Discussion

Morphologically almost identical species are likely to compete for resources, and therefore offer good case studies to understand processes that drive species coexistence. We find evidence of reduced trophic niche overlap in recently separated cryptic bat species in sympatric locations relative to allopatric ones based on DNA metabarcoding and high throughput sequencing. The functional analysis suggests that the subtle trophic shift seen may be driven by differential foraging mode. Our results support niche theory predictions of the role of biotic interactions in driving species assemblages (Schoener, 1974). Trophic resource partitioning was only evident at the fine spatial scale, within areas of range overlap, suggesting that fine-scale mechanisms of coexistence could have implications for the maintenance of broad- scale diversity patterns (Godsoe, Murray, & Plank, 2015).

Trophic ecology of Myotis escalerai and Myotis crypticus

Our results reveal that the two bat species have a broad generalist diet, tend to consume prey relative to their availability and use gleaning to a high extent. We found high similarity in their trophic ecology in terms of both order and prey species composition. Both bat species’ diets are mostly composed of Lepidoptera, Diptera and Araneae, but also include several other prey orders. However, M. escaleraiconsumes a higher percentage of Hemiptera, while M. crypticu s Diptera. Functionally, the two bat species consume an equally high proportion of prey items that are not nocturnally volant, which suggests that both bats predominantly glean prey from vegetation.
The trophic ecology of these two recently described bat species is very similar to their cryptic sister-species M. nattereri, which also feeds mostly on Lepidoptera, Diptera and Araneae (Hope et al., 2014; Swift, 1997; Swift & Racey, 2002; Vaughan, 1997) and is known to catch a high proportion of its prey through gleaning (Arlettaz, 1996; Hope et al., 2014; Shiel, McAney, & Fairley, 1991; Swift, 1997; Swift & Racey, 2002). Similarly to our study, Shiel et al., (1991) estimated that 68% of M. nattereri ’s diet is made up of arthropod families that are not active at night. The row of hairs in the uropatagium border is a characteristic trait of members of the M. nattereri species complex that is thought to be functionally linked with gleaning (Czech, Klauer, Dehnhardt, & Siemers, 2009). Although the presence of more developed hairs in M. escalerai is one of the characters which separates these taxa (Juste et al., 2018), we found no difference in the extent of gleaning between the two bat species.

Trophic partitioning across spatial scales

Despite overall high trophic niche similarity between the two bat species at the prey order level, we detect a signature of trophic shift at the prey species (BIN) level, whereby diet overlap is lower in locally sympatric compared to locally allopatric locations at the fine-scale. This supports the contribution of trophic partitioning to species coexistence even when overall trophic niche overlap is high. A similar trend was seen at the functional level, whereby the proportion of prey items that are not nocturnally volant is borderline different between bat species only when locally sympatric. This suggests that the differentiation in diet composition seen at the prey species level when locally sympatric may be driven by a shift in foraging strategy (e.g. Krüger et al., 2014), through M. escalerai decreasing its extent of gleaning. However, our inference is limited by small sample sizes, which reduced the power of the analysis. At the arthropod order level, we find differences in the use of some arthropod orders among allopatric regions, likely due to differences in arthropod availability between the Mediterranean region, where only M. escalerai is found and the Atlantic region, where M. crypticus is present.
Several studies have identified trophic niche shifts from allopatry to sympatry, for instance between morphologically similar fish (Gkenas, Magalhães, Cucherousset, Orjuela, & Ribeiro, 2019; Schmitt & Coyer, 1983) and reptile species (Huey, Pianka, Egan, & Coons, 1974; Klawinski et al., 1994). However, in bats, previous coexistence studies looking at trophic ecology only focused on sympatric populations, and rarely found evidence of trophic resource partitioning. A few exceptions are the gleaning bats M. nattereri , Plecotus auritus andMyotis bechsteini in a sympatric population in central Europe (Andreas, Reiter, & Benda, 2012a), and evidence of low dietary overlap between sympatric P. auritus and Plecotus macrobullaris(Ashrafi, Beck, Rutishauser, Arlettaz, & Bontadina, 2011).
The observed trophic shift, albeit subtle, suggests that the two bat species are likely competing for food resources. It has been previously hypothesised that arthropods are abundant and do not constitute a limiting resource for bats (Arlettaz, 1999; Krüger et al., 2014). However, exclusion experiments in both tropical (Kalka, Smith, & Kalko, 2008) and temperate forests (Böhm, Wells, & Kalko, 2011) show that bats can control the abundance of arthropods, and therefore arthropods could be a limiting resource to competitors (Salinas‐Ramos et al., 2020).
Our study does not refute the possibility that other coexistence mechanisms, such as habitat or temporal partitioning (Schoener, 1974), occur among these two species, or the role of environmental variability in facilitating coexistence (Chesson & Warner, 1981). Spatial partitioning is frequently cited as a key mechanism of coexistence in other bat studies (e.g. Arlettaz, 1999; Emrich, Clare, Symondson, Koenig, & Fenton, 2014; Kunz, 1973; Russo et al., 2014). Although in many cases, contrary to our study, spatial partitioning may be driven by slight differences in bat morphology (e.g. Salsamendi et al., 2008, 2012), which would affect their performance in different habitats (Norberg, 1994). However, in our study the two species were caught in the same sampling sites, some of which were forests, where they are known to forage, suggesting they may share the same foraging sites.
A better understanding of the spatial scales of species coexistence is an important advance in our understanding of the maintenance of diversity (Hart et al., 2017). Our finding that trophic partitioning only occurs at the fine spatial scale is consistent with other bat (Peixoto et al., 2018), ant (Albrecht & Gotelli, 2001), parasitoid insects (Harvey, Snaas, Malcicka, Visser, & Bezemer, 2014) and bobcat (Lewis et al., 2015) studies, showing that interspecific interactions are more important for shaping community structure at fine rather than broad spatial scales. However, this pattern is not universal (e.g Harmáčková, Remešová, & Remeš, 2019). Fine-scale coexistence mechanisms could prevent competitive effects from scaling-up (Godsoe et al., 2015), which in our study system could contribute to enabling broad-scale range overlap across the north of the Iberian Peninsula.

Prey consumption relative to availability

The diet of a species is a function of both consumer selection and trophic resource availability within the foraging habitat (Lawlor, 1980). Therefore, considering resource availability allows for a better inference of species trophic preferences. Previous studies comparing bat prey consumption with prey availability pointed to selection of certain prey orders, such as Coleoptera by Eptesicus fuscus (Agosta, Morton, & Kuhn, 2003), chironomid flies by Myotis daubentoni(Vesterinen et al., 2016) and certain prey traits like moth size byBarbastella barbastella (Andreas, Reiter, & Benda, 2012b). Similarly, M. nattereri was found to over-select arachnids, Opiliones, Coleoptera, and several Diptera families, and under-select Hemiptera (Swift & Racey, 2002). In this study we do not detect clear trends of over-selection for specific prey orders matching the generalist broad trophic niche of the studied bats. However, diet selection results should be interpreted with caution due to the difficulty of obtaining a representative estimation of arthropod availability. Any arthropod sampling technique is biased towards certain types of arthropods (Cooper & Whitmore, 1990) and the habitats sampled and their respective sampling effort may not adequately represent where bats actually forage, especially given that they can use large areas and arthropod communities change depending on habitat type (Lamarre et al., 2016) and vertical stratification (Ulyshen, 2011). Our molecular diet analysis results confirm that the two studied bat species indeed glean prey from the vegetation, and therefore the arthropod community sampled using sweep nets likely represents at least part of the prey resources available to the bats.

Methodological considerations and study limitations

Primer bias towards certain taxonomic groups is a major issue in metabarcoding studies (Elbrecht et al., 2019). In this study, prey items were frequently recovered by only one of the primers, and differences existed in the recovery of the different arthropod orders. This supports previous studies that suggest that more than one set of primers should be used when the expected diet covers a broader taxonomical spectrum (Alberdi et al., 2018). The inclusion in this study of a set of samples with known composition based on morphological analysis (albeit only at the order level) gives us some idea of potential biases in the molecular identification. Opiliones, in particular, were morphologically identified in several sweep net samples and are known to be present in the diet of M. nattereri (Galan et al., 2018; Swift, 1997; Swift & Racey, 2002), but were absent from the molecularly-characterised diets of the two bats. Thus, their absence in this study is likely the result of primer amplification bias.
Parameter choice during bioinformatic analysis can modify the diet composition recovered (Alberdi et al., 2018). The strong match between the inferences drawn using the additive and the conservative approach of dealing with PCR replicates (Fig. S9), show that the results are robust to that choice, and mirror other studies showing that parameters choice does not change ecological conclusions (Clare, Chain, Littlefair, Cristescu, & Deiner, 2016). In the same manner, high similarity in diet composition based on weighted Percent of Occurrence (wPOO) and Relative Read abundance (RRA), indicates that the results are also robust to the measure used.
Because prey development stage cannot be identified using the metabarcoding approach, some of the prey species (BINs) classified as nocturnally volant may correspond to non-flying larval stages. This could be important in Lepidoptera, and could increase the inferred importance of the gleaned behaviour of both species because larval stages are known to be consumed by M. nattereri (Hope et al., 2014). More generally, nocturnal aerial activity is not directly quantifiable, and therefore our classification is subject to a certain degree of subjectivity. However, because arthropod orders most difficult to categorise due their diversity, like Coleoptera, are consumed in similarly low proportion by both bat species, potential classification biases are expected to be low and standardised across species. Nevertheless, due to potential classification biases and low sample sizes in sympatric locations, interpretations of functional prey shift should be considered with caution.

Conclusions

In line with niche theory predictions, we show that coexistence among morphologically identical (cryptic) species can be facilitated through fine-scale mechanisms of resource partitioning, despite high levels of trophic similarity at the broad-scale, even in sympatric regions. Hence, this study highlights the importance of using appropriated spatial scales when studying impacts of biotic interactions on community assembly (Viana & Chase, 2019). Our findings that trophic resource partitioning is only evident at the fine spatial scale, within areas of range overlap, suggest that fine-scale mechanisms of coexistence could have implications for the maintenance of broad- scale diversity patterns. This is the first study to identify a trophic shift between allopatric and sympatric populations of insectivorous bats, supporting the role of trophic resource partitioning in enabling species co-occurrence in the same foraging site. It thereby addresses some of the key limitations identified in a recent review of interspecific competition in bats (Salinas‐Ramos et al., 2020). We highlight the importance of using high taxonomic resolution and allopatric populations at meaningful spatial scales for identifying patterns of niche shift, and the utility of using a functional approach that better links mechanistically with species trophic ecology. Understanding mechanisms of coexistence is essential for predicting species vulnerability under climate change because range shifts will result in new community assemblages and competitive interactions (HilleRisLambers, Harsch, Ettinger, Ford, & Theobald, 2013). This is particularly relevant in our study system as both species are restricted to the Mediterranean region, where climate change is predicted to be particularly severe (Sala et al., 2000), and both are predicted to experience range shifts and changes in range overlap under climate change (Razgour et al., 2019).