Introduction

In the context of climate change and insect decline (see Harvey et al. 2022), the alpine fauna may be especially vulnerable compared to those of other environments, having a higher proportion of less resilient specialist species, facing higher biogeographic isolation, and witnessing a faster pace of modifications (Hochkirch 2016; Wagner et al. 2021; Lencioni and Gobbi 2021). The ability of insect species to modify their distribution range to track the geographic shifts of tolerable climates is crucial to determine their chances to survive climatic changes: it depends on their natural history traits and on their ability to adapt their key traits quickly enough to the pace of the environmental modifications (Kellermann and van Heerwaarden 2019; Poloni et al. 2022). This can be especially difficult for alpine species that already live close to their physiological limits (Dahlhoff et al. 2019; Yadav et al. 2021). Upward elevational shifting is often the only possible response of alpine species facing increasing temperatures, yet it may cost a considerable reduction and fragmentation of the distribution range, and consequently of the population size; at the same time, species that already live on mountain peaks do not have this vital opportunity (Hodkinson 2005; Wilson et al. 2007; Shah et al. 2020; Halsch et al. 2021; Vitasse et al. 2021). Under current predictions, one third to more than a half of alpine and subalpine species may lose over 80% of suitable habitat by 2070 to 2100 (Engler et al. 2011), in addition, local extinctions of some taxa have been already documented (Panza and Gobbi 2022). However, baseline data on distribution and diversity in the European Alps, essential to monitor these trends, are still missing for several arthropod groups (Sommaggio et al. 2022).

Ants (Hymenoptera: Formicidae) are widespread and diverse across most terrestrial habitats of the globe often representing a remarkably high biomass, have a highly impactful ecological role through a multitude of relationships with animals, plants, fungi, and microbes, and can be used as ecological indicators (Hölldobler and Wilson 1990; Lach et al. 2010; Parker and Kronauer 2021; Schultheiss et al. 2022). Excluding the Mediterranean regions, the European fauna is very well resolved taxonomically, which represents an ideal context for ecological studies (Seifert 2018). Still, much of the existing ecological data have been produced under an outdated taxonomic framework (Seifert 2017). The European Alps are a key area for conservation biology due to multiple biogeographic aspects and its biota is threatened by rapid climatic alteration (Theurillat and Guisan 2001; Schwager and Berg 2019; Menchetti et al. 2021). Besides historic records, recent assessments on the European Alps ant fauna focus only on specific groups or deal with a narrow range of the elevational gradient (Glaser et al. 2012; Spotti et al. 2015; Krapf et al. 2018; Wagner 2020; Guariento and Fiedler 2021). Moreover, while there are literature data on the elevational range of most species, these are often based on decades-old studies (Seifert 2018). More generally, the role of climate change on the Alpine ant fauna has never been examined and no conservation priorities have been set.

As part of a wider long-term monitoring effort on the arthropod fauna of the Italian National Parks funded by the Italian Ministry for the Environment, we carried out a systematic investigation of ant diversity and distribution along elevational transects, aiming (i) to provide an updated reference on species elevational limits, (ii) to define the main diversity patterns and relationships with abiotic variables, as well as (iii) laying the groundwork for future conservation assessments by examining possible vulnerabilities to the ongoing climate change.

Materials and methods

Sampling was conducted in two Italian National Parks located in the Southern Alps, the Dolomiti Bellunesi National Park (DBNP) and the Stelvio National Park (SNP) during the years 2018–2019 (Fig. 1). Elevational transects consisted of a variable number of sampling stations at different elevations, each represented by a group of five pitfall traps. The traps were built with plastic glasses of c. 6 cm in diameter and c. 7 cm in height, buried in the ground, filled with c. 150 ml cc of a preserving mixture of white vinegar, sodium chloride, and a drop of detergent as a surfactant, as described by Gobbi (2020). Traps were replaced every 16–18 days from May/June to September/October, resulting in 5 to 10 sampling sessions per year (a shorter sampling season and number of sessions per year correspond to higher elevations where the snow coverage lasts longer). Sampling covered elevations from 780 to 2130 m asl in the DBNP (2 elevational transects, 11 sampling stations, 55 traps) and from 1400 to 2690 m asl in the SNP (6 elevational transects, 30 sampling stations, 150 traps), for a total of 205 traps (Fig. 1). In both Parks, traps of the same sampling station were normally placed in the same habitat type, while different sampling stations covered the widest variety habitat types and were placed at an elevational difference of about 150–200 m from the next ones within the same transect. For each trap, we recorded the habitat type as well as the temperature throughout the sampling season using data loggers (Thermochrom i-Button), placed at the center of each sampling station. Habitats were grouped in three macro-categories for the analyses: woodlands (Fagus sylvatica, Larix decidua, Pinus sylvestris, and Picea abies forests; 52 traps from 17 sampling stations), grassland/shrublands (grasslands, shrublands, wetlands; 145 traps from 34 sampling stations), and scree slopes (10 traps from 3 sampling stations). The exposure duration of each trap, accounting for the lost session due to malfunctioning or lost samples, is described in the Supplementary table S1.

Fig. 1
figure 1

Sampling effort in the Dolomiti Bellunesi National Park (DBNP) and Stelvio National Park (SNP), in the Italian Alps. Each point in the map indicates a sampling station (11 sampling stations in the DBNP, 30 in the SNP), while the number of traps used at each elevational range is illustrated in the barplot. Map image from Stadia Maps and Stamen Design, modified

Full size image

Ants were identified under a Zeiss Stemi 508 stereoscopic microscope, taking morphometric measurements with the aid of an Axiocam Erc 5s camera and the Zeiss Zen Core Software. Identification was based on the taxonomic keys provided by Radchenko and Elmes (2010), Wagner et al. (2017), and Seifert (2018). For what concerning the Tetramorium caespitum cryptic complex, a safe morphological identification should be based on quantitative morphological data from at least three workers belonging to the same colony, which cannot be ensured with pitfall trapping (Wagner et al. 2017). Therefore, we made an approximation suited to the purposes of this study, applying the key by Wagner et al. (2017) to up to 3 workers per trap.

Statistical analyses were based on ant workers only, thus excluding dispersing sexuals that may occur far from the suitable habitats of their species, except for queens of workerless social parasites collected alongside workers of their host (Schifani et al. 2021). Before all analyses, the number of workers per traps was divided by the number of days the traps were active in order to standardize the values based on the sampling effort. Simpson’s Index (1-D) values were computed for each trap to have a measure of diversity in addition to genus and species richness (Simpson 1949). Analyses on the number of individuals were conducted after logarithmically transforming their numbers to remove outliers that may have resulted from traps accidentally placed too close to ant nests. Species accumulation curves were computed using the function specaccum from the R package vegan (Oksanen et al. 2022) to compare the exploration of the eight transects.

Differences in genus and species richness, Simpson’s Index scores, and number of individuals between the three main habitat types (grasslands/shrubland, woodland, and scree) were assessed using mixed-effect models in which the Park was set as a random factor to account for the differences between distinct Alpine regions. Tukey’s post hoc tests using Bonferroni’s correction were used for pairwise comparisons.

We investigated whether species distribution followed Bergmann’s rule at the assemblage level (species of colder environments having a larger size, see Bergmann 1847) with a mixed-effect model with polynomial terms using the lmer function from the lme4 R package (Bates et al. 2015). We set the sampling station as the random factor while studying the relationship between mean temperature and the average cephalic size (the most used as a proxy for body size) among the ant species found in the traps (using community-weighted means, i.e. based on the occurrence and not the abundance of each species per trap), assigning to each species the corresponding mean value of cephalic size based on literature data (Seifert 2017, 2018). Cephalic size was calculated as the arithmetic mean between cephalic width and cephalic length (Seifert 2018).

Pearson’s correlation tests were used to verify whether species distribution abided to the Rapoport’s rule (species found at higher elevations having a smaller elevational range Stevens 1992): We correlated the maximum elevational limit with the difference between the maximum and minimum elevational ranges that we recorded for each species.

Finally, using the lmer function from the lme4 R package (Bates et al. 2015), mixed-effect models with polynomial terms were used to assess potential non-linear relationships between diversity measures (number of species, number of genera, Simpson’s Diversity Index, and number of individuals) and environmental variables (elevation, minimum, mean, and maximum temperature—using a model for each variable), while setting the sampling station as a random factor. This design was preferred to the use of a nested design with transect set as a nested factor within park, as well as a design with park (instead of the sampling station) as the sole random factor. The former resulted in similar AIC and BIC values while leading to singularity and convergence issues, while the latter resulted in significantly higher AIC and BIC values. A correlation plot illustrating the relationship between the four environmental variables (elevation, minimum, mean, and maximum temperature) is provided in the Supplementary Figure S2.

All statistical analyses were carried out using the software R v4.2.0 and RStudio-2022.02.2-485 (R Core Team 2022).

Results

A full dataset illustrating for each trap the number of ant specimens collected and their identity, as well as the environmental variables recorded, is provided in Supplementary Table S1. Species accumulation curves suggested that a satisfactory coverage was reached in all transects except for the transect MG of the DBNP, which did not fully reach a proper plateau (Fig. 2).

Fig. 2
figure 2

Species accumulation curves illustrating the coverage of the eight sampling transects from the Stelvio National Park (SNP) and Dolomiti Bellunesi National Park (DBNP) based on sampling effort and species richness

Full size image

A total of 262,693 worker ant specimens were processed and identified during this study. They belonged to 40 species, 12 genera, and 3 subfamilies (Dolichoderinae, Formicinae, and Myrmicinae). Of these, the only finding of Formicoxenus nitidulus consisted of one winged queen which was excluded from the analysis. The most species-rich genera were Formica and Myrmica, both with 11 species, followed by Lasius, with 6 species. The most abundant species were Formica aquilonia (158,513 specimens) and Formica lugubris (75,698 specimens), while the most frequent were F. lugubris (found in 68 traps) and Tetramorium. alpestre (found in 42 traps). The species occurring at the highest elevations were Formica lemani, Myrmica lobulicornis, and T. alpestre, while Formica gagates, Plagiolepis pygmaea, and Lasius mixtus occurred only at the lowest elevations (Fig. 3).

Fig. 3
figure 3

Elevational distribution of the ant species found in the traps along our sampling transects, excluding dispersing sexuals from the analysis (Formicoxenus nitidulus is excluded since only a single dispersing queen was collected). Species are listed in an increasing order based on their upper elevational range

Full size image

The highest diversity of species was found at intermediate elevations, between 1400 and 1800 m asl (Fig. 4). From 1600 to 2200 m asl most specimens belonged to mound-building Formica species (F. exsecta, F. truncorum, and the F. rufa group, including F. aquilonia, F. lugubris, and F. rufa), while at higher elevations other Formica species were found (mainly F. cunicularia, F. lemani or F. rufibarbis, Fig. 4). The highest elevation was reached by F. lemani, found at 2631 m asl. At the species-level, the prevalent genera were Formica and Myrmica across all elevations, while a significant portion of Lasius existed until 1800 m and Tetramorium was one of the three genera recorded above 2400 m (Fig. 4). On the other hand, at the individual level, after 1600 m, more than 95% of the captured specimens belonged to Formica, with mound-building species being replaced by F. lemani above 2400 m (Fig. 4).

Fig. 4
figure 4

Abundance and diversity of the ant genera recovered at each elevational range based on our sampling transects given as percentage values. For each elevational range, the total number of species (above) and individuals (below) is indicated on the top of the bar

Full size image

Concerning the comparison among habitat types, statistically significant differences were found for the number of species (F(2202) = 8.62, p < 0.001), genera (F(2201) = 7.97, p < 0.001), and individuals (F(2202) = 21.63, p < 0.001), as well as for the Simpson’s Index values (F(2201) = 5.93, p < 0.001) (Fig. 5). Pairwise comparisons of the number of species revealed significant differences between grassland/shrubland, having the highest species richness values, and both scree (p = 0.002) and woodlands (p = 0.040), but not between woodlands and scree (p = 0.101). For the number of genera, scree had the lowest values and differed significantly from both grasslands/shrublands (p = 0.003) and woodlands (p = 0.004), while grasslands/shrublands was not different from woodland (p = 0.573). Woodlands and scree were again not significantly different for Simpson’s Index values, having the lowest values (p = 0.175), while grasslands/shrublands differed from scree (p = 0.009) but not from woodlands (p = 0.093). Finally, the number of individuals showed a discordant trend, with all habitats significantly different from one another (p < 0.001 for all comparisons), woodlands having the highest numbers, scree the lowest, and grasslands/shrublands holding an intermediate position (Fig. 5).

Fig. 5
figure 5

Differences in abundance and diversity of ants among habitat macro-categories based on our sampling transects (n traps: scree = 10; woodlands = 52; grasslands/shrublands = 143). Habitats that are significantly different based on pairwise comparisons (Tukey’s post hoc tests) are marked with different letters

Full size image

Concerning Bergmann’s rule, a significant relationship was found between minimum temperature increase and the average cephalic size of ant assemblages (F(238) = 3.79, p < 0.031, R2 = 0.45; estimate1: 389, p = 0.59; estimate2: − 1765, p = 0.02). Ants associated with intermediate temperature ranges, instead of coldest temperatures, showing the larger size (Fig. 6). For Rapoport’s rule, we found a strong positive correlation between species’ elevational range and their maximum elevational limit (t(37) = 6.30, p < 0.001, R = 0.72) (Fig. 6).

Fig. 6
figure 6

Relationship between ant size (with cephalic size as a proxy) of the collected ants and minimum temperature recorded at the corresponding sampling station (on the left), and relationship between the maximum elevational limit and the elevational range of each species (on the right). According to the Bergmann’s rule, size of organisms should increase at lower temperatures, while Rapoport’s rule predicts that species found at higher elevations are characterized by a larger elevational range

Full size image

Significant relationships were found between each diversity measure (number of species, number of genera, Simpson’s Diversity Index, and number of individuals) and environmental variable (elevation, minimum, mean, and maximum temperature) (0.034 < p < 0.001), except for the number of individuals and maximum temperature (p = 0.117) (see Supplementary Table S3, Fig. 7). The comparison of AIC and BIC values between models that included different environmental variables revealed the lowest scores for the effect of mean temperature except for the models on the number of individuals, for which the lowest AIC and BIC scores corresponded to the minimum temperature. The highest AIC and BIC scores always corresponded to the models with the maximum temperature (Supplementary Table S3). Ant diversity (number of species, number of genera, Simpson’s diversity index) and abundance (number of individuals) mostly decreased as expected and increased with temperature, but in most cases an opposite trend was observed at the lowest altitudes or highest temperatures recorded, resulting in a non-linear relationship (Fig. 7).

Fig. 7
figure 7

Relationships between diversity measures (number of species, number of genera, Simpson’s Diversity Index, and number of individuals), and environmental variables (elevation, minimum, mean and maximum temperature)

Full size image

Discussion

Our study provides a detailed reference dataset as an essential tool to monitor ant elevational distribution in the European Alps and observe future shifts in response to climatic change, also laying the ground to discuss conservation priorities and better-include ants in management plans. Overall, the species diversity we found represents approximately 21% of the central and Northern European fauna and 15% of the Italian fauna (Seifert 2018; Schifani 2022).

The European Alps ant fauna appears characterized by a remarkable proportion of the European diversity of the cold-adapted genera Formica and Myrmica (Seifert 2018), and a very significant representation of social parasitism (Buschinger 2009). All socially parasitic strategies were represented: dulosis (F. sanguinea); temporary social parasitism (F. aquilonia, F. exsecta, F. sanguinea, F. lugubris, F. rufa, F. truncorum, L. mixtus, L. umbratus) and hyperparasitism (L. fuliginosus); inquilinism (F. nitidulus, M. microrubra, M. myrmicoxena) (Buschinger 2009; Seifert 2018). Inquiline social parasite species are unlikely to be captured with pitfall trapping, and parasite Temnothorax, Tetramorium or Strongylognathus may have been present but were not captured. At the same time, compared to what is observed in more thermophilous regions of Europe (e.g., Castracani et al. 2010; Bazzato et al. 2022), species diversity per site was lower, and there were no diet specialists (such as granivorous species or specialized predators) (Seifert 2018). Myrmica myrmicoxena, listed as Vulnerable in the IUCN 2.3 Red List together with F. nitidulus, was the only Alpine endemic species collected (Social Insects Specialist Group 1996; Schifani et al. 2021), while all other ant species are characterized by wide distribution ranges, mostly covering continental Europe and in many cases extending to other high-latitude Palearctic regions (Seifert 2018; Schär et al. 2018; Wang et al. 2022). We detected a complex relationship between ant size and temperature, not clearly following Bergmann’s rule. In areas with the lowest temperatures many small sized genera did not occur (e.g. Leptothorax, Plagiolepis, Tapinoma, Temnothorax), but also the largest sized taxa were absent (Camponotus spp., mound-building Formica species), while ant size was larger at intermediate temperature range, a pattern largely driven by mound-building Formica species. Previous studies on different scales found mixed support for the Bergmann’s rule in ants, with general trends at least less pronounced than in other animal groups (Kaspari and Vargo 1995; Geragthy et al. 2007), and in some cases the observed trend being the opposite to that described by the Bergmann’s rule (Brassard et al. 2020). Furthermore, it may still be possible that intraspecific patterns conform to Bergmann’s rule better than interspecific ones or that the rule gains support only by sampling a much wider thermal gradient.

The abundant presence of the iconic mound-building Formica species, mostly belonging to the F. rufa group (red wood ants), had a major role in the data we collected. Since their normal colony size is several times larger than in most other alpine ant species (Seifert 2018), their frequent presence in woodlands at intermediate elevations determined the highest absolute number of ants to be found in these habitats. These ants are behaviorally dominant to other ants (Lach et al. 2010), and are considered ecological keystones acting as important forest predators and trophobionts, hosting a variety of associated myrmecophilic organisms, but at the same time significantly lowering the diversity and abundance of many other arthropods in their territory, including other ants, as a consequence of competition or territorialism (Ślipiński et al. 2014; Stockan and Robinson 2016; Trigos-Peral et al. 2016; Seifert 2017; 2018; Parmentier and Claus 2019; Castellucci et al. 2022). Red wood ants have historically attracted higher conservation attention than any other European ant, and efforts to monitor their presence are frequent, yet no specific concerns for the alpine populations have been identified so far (Antonova and Marinov 2021; Sorvari 2022; Van Buggenum 2022; Balzani et al. 2022).

Ant abundance and diversity expectedly decreased with elevation and increased with temperature (Reymond et al. 2013). The relationship tended to be stronger with mean temperature and always weaker with the maximum temperature. Exploring the role of thermal limits in ant biogeographic patterns is an important and rapidly developing field with obvious relevance to climate change scenarios (Nascimento et al. 2022; Willot et al. 2022; 2023; Araujo et al. 2023). Species adapted to tolerate the more extreme elevations were characterized by larger elevational ranges, in accordance with the Rapoport’s rule (Stevens 1992). In this sense, alpine ants appear to be less susceptible to the direct effects of temperature increases compared to organisms of other groups (Gobbi 2020; Pallarés et al. 2020; Poloni et al. 2022). At the same time, the expansion of dominant forest ants favored by a general increase of forested areas may have significant impacts on the diversity of both ants and other arthropod groups (Bebi et al. 2017). In general, ant communities became very species-poor above the tree line (> 2200 m asl), as only three species (F. lemani, M. lobulicornis, T. alpestre) were found in the highest grasslands and screes, where energy nutrients may be particularly limiting (Guariento et al. 2021)—each of them occurring across an elevational range of at least 1000 m. Maximum elevational ranges recovered in this study mostly corresponded to those reported in the literature, with no indication of rapid upward range shifts taking place as in other groups (Brambilla and Gobbi 2014; Gilgado et al. 2022; Panza and Gobbi 2022). Open habitats like grasslands and shrublands were those that hosted the highest species diversity. This is likely a combined result of higher topsoil temperatures due to reduced shade, and reduced presence of territorial and behaviorally dominant ant species (Seifert 2017). Therefore, preserving open habitats appears to be fundamental for the conservation of alpine ants as it happens with other arthropod groups (Bussan 2022; Gobbi et al. 2015, 2022; Walcher et al. 2019; Schwarz and Fartmann 2021). Dispersal ability may play a crucial role in future scenarios for alpine organisms but very little is known in this regard for most European ant species. However, all the species we detected are capable of flight dispersal (Seifert 2018). Interestingly, we once found a L. fuliginosus queen at 2400 m asl, several hundred meters above the highest elevation recorded for the colonies of this parasitic species or its hosts: upward winds are likely to favor occasional dispersal of alates above the suitable elevational limit of their species.

In conclusion, the ant fauna we investigated appears to be well-characterized by the richness of cold-adapted taxa and the peculiar prominent role of red wood ants near woodlands. The scarcity of endemics or species with narrow distribution ranges, including at the highest elevations, as well as the absence of diet specialists and the presumably good dispersal capabilities of the occurring species, may perhaps grant Alpine ants some more resilience to climate-driven environmental changes compared to other arthropod groups. The cold climate conditions also prevent the establishment of most invasive ant species that significantly threaten ant diversity elsewhere in Europe (Menchetti et al. 2023; Schifani 2019). On the other hand, strong interconnections between species in the form of widespread social parasitism may represent a special weakness to potential ecological cascade effects. Further efforts are required to monitor elevational shifts over time and genetic diversity along elevational gradients, as well as the ecological relationships between ants and other key arthropod groups in the Alpine environment.