Coprophagous beetles are essential for fecal matter removal and are thus considered key ecosystem services providers. Yet, our knowledge of these beetles’ distribution and ecology remains very limited. Here, we used Species Distribution Models (SDM) to investigate the species-environment relationships (i.e. their niche) and predict the geographic distribution of coprophagous beetles in the Western Swiss Alps. We used our own sampled data and existing national data from the Swiss faunal database to calibrate, for each species, a regional and a national SDM respectively. In both models, the best predictors were temperature and rock cover proportion, while a soil characteristic (∂¹³C) indicating its organic content and texture was important in the regional models and precipitations in the Swiss models. The model performed better for species specialized on low or high altitudes than for generalist species occurring in a large altitudinal range. The model performances were neither influenced by the size, nor by the nesting behavior (laying eggs inside or below the excrements) of the species. We also showed that species richness decreased with altitude. This study opens new perspective for a better knowledge of coprophagous beetle’s ecology and a useful tool for their conservation in mountain regions.

Dung beetles, Species distribution modeling (SDM), Ensemble of Small Models (ESMs), Hydrophilidae, Geotrupidae, Scarabaeidae

Coprophagous beetles are part of a specialized entomofauna feeding on the droppings of mammals (Hanski 2016). Some taxa have coprophagous adults and predaceous larvae, which are chasing fly larvae from dung patches (Hydrophilidae, Sphaeridinae), while other have coprophagous adults and larvae. In the latter case, some species lay their eggs directly in the dung (non-nesters: Scarabaeidae, Aphodiinae) and other dig simple wells or sophisticated network of tunnels and rooms where they stock dung and lay their eggs (paracoprids: Geotrupidae and Scarabaeidae, Scarabeinae) as a strategy to avoid the harsh intra- and inter-specific competition to exploit dung patches before they dry (Hanski 2016). By feeding on excrements and burying it, coprophagous beetles are essential for dung decomposition (Gittings et al. 1994). They avoid the accumulation of excrements, preventing pasture surface loss (Beynon et al. 2012b) and supplementary expenses for dung removal (Fincher 1981; Losey and Vaughan 2006; Beynon et al. 2015) and are therefore considered as key “Ecosystem Service Providers” (Nichols et al. 2008). In addition, coprophagous beetles represent a part of the food for some insectivorous animals such as birds (in particular corvids) (Lumaret and Stiernet 1990) or mammals (e.g. greater horseshoe bat (Rhinolophus ferrumequinum)) (Beynon et al. 2015). The economic and ecological importance of coprophagous beetles coupled with the possibility to characterize the whole species assemblages found at a given location (dung patch) in a given time point (Finn and Giller 2000; Hanski 2016) make them an adequate group to study biogeography (Lumaret 1979) and animal communities (Hanski and Koskela 1977). In Europe, the species assemblages of coprophagous beetles and their relative abundance have already been investigated (Lumaret and Stiernet 1984; Lumaret and Stiernet 1989; Errouissi et al. 2004; Negro et al. 2011) and the importance of climatic and edaphic factors have been shown at a coarse level (Hortal et al. 2001; Lobo and Martin-Piera 2002; Lumaret and Jay-Robert 2002). However, ecological needs and fine geographic distribution of single coprophagous beetle species remains an understudied topic.

The study of the realized environmental niche of species, adaptation to local conditions and interspecific interactions (Hutchinson 1957) allows a better understanding of the distribution of species (see Niche-Geography duality: Colwell and Rangel 2009), which is crucial to overcome Wallacean (knowledge about the geographical distribution of species) and Hutchinsonian (knowledge about the tolerance of species to abiotic factors) shortfalls concerning biodiversity (Hortal et al. 2015). The development of statistical species distribution models (SDM; also called ‘habitat suitability’ or ‘ecological niche’ models; see Franklin 2010; Peterson et al. 2011; Guisan et al. 2017) to quantify the niche and derive geographic predictions have brought powerful perspectives to better understand, compare and quantify the relationship between organism and their environment (i.e. their environmental niche), but also to predict their distribution in space and time (Guisan and Zimmermann 2000; Guisan and Thuiller 2005). These tools can therefore bring essential knowledge about the ecology of understudied taxonomical groups like most arthropods (Hochkirch et al. 2021). SDMs have been used to study various groups of insects (Pellissier et al. 2012; Pradervand et al. 2014; D’Amen et al. 2015; Descombes et al. 2016; Mata et al. 2017) but there are only few examples of single modeled coprophagous beetles (e.g. Chefaoui et al. 2005; Lobo et al. 2010).

The aim of this study was to bring a better understanding of the factors influencing the distribution of coprophagous beetle species in temperate mountain environments using a SDM approach. In order to obtain a sufficient number of accurate species data to quantify species-environment relationships, we sampled coprophagous beetles throughout the Western Swiss Alps in a random stratified manner. We additionally obtained all the occurrences available in Switzerland for the beetles of interest (Hydrophilidae, Geotrupidae and Scarabaeidae) from the Swiss national database (www.cscf.ch). This allowed us to compare fine-scale models calibrated in the study area using our precisely sampled data (regional model) and large-scale models calibrated at the Swiss level using national occurrences and our data (Swiss model). We expected the latter to reduce the risk, while calibrating the SDMs, of truncating the species’ environmental niche, which can happen when the complete extent of the species’ geographic distributions and environmental requirements are not covered in an analysis (Pearson et al. 2004; Thuiller et al. 2004; Hannemann et al. 2016; Guisan et al. 2017; El-Gabbas and Dormann 2018; G. Mateo et al. 2019; Chevalier et al. 2021). Here, we particularly focused on the climatic, land-use and edaphic factors as environmental predictors of the species’ presence. In addition, we investigated the effects of species characteristics such as the altitudinal amplitude where they occur, their nesting behavior and their body size on the SDMs performances. Finally, we assessed whether the stacking of all species predictions produced meaningful richness maps of coprophagous beetles in the study area.

We investigated the influence of various factors on the distributions of single coprophagous beetle species in the Western Swiss Alps using correlative species distribution modeling (SDM) approaches based on quantifying habitat suitability (Guisan et al. 2017). Given the high number of species with small number of occurrences, we used a particular approach recently developed for small sample sizes: ensemble of small models (ESMs; Lomba et al. 2010; Breiner et al. 2015, 2018). In all our models, the predictors with the greatest importance included climatic variables (Fig. 3), like in many SDM studies (Austin and Van Niel 2011; Pradervand et al. 2013; Mod et al. 2016; Scherrer et al. 2019). Interestingly, in both models (i.e. Swiss, Fig. 3A and Regional, Fig. 3B) the proportion of rock and bare soil cover (Rock) was an important predictor. In the study region as in the rest of the Alps, the altitude is correlated with the proportion of rocky surfaces, which could act as a confounding factor. As specialized species are better modeled than the widespread species covering a large elevation range (Guisan and Hofer 2003), the importance of the rocky surface variable in the models might be artificially high. It is also possible that more species are present in pastures of high ecological value (i.e. with a high overall biodiversity) including grasslands with discontinuous vegetation cover (Delarze et al. 2015). Furthermore, the superficial rock cover proportion could be an indirect way of quantifying the heterogeneity of the landscape, such as the proportion of forest edges, which was also an important predictor (Fig. 3) in our model. Indeed, Negro et al. (2011) suggested that habitat heterogeneity, especially the presence of natural forested areas next to pastures, plays an important role in increasing coprophagous beetle species richness. In contrast, some variables had little influence in our models. These were often land cover or land use variables spatially restricted in Switzerland (e.g. wet habitats) or in the study area (e.g. cultivations) but it is difficult to know if it is the low frequency of these variables over the landscape that induces their smaller influence in the models or if they really do not have an influence on species distributions.

Our ESMs had very variable predictive performances as measured by the maximized TSS (see Jimenez-Valverde 2014; Guisan et al. 2017), with values ranging from 0.27 to 0.93 for the Swiss models (Fig. 2A) and from 0.40 to 0.85 for the regional models (Fig. 2B). Note that the use of threshold independent discrimination metrics, such as the maxTSS (or the classical AUC) for the evaluation of presence/absence models (i.e. our Regional models) may be problematic because of a non-linear asymptotic relationship between discrimination metrics and true model accuracy, and that it might be accordingly difficult to distinguish between models with high AUC value (Jimenez-Valverde 2014). Nonetheless, maxTSS from models calibrated in the same area can still correctly inform on the ranking of accuracy between models (e.g. between poor, useful or good models), except among very high TSS values (i.e. calling for caution in the ranking among good models). We found that ubiquitous species present over a wide altitudinal range had weaker models compared with specialized species occurring in narrower altitudinal amplitude (Fig. 4A). Our results are in line with those of Guisan and Hofer (2003) and Grenouillet et al. (2011), who showed that the distributions of generalist reptile and fish species, respectively, are more difficult to predict, and with those of Tessarolo et al. (2021), who found that niche marginality has a major influence on the models’ quality for dung beetles in Spain. On the other hand, we found no influence of the nesting behavior (Fig. 4B) nor the size of the species (Fig. 4C) on the maxTSS of the models, meaning that these biological traits seem not relevant to explain models’ quality.

When looking at the expected species richness of the coprophagous beetle communities based on the stacking of single species environmental suitability values over the study region, the global trend shows a diminution of the number of species with increasing altitude (Fig. 5). This result was also observed in other taxa in the same region (Dubuis et al. 2011 for plants; Pradervand et al. 2013 for orthopterans; Reymond et al. 2013 for ants; Pellissier et al. 2013 for butterflies; Scherrer et al. 2019 for bats; Seppey et al. 2020 for protists), for which the climatic predictors, especially temperature, were also of great importance. However, it is important to notice that for coprophagous beetles, the decrease in species richness depends on the nesting behavior: the paracoprids (Fig. 5B) and the Hydrophilidae (Fig. 5D) show a steep decrease with increasing elevations, while the non-nesters (Fig. 5C) show a gentler one. This latter group forms the biggest part of the coprophagous beetle diversity at high altitude where almost no paracoprids and Hydrophilidae are found (Fig. 5B, C). This result, consistent with those of Lobo et al. (2007) is explained by the ecology of non-nester beetles, which are more tolerant to cold and are outcompeted by paracoprids in thermophilous places (Hanski 2016).

Many of the studies focusing on the coprophagous fauna use dung-baited trap to get an exhaustive species list in addition to data on the phenology and abundance (see for example Lumaret 1978). No large-scale trapping campaign was performed in Switzerland where the ecology of coprophagous beetles is still poorly studied. As a first step in a better knowledge of these taxa and since we were not interested in phenological or abundance data, we made the choice to perform an active sampling, less constraining logistically (transport of traps and dung), more efficient to visit a large number of sites (what we needed to build models) and allowing to select the individuals to collect sparing identification time. Moreover, the active search has been shown to be a very efficient way to get species inventories for beetles traditionally caught with traps (Chittaro and Marggi 2016). Our active field sampling designed in a random stratified manner permitted to be representative of the various environment of the study area and likely allowed us to find most of the coprophagous beetle species known from the study area (Agolius abdominalis and Neagolius montanus only were missing). Nevertheless, it is possible that we missed the occurrence of some species with low detectability in some plots. Future studies aim to correct for this bias, for example by implementing methods to estimate the completeness of plot inventory, or combining models of α- and β-diversity to predict the spatial community composition in a region, and areas of incompleteness within it (Mokany et al. 2011).

From a faunistical point of view, our study brings valuable new records for beetles, an under-sampled taxon in comparison to other insect groups such as orthopterans, butterflies, and even more vertebrates (Troudet et al. 2017), with the perspective to improve predictions of global change impact on biodiversity in mountain areas (Guisan et al. 2019) and better support conservation decisions (Guisan et al. 2013). Indeed, the data sampled in our study represents now 17.9% (N=1120, Swiss database info fauna-CSCF) of all precise occurrences existing for these 47 coprophagous beetle species in Switzerland (N=6258). An important part (42%) of all the coprophagous beetle species of Switzerland is found in the Vaud Alps (info fauna-CSCF), reinforcing the status of biodiversity hotspot of this study region in the European Alps (Lassen and Savoia 2005). Future studies should investigate more of such under-sampled taxa, like other invertebrate groups, to allow more robust comparative studies and produce better global biodiversity assessments within a same study area (Mod et al. 2020).

We particularly thank: Yannick Chittaro and Andreas Sanchez for their advises in entomology, for providing us the Swiss occurrences dataset and for their comments on the manuscript; Tatiana Zingre for her help during the field sampling; Paul Béziers, Robin Séchaud and the reviewers for their constructive remarks on the manuscript, which permitted to improve it substantially. Finally, we thank the researchers/curators of the Natural History Museum of Geneva, Giulio Cuccodoro, Bernard Landry and Emmanuel Toussaint for providing the photographic material. AG received support from the Federal Office of the Environment through the ValPar.CH project.