Ecosystem-size relationships of river populations and communities.

Knowledge of ecosystem-size influences on river populations and communities is integral to the balancing of human and environmental needs for water. The multiple dimensions of dendritic river networks complicate understanding of ecosystem-size influences, but could be resolved by the development of scaling relationships. We highlight the importance of physical constraints limiting predator body sizes, movements, and population sizes in small rivers, and where river contraction limits space or creates stressful conditions affecting community stability and food webs. Investigations of the scaling and contingency of these processes will be insightful because of the underlying generality and scale independence of such relationships. Doing so will also pinpoint damaging water-management practices and identify which aspects of river size can be most usefully manipulated in river restoration.

Predicting biomass of resident kokopu (Galaxias) populations using local habitat characteristics.

With the global decline of freshwater fishes, quantifying the body size-specific habitat use of vulnerable species is crucial for accurately evaluating population health, identifying the effects of anthropogenic stressors, and directing effective habitat restoration. Populations of New Zealand's endemic kokopu species (Galaxias fasciatus, G. argenteus, and G. postvectis) have declined substantially over the last century in response to anthropogenic stressors, including habitat loss, migratory barriers, and invasive species. Despite well-understood habitat associations, key within-habitat features underpinning the reach-scale biomass of small and large kokopu remain unclear. Here, we investigated whether the total biomass of large (> 90 mm) size classes of each kokopu species and the composite biomass of all small (≤ 90 mm) kokopu were associated with components of the physical environment that provided refuge and prey resources across fifty-seven 50-m stream reaches. Because kokopu are nocturnal, populations were sampled by removal at night using headlamps and hand-nets until reaches were visually depleted. Based on Akaike's information criterion, greater large banded kokopu biomass was most parsimoniously explained by greater pool volume and forest cover, greater large giant kokopu biomass by greater bank cover and pool volume, and greater large shortjaw kokopu biomass by greater substrate size and pool volume. In contrast, greater composite small kokopu biomass was best explained by smaller substrate size, reduced bank cover, and greater pool volume. Local habitat associations therefore varied among kokopu species and size classes. Our study demonstrates the importance of considering the ontogenetic shift in species' habitat use and provides an effective modelling approach for quantifying size-specific local habitat use of stream-dwelling fish.

Quantifying the relative contributions of habitat modification and mammalian predators on landscape-scale declines of a threatened river specialist duck.

Habitat modification and introduced mammalian predators are linked to global species extinctions and declines, but their relative influences can be uncertain, often making conservation management difficult. Using landscape-scale models, we quantified the relative impacts of habitat modification and mammalian predation on the range contraction of a threatened New Zealand riverine duck. We combined 38 years of whio (Hymenolaimus malacorhynchos) observations with national-scale environmental data to predict relative likelihood of occurrence (RLO) under two scenarios using bootstrapped boosted regression trees (BRT). Our models used training data from contemporary environments to predict the potential contemporary whio distribution across New Zealand riverscapes in the absence of introduced mammalian predators. Then, using estimates of environments prior to human arrival, we used the same models to hindcast potential pre-human whio distribution prior to widespread land clearance. Comparing RLO differences between potential pre-human, potential contemporary and observed contemporary distributions allowed us to assess the relative impacts of the two main drivers of decline; habitat modification and mammalian predation. Whio have undergone widespread catastrophic declines most likely linked to mammalian predation, with smaller declines due to habitat modification (range contractions of 95% and 37%, respectively). We also identified areas of potential contemporary habitat outside their current range that would be suitable for whio conservation if mammalian predator control could be implemented. Our approach presents a practical technique for estimating the relative importance of global change drivers in species declines and extinctions, as well as providing valuable information to improve conservation planning.

Regulation of open populations of a stream insect through larval density dependence.

In organisms with complex life cycles, the various stages occupy different habitats creating demographically open populations. The dynamics of these populations will depend on the occurrence and timing of stochastic influences relative to demographic density dependence, but understanding of these fundamentals, especially in the face of climate warming, has been hampered by the difficulty of empirical studies. Using a logically feasible organism, we conducted a replicated density-perturbation experiment to manipulate late-instar larvae of nine populations of a stream caddisfly, Zelandopsyche ingens, and measured the resulting abundance over 2 years covering the complete life cycle of one cohort to evaluate influences on dynamics. Negative density feedback occurred in the larval stage, and was sufficiently strong to counteract variation in abundance due to manipulation of larval density, adult caddis dispersal in the terrestrial environment as well as downstream drift of newly hatched and older larvae in the current. This supports theory indicating regulation of open populations must involve density dependence in local populations sufficient to offset variability associated with dispersal, especially during recruitment, and pinpoints the occurrence to late in the larval life cycle and driven by food resource abundance. There were large variations in adult, egg mass and early instar abundance that were not related to abundance in the previous stage, or the manipulation, pointing to large stochastic influences. Thus, the results also highlight the complementary nature of stochastic and deterministic influences on open populations. Such density dependence will enhance population persistence in situations where variable dispersal and transitioning between life stages frequently creates mismatches between abundance and the local availability of resources, such as might become more common with climate warming.

Habitat size influences community stability.

Mechanisms linked to demographic, biogeographic, and food-web processes thought to underpin community stability could be affected by habitat size, but the effects of habitat size on community stability remain relatively unknown. We investigated whether those habitat-size-dependent properties influenced community instability and vulnerability to perturbations caused by disturbance. This is particularly important given that human exploitation is contracting ecosystems, and abiotic perturbations are becoming more severe and frequent. We used a perturbation experiment in which 10 streams, spanning three orders of magnitude in habitat size, were subjected to simulated bed movement akin to a major flood disturbance event. We measured the resistance, resilience, and variability of basal resources, and population and community-level responses across the stream habitat-size gradient immediately before, and at 0.5, 5, 10, 20, and 40 d post-disturbance. Resistance to disturbance consistently increased with stream size in all response variables. In contrast, resilience was significantly higher in smaller streams for some response variables. However, this higher resilience of small ecosystems was insufficient to compensate for their lower resistance, and communities of smaller streams were thus more variable over time than those of larger streams. Compensatory dynamics of populations, especially for predators, stabilized some aspects of communities, but these mechanisms were unrelated to habitat size. Together, our results provide compelling evidence for the links between habitat size and community stability, and should motivate ecologists and managers to consider how changes in the size of habitats will alter the vulnerability of ecosystems to perturbations caused by environmental disturbance.

Negative resistance and resilience: biotic mechanisms underpin delayed biological recovery in stream restoration.

Traditionally, resistance and resilience are associated with good ecological health, often underpinning restoration goals. However, degraded ecosystems can also be highly resistant and resilient, making restoration difficult: degraded communities often become dominated by hyper-tolerant species, preventing recolonization and resulting in low biodiversity and poor ecosystem function. Using streams as a model, we undertook a mesocosm experiment to test if degraded community presence hindered biological recovery. We established 12 mesocosms, simulating physically healthy streams. Degraded invertebrate communities were established in half, mimicking the post-restoration scenario of physical recovery without biological recovery. We then introduced a healthy colonist community to all mesocosms, testing if degraded community presence influenced healthy community establishment. Colonists established less readily in degraded community mesocosms, with larger decreases in abundance of sensitive taxa, likely driven by biotic interactions rather than abiotic constraints. Resource depletion by the degraded community likely increased competition, driving priority effects. Colonists left by drifting, but also by accelerating development, reducing time to emergence but sacrificing larger body size. Since degraded community presence prevented colonist establishment, our experiment suggests successful restoration must address both abiotic and biotic factors, especially those that reinforce the 'negative' resistance and resilience which perpetuate degraded communities and are typically overlooked.

Cross-ecosystem bottlenecks alter reciprocal subsidies within meta-ecosystems.

Reciprocal subsidies link ecosystems into meta-ecosystems, but energy transfer to organisms that do not cross boundaries may create sinks, reducing reciprocal subsidy transfer. We investigated how the type of subsidy and top predator presence influenced reciprocal flows of energy, by manipulating the addition of terrestrial leaf and terrestrial insect subsidies to experimental freshwater pond mesocosms with and without predatory fish. Over 18 months, fortnightly addition of subsidies (terrestrial beetle larvae) to top-predators was crossed with monthly addition of subsidies (willow leaves) to primary consumers in mesocosms with and without top predators (upland bullies) in a 2 × 2 × 2 factorial design in four replicate blocks. Terrestrial insect subsidies increased reciprocal flows, measured as the emergence of aquatic insects out of mesocosms, but leaf subsidies dampened those effects. However, the presence of fish and snails, consumers with no terrestrial life stage, usurped and retained the energy within in the aquatic ecosystem, creating a cross-ecosystem bottleneck to energy flow. Thus, changes in species composition of donor or recipient food webs within a meta-ecosystems can alter reciprocal subsidies through cross-ecosystem bottlenecks.

Mechanisms of trophic niche compression: Evidence from landscape disturbance.

Natural and anthropogenic disturbances commonly alter patterns of biodiversity and ecosystem functioning. However, how networks of interacting species respond to these changes remains poorly understood. We described aquatic food webs using invertebrate and fish community composition, functional traits and stable isotopes from twelve agricultural streams along a landscape disturbance gradient. We predicted that excessive deposition of fine inorganic sediment (sedimentation) associated with agricultural activities would negatively influence aquatic trophic diversity (e.g. reduced vertical and horizontal trophic niche breadths). We hypothesized that multiple mechanisms might cause trophic niche 'compression', as indicated by changes in realized trophic roles. Food-web properties based on consumer stable isotope data (δ13 C and δ15 N) showed that increasing sediment disturbance was associated with reduced trophic diversity. In particular, the aquatic invertebrate community occupied a smaller area in isotopic niche space along the sedimentation gradient that was best explained by a narrowing of the invertebrate community δ13 C range. Decreased niche partitioning, driven by increasing habitat homogeneity, environmental filtering and resource scarcity all seemingly lead to greater trophic equivalency caused by the collapse of the autochthonous food-web channel. Bayesian mixing-model analyses supported this contention with invertebrate consumers increasingly reliant on detritus along the sedimentation gradient, and predatory invertebrates relying more on the prey using these basal resources. The narrowing of the fish community δ13 C range along the sedimentation gradient contributed to an apparent 'trophic shift' towards terrestrial carbon, further indicating the loss of the autochthonous food-web channel. On the vertical trophic niche axis, fish became increasingly separated from aquatic invertebrates with an increase in their estimated trophic position. In combination, these responses were most likely mediated through reduced fish densities and a diminished reliance on aquatic prey. Although species losses remain a major threat to ecosystem integrity, the functional roles of biota that persist dictate how food webs and ecosystem functioning respond to environmental change. Sedimentation was associated with nonlinear reductions in trophic diversity which could affect the functioning and stability of aquatic ecosystems. Our study helps explain how multiple mechanisms may radically reshape food-web properties in response to this type of disturbance.

Disturbance-mediated consumer assemblages determine fish community structure and moderate top-down influences through bottom-up constraints.

Disturbance is a strong structuring force that can influence the strength of species interactions at all trophic levels, but controls on the contributions to community structure of top-down and bottom-up processes across such gradients remain poorly understood. Changes in the composition of predator and consumer assemblages, and their associated traits, across gradients of environmental harshness (e.g. flooding), are likely to be a particularly important influence on the strength of top-down control and may drive bottom-up constraints. We examined how consumers with particular traits, and the predators that consumed them, varied across a gradient of stream flooding disturbance and used experiments to assess the predation impact on those contrasting consumer communities (ultimately quantifying how flood disturbance altered the strength of top-down control). Consumer community composition and mobility were strongly related to flood disturbance; the biomass and drift of protected primary consumers (i.e. those with morphological defences) decreased with increasing flood disturbance. Predatory fish species had different disturbance niches, and path analysis identified that both direct flood disturbance effects and indirect bottom-up constraints of flood disturbance on consumers influenced predatory fish composition and biomass. Fish generally fed most effectively on consumer types associated with their particular niche, but all fish were strongly size-selective when feeding on protected consumers. Although protected consumers did not grow large enough to escape predation, an in situ experiment showed protected consumers were at a reduced risk of predation as disturbance increased compared to unprotected consumers. Overall, top-down control declined with flood disturbance, but the effect depended on consumer traits. Predatory fish were only capable of exerting top-down control on protected consumers in benign habitats but impacted unprotected consumers across a larger range of the disturbance gradient. Collectively our findings suggest that a shift towards a more disturbed state will probably result in reduced predator impacts and a weakening of top-down control. Moreover, predicted increases in the frequency and intensity of climatic events causing disturbance, such as flooding, are likely to result in a community shift that disproportionately impacts protected consumers and the predators that utilize them as prey through the subsequent bottom-up constraints.

Springs drive downstream nitrate export from artificially-drained agricultural headwater catchments.

Excessive nutrient loading from small agricultural headwaters can substantially degrade downstream water quality and ecological conditions. But, our understanding of the scales and locations to implement nutrient attenuation tools within these catchments is poor. To help inform farm- and catchment-scale management, we quantified nitrate export in nine one-kilometre-long lowland agricultural headwaters fed by tile and open tributary drains in a region with high groundwater nitrate (<1 to >15 mg L-1 NO3-N) over four years. Across-catchment differences in upstream spring water nitrate concentrations predicted differences in annual nitrate loads at catchment outlets (range <1-72 megagrams NO3-N 365 d-1), and nitrate loads were higher in wet seasons and wet years, reflecting strong groundwater influences. Partitioning the sources of variability in catchment nitrate fluxes revealed that ~60% of variation was accounted for by a combination of fluxes from up-stream springs and contributions from tile and open tributary drains (46% and 15%, respectively), with ~40% of unexplained residual variation likely due to groundwater upwellings. Although tile and open tributary drains contributed comparatively less to catchment loads (tile drains: <0.01 and up to 50 kg NO3-N d-1; open drains: <5 kg and up to 100 kg NO3-N d-1), mitigation targeted at these localised, farm-scale sources will contribute to decreasing downstream nitrate fluxes. However, high nitrate loads from groundwater mean current NO3-N waterway management and rehabilitation practices targeting waterway stock exclusion by fencing alone will be insufficient to reduce annual NO3-N export. Moreover, managing catchment nutrient fluxes will need to acknowledge contributions from groundwater as well as farm-scale losses from land. Overall, our results highlight how nutrient fluxes in spring-fed waterways can be highly dynamic, dominated more by groundwater than local run-off, and point to the scales and locations where nitrate attenuation tools should be implemented.

Capacity to support predators scales with habitat size.

Habitat reduction could drive biodiversity loss if the capacity of food webs to support predators is undermined by habitat-size constraints on predator body size. Assuming that (i) available space restricts predator body size, (ii) mass-specific energy needs of predators scale with their body size, and (iii) energy availability scales with prey biomass, we predicted that predator biomass per unit area would scale with habitat size (quarter-power exponent) and prey biomass (three-quarter-power exponent). We found that total predator biomass scaled with habitat size and prey resources as expected across 29 New Zealand rivers, such that a unit of habitat in a small ecosystem supported less predator biomass than an equivalent unit in a large ecosystem. The lower energetic costs of large body size likely mean that a unit of prey resource supports more biomass of large-bodied predators compared to small-bodied predators. Thus, contracting habitat size reduces the predator mass that can be supported because of constraints on predator body size, and this may be a powerful mechanism exacerbating reductions in biodiversity due to habitat loss.

Metabolism drives distribution and abundance in extremophile fish.

Differences in population density between species of varying size are frequently attributed to metabolic rates which are assumed to scale with body size with a slope of 0.75. This assumption is often criticised on the grounds that 0.75 scaling of metabolic rate with body size is not universal and can vary significantly depending on species and life-history. However, few studies have investigated how interspecific variation in metabolic scaling relationships affects population density in different sized species. Here we predict inter-specific differences in metabolism from niche requirements, thereby allowing metabolic predictions of species distribution and abundance at fine spatial scales. Due to the differences in energetic efficiency required along harsh-benign gradients, an extremophile fish (brown mudfish, Neochanna apoda) living in harsh environments had slower metabolism, and thus higher population densities, compared to a fish species (banded kokopu, Galaxias fasciatus) in physiologically more benign habitats. Interspecific differences in the intercepts for the relationship between body and density disappeared when species mass-specific metabolic rates, rather than body sizes, were used to predict density, implying population energy use was equivalent between mudfish and kokopu. Nevertheless, despite significant interspecific differences in the slope of the metabolic scaling relationships, mudfish and kokopu had a common slope for the relationship between body size and population density. These results support underlying logic of energetic equivalence between different size species implicit in metabolic theory. However, the precise slope of metabolic scaling relationships, which is the subject of much debate, may not be a reliable indicator of population density as expected under metabolic theory.

The scaling of population persistence with carrying capacity does not asymptote in populations of a fish experiencing extreme climate variability.

Despite growing concerns regarding increasing frequency of extreme climate events and declining population sizes, the influence of environmental stochasticity on the relationship between population carrying capacity and time-to-extinction has received little empirical attention. While time-to-extinction increases exponentially with carrying capacity in constant environments, theoretical models suggest increasing environmental stochasticity causes asymptotic scaling, thus making minimum viable carrying capacity vastly uncertain in variable environments. Using empirical estimates of environmental stochasticity in fish metapopulations, we showed that increasing environmental stochasticity resulting from extreme droughts was insufficient to create asymptotic scaling of time-to-extinction with carrying capacity in local populations as predicted by theory. Local time-to-extinction increased with carrying capacity due to declining sensitivity to demographic stochasticity, and the slope of this relationship declined significantly as environmental stochasticity increased. However, recent 1 in 25 yr extreme droughts were insufficient to extirpate populations with large carrying capacity. Consequently, large populations may be more resilient to environmental stochasticity than previously thought. The lack of carrying capacity-related asymptotes in persistence under extreme climate variability reveals how small populations affected by habitat loss or overharvesting, may be disproportionately threatened by increases in extreme climate events with global warming.

Thinking beyond the Bioreactor Box: Incorporating Stream Ecology into Edge-of-Field Nitrate Management.

Around the world, artificially drained agricultural lands are significant sources of reactive nitrogen to stream ecosystems, creating substantial stream health problems. One management strategy is the deployment of denitrification enhancement tools. Here, we evaluate the factors affecting the potential of denitrifying bioreactors to improve stream health and ecosystem services. The performance of bioreactors and the structure and functioning of stream biotic communities are linked by environmental parameters like dissolved oxygen and nitrate-nitrogen concentrations, dissolved organic carbon availability, flow and temperature regimes, and fine sediment accumulations. However, evidence of bioreactors' ability to improve waterway health and ecosystem services is lacking. To improve the potential of bioreactors to enhance desirable stream ecosystem functioning, future assessments of field-scale bioreactors should evaluate the influences of bioreactor performance on ecological indicators such as primary production, organic matter processing, stream metabolism, and invertebrate and fish assemblage structure and function. These stream health impact assessments should be conducted at ecologically relevant spatial and temporal scales. Bioreactors have great potential to make significant contributions to improving water quality, stream health, and ecosystem services if they are tailored to site-specific conditions and implemented strategically with land-based and stream-based mitigation tools within watersheds. This will involve combining economic, logistical, and ecological information in their implementation.

Drought survival is a threshold function of habitat size and population density in a fish metapopulation.

Because smaller habitats dry more frequently and severely during droughts, habitat size is likely a key driver of survival in populations during climate change and associated increased extreme drought frequency. Here, we show that survival in populations during droughts is a threshold function of habitat size driven by an interaction with population density in metapopulations of the forest pool dwelling fish, Neochanna apoda. A mark-recapture study involving 830 N. apoda individuals during a one-in-seventy-year extreme drought revealed that survival during droughts was high for populations occupying pools deeper than 139 mm, but declined steeply in shallower pools. This threshold was caused by an interaction between increasing population density and drought magnitude associated with decreasing habitat size, which acted synergistically to increase physiological stress and mortality. This confirmed two long-held hypotheses, firstly concerning the interactive role of population density and physiological stress, herein driven by habitat size, and secondly, the occurrence of drought survival thresholds. Our results demonstrate how survival in populations during droughts will depend strongly on habitat size and highlight that minimum habitat size thresholds will likely be required to maximize survival as the frequency and intensity of droughts are projected to increase as a result of global climate change.

Increases in disturbance and reductions in habitat size interact to suppress predator body size.

Food webs are strongly size-structured so will be vulnerable to changes in environmental factors that affect large predators. However, mechanistic understanding of environmental controls of top predator size is poorly developed. We used streams to investigate how predator body size is altered by three fundamental climate change stressors: reductions in habitat size, increases in disturbance and warmer temperatures. Using new survey data from 74 streams, we showed that habitat size and disturbance were the most important stressors influencing predator body size. A synergistic interaction between that habitat size and disturbance due to flooding meant the sizes of predatory fishes peaked in large, benign habitats and their body size decreased as habitats became either smaller or harsher. These patterns were supported by experiments indicating that habitat-size reductions and increased flood disturbance decreased both the abundance and biomass of large predators. This research indicates that interacting climate change stressors can influence predator body size, resulting in smaller predators than would be predicted from examining an environmental factor in isolation. Thus, climate-induced changes to key interacting environmental factors are likely to have synergistic impacts on predator body size which, because of their influence on the strength of biological interactions, will have far-reaching effects on food-web responses to global environmental change.

Habitat loss drives threshold response of benthic invertebrate communities to deposited sediment in agricultural streams.

Agricultural land uses can impact stream ecosystems by reducing suitable habitat, altering flows, and increasing inputs of diffuse pollutants including fine inorganic sediment (< 2 mm). These changes have been linked to altered community composition and declines in biodiversity. Determining the mechanisms driving stream biotic responses, particularly threshold impacts, has, however, proved elusive. To investigate a sediment threshold response by benthic invertebrates, an intensive survey of 30 agricultural streams was conducted along gradients of deposited sediment and dissolved nutrients. Partial redundancy analysis showed that invertebrate community composition changed significantly along the gradient of deposited fine sediment, whereas the effect of dissolved nitrate was weak. Pollution-sensitive invertebrates (%EPT, Ephemeroptera, Plecoptera, Trichoptera) demonstrated a strong nonlinear response to sediment, and change-point analysis indicated marked declines beyond a threshold of -20% fine sediment covering the streambed. Structural equation modeling indicated that decreased habitat availability (i.e., coarse substrate and associated interstices) was the key driver affecting pollution-sensitive invertebrates, with degraded riparian condition controlling resources through direct (e.g., inputs) and indirect (e.g., flow-mediated) effects on deposited sediment. The identification of specific effects thresholds and the underlying mechanisms (e.g., loss of habitat) driving these changes will assist managers in setting sediment criteria and standards to better guide stream monitoring and rehabilitation.

Limitation and facilitation of one of the world's most invasive fish: an intercontinental comparison.

Purposeful species introductions offer opportunities to inform our understanding of both invasion success and conservation hurdles. We evaluated factors determining the energetic limitations of brown trout (Salmo trutta) in both their native and introduced ranges. Our focus was on brown trout because they are nearly globally distributed, considered one of the world's worst invaders, yet imperiled in much of their native habitat. We synthesized and compared data describing temperature regime, diet, growth, and maximum body size across multiple spatial and temporal scales, from country (both exotic and native habitats) and major geographic area (MGA) to rivers and years within MGA. Using these data as inputs, we next used bioenergetic efficiency (BioEff), a relative scalar representing a realized percentage of maximum possible consumption (0-100%) as our primary response variable and a multi-scale, nested, mixed statistical model (GLIMMIX) to evaluate variation among and within spatial scales and as a function of density and elevation. MGA and year (the residual) explained the greatest proportion of variance in BioEff. Temperature varied widely among MGA and was a strong driver of variation in BioEff. We observed surprisingly little variation in the diet of brown trout, except the overwhelming influence of the switch to piscivory observed only in exotic MGA. We observed only a weak signal of density-dependent effects on BioEff; however, BioEff remained < 50% at densities > 2.5 fish/m2. The trajectory of BioEff across the life span of the fish elucidated the substantial variation in performance among MGAs; the maximum body size attained by brown trout was consistently below 400 mm in native habitat but reached approximately 600 mm outside their native range, where brown trout grew rapidly, feeding in part on naive prey fishes. The integrative, physiological approach, in combination with the intercontinental and comparative nature of our study, allowed us to overcome challenges associated with context-dependent variation in determining invasion success. Overall our results indicate "growth plasticity across the life span" was important for facilitating invasion, and should be added to lists of factors characterizing successful invaders.

Top-down control of prey increases with drying disturbance in ponds: a consequence of non-consumptive interactions?

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 non-consumptive effects of predators in complex food webs will likely improve the understanding of biotic interactions across environmental gradients.