Temperature-induced trophic cascades in stream food webs

The study examines how warming and the presence of apex fish predators jointly influence community biomass, decomposition rates, and food web structure in natural stream ecosystems. Building on prior observations that warming can alter body size distributions, community biomass, and ecosystem functioning, the authors test three hypotheses: H1, that higher temperatures directly increase community biomass and decomposition rates but simplify food webs; H2, that the presence of apex predators (brown trout) induces trophic cascades (reducing invertebrate biomass and increasing algal biomass), suppresses decomposition via reduced detritivores, and simplifies food webs; and H3, that warming accentuates apex predator effects, strengthening cascades, further suppressing decomposition, and simplifying food webs. The study aims to disentangle direct physiological effects of temperature from indirect multi-trophic interactions using a field manipulation across geothermally warmed and cold streams.

Prior laboratory and mesocosm studies show warming can alter food-web structure and ecosystem metabolism, often strengthening top-down control in colder regions. In the Hengill geothermal stream system, previous work has documented thermal dependencies of biomass, structure, and functioning, with evidence that warming can reduce predator-free space and potentially simplify communities through increased top-down control. Meta-analyses suggest geographical context matters, with warming strengthening top-down control mainly in cold regions. However, manipulative field experiments in complex natural systems remain scarce, and integrating multi-trophic interactions is critical to predict warming impacts beyond direct physiological responses.

Field experiment in six geothermally heated streams (Hengill Valley, Iceland), using a split-plot design with two temperature categories (cold ~6.8 ± 1.4 °C; warm ~13.5 ± 2.4 °C) crossed with fish manipulation (presence/absence of brown trout, Salmo trutta). Three streams per temperature category; within each stream, fenced reaches established paired 'Fish' and 'No fish' treatments (mesh allowed invertebrate drift). Brown trout were removed from all reaches, then 42 fish (mean fork length 17.5 ± 3.3 cm) were added evenly to 'Fish' reaches (~0.3 m−2); densities were checked and replenished mid-experiment; experiment duration 5 weeks. Sampling: invertebrates and benthic algae collected immediately before fish addition and just before fish removal; invertebrates via five Surber samples per reach; algae via paired rock scrapes for diatom ID and chlorophyll. Decomposition quantified using coarse (5 mm) and fine (250 µm) mesh litter bags with dried Carex spp.; microbial decomposition from fine bags; invertebrate-mediated decomposition from coarse–fine differences. Biomass estimation: invertebrates and diatoms identified to species; abundances scaled to area/volume; body sizes measured from images; biomass calculated via length–weight and volume–carbon relationships. Food webs: five localized webs per reach constructed from >49,000 gut content observations supplemented by literature; included only links between co-occurring species in paired samples; metrics computed included species richness, link richness, connectance (L/S^2), linkage density, mean trophic level, and consumer:resource richness ratio. Statistical analysis: for each response, end-minus-start changes were analyzed using linear mixed-effects models with temperature and fish as fixed effects and fish treatment within stream as a random effect; variance structures compared by AIC. Bioenergetic model: two-trophic-group model (diatoms and invertebrates) with fish consuming invertebrates; parameters for growth, carrying capacity, and consumption scaled with temperature and body mass (Arrhenius and allometric relations) using independent experiments; equilibrium analyses and Monte Carlo simulations (n=1000) generated mean and 95% CIs; sensitivity analysis assessed parameter contributions to cascade magnitude and thermal sensitivity.

- No main effect of temperature on invertebrate or algal biomass (contrary to H1a,b). Invertebrate biomass was significantly reduced by fish (supports H2a), particularly in warm streams (supports H3a): Δ invertebrate biomass, Fish main effect F=4.640, P=0.040; Temperature×Fish interaction F=12.06, P=0.002. - Algal responses showed trophic cascade in warm streams with fish: Δ diatom biomass Temperature×Fish F=6.971, P=0.016; Δ chlorophyll Temperature×Fish F=7.045, P=0.013; Fish main effect significant for chlorophyll (F=5.088, P=0.032) and marginal for diatoms (F=3.139, P=0.092). These increases were strongest in Warm-Fish treatments (supports H2b, H3b). - Decomposition: Invertebrate-mediated decomposition showed no significant main or interaction effects (contrary to H1–H3c). Microbial decomposition was suppressed in Warm-Fish (supports H3c): Temperature×Fish F=6.760, P=0.019; no significant main effects (Temperature F=2.112, P=0.220; Fish F=2.189, P=0.158). SEM suggested positive influence of invertebrates on microbial decomposition, consistent with reduced nutrients/excretion in Warm-Fish. - Food web structure: No main effects of temperature or fish alone on connectance, mean trophic level, or consumer:resource ratio (contrary to H1–H2d), but significant simplification in Warm-Fish (supports H3d): Δ connectance Temperature×Fish F=4.653, P=0.043; Δ mean trophic level Temperature×Fish F=5.603, P=0.028; Δ consumer:resource ratio Temperature×Fish F=4.471, P=0.047. Simplification was driven by a disproportionate loss of omnivorous invertebrate consumers in Warm-Fish, reducing consumer:resource ratios and vertical diversity. - Bioenergetic model reproduced observed trophic cascade with warming in the presence of fish. Sensitivity analysis indicated that parameters linked to consumption rates of invertebrates (I_na) and fish (I_b) primarily determined cascade magnitude, while invertebrate consumption and body size parameters (I_na, E_n) most influenced thermal sensitivity, implicating strengthened biotic interactions rather than direct metabolic rate changes as key drivers.

Findings indicate that direct warming effects over five weeks did not alter biomass, decomposition, or food-web structure on their own, but warming strengthened apex predator impacts, yielding a temperature-induced trophic cascade: fish suppressed invertebrate biomass more in warm streams, releasing diatoms and elevating chlorophyll. Microbial decomposition decreased in Warm-Fish despite exclusion of larger consumers from fine bags, potentially due to reduced nutrient recycling via invertebrate excretion; SEM supported a positive link between invertebrates and microbial decomposition. Food webs became less connected with lower mean trophic level and reduced consumer:resource ratios in Warm-Fish, indicating reduced robustness and vertical diversity, consistent with loss of larger, omnivorous invertebrates. Model–data integration supports that strengthened consumer-resource interactions under warming, especially consumption rates, mediate these outcomes more strongly than direct physiological temperature effects. Results underscore that multi-trophic interactions are critical to predicting warming responses, particularly in cold, high-latitude systems where apex predators (e.g., brown trout) perform well within the studied temperature range.

The study demonstrates that warming can amplify top-down control by apex predators, producing trophic cascades and simplifying food-web structure in natural stream ecosystems, even when direct warming effects on biomass and processes are undetectable over short timescales. By combining a replicated field manipulation with a bioenergetic model, the authors show that interaction strengths, especially consumption rates, dominate warming impacts on communities. These results caution against extrapolating from reductionist or single-trophic-level studies and highlight the need to incorporate network structure into conservation and biomonitoring. Future research should: (1) conduct multi-trophic warming experiments across biogeographic contexts, especially at lower latitudes and near predator thermal limits; (2) experimentally manipulate temperature (beyond space-for-time) to capture transient dynamics and adaptation; (3) test mechanistic links between nutrient recycling, microbial activity, and algal–microbial interactions; and (4) evaluate how predator diversity modulates warming-induced cascades.

- Space-for-time substitution: streams have been geothermally heated for decades, potentially allowing thermal adaptation and dampening direct temperature effects over the 5-week experiment. - Short duration and scale constraints typical of field experiments may miss longer-term or seasonal dynamics. - Fencing allowed natural invertebrate drift, introducing background noise across treatments; fish were replenished mid-experiment, and final counts varied by reach. - Diets of invertebrates were not compared between fish treatments, limiting inference on behavioural diet shifts due to predation risk. - Nitrogen limitation and absence of riparian leaf litter may reduce detritivore biomass, affecting detectability of invertebrate-mediated decomposition responses. - Publication focuses on cold, high-latitude streams with a single apex predator; generalizability to warmer regions or systems with diverse predators may be limited.