

Materials and methods study sites and natural history Lentic habitats in the Waimakariri and Rakaia River catchments, South Island, New Zealand range in size from 2-m diameter ephemeral pools to large lakes (up to 179 ha) that lie within fluvio-glacial depressions (Wissinger, Greig & McIntosh 2009). Pond hydroperiod (duration of inundation) strongly influences species richness and community composition in the ponds, and predator species richness, size and biomass increase with pond permanence (Wissinger, Greig & McIntosh 2009). Most permanent ponds contain small (< 200 mm) predatory benthic fish (Galaxias brevipinnis G€ unther, Gobiomorphus breviceps Stokell), and the larger lakes support low-density populations of longfin eels (Anguilla dieffenbachia Grey), introduced brown (Salmo trutta L.) and rainbow trout (Onchorynchus mykiss Walbaum) (Jeppesen et al. 1997; Wissinger, McIntosh & Greig 2006). The biomass of predatory invertebrate guilds is dominated by odonates in permanent ponds, and beetles and small Anisops spp. backswimmers in temporary habitats (Wissinger, Greig & McIntosh 2009).

assays of predator impact in natural ponds Twelve ponds were selected for predator manipulations based on their likelihood of retaining water throughout the duration of experiments, and the presence of submerged vegetation in shallow water (<60 cm deep). Five of those ponds contained predatory fish (Table S1, Supporting Information). Visual monitoring of pond hydrology began March 2005 when temporary ponds were dry and from 12 October 2005 water depth was recorded hourly in each pond with stage height data loggers (HT-100; TruTrack Ltd., Christchurch, New Zealand) placed in the deepest possible point of the pond basin. Pond permanence was quantified with a multivariate index derived from principal components analysis of three aspects of pond hydrology: duration (days) of inundation following winter refilling, maximum proportion of total depth lost over the observation period, and number of days inundated per annum (to account for multiple drying and refilling events; Table S1, Supporting Information). In November 2005, we manipulated predator biomass in cages (1
5 mm mesh on a steel wire frame, 0
25 m2 surface area, 50 cm height) whose bottom edges were pushed into the sediment within macrophyte beds and sealed with clay and fine gravel. Three cages were placed 1 m apart in shallow water (mean depth  SE: 26  3 cm) in each pond, with each cage representing one replicate of three different treatments: a ‘all predator reduction’ where all fish and all predatory invertebrates captured in five sweeps

© 2013 The Authors. Journal of Animal Ecology © 2013 British Ecological Society, Journal of Animal Ecology, 82, 598–607

600 H. S Greig, S. A. Wissinger & A. R. McIntosh were removed from the cages, a ‘fish removal’, and an open ‘control’ cage (steel frame only) that allowed access to both predatory fish and invertebrates. Predators were removed with five successive sweeps of a 1-mm mesh D-net through the water column and benthic substrates. The contents of each sweep were transferred to sorting trays, predatory taxa were removed by hand, and the remaining invertebrates and detritus were returned to the cage. This procedure left submergent vegetation intact and removed all fish biomass and 56  10% (mean  SE) of predatory invertebrate biomass when compared with cages without predator removal. The efficacy of predator removal was not influenced by pond permanence (linear regression on % predator biomass reduction, F1,10 = 1
74, P = 0
24). Predatory invertebrates were classified as species that obtain the majority of their energy through the consumption of other macroscopic animals. Smallbodied predators that could move through the cage mesh (Liodessus spp. beetles, mites and cyclopoid copepods) were not removed in the manipulation. To control for prey mortality during the predator manipulations, open-sided cages which represented ambient levels of predation were covered in a temporary sleeve of mesh and disturbed using the same procedures as above but without the removal of predators. After 2 weeks, small-bodied invertebrates (crustaceans, chironomids, worms and molluscs) were subsampled in each cage with five sweeps of a 500-lm mesh net through a PVC pipe (0
020 m2) pushed into the substrate. Those subsamples were preserved in 70% ETOH and sorted in the laboratory under 109 magnification. The remaining taxa in the cage were sampled with five sweeps of a 1-mm mesh D-net, sorted on-site and preserved in 70% ETOH. Invertebrates were identified to tribe for Chironomidae, family or genus for Crustacea, and to the lowest possible taxonomic level (usually species) for remaining taxa. Biomass for each taxon was calculated using length-dry weight regressions (Benke et al. 1999; Stoffels, Karbe & Paterson 2003; H. S. Greig unpublished data) on the body length measurements (ocular micrometre at 10–209 magnification) of a random subset of 10 individuals from each sample. Dry weights were converted to ash-free dry mass (AFDM) using taxon-specific estimates of % ash (Benke et al. 1999, H. S. Greig unpublished data). Predator impact (PI) on prey biomass was calculated using log ratio of effect size (Berlow et al. 2004) calculated from the ratio of prey biomass in unmanipulated predator treatments to prey biomass in predator reduction treatments (Table 1). Fish PI was determined by comparing invertebrate biomass in the open control cages with invertebrate biomass in the fish removal cages. Predatory invertebrate PI was determined by comparing nonpredatory invertebrate biomass in fish removal cages with that from the all predator reduction cages. Finally, comparisons of non-predatory biomass between the open control cage and the all predator reduction cages produced total PI (Table 1).

The effect of fish on prey biomass in permanent habitats was analysed by comparing deviations of predator impact from zero using a one-sample t-test for each of the three prey categories: all invertebrates, all non-predatory invertebrates and ‘unprotected’ non-predators that did not have cases or shells. The effect of pond permanence on the strength of predatory invertebrate impact on all primary consumers and unprotected primary consumers was analysed with linear regression. Finally, we used linear regression on the biomass of predatory invertebrates in unmanipulated control cages to assess whether ambient predator biomass was influenced by pond permanence.

mesocosm experiment 1 : manipulation of predatory fish and submergent vegetation Whether habitat refugia could explain the apparent weak consumptive effects of fish in permanent ponds was examined by manipulating the presence and/or absence of complex vegetation structure and predatory fish in mesocosms (0
8 m2 oval tanks) in a 2 9 2 factorial design with four replicates of each treatment combination. The outdoor mesocosms, which were housed at University of Canterbury’s Cass Field Station, were filled with groundwater regulated to 25 cm deep throughout the experiment. Clumps of Myriophyllum and Carex (surface area: 0
1 m2) were added to the mud and fine gravel substrate of all tanks, and were cut to soil level for the no-vegetation treatment. Koaro (Galaxias brevipinnis) were chosen as predators as they are an abundant native predatory fish in ponds and lake littoral zones in the study area and they feed on a large range of macroinvertebrates (Rowe, Konui & Christie 2002). The prey community in each mesocosm consisted of 11 macroinvertebrate species that are common in nearby permanent and temporary ponds (Fig. S1, Supporting Information). The total biomass of each prey taxa was approximately equal, and combined biomass of all prey fell within the range observed in natural ponds (H. S. Greig unpublished data). Prey were added on the 28th February 2007 10 h prior to the addition of fish (one koaro per tank, mean fork length: 120  5
4 mm). After 2 weeks, fish were removed from tanks and one pipe sample (0
020 m2) was taken within the largest clump of Myriophyllum to subsample small-bodied prey (chironomids and crustaceans) in the same fashion as the predator assay. Following this procedure, all tank contents were tipped into a 1-mm mesh net whose contents were then transferred to a tray for sorting. Soil clumps were also searched for macroinvertebrates. Invertebrates were preserved and processed in the same manner as the predator assay. The effects of fish and vegetation structure on total prey biomass were evaluated using 2 9 2 factorial ANOVA with fish and vegetation as fixed effects and loge-transformed total invertebrate

Table 1. Derivation of predatory impact indices for the field manipulation of predator abundance. Predator impact (PI) was calculated as the log-ratio effect size PI = ln(Ba/Br) where Ba is the prey biomass in ambient cages and Br is the biomass in predator removal or reduction cages. Contrasts for PI Predator impact type

Prey type

Ba

Br

Fish PI Predatory invertebrate PI Total PI

All invertebrates Non-predatory inverts Non-predatory inverts

Open cages Fish removal Open cages

Fish exclusion All predator reduction All predator reduction

© 2013 The Authors. Journal of Animal Ecology © 2013 British Ecological Society, Journal of Animal Ecology, 82, 598–607

Disturbance increases top-down control

601

of fish chemical cues was also tested. Each of the 11 treatment combinations (Fig. 1b), including predator-free controls with and without fish cues, was replicated four times. Manipulations were conducted in 0
35 m2 white plastic tubs housed outdoors at the Cass Field Station. Mesocosms were filled to a depth of 18 cm with groundwater, and three 15-cm high clumps of plastic aquarium plants, three large cobbles and a layer of fine sand were added as substrate (Fig. S2, Supporting Information). Predator density varied between 4 and 12 individuals per mesocosm, reflecting the range of predator biomass observed in natural ponds. The combined cues of two fish species were added to fish treatments by dripping water (6
2  1 L h 1) from a single tank housing one koaro (145-mm fork length [FL]) and one rainbow trout (140 mm FL). Fish were fed ad libitum on all three species of invertebrates used in the experiment to ensure the complete range of cues from fish feeding was present. A control drip from an identical tank without fish or invertebrates was distributed at the same rate into the remaining tanks. Prey were late instar Chironomus zelandicus Hudson larvae (100 per mesocosm), collected from a nearby pond. Chironomus are found in almost every lentic habitat in the landscape and are an important component of predator diets. Prey were added at 1600 h on the 31st October 2007 and allowed to acclimate for 24 h before predators were introduced. Mesocosms were covered with 1 mm mesh throughout the experiment to contain prey emergence. After 5 days, adult chironomids that had emerged were handcollected, cobbles and artificial plants were washed and removed, and invertebrates were collected by repeated elutriation of the tank contents through a 250-lm mesh net followed by an inspection of the remaining sand. Invertebrates were preserved in 70% ETOH, enumerated in the laboratory and AFDM

AFDM the response variable. Treatment effects on prey community composition were assessed with PERMANOVA (PERMANOVA 6; Anderson 2001) on a Bray-Curtis dissimilarity matrix created from untransformed AFDM of each of the 11 macroinvertebrate taxa. Significance (P < 0
05) was tested with 999 permutations.

mesocosm experiment 2 : multiple predators The second mesocosm experiment manipulated predator diversity (within and between trophic levels) to determine if increasing food-web complexity dampened top-down control in permanent ponds. Predator diversity was examined in mesocosms using a hybrid design that included both additive and substitutive manipulations of predator richness (Byrnes & Stachowicz 2009). This design enabled us to investigate the changes in intraspecific and interspecific interactions associated with increases in the density and complexity of the predator guild with pond permanence (Fig. 1a). We manipulated three predatory invertebrate taxa including a common beetle, damselfly and dragonfly larva and the non-consumptive effects of a native and an introduced predatory fish species was also included in one comparison (Fig. 1b). Treatments were limited to the subsets of possible combinations that were relevant to natural predator combinations across the permanence gradient. The design enabled three models to be tested (indicted by small cased letters beneath treatments in Fig. 1b): (1) density manipulations (d) of a single widespread predator species, (2) additive manipulations (a) of multiple predator species in combination with all possible single species at low density treatments to enable the calculation of expected predation rates and (3) a substitutive model (s) at high predator densities in which the effects

Food web complexity

(a)

P R

R

P

R

X

X

R

X

(b)

R R

Control Control

R

X

P

d,a

a

a

R X

X d

R

a

d,s

s

R

R

X

X

P

P

a,s

a,s

Predatory invertebrate density

Pond permanence

Fig. 1. (a) Change in the complexity of predator guilds across the gradient of water permanence from temporary to permanent lentic habitats in Canterbury, New Zealand. Letters denote representative beetle (R, Rhantus suturalis Macleay), damselfly (X, Xanthocnemis zelandica Mclauchlan) and dragonfly taxa [P, Procordulia grayi (Selys)]. Solid and dashed arrows represent consumptive and non-consumptive effects, respectively. (b) Hybrid experimental design used to investigate multiple predator effects along the permanence gradient. In (b) each box represents one treatment and box size is proportional to predatory invertebrate density. Capital letters correspond to species in (a) and shading indicates the presence of chemical cues from fish (Oncorhynchus mykiss and Galaxias brevipinnis). Lower case letters denote treatment combinations used in the single-species density manipulation (d), and the additive (a) and substitutive (s) diversity manipulations. Two predator-free controls (with and without fish cues) were also included. Prey density was constant across all treatments. © 2013 The Authors. Journal of Animal Ecology © 2013 British Ecological Society, Journal of Animal Ecology, 82, 598–607

602 H. S Greig, S. A. Wissinger & A. R. McIntosh determined in the manner described earlier. Treatment combinations were grouped according to types of predator manipulation: (1) Rhantus beetle density treatment, (2) additive diversity manipulations and (3) substitutive diversity manipulations (Fig. 1b). Significance of each predator manipulation was tested using one-way ANOVA with tanks as replicates, followed by Tukey’s post-hoc comparisons of treatment means. Multiple predator effects were tested in the additive model by calculating expected predation rates for the two- and three-species treatments from the low-density single-species treatments using an additive probability model assuming finite prey resources (Soluk & Collins 1988). The model was modified for a three species guild (following Miller 2006:62), whereby expected prey consumption = 100 9 (Pr + Px + Pp PrPx PxPp + PrPxPp), where Pr, Px and Pp are the proportions of initial prey abundance eaten in the Rhantus, Xanthocnemis and Procordulia single-predator treatments respectively. The effect of density on intraspecific interactions in Rhantus single-species treatments was also tested using these models. In this case, expected values were based on Rhantus low density treatments. Significant differences between observed predation rates and expected values were assessed with one-sample t-tests. All analyses for the field and mesocosm experiments (except PERMANOVA) were conducted in Statistica 8 (StatSoft 2008).

Results assays of predator impact in natural ponds No consistent effects of fish on the biomass of prey were observed in the in situ assays of predator impact in natural ponds (Fig. S3, Supporting Information). The biomass of all invertebrates and primary consumers in cages that allowed fish access did not differ significantly from cages that prevented fish access (PI not significantly different from zero; one-sample t-tests: t4 < 1
14, P > 0
31). The same pattern was evident when only unprotected prey biomass (species without cases or shells) was considered (one-sample t-tests: t4 = 1
21, P = 0
29; Fig. S3, Supporting Information). The impact of predatory invertebrates on prey biomass decreased with increasing pond permanence (Fig. 2a). This pattern strengthened when only prey without morphological defences were considered, owing to an increase in PI in more temporary ponds (Fig. 2b). The biomass of predatory invertebrates in the unmanipulated cages did not vary significantly with pond permanence when expressed as either biomass per cage, or proportion of total invertebrate biomass (Fig. S4, Supporting Information; P > 0
15).

mesocosm experiment 1 : manipulation of fish and aquatic vegetation There was no detectable effect of predatory fish on prey biomass in the mesocosms (Fig. 3). Total prey biomass was not significantly different between fish and fishless treatments (F1,12 = 0
29, P = 0
60; Fig. 3), or between complex and simple vegetation treatments (F1,12 = 2
74, P = 0
12), and there was no fish by vegetation interaction

(F1,12 = 1
03, P = 0
33). Similarly, PERMANOVA indicated that there were no treatment effects on the composition of the prey community (fish: F1,12 = 0
84, P = 0
51; vegetation: F1,12 = 1
95, P = 0
11; interaction: F1,12 = 1
17, P = 0
28). Lastly, fish had no effect on the total biomass of large-bodied (>4 mg DW) predatory taxa (ANOVA, fish: F1,12 = 0
52, P = 0
48, interaction: F1,12 = 0
87, P = 0
37).

mesocosm experiment 2 : multiple predators Predation rates of Rhantus beetles as the sole predator increased significantly with density (one-way ANOVA: F2,9 = 7
26, P = 0
013; Fig. 4a), but observed predation rates in high density treatments were significantly lower than expected based on additive predation at low densities (Fig. 4a). The additive manipulation of predator diversity had significant impacts on the consumption of Chironomus (Fig. 4b, One-way ANOVA: F5,18 = 17
9, P < 0
0001). In treatments with single predator species, predation rates of Rhantus beetle larvae and Procordulia dragonflies were almost identical, but Xanthocnemis damselflies were a significantly weaker predator on Chironomus (Fig. 4b). The two-predator species treatment resulted in significantly higher predation rates than by Xanthocnemis alone, but was not significantly different from rates for Procordulia or Rhantus alone. In contrast with the single-species density manipulation, increasing predator density by increasing diversity led to predation rates almost identical to expected values calculated from the single-species treatments (Fig. 4b). However, the addition of fish chemical cues significantly decreased predation rates in the threespecies treatment to a level similar to the two-species treatment and not significantly higher than for Rhantus and Procordulia alone at low density (Fig. 4a). The substitutive manipulation of predator diversity, in which total predator density was kept constant, significantly influenced predation on Chironomus (Fig. 4c, oneway ANOVA: F3,12 = 4
29, P = 0
028); however, this was due to a reduction in predation rates with fish cues in the three-species treatment (Fig. 4c). Post-hoc tests indicated no significant difference between the one-, two- and threespecies treatments in the absence of fish cues (P > 0
34).

Discussion Decreases in the intensity of competition and predation with increasing abiotic stress or disturbance are central to several models of community organization across environmental gradients (Menge & Sutherland 1987; Wellborn, Skelly & Werner 1996; Chesson 2000). Our field assays of predator–prey interactions across a pond drying gradient revealed patterns opposite to these predictions. Top-down control of prey biomass decreased with pond permanence despite an increase in the size, biomass and diversity of predators in increasingly permanent ponds (Wissinger, Greig & McIntosh 2009). Our mesocosm experiment suggested that those patterns could result from the non-con-

© 2013 The Authors. Journal of Animal Ecology © 2013 British Ecological Society, Journal of Animal Ecology, 82, 598–607

Disturbance increases top-down control (a) Total prey biomass

Total prey biomass (mg AFDM per tank)

–3

–2

–1

Predatory invertebrate impact

0

1

–3

P = 0·035 R2 = 0·37

603

Fish Fishless

400

300

200

100

0

Present

Absent

Vegetation cover

(b) Unprotected prey biomass

Fig. 3. Effect of predatory fish (Galaxias brevipinnis) and submerged vegetation on total biomass of 11 invertebrate species after 14 days in a mesocosm experiment. Means (SE) were calculated with tanks as replicates. Vegetation cover indicates the presence or absence of submergent and emergent vegetation in mesocosms.

–2

–1

0

1 P = 0·005 R2 = 0·57

–1·5

–0·5 0·5 1·5 Pond permanence index

2·5

Fig. 2. Predatory invertebrate impact index for (a) total prey biomass and (b) unprotected prey biomass (i.e. snails and cased caddis excluded) in ponds varying in permanence due to drying. Negative predator impacts indicate a decrease in prey biomass relative to predator removal treatments; higher values on the X-axis indicate more permanent ponds. Ponds with and without fish are indicated by triangles and circles, respectively. Regression equations are y = 0
51 + 0
59x and y = 0
56 + 0
92x for total prey biomass and undefended prey biomass respectively.

sumptive effects of fish weakening predation by invertebrates in the more complex predator guilds of permanent ponds. Below we evaluate these unexpected patterns and their implications for understanding top-down control in food webs.

consumptive effects of predatory fish Long-lived predatory vertebrates such as fish and paedomorphic salamanders can only persist in perennial freshwater habitats. Consequently, they are often implicated as the key drivers of community shifts across pond permanence gradients (Wellborn, Skelly & Werner 1996; Wissinger 1999). Unexpectedly, our results revealed negligible consumptive effects of predatory fish on invertebrate biomass in permanent ponds. We observed no measurable effect of fish exclusion on prey biomass in cages within permanent ponds, but variability between ponds led to low power to detect a significant mean effect of fish across

all fish containing ponds. (minimum detectable effect sizes at b = 0
8 was 4–10 times greater than those presented in Fig. S3, Supporting Information). Nevertheless, our first mesocosm experiment revealed that even in spatially confined environments, predatory fish did not depress prey biomass. In this experiment, power was sufficient (at b = 0
8) to detect a 15% difference in prey biomass between fish and fishless tanks which is lower than effect sizes observed in other studies of fish predation on littoral benthic invertebrates (Morin 1984; Hershey & Dodson 1985; Diehl 1992). The absence of fish effects were surprising given that (a) koaro of the size used in the experiment regularly feed on the taxa and size ranges of invertebrates added to the mesocosms (Rowe, Konui & Christie 2002), and (b) we observed koaro feeding in the experiment, which was confirmed by the presence of Rhantus and Xanthocnemis larvae in the gut contents of several fish sacrificed at the conclusion of the experiment. Several hypotheses could explain the apparent absence of consumptive effects of predatory fish in our system. First, the structural complexity provided by dense stands of aquatic vegetation provides prey refugia that can lower the foraging efficiency of fish and reduce top-down control of prey (Diehl 1992; Pierce & Hinrichs 1997). Those refugia are likely to be more effective for prey avoiding large, actively foraging, visual predators like fish (Swisher, Soluk & Wahl 1998), than predatory invertebrates that are similar in body size and habitat use to prey. However, in our experiment the presence of refugia alone could not explain weak consumptive fish effects, as there was no difference in predation between mesocosms with or without submergent vegetation. A second potential explanation for minimal consumptive effects of fish is that size-selective foraging by fish on predatory invertebrates released primary consumers from consumption by mesopredators (Wooster 1994; Meissner & Muotka 2006). However, we

© 2013 The Authors. Journal of Animal Ecology © 2013 British Ecological Society, Journal of Animal Ecology, 82, 598–607

604 H. S Greig, S. A. Wissinger & A. R. McIntosh 80

(a)

** 60

b

*

ab a

40

20

0 4

8

12

Chrionomus predation per tank in 5 days

Rhantus density per tank 80

(b) c *

60

a

a a

40

a

b

20

0 R

X

P

RX

RXP RXP fish

Additive predator diversity 80

(c) a ab

60

ab b

did not observe any effects of fish on the biomass of predatory invertebrates in either the field assay or in the mesocosm experiment, providing little support for this hypothesis. Similarly, predatory invertebrate biomass was not negatively correlated with permanence, which would be expected if fish consumption suppressed large-bodied invertebrates. Finally, the effect of fish manipulations may be limited if the prey community is a preselected subset of the regional species pool that is resistant to fish predation (Allan 1982; Thorp 1986). However, the prey community in the mesocosms consisted of species collected from habitats with and without fish. Furthermore, there was no consistent effect of fish on unprotected prey species in the assay, indicating that morphologically defended prey were not responsible for the lack of fish effects. The most likely explanation for negligible fish effects is that weak consumption by fish was balanced by strong interactions within the invertebrate community in fishless mesocosms. For example, intraguild predation and cannibalism within prey communities is generally stronger in the absence of risk-sensitive behavioural responses to predators (Schmitz 2007; Rudolf 2008). Regardless of the mechanism, our results and those in previous studies (Wissinger, McIntosh & Greig 2006; Wissinger, Greig & McIntosh 2009), indicate that fish feeding has weak impact at best on epibenthic invertebrate community composition in the lentic habitats of our study region. Importantly, however, fish may still exert a strong influence on food-web interactions through non-consumptive effects on prey individuals (McPeek & Peckarsky 1998; Werner & Peacor 2003; Peckarsky et al. 2008), and these interaction modifications were evident in the multiple predator experiment.

40

the influence of food web complexity on top-down control

20

A second experiment with multiple predators provided strong evidence that the non-consumptive presence of fish (as indicated by fish odour) reduced prey consumption by predatory invertebrates. The presence of fish cues reduced predation on chironomids in the three-species treatment to levels more similar to those found in low-density single-species treatments. Thus, if the consumptive effects of fish are as weak, as indicated by the mesocoms experiments and in situ manipulations, the presence of fish in permanent ponds in our study sites should result in a net decrease in top-down control of benthic invertebrates. We contend that the non-consumptive effects of fish weakened the effects of predatory invertebrates in permanent ponds leading to a decrease in the strength of predator– prey interactions with increasing pond permanence. The suppression of predatory invertebrate foraging also provides a mechanism for the negligible effects of fish on invertebrate prey communities in the mesocosm experiment. Furthermore, these results provide support for the

0

RXP

R

RX

RXP fish

Substitutive predator diversity

Fig. 4. Effect of (a) larval Rhantus density, (b) additive predator diversity and (c) substitutive predator diversity on mean (SE) consumption of Chironomus larvae during a 5-day mesocosm experiment. X-axis labels for (b) and (c) are as follows: R, Rhantus larvae; X, Xanthocnemis larvae; P, Procordulia larvae; fish, presence of chemical cues from both rainbow trout (Onchorynchus mykiss) and koaro (Galaxias brevipinnis). See Fig. 1 for further description of treatments. Means were calculated with tanks as replicates; letters indicate significant differences (P < 0
05) between treatments (Tukey’s post-hoc comparisons). Dashed lines indicate expected Chironomus predation calculated from additive probability models (adjusted for finite prey resources) of low density conspecifics in (a), and from single-species treatments in (b). Asterisks indicate significant departures from the expected model (*0
1 < P < 0
05, **P < 0
05) based on one-sample t-tests.

© 2013 The Authors. Journal of Animal Ecology © 2013 British Ecological Society, Journal of Animal Ecology, 82, 598–607

Disturbance increases top-down control hypothesis that non-consumptive effects of fish in freshwater food webs often outweigh their consumptive effects (e.g. McPeek & Peckarsky 1998). Our experiments focused on the most commonly observed predatory invertebrates across the permanence gradient, and indicate that increased invertebrate predator diversity in the absence of fish cues led to small positive effects of diversity on predator impacts, rather than antagonistic effects among species. The additive model in particular revealed that increases in predator density by enhancing diversity may reduce negative intraspecific interactions apparent in high densities of a single predator taxa (as in Griffin et al. 2008). This is consistent with the assertion that intraspecific interactions among predators are stronger than interspecific interactions (Byrnes & Stachowicz 2009). Predation rates of Rhantus beetle larvae at high densities were significantly lower than the expected additive predation rate calculated from low density treatments, but predation rates in high density treatments of multiple predators were almost identical to expected rates. The Rhantus only treatment is representative of predator guilds in the most temporary ponds in our study area (< 1 month inundated), which we were unable to include in the field experiment because of the logistical constraints of their short duration. The small positive effects of predatory invertebrate diversity we observed suggest that predator impact may follow a hump-shaped relationship with pond permanence when the full range of pond duration is considered. Differences we observed in the effects of diversity by adding trophic levels or adding species within trophic levels reinforce the contention that horizontal and vertical diversity components of food webs have different but interactive effects on species interactions (Duffy et al. 2007; Srivastava & Bell 2009). Vertical diversity can increase top-down control, for example, when top predators increase spatial overlap of mesopredators and prey (Grabowski & Kimbro 2005). However, in many cases, the addition of predator trophic levels should dampen top-down control. Body size generally increases with trophic level (Woodward et al. 2005), so increasing vertical diversity in food webs generates size asymmetry between predator species that promotes intraguild predation and density- and trait-mediated foraging suppression of mesopredators. This is analogous to the effects of introducing size-structure within predator populations (Rudolf 2007). These negative interactions among trophic levels may overwhelm positive effects of within-trophic level diversity, as observed in our study when fish cues reduced the foraging of three species predator guilds to rates similar to single species guilds. Several aspects of food web complexity in addition to diversity may have contributed to the decreased predator impact with more permanent ponds. Many permanent pond species have semivoltine life histories, which results in overlapping cohorts at a given time. Consequently, size-structured interactions such as cannibalism and intra-

605

guild predation that can weaken top-down control (Polis 1991; Padeffke & Suhling 2003) are likely to be more prevalent in permanent ponds than in temporary ponds where drying and refilling promotes developmental synchrony within populations. Permanent ponds also often contain a higher proportion of morphologically defended prey species (e.g. snails and cased caddisflies) than temporary ponds, although the negative relationship between predator impact and permanence actually strengthened when we excluded morphologically defended prey from the analysis (Fig. 3b). Finally, prey in resource-limited or time-constrained habitats such as temporary ponds may be less likely to exhibit risk-sensitive foraging and other antipredatory responses, which should strengthen top-down control.

Conclusions Interactions between the same sets of species can reverse in direction or change in strength depending on environmental and consumer context (Crain 2008). Our study provides evidence that changes in the properties of food webs along environmental gradients from structurally simple, chain-like food webs in disturbed habitats, to complex, reticulate food webs in physically benign habitats are likely to dampen top-down control and generate unexpected relationships between abiotic stress and the strength of species interactions. These results emphasize that considering food web complexity, especially the nonconsumptive effects of top predators, is essential to understand species interactions across environmental gradients.

Acknowledgements We thank A. Klemmer, P. Jellyman, E. Isherwood, C. Ross and M. Galatowitsch for field and laboratory assistance. N. Etheridge and V. Greig constructed the cages, and M. Fraundorfer, R. Hill, J. Westenra, R. Smith and the Dept. of Conservation granted access to research sites. We thank B. Peckarsky, R. Didham, B. Sorrell, M. Winterbourn, A. Hildrew and G. Closs for valuable comments and discussion. The research was funded by the Miss E.L. Hellaby Indigenous Grassland Research Trust. HSG was supported by a Top Achiever Doctoral Scholarship and a Foundation for Science, Research and Technology Postdoctoral Fellowship and SAW was supported by an Erskine Fellowship from the University of Canterbury.

References Agrawal, A.A., Ackerly, D.D., Adler, F., Arnold, A.E., Caceres, C., Doak, D.F., Post, E., Hudson, P.J., Maron, J., Mooney, K.A., Power, M., Schemske, D., Stachowicz, J., Strauss, S., Turner, M.G. & Werner, E. (2007) Filling key gaps in population and community ecology. Frontiers in Ecology and the Environment, 5, 145–152. Allan, J.D. (1982) The effects of reduction in trout density on the invertebrate community of a mountain stream. Ecology, 63, 1444–1455. Anderson, M.J. (2001) A new method for non-parametric multivariate analysis of variance. Austral Ecology, 26, 32–46. Benke, A.C., Huryn, A.D., Smock, L.A. & Wallace, J.B. (1999) Lengthmass relationships for freshwater macroinvertebrates in North America with particular reference to the southeastern United States. Journal of the North American Benthological Society, 18, 308–343. Bilton, D.T., Foggo, A. & Rundle, S.D. (2001) Size, permanence and the proportion of predators in ponds. Archiv Fur Hydrobiologie, 151, 451–458.

© 2013 The Authors. Journal of Animal Ecology © 2013 British Ecological Society, Journal of Animal Ecology, 82, 598–607

606 H. S Greig, S. A. Wissinger & A. R. McIntosh Berlow, E.L., Neutel, A.M., Cohen, J.E., de Ruiter, P.C., Ebenman, B., Emmerson, M., Fox, J.W., Jansen, V.A.A., Jones, J.I., Kokkoris, G.D., Logofet, D.O., McKane, A.J., Montoya, J.M. & Petchey, O. (2004) Interaction strengths in food webs: issues and opportunities. Journal of Animal Ecology, 73, 585–598. Brendonck, L., Michels, E., De Meester, L. & Riddoch, B. (2002) Temporary pools are not ‘enemy-free’. Hydrobiologia, 486, 147–159. Byrnes, J.E. & Stachowicz, J.J. (2009) The consequences of consumer diversity loss: different answers from different experimental designs. Ecology, 90, 2879–2888. Chesson, P. (2000) Mechanisms of maintenance of species diversity. Annual Review of Ecology and Systematics, 31, 343–366. Chesson, P. & Huntly, N. (1997) The roles of harsh and fluctuating conditions in the dynamics of ecological communities. American Naturalist, 150, 519–553. Connell, J.H. (1975) Some mechanisms producing structure in natural communities: a model and evidence from field experiments. Ecology and Evolution of Communities (eds M.L. Cody & J.M. Diamond), pp. 460– 490. Bclknap, Harvard, Cambridge, MT. Crain, C.M. (2008) Interactions between marsh plant species vary in direction and strength depending on environmental and consumer context. Journal of Ecology, 96, 166–173. Crain, C.M., Silliman, B.R., Bertness, S.L. & Bertness, M.D. (2004) Physical and biotic drivers of plant distribution across estuarine salinity gradients. Ecology, 85, 2539–2549. Diehl, S. (1992) Fish predation and benthic community structure: the role of omnivory and habitat complexity. Ecology, 73, 1646–1661. Duffy, J.E., Cardinale, B.J., France, K.E., McIntyre, P.B., Thebault, E. & Loreau, M. (2007) The functional role of biodiversity in ecosystems: incorporating trophic complexity. Ecology Letters, 10, 522–538. Edwards, K.F. & Stachowicz, J.J. (2010) Multivariate trade-offs, succession, and phenological differentiation in a guild of colonial invertebrates. Ecology, 91, 3146–3152. Emmerson, M. & Yearsley, J.M. (2004) Weak interactions, omnivory and emergent food-web properties. Proceedings of the Royal Society of London Series B-Biological Sciences, 271, 397–405. Finke, D.L. & Denno, R.F. (2004) Predator diversity dampens trophic cascades. Nature, 429, 407–410. Gerhardt, F. & Collinge, S.K. (2007) Abiotic constraints eclipse biotic resistance in determining invasibility along experimental vernal pool gradients. Ecological Applications, 17, 922–933. Grabowski, J.H. & Kimbro, D.L. (2005) Predator-avoidance behavior extends trophic cascades to refuge habitats. Ecology, 86, 1312–1319. Griffin, J.N., De la Haye, K.L., Hawkins, S.J., Thompson, R.C. & Jenkins, S.R. (2008) Predator diversity and ecosystem functioning: density modifies the effect of resource partitioning. Ecology, 89, 298– 305. Grime, J.P. (1977) Evidence for existence of 3 primary strategies in plants and its relevance to ecological and evolutionary theory. American Naturalist, 111, 1169–1194. Hershey, A.E. & Dodson, S.I. (1985) Selective predation by a sculpin and a stonefly on 2 chironomids in laboratory feeding trials. Hydrobiologia, 124, 269–273. Jenkins, B., Kitching, R.L. & Pimm, S.L. (1992) Productivity, disturbance and food web structure at a local spatial scale in experimental container habitats. Oikos, 65, 249–255. Jeppesen, E., Lauridsen, T., Mitchell, S.F. & Burns, C.W. (1997) Do planktivorous fish structure the zooplankton communities in New Zealand lakes? New Zealand Journal of Marine and Freshwater Research, 31, 163–173. Lubchenco, J. (1980) Algal zonation in the New England rocky intertidal community: an experimental analysis. Ecology, 61, 333–344. McAbendroth, L., Foggo, A., Rundle, S.D. & Bilton, D.T. (2005) Unravelling nestedness and spatial pattern in pond assemblages. Journal of Animal Ecology, 74, 41–49. McPeek, M.A. & Peckarsky, B.L. (1998) Life histories and the strengths of species interactions: combining mortality, growth, and fecundity effects. Ecology, 79, 867–879. Meissner, K. & Muotka, T. (2006) The role of trout in stream food webs: integrating evidence from field surveys and experiments. Journal of Animal Ecology, 75, 421–433. Menge, B.A. & Farrell, T.M. (1989) Community structure and interaction webs in shallow marine hard-bottom communities: tests of an environmental-stress model. Advances in Ecological Research, 19, 189– 262.

Menge, B.A. & Sutherland, J.P. (1987) Community regulation: variation in disturbance, competition, and predation in relation to environmentalstress and recruitment. American Naturalist, 130, 730–757. Miller, G.K. (2006) Probability: Modelling and Applications to Random Processes. Willey, Hoboken, USA. Morin, P.J. (1984) The impact of fish exclusion on the abundance and species composition of larval odonates: results of short-term experiments in a North Carolina farm pond. Ecology, 65, 53–60. Padeffke, T. & Suhling, F. (2003) Temporal priority and intra-guild predation in temporary waters: an experimental study using Namibian desert dragonflies. Ecological Entomology, 28, 340–347. Peckarsky, B.L. (1983) Biotic interactions or abiotic limitations? A model of lotic community structure. Dynamics of Lotic Ecosystems (eds T.D. Fontaine III & S.M. Bartell), pp. 303–323. Ann Arbor Science Publications, Ann Arbor, USA. Peckarsky, B.L., Kerans, B.L., Taylor, B.W. & McIntosh, A.R. (2008) Predator effects on prey population dynamics in open systems. Oecologia, 156, 431–440. Pierce, C.L. & Hinrichs, B.D. (1997) Response of littoral invertebrates to reduction of fish density: simultaneous experiments in ponds with different fish assemblages. Freshwater Biology, 37, 397–408. Polis, G.A. (1991) Complex trophic interactions in deserts: an empirical critique of food-web theory. American Naturalist, 138, 123–155. Power, M.E., Parker, M.S. & Wootton, J.T. (1996) Disturbance and food chain length in rivers. Food Webs: Contemporary Perspectives (eds G.A. Polis & K.O. Winemiller), pp. 286–297. Chapman and Hall, New York, USA. Rowe, D.K., Konui, G. & Christie, K.D. (2002) Population structure, distribution, reproduction, diet, and relative abundance of koaro (Galaxias brevipinnis) in a New Zealand lake. Journal of the Royal Society of New Zealand, 32, 275–291. Rudolf, V.H.W. (2007) Consequences of stage-structured predators: cannibalism, behavioral effects, and trophic cascades. Ecology, 88, 2991–3003. Rudolf, V.H.W. (2008) The impact of cannibalism in the prey on predator-prey systems. Ecology, 89, 3116–3127. Schmitz, O.J. (2007) Predator diversity and trophic interactions. Ecology, 88, 2415–2426. Schneider, D.W. & Frost, T.M. (1996) Habitat duration and community structure in temporary ponds. Journal of the North American Benthological Society, 15, 64–86. Soluk, D.A. & Collins, N.C. (1988) Balancing risks: responses and nonresponses of mayfly larvae to fish and stonefly predators. Oecologia, 77, 370–374. Srivastava, D.S. & Bell, T. (2009) Reducing horizontal and vertical diversity in a foodweb triggers extinctions and impacts functions. Ecology Letters, 12, 1016–1028. StatSoft (2008) STATISTICA (data analysis software system), Version 8. Tulsa, Oklahoma. Stoffels, R.J., Karbe, S. & Paterson, R.A. (2003) Length-mass models for some common New Zealand littoral-benthic macroinvertebrates, with a note on within-taxon variability in parameter values among published models. New Zealand Journal of Marine and Freshwater Research, 37, 449–460. Swisher, B.J., Soluk, D.A. & Wahl, D.H. (1998) Non-additive predation in littoral habitats: influences of habitat complexity. Oikos, 81, 30–37. Thompson, R.M. & Townsend, C.R. (1999) The effect of seasonal variation on the community structure and food-web attributes of two streams: implications for food-web science. Oikos, 87, 75–88. Thorp, J.H. (1986) Two distinct roles for predators in fresh-water assemblages. Oikos, 47, 75–82. Urban, M.C. (2004) Disturbance heterogeneity determines freshwater metacommunity structure. Ecology, 85, 2971–2978. Walters, A.W. & Post, D.M. (2008) An experimental disturbance alters fish size structure but not food chain length in streams. Ecology, 89, 3261–3267. Wellborn, G.A., Skelly, D.K. & Werner, E.E. (1996) Mechanisms creating community structure across a freshwater habitat gradient. Annual Review of Ecology and Systematics, 27, 337–364. Werner, E.E. & Peacor, S.D. (2003) A review of trait-mediated indirect interactions in ecological communities. Ecology, 84, 1083–1100. Williams, D.D. (1996) Environmental constraints in temporary fresh waters and their consequences for the insect fauna. Journal of the North American Benthological Society, 15, 634–650. Winemiller, K.O. (1990) Spatial and temporal variation in tropical fish trophic networks. Ecological Monographs, 60, 331–367.

© 2013 The Authors. Journal of Animal Ecology © 2013 British Ecological Society, Journal of Animal Ecology, 82, 598–607

Disturbance increases top-down control Wissinger, S.A. (1999) Ecology of wetland invertebrates: synthesis and applications for conservation and management. Invertebrates in Freshwater Wetlands of North America (eds D.P. Batzer, R.B. Rader & S.A. Wissinger), pp. 1043–1086. Willey, New York. Wissinger, S.A., Greig, H. & McIntosh, A. (2009) Absence of species replacements between permanent and temporary lentic communities in New Zealand. Journal of the North American Benthological Society, 28, 12–23. Wissinger, S.A., McIntosh, A.R. & Greig, H.S. (2006) Impacts of introduced brown and rainbow trout on benthic invertebrate communities in shallow New Zealand lakes. Freshwater Biology, 51, 2009–2028. Wissinger, S.A., Bohonak, A.J., Whiteman, H.H. & Brown, W.S. (1999) Habitat permanence, salamander predation and invertebrate communities. Invertebrates in Freshwater Wetlands of North America: Ecology and Management (eds D.P. Batzer, R.B. Rader & S.A. Wissinger), pp. 757–790. John Wiley & Sons, Inc, New York, USA. Wissinger, S.A., Whissel, J.C., Eldermire, C. & Brown, W.S. (2006) Predator defense along a permanence gradient: roles of case structure, behavior, and developmental phenology in caddisflies. Oecologia, 147, 667– 678. Woodward, G., Ebenman, B., Ernmerson, M., Montoya, J.M., Olesen, J.M., Valido, A. & Warren, P.H. (2005) Body size in ecological networks. Trends in Ecology & Evolution, 20, 402–409. Wooster, D. (1994) Predator impacts on stream benthic prey. Oecologia, 99, 7–15.

607

Supporting Information Additional Supporting Information may be found in the online version of this article. Table S1. A list of environmental variables for each pond, including hydrological parameters underlying the multivariate pond permanence index, and the identity of predatory fish. Figure S1. Photos and description of the mesocosm experiment manipulating the presence and absence of fish and vegetation. Figure S2. Photo of the mesocosms used in the predator diversity manipulation. Figure S3. The impact of predatory fish on invertebrate biomass in the in situ cage experiment. Figure S4. Regression of predatory invertebrate biomass in unmanipulated cages against pond permanence for the in situ cage experiment.

Received 30 July 2012; accepted 29 November 2012 Handling Editor: Karl Cottenie

© 2013 The Authors. Journal of Animal Ecology © 2013 British Ecological Society, Journal of Animal Ecology, 82, 598–607