INTRODUCTION
Climate warming, extreme precipitation events and temperature variability are all intensifying at unprecedented rates (IPCC Climate Change, 2014; Donat et al. , 2016; Bathiany et al ., 2018). Such changes cause species–environment mismatches, and have already led to shifts in the geographic ranges and elevational or vertical distributions of species as they track their climatic niche (Vázquezet al ., 2017; Essl et al ., 2019; Trisos et al ., 2020). As other means to avoid adverse conditions, terrestrial species may seek microclimatic refuges (Scheffers et al ., 2014; Suggittet al ., 2018) and even manipulate their own microclimate (Pincebourde and Casas, 2019). Microclimatic refuges provided by a variety of habitat structural components – including plant architectural complexity, tree bark, downed woody debris, leaf litter, and leaf shelters – reduce the exposure of their inhabitants to climate variability and extremes (Scheffers et al ., 2014; Suggittet al ., 2018; Pincebourde and Casas, 2019; Pinsky et al ., 2019). Such refuges can be particularly important to ectotherms (Pinskyet al ., 2019), such as arthropods. Ectotherms depend on external energy to thermoregulate and are amongst the taxa most threatened by global change (García-Robledo et al ., 2016; Warren et al.,2018; Wagner, 2020; Wagner et al ., 2021). The availability of these refuges, and their impact on insect use, may be one factor mitigating the various negative climatic factors leading to insect decline.
Protection against harsh environmental conditions is typically provided by ecosystem engineers, which are organisms that modify their habitats by creating physical structures that can be used as structural refuges by other organisms (Jones et al. Hastings et al ., 2007; Romeroet al ., 2015). In terrestrial ecosystems, bark beetles and stem borers make stem and trunk galleries, gallers manipulate plant physiology to produce galls, and various arthropods (e.g., caterpillars, aphids, mites and thrips) build leaf shelters, such as leaf rolls and ties, thus providing refuge to many other plant-dwelling organisms (Wanget al., 2012; Vieira and Romero, 2013; Cornelissen et al ., 2015; Priest et al ., 2021). The role of leaf-rolling engineers as biodiversity amplifiers can be even stronger in dry seasons, and extend to the whole plant, thus influencing arthropod assemblages at temporal and microspatial scales (Vieira and Romero, 2013). After such shelters are abandoned by their creators, they become available for other arthropods for a long period of time (Vieira and Romero, 2013). Similar to many other biotic interactions, which are stronger at lower latitudes (Schemske et al., 2009), the facilitative effects of vertebrate and invertebrate ecosystem engineers seem particularly pronounced in the wet tropics, but also in arid regions (Romero et al ., 2015). Therefore, these latitudinal, regional (Romero et al ., 2015) and temporal patterns (Vieira and Romero, 2013) suggest that climatic conditions may be a common driver underlying such facilitative interactions.
Large-scale patterns in climatic signatures have been recently investigated using standardized replicated experiments, which can differentiate the direct effects of climate from the indirect effects of latitude or elevation on biotic interactions (LaManna et al ., 2017; Roslin et al ., 2017; Romero et al ., 2018). Most of these macroecological studies have focused on antagonistic interactions, such as predation and competition (LaManna et al ., 2017; Roslinet al ., 2017; Romero et al ., 2018). Although the importance of facilitative interactions among organisms is expected to increase under stressful conditions (He et al ., 2013; Romeroet al ., 2015), no empirical studies to date have examined the impacts of current climate and climate variability, or predicted the impacts of future climates, on facilitative interactions at global scale.
Here, we test the effects of current and future climatic scenarios on facilitation mechanisms provided by leaf-rolling ecosystem engineers. We compare how current and projected climate affects the usage of leaf rolls by terrestrial arthropods, and how shelter usage is affected by body size and trophic position. We hypothesize that abundance, species richness and biomass of arthropods within leaf rolls will increase with increasing climate variability and drought. We also expect that species’ responses to climate will vary with their trophic position and body size, both of which are known to influence the metabolic requirements of ectotherms (Daufresne et al., 2009) and their capacity to dissipate heat (Rubalcaba et al., 2019). Competition at higher trophic levels (predators) selects for larger body size. Larger-bodied organisms typically require more favourable climatic conditions than their prey (Petchey et al., 1999; Voigt et al., 2003; Brose et al., 2012). Therefore, we expect larger predators to be dominant in refuges especially in arid and climatically variable regions (Fig. 1). Finally, with continuing climate change, we expect current climatic dependencies to reflect future changes in geographic patterns of refuge use by arthropods.
To test the above predictions, we conducted a global, coordinated experiment at 52 sites across an 11,790 km latitudinal gradient (from 45.8° S to 60.2° N, Fig. 1) and elevation spanning from 5 to 2900 m above sea level. To measure the beneficial effects of leaf rolls, we recorded abundance and diversity of arthropods colonizing manually rolled and control (unrolled) leaves, and calculated the standardized mean difference (Hedges’ d) between arthropod abundance, species richness, biomass and body size in rolled and control leaves for each site. To test how climate influences shelter colonization, we investigated the effects of different categories of moderators (predictor variables) on the Hedges’ d effect size, including both geographic distance (absolute latitude, elevation) and direct measures of climate (temperature and precipitation). Finally, we provide geographical interpretations of climate change scenarios on the patterns of refuge use and predict future changes in refuge use by projecting effect sizes to future (2070) climatic conditions at the study locations.