Landscape models can predict the distribution of aquatic insects across agricultural areas

Context Adult aquatic insects are major vectors of aquatic subsidies to terrestrial ecosystems, providing substantial ecosystem services to agriculture. Distance to water is reported as the prime factor to model their spatial distribution, but the role of local drivers, either terrestrial or aquatic habitat features, is not well established, notably in highly heterogeneous landscapes. Objectives We assess the distribution and prioritise explanatory factors of aquatic insects in agricultural landscapes. Methods Adult aquatic insects (Ephemeroptera, Plecoptera, Trichoptera) were caught with sticky traps on a regular grid-pattern set across a heterogeneous agricultural landscape during two years. Using innovative landscape variables related to water, woodland, elevation and insolation (derived from drone data), classical random forest models were run to predict the distribution of insects. Results Variables that significantly explained insect distribution differed between insect orders, but the distance to water was consistently amongst the three most informative variables in models. Ephemeroptera occurrence was influenced by woodland density, while Plecoptera and Trichoptera were more influenced by potential insolation. Predictions revealed a low distribution pattern for Ephemeroptera and a high probability of occurrence for Plecoptera across terrestrial ecosystems. Predictions were poor for Trichoptera, perhaps reflecting various dispersal patterns among species that emerged earlier or later in spring and from a wide range of aquatic habitats. Conclusions This work is the first to predict the probability of occurrence of aquatic insects across a spatially heterogeneous agricultural landscape. Prediction maps suggest that the aquatic footprint, defined as the occurrence of aquatic insects across a terrestrial landscape, is higher than previously thought.