A demographic and ecological study of an Italian population of Polyommatus ripartii: the ESU Polyommatus exuberans

Polyommatus exuberans is an evolutionarily significant unit (ESU) of the lycaenid butterfly Polyommatus ripartii. This ESU is known to survive at only two sites in the Susa Valley (NW Italy). Lack of correct management, reforestation and frequent wildfires severely threaten this ESU (listed as endangered species in the most recent IUCN Italian Red List). Although the taxonomic rank of this taxon is still debated, current threats could cause extinction of its two remaining populations before its taxonomic rank and its ecology are clarified. We collected data for the first time on this population at the small site of Mompantero (ab. 10 ha). We used butterfly GPS-positioning and the mark-release-recapture (MRR)-method to estimate its population size (269 individuals), sex-ratio (1.36 M/F), lifespan (4.76 days), density (47/ha) and mobility (median 153 and 33 m for males and females, respectively). Both sexes are equally catchable. Catchability increases around midday and decreases during overcast weather. While the size and density of this small population are comparable to those of other endemic Polyommatus species (such as P. humedasae and P. gennargenti), scarce mobility makes its populations isolated and even more seriously threatened. We suggest that implementing an active management plan, including mowing before July and/or in autumn, and supporting ant diversity, is of immediate importance. Management should be extended to road verges, where the larval host plant (Onobrychis sp.) is abundant, and would thus also serve as corridors to favour dispersion between sites. Our research is the first study to investigate this taxon, thus shedding some light on the ecological and biological aspects that are crucial for long-term survival.


Introduction
Recent research has emphasized a global decline in insect populations during the last 30 years (Collen et al. 2012;Dirzo et al. 2014;Hallmann et al. 2017;Vogel 2017; Lister and Garcia 2018; Sánchez-Bayo and Wyckhuys 2019; Crossley et al. 2021). Butterflies are included in this phenomenon (Thomas et al. 2004;Forister et al. 2011;Habel et al. 2016;van Strien et al. 2019;Crossley et al. 2021). To oppose this trend and to plan ad hoc conservation measures, it is crucial to understand the subtended mechanisms of interactions between species and their ecosystems. However, not all species are well studied, and there is a disjunction between scientific research and conservation needs (e.g. Di Marco et al. 2017). Indeed, research studies are not equally distributed among species; only a few species, such as the monarch butterfly Danaus plexippus (Linné 1758) (Wells and Wells 1992;Agrawal 2019), or, among lycaenid butterflies, the five European species of the genus Maculinea (Settele and Kühn 2009), have obtained much more attention than other very rare, scattered, fragmented and endemic species, such as Erebia christi Rätzer, 1890, Polyommatus humedasae (Toso and Balletto 1976) or P. ripartii (Freyer 1830) (Dincă et al. 2013;Bonelli et al. 2018). These butterflies are among the least studied butterflies in Europe; when a literature research on the ISI Web of Knowledge is performed, Erebia christi results in having three studies, of Enrico Parile and Irene Piccini have contributed equally to the research.
1 3 which one investigated the population size (Battisti et al. 2021); while for P. humedasae, there was only one study. The ISI Web of Knowledge research shows 24 studies for P. ripartii, of which only three related to the ecology and biology of P. ripartii (Lafranchis and Lafranchis 2012;Przybylowicz 2014, Pérez-Fernández et al. 2019. P. ripartii has scattered populations across southern Europe. These populations may show local adaptations such as having different host plants (Onobrychis spp.) or different ant species hosts. Thus, little is known about these butterfly species, even if the contribution of rare and endemic species to ecosystems is generally crucial due to their role in ecosystem functioning, which is usually different from common species (Kunin and Gaston 1993;Gaston 2012). Moreover, rare, scattered and endemic species are usually more prone to extinction due to characteristics such as lower overall reproduction and poorer dispersal abilities (Kunin and Gaston 1993;Işik 2011). There is, therefore, a high risk of losing these species without even noticing it (McGarrahan 1997;Bonelli et al. 2011;Theng et al. 2020). Conservation actions, such as habitat protection and restoration, are likely to be effective in opposing the decline of local insect populations (Samways 2007;Forister et al. 2019;Haddaway et al. 2020;Crossley et al. 2021). To be effective, policy and management must be supported by reliable knowledge accumulated from diverse sources of evaluated evidence on different aspects of species ecology and phenology.
Polyommatus ripartii presents a fragmented distribution, with scattered populations from Spain-in the west-to Mongolia-in the east-probably representing a relict distribution (Dincă et al. 2013). In Italy, P. ripartii is found in the Ligurian Alps at ten locations; in the Pollino and Orsomarso Massifs formerly known as P. galloi (Baletto and Toso 1979) and in the Susa Valley at two locations (Bonelli et al. 2018). The population of the Susa Valley is known in Italy by the name of Polyommatus exuberans (see below for more information on its taxonomic status). According to Dincă et al. (2013), the highly fragmented populations of P. ripartii in the Susa Valley can be considered and studied as separate Evolutionarily Significant Units of Conservation (ESU; see Ryder 1986;Moritz 1994;Casacci et al. 2014). This approach has been adopted for other butterflies with similar distributional characteristics, such as Lycaena dispar (Haworth 1803) (Lai and Pullin 2004), and Maculinea rebeli (Hirschke 1905) (Casacci et al. 2014). Hereinafter, we consider P. exuberans as the Italian ESU of P. ripartii located in the Susa Valley (hereinafter P. exuberans). In the last IUCN Italian Red List (Bonelli et al. 2018), which follows Balletto et al. (2014), P. exuberans is considered as a separate species within a specific category: EN (Endangered). On the other hand, P. ripartii is classified as NT (Near Threatened) in Italy (Bonelli et al. 2018) and LC (Least Concern) at a European and Mediterranean level (van Swaay et al. 2014) according to the IUCN Red List criteria. No information on the life cycle or the ethology of P. exuberans is currently available, and other information about this ESU is scarce. P. exuberans is-overall-in decline, following the extinction of six of its former populations, including at its type locality ("Oulx"), due to various reasons, corresponding to a loss of nearly 75% of its original distribution area . It currently survives at only two locations, where it is severely threatened mainly by the abandonment of traditional agro-pastoral activities and consequent reforestation, as well as by recurrent brush fires (the most recent took place at Mompantero during the last week of October 2017) and by over-collection (Sindaco et al. 2008;Bonelli et al. 2018). Its Extent Of Occurrence (EOO) is now much less than 5000 km 2 , while its Area Of Occupancy (AOO) is less than 500 km 2 (Bonelli et al. 2018). Munguira et al. (1993) previously evaluated this lycaenid as endangered, which was recently confirmed (EN), based on stricter IUCN criteria (Bonelli et al. 2018).
We decided to study P. exuberans as information about this ESU is fragmented and unclear, as its population is in decline and because, despite being inside the Natura 2000 network, the species is suffering from lack of management and uncontrolled fires. Specific management actions should be quickly adopted for conservation purposes. Collecting information about the ecology and biology of P. exuberans would corroborate the use of the ESU concept (Casacci et al. 2014), and should improve conservation management and policy (Haddaway et al. 2020). Here, we present the results of a pilot one-season investigation of the adult population located at Mompantero (NW Italy), which falls into two SAC (Special Areas of Conservation IT1110039 and IT1110030). The aims of our study, which represents a starting point for future investigations, were to (1) estimate the size of the adult population, its density, the sex ratio, and the mean lifespan; (2) analyse how catchability was influenced by the time of day; (3) evaluate adult mobility in relation to sex and weather conditions; (4) describe the use of nectar source by the two sexes.

Taxonomic status of P. exuberans
Polyommatus exuberans (Hirsutina admetus exuberans) was originally described by Roger Verity in 1926. The name was chosen due to the exceptional wing size of the specimens of this butterfly collected by the author during the summer of 1925 in Oulx (Verity 1926). Locus typicus: "Cottian Alps, Oulx (July 25th-August 5th)". While other authors classify P. exuberans as a member of the separate genus Agrodiaetus (Bertaccini 2003;Vila et al. 2010), in the present study we follow the most usual taxonomic approach (e.g. Wiemers et al. 2009;Przybylowicz 2014;Lukhtanov et al. 2015 and many others), where Agrodiaetus is treated as a subgenus of the broader genus Polyommatus. Species of Agrodiaetus are all of recent origin, since their most recent common ancestor dates back to ab. 0.87 Mya (Dincă et al. 2013). This makes these species poorly differentiated both at the morphological (they share identical genitalia) and the genetic level (Vila et al. 2010;Dincă et al. 2013;Lukhtanov et al. 2015).
Indeed, the enzyme electrophoretic study by Mensi et al. (1994) showed that P. exuberans and P. ripartii have allozyme profiles extremely similar to a Nei (1972) distance of only 0.074. Moreover, no diagnostic loci were found.
In accordance with Mensi et al. (1994), Vila et al. (2010) demonstrated that the two latter taxa share very similar mitochondrial and nuclear DNA (COI + COII + ITS-2) and did not support P. exuberans as a distinct species. Consequently, this taxon is excluded from the latest checklists of European butterflies, and it is likely considered as part of P. ripartii (Wiemers et al. 2018). Dincă et al. (2013) proposed considering the geographically highly fragmented genetic lineages of the P. ripartii group (including P. exuberans = P. ripartii susae Bertaccini, 2003) as separate Evolutionarily Significant Units of Conservation (ESU: see Ryder 1986;Moritz 1994;Casacci et al. 2014), while Lukhtanov et al. (2015), working on a set of two genetic markers (the mitochondrial COI barcode and the nuclear spacer ITS-2), demonstrated that the P. ripartii clade remains fundamentally unresolved. Therefore, protecting P. exuberans is greatly important for future studies of the phylogeny and zoogeographic history of P. ripartii's clade and of taxonomically related species (Przybylowicz 2014).

Study species
The P. ripartii ESU is a highly localized butterfly of the Susa Valley in NW Italy (Hellmann and Bertaccini 2004). It is a monovoltine, heliophilous butterfly of the xerothermophilous grasslands, and is present at elevations between 500 and 1200 m (Hellmann and Parenzan 2010). The flying period of this butterfly lasts from mid-July to mid-August (Hellmann and Parenzan 2010). Males and females are similar in size, with a wingspan of 29-31 mm and 26-29 mm, respectively (Bertaccini 2003). Larvae of P. exuberans develop on sainfoin (complex of Onobrychis viciifolia Scop.). The females usually lay a single egg on the upper margin of a floral bract, and rarely on other parts of the host plant (Hellmann and Bertaccini 2004).

Study area
The monitored area was in North-Western Italy (45° 09′ N 7° 5′ E, between 1060 and 1250 m a.s.l.), in the Susa Valley ( Fig. 1), and is situated on the slopes of Mt Rocciamelone (3.538 m). The area extends for ca. 3.8 ha and is part of the SAC IT1110039 "Rocciamelone" and of the SAC IT1110030 "Xerothermic oasis of the Susa Valley-gorges of Chianocco and Foresto" within the Mompantero municipality. This area is occupied by spinneys with Pinus sylvestris and by xerophilous grassland with Festuca vallesiaca, and extends from west to east. The area is crossed by a dirt road leading to the Chiamberlando hamlet, and results in an ecotonal area. Host plants of P. exuberans (sainfoins) were abundant, especially on road verges, which are occasionally managed. Mowing of the road edges inside the monitored area took place during the last week of our monitoring, around mid-August.

Mark-release-recapture (MRR)
Excursions were made periodically to verify the presence of the lycaenid in the monitored area. Capture events occurred every 2 days from 12th July to 17th August 2019. In the last two events (15th and 17th August) we did not find any individuals of P. exuberans. Two longer intervals between 5 to 8th August and from 10 to 13th August were due to unsuitable weather conditions. Altogether, the population was sampled 18 times. Each butterfly was netted and individually marked with a consecutive number on the underside of the right hindwing using a non-toxic violet fine-tip permanent marker, and immediately released at the same location. Recaptures of previously marked butterflies were recorded separately. Before releasing each specimen, we noted its individual number, sex, habitus, behaviour before netting, GPS position (Garmin® eTrex 20 with precision of ± 3 m) and time (date, hour) of the capture/recapture event. Behaviour was scored into six different categories: flying, basking, feeding, resting, copula and oviposition. When individuals were found nectaring, the plant species was also recorded and identified at the genus level. Weather conditions were recorded on a scale from 0 to 3 for both wind intensity (0 = absent, 1 = weak, 2 = medium, 3 = strong) and cloud cover (0 = no clouds, 1 = few clouds, 2 = many clouds, 3 completely overcast). To evaluate mobility and behaviour, we registered individuals that were captured more than once during the same day of activity.

Demography
The data derived from the MRR method were analysed according to the Cormack-Jolly-Seber type constrained models (Schwarz and Seber 1999) with program MARK® 2.1 (White and Burnham 1999). The best model was chosen on the basis of Akaike's information criterion (AIC). The model with the lowest AIC is taken as the best model for data fitting. Models with the AIC Delta < 2 and with the smallest number of parameters were considered as potential good model candidates (e.g. Burnham and Anderson 1998;Schtickzelle et al. 2002).
The program MARK estimates survival probability ( ; combining mortality and emigration) and capture probability (p). These parameters may be constant (·), dependent on sex (g) or on time (t).
In MRR studies, the average life span of butterflies that emerge discontinuously from pupae is calculated as ê = 1 1− − 0.5 , derived from Deevey's (1947) formula for life expectancy of new-born individuals under the assumption of age-independent survival (Nowicki et al. 2005).

Catchability
To define whether catchability varied during the day, we divided the time of capture (hour) data into three time slots: morning (8:30-11:30), midday (11:30-14:30) and afternoon (14:30-17:30). We modelled the number of captured and recaptured individuals, subdivided by sex, in a generalized linear model (GLM, Poisson, maximum likelihood fit) where time slot and sex, as categorical variables, were used as additive explanatory variables. We tested the model for over/ under-dispersion using the DHARMAs package in R (Obs/ Sim = 0.91, p = 0.832). Moreover, we evaluated if catchability changed according to weather conditions, specifically in relation to clouds and wind. We modelled the number of captures and recaptures, subdivided by sex, in a GLM (Poisson, Fig. 1 Special areas of conservation of piedmont and a focus on the studied area with minimal convex hull, captures divided by sex and host plant locations. The precise location of the studied area is vague for reasons of conservation maximum likelihood fit), where cloudiness, as the ordinal variable (from 0 to 3), was added as an additive explanatory variable. We tested for over/under-dispersion of the model using DHARMAs package in R (Obs/Sim = 0.61, p = 0.176). Similarly, we modelled the number of captures and recaptures in GLM (GLM, Negative Binomial, maximum likelihood fit) where wind, as the ordinal variable, was added as an additive explanatory variable. Considering that the model with the Poisson distribution family was over/under dispersed (dispersion test: Obs/Sim = 4.10, p < 0.001***), we used the Negative Binomial distribution family (according to Zuur et al. 2009).

Mobility
The distance between consecutive captures on different days (D) was calculated as a straight line connecting two consecutive points of butterfly capture (QGIS 10.0; ESRI, Redlands, CA, USA). According to Jugovic et al. (2017); we calculated the mean, median and maximum distances separately for males and females. Moreover, we modelled distances (D) in a linear model (LM, Gaussian, maximum likelihood fit), where sex was used as an explanatory variable. We tested the models for normality of the residuals (Shapiro-Wilk test: p = 0.07).
The movements of adults were estimated separately for males and females, using distance between consecutive captures (D). To understand their dispersal abilities (Pennekamp et al. 2014), we estimated the probabilities of movements with respect to distance beyond those covered during the MRR study. Thus, we calculated the negative-exponential function (NEF) and the inverse-power function (IPF), according to Hill et al. (1996).
All distances were grouped into 50 m classes, separately for each sex. The probability (P) of an individual moving a certain distance (D), beyond that covered during the study, was calculated as: P NEF = ae -kD , P IPF = bD -n . Parameters a, b, k and n are typical of the species, and k is a species-specific dispersal constant describing the shape of the exponential curve; k is equivalent to 1/D′, see Harrison et al. (1988), where D′ represents the average distance moved by individuals (Hill et al. 1996). Considering that log-transformed formulas can be expressed as linear relationships [i.e. ln(P NEF ) = ln(a) − kD and ln(P IPF ) = ln(b) − n ln(D)], and following Čelik (2012), the a, b, k and n parameters were estimated by LMs in R. The natural logarithms of the inverse cumulative proportions (ICP) of individuals moving certain distances (lnP) were regressed on distances (meters). Therefore, to estimate a, b, k and n parameters, we modelled natural logarithms of ICP into a LM (Gaussian, maximum likelihood fit) where sex (as a categorical variable), meters for NEF function, or natural logarithm of meters for IPF function (as continuous variables), and their interaction (sex × distance) were used as additive explanatory variables. We tested for the normality of the residuals for both models (Shapiro-Wilk test: p = 0.17 and p = 0.11 for NEF and IPF functions, respectively).

Behaviour
According to Jugovic et al. (2017), we evaluated which behaviours were exhibited by individuals immediately before capture, as percentage values of the six observed behaviours (flying, basking, feeding, resting, oviposition and copula), separately for each sex. Within the feeding behaviour, nectar plant choices are also presented here as percentages, separately for each sex.
To assess if males and females exhibited different behaviours in relation to different weather conditions, we modelled behaviour frequencies in a LM (Gaussian, maximum likelihood fit), where sex and behaviours, as categorical variables, and weather conditions (separately: wind and clouds), as ordinal variables, were used as explanatory variables. We tested for normality of the residuals for both models (Shapiro-Wilk test: p = 0.412 and p = 0.150 for clouds and wind, respectively).

Ant communities
Fiedler (1991) described P. ripartii as "steadily myrmecophilus", with larvae presenting a complete set of organs for myrmecophily (a dorsal nectary organ and a pair of eversible tentacle organs), while there are no specific data for ESU P. exuberans. The ants which were observed to be associated with P. ripartii belong to genera Camponotus, Plagiolepis, Crematogaster, Lasius, Lepisiota and Tapinoma (Lafranchis et al. 2007).
As a pilot study, we evaluated the ant communities to see if the ant genara used by P. ripartii were present in the area. To explore the ant communities present in the study area, we prepared five 5 × 5 m 2 plots, and we placed five pitfall traps within each plot. Traps were collected after 2 days of activation.
The collected ants were preserved in 70% alcohol inside a 2 ml microtube with a screw-cap. We identified ant samples at the genus level. To this end, we inspected the ants with a stereomicroscope Leica EZ4 D, and we followed the identification keys from "The ants (Hymenoptera, Formicidae) of Poland" (Czechowski et al. 2002), "Clé de reconnaissance des fourmis Françaises" (http:// cle. fourm is. free. fr/) and "Ants of Britain and Europe. A Photographic Guide" (Lebas et al. 2019).

Demography
In total, 59 individuals (males:25, females:34) were marked, and 27 (46%) of them were recaptured (Table 1), of which 12 (44%) were recaptured more than once. Males were recaptured more often than females, respectively, 48% males (total number 12) and 44% of females (total number 14). In relation to this result, we calculated a sex ratio of 1.36, in favour of females.
The best model, evaluated by the Cormack Jolly-Seber (CJS) model (Online Appendix Table 2), indicates that survival probability (ϕ) and capture probability (p) were not sex-specific and varied over time (Online Appendix  Table 3).
Using Mark parameters, total population size (N) was estimated as the sum of daily captured individuals (excluding recaptures), divided by p(t). Daily population size (Ni) was calculated as the number of captured and recaptured individuals, divided by p(t) along the flight period. For N 0 (on 12/7/2019), we divided the number of captured individuals by p(1). Standard errors were calculated as numbers of captured and recaptured males and females divided by p ± p SE .
According to Nowicki et al. (2005), daily recruitment (Bi), i.e. the number of individuals entering the population between capturing sessions, was calculated separately for each sex as (1) N i+1 − N i × ϕ d , where d is the length of the interval between capturing occasions, measured in days. According to Jugovic et al. (2017), negative values were corrected to 0. Similarly to what was carried out for population size, standard errors of daily recruitment values were calculated by the same formula (1), using ϕ ± ϕ SE instead of ϕ, and then subtracting the result from (1) (resulting in different amounts of positive and negative standard errors).
The estimated total population size was 269 individuals (ranging from 187 to 467 individuals), including 150 estimated females (ranging from 106 to 247) and 119 estimated males (ranging from 80 to 220). The minimal convex hull was 5.69 ha wide (Fig. 1), so that butterfly density was about 47 estimated individuals/ha, and 10 captured individuals /ha. Daily recruitment was calculated along the flight period (Fig. 2a), and was in line with the trend of daily population size (Ni), over time (Fig. 2b). Analysing the flight curve (Fig. 2b), flights of males reached a peak on July 20th. While males decreased after this peak, females continued to increase, reaching their first peak on July 28th, and a second peak on August 8th.
Using the average value of daily survival (ϕ = 0.81 ± 0.07), we estimated an average lifespan of 4.76 days. The mean and maximum number of days between the first and the last captures of the same individual were: 6.89 and 19, respectively.

Catchability
Catchability differed between time slots. During the midday slot we captured significantly more butterflies than in the morning slot (Z value = 3.38; p < 0.001***), while butterfly captures in the afternoon were not significantly different from in the morning (Z value = 0.85; p = 0.397; Fig. 3a). We did not find any differences in catchability (Z value = − 1.38; p = 0.167) between sexes. In contrast, catchability decreased significantly with increasing cloud cover (Z value = − 6.64; p < 0.001***; Fig. 3b), while it did not significantly change in relation to wind conditions (Z value = − 1.17; p = 0.288).

Mobility
Median, mean and max distances recorded for males were 153 m, 180 m and 522 m, respectively. For females, these  Fig. 7). The natural logarithmic inverse cumulative proportion of moving butterflies decreased with increasing distances and was different for each sex, for NEF function (Sex × Distance: t value = 4.52, p = 0.002**; Fig. 4; see Online Appendix Table 4), but not significantly different for IPF function (Sex × ln(Distance): t value = 1.66, p = 0.13; see Online Appendix Table 5). Moreover, the natural logarithm of inverse cumulative proportion decreased with the increase of distances covered for both functions: NEF (Distance: t value = − 11.19, p < 0.001***; see Online Appendix Table 4) and IPF (ln(Distance): t value = − 7.04, p < 0.001***; see Online Appendix Table 5).
Fitting NEF to mobility data resulted in the following equations: P m = 1.63 e -0.009D for males and P f = 2.01e -0.017D for females (where P is a probability of moving a certain distance, D). The estimated distances (D′ = 1/k) between consecutive captures were 112 m and 61 m, respectively, for males and females (the data averages were 117 m for males, and 52 m for females). The estimated probabilities (in % of individuals) of long-distance movements were 2% (males) and 0.05% (females) for 500 m, 0.02% (males) and 0 (1.3 e -5 %) (females) for 1 km, and approximately 0 for both sexes for 2 km (2.7 e -6 % males and 8.9 e -13 % females).
Fitting IPF to mobility data (Online Appendix Table 5) resulted in the following equations: P m = 445 D −1.45 for males and P f = 3673 D −2.05 for females. The estimated probabilities (in % of individuals) of long-distance movements based on the IPF were: 5% (males) and 1% (females) for 500 m, 2% (males) and 0.3% (females) for 1 km, and 0.7% (males) and 0.06% (females) for 2 km. The regression slope was steeper for females (t = 5.91, df = 94, p < 0.001). IPF predicted a higher probability of movement for distances of over 300 m (males) and of over 200 m (females).
Butterflies were found to exhibit different behaviours in relation to cloud coverage, but not in relation to sex. In particular, increasing cloud presence decreased the percentage of butterflies in flight (behaviour × cloud coverage: t value = − 3.90; p < 0.001***; Fig. 6). We did not find any differences between the other behaviours (feeding and resting) in relation to cloud coverage (Online Appendix Table 6). In addition, we did not find any significant differences between sexes and behaviours (basking was the reference category) in relation to wind (Online Appendix Table 7).

Ant communities
We found 1479 ants belonging to 11 different genera  Table 8). Camponotus, Plagiolepis, Lasius, Lepisiota and Tapinoma are used by P. ripartii. Therefore, these five genera of ants are candidates for use by P. exuberans.

Discussion
The population of P. exuberans of the Mompantero site is small, restricted to a limited area (5.69 ha). We found that (1) the population size was estimated at about 269 individuals, i.e. 150 females and 119 males. Butterfly density was also low, about 47 individuals/ha, and comparable to those of other Polyommatus populations, equally confined to small sites, such as P. humedasae (Toso and Balletto 1976;Balletto 1993a). The sex ratio was slightly in favour of females (1.36). The estimated mean lifespan was 4.76 days. (2) As expected, catchability was significantly higher during midday hours, and significantly decreased under worsening weather conditions, specifically when cloud coverage increased. (3) Males covered longer distances than females, but both sexes only moved short distances (maximum 500 m). In both NEF and IPF estimations, approximately none of the individuals covered distances longer than 2 km. Fig. 5 a Number and percentage of the different behaviours (flying, feeding, resting and basking) recorded for males (inner circle) and females (outer circle). b Within feeding behaviours (N = 27), number of visits on nectar plant genera by P. exuberans was recorded for both sexes Fig. 6 Results from the LM of butterflies in relation to cloud coverage (from 0 to 3) captured while exhibiting different behaviours. Conditional plots are shown, drawn with the visreg package in R (4) Butterflies decreased their flying behaviour when clouds increased. Both male and female butterflies preferably feed on Origanum sp., and Allium sp., but only males were found to feed on the host plant Onobrychis sp.; females, instead, prefer Lavandula sp.

Demography
The presence of the first adults at the Mompantero site was observed on 12th July 2019, and continued throughout the whole flying season; no butterflies were detected after 13th August. Captures and recaptures were only 59 and 43 respectively, which are low numbers, but comparable with other studies on local rare lycaenidae populations (Marschalek and Deutschman 2008;Marschalek and Klein 2010). The probability of recapture (46%) was relatively high (Morton 1982;Polic et al. 2014), probably due to the small area where the population occurs, and to the low mobility of the butterflies. The number of marked males was lower than that of females, a rather atypical occurrence in MRR studies carried out on butterflies (Čelik 2012;Jugovic et al. 2017), resulting in an unusual female-biased sex ratio of 1.36. However, the high number of recaptures and the low individual mobility observed in this study show that individuals of either sex disperse rarely, and most of them occur together at the same site.
The longest recorded life span was 19 days, which is higher than that found for the lycaenid Phengaris (Maculinea) arion (Linné 1758), where it was shown to range from 7 to 17 days (Osváth-Ferencz et al. 2017). The mean lifespan value, calculated based on ϕ (output of Software Mark), was about 4.76 days, which is slightly higher than that reported for Pseudophilotes bavius hungarica (Dioszegy 1913) (2.4-5.4 days; Crişan et al. 2014), and similar to that observed in Zerynthia polyxena ([Denis and Schiffermüller] 1775) (4.4 days; Čelik 2012). The lifespan was found to be lower, in contrast, compared to other univoltine, big-sized, butterflies, such as Aporia crataegi (Linné 1758) (about 7 days; Jugovic et al. 2017), which are known for having generally longer adult lives. Each individual specimen of Parnassius apollo was observed to survive for an average of about 2-4 weeks (Lafranchis et al. 2015).
The flight graph shows a non-parabolic trend in daily population sizes, which is atypical for univoltine butterfly species (Jugovic et al. 2017). The presence of a peak (28/7), followed by a higher peak (08/8) during the flying period is generally not common for butterflies (Clarke and Dennis 2020), while protandry is instead a common characteristic (Baguette and Schtickzelle 2003;Fric et al. 2010;Čelik 2012;Jugovic et al. 2017). After the rainy day of 7th August, an increase in daily population size was detected. The subsequent decline in daily population size was shortly followed (2-3 days) by another rainy day on 12th August, and by the start of mowing activities. Both bad weather and mowing could have influenced the complete lack of recaptures after 10th August, and may have, therefore, accelerated the end of the flight period. Mowing in particular, may have forced the remaining adults to emigrate perhaps due to disturbance (Mouquet et al. 2005, Jugovic et al. 2017). However, although mowing only occurred on road verges, most nectar sources and still available host plants were concentrated at these sites.

Catchability
Our results support the classification of P. exuberans as heliophilous (Hellmann and Parenzan 2010), thus it seems likely to find P. exuberans in xerothermic biotopes on the south-facing northern slope of the Susa Valley. Midday seems to be the optimal time to observe this butterfly. In addition, we did not detect any differences in catchability between sexes, which is a crucial factor for the reliability of results (Nowicki et al. 2008). Weather conditions have to be favourable too, with minimal cloud covering and light winds (Online Appendix Figs. 8 and 9).

Mobility
The observed mobility parameters recorded for P. exuberans suggest that individuals from this population are particularly sedentary. The maximum distance recorded was higher for males (522 m) than for females (242 m). The opposite (298 m for males, 600 m for females) was reported for Lycaena helle ([Denis and Schiffermüller] 1775), another extremely sedentary lycaenid (Modin and Öckinger 2020). The maximum distance recorded for P. exuberans (522 m) is comparable to that observed by Kuussaari et al. (2014) for other Polyommatus species, namely 520 m for P. amandus (Schneider 1792), 510 m for P. icarus (Rottemburg 1775) and 520 m for P. semiargus (Rottemburg 1775). As a caveat, we recognize that, even though we examined the surrounding areas of the site, this dispersal analysis is influenced by the size of the site (Schneider 2003).
According to Čelik (2012), the longest movement observed for males was about equal to the distance between the maximum opposite angles of the minimal convex hull (530 m). Thus, males were able to cover the area on which the population occurred but rarely covered longer distances. Indeed, based on NEF and IPF estimations, the percentage of males that covered longer distances, and migrated to other areas was estimated as being very low. According to Čelik (2012), the NEF predicted a lower probability of movement, indeed 0.2% of males dispersed 1 km. The IPF function predicted that 2% of males disperse 1 km. Movement was even scarcer in the case of females, with the maximum distance covered by females being 242 m, and the percentage of females dispersing up to 1 km was estimated to be almost 0, for both NEF and IPF functions. The scarce mobility observed for males and the really low mobility of females could be dangerous if unpredictable events were to destroy the site at the moment when adults begin to fly, such as wildfires, which frequently occur in the area. In most species of Agrodiaetus, pupation occurs in the litter.
According to the IUCN Italian Red List (Bonelli et al. 2018), P. exuberans only occurs in two localities that are several kilometres away from each other. Considering the population size and the area in which P. exuberans is present, it is unlikely that the Mompantero population would show a system of interconnected subpopulations, like that of L. helle (Fischer et al. 1999;Bauerfeind et al. 2009;Modin and Öckinger 2020), but only genetic analysis can confirm this hypothesis.

Behaviour
Males and females showed similar behaviours, with flying being the most frequent activity. Behaviour of males could be explained as a way to patrol the area in order to search for females or resting places such as mud puddles. In fact, males drinking on the sides of water puddles along with other lycaenid species were observed on three occasions, but this behaviour was not included in the comparisons. Mudpuddling is common among adult male butterflies looking for water and minerals (Arms et al. 1974;Beck et al. 1999). For females, the observed flight activity that we recorded may be due to their higher density, which would result in more time spent looking for available host plants, likely using oviposition-deterring pheromones to avoid deposition in spots already occupied (as already recorded for Phengaris (Maculinea) teleius: Sielezniew and Stankiewicz-Fiedurek 2013;and for P. ripartii: Przybylowicz 2014).
The utilization of nectar from 11 plant genera suggests that, like most other butterflies, adults of P. exuberans are opportunistic feeders. For females, Origanum sp. was the most utilized nectar source. This was also the most abundant plant that was flowering during the flying period. For males, in addition to Origanum sp., Onobrychis sp. and Allium sp. were also equally visited. Since Onobrychis sp.is the host plant of preimaginal stages of P. exuberans, this may be connected to female research or male individuals recently emerged from pupae. This high number of visits shown only by males suggests that the host plant may be used as a proxy of the presence of male adults during the flying period. Origanum sp. is reported as a favourite nectar source for P. ripartii in Greece (Lafranchis et al. 2007). The genus Allium sp. is known for including species appreciated by butterflies (Hantson and Baz 2013).

Ant communities
The ants seen in association with P. ripartii belong to the genera Camponotus, Plagiolepis, Crematogaster, Lasius, Lepisiota and Tapinoma (Lafranchis et al. 2007). Three of these genera in the studied area (Camponotus, Lasius and Tapinoma) were found in all the investigated plots, while the genus Plagiolepis was present in only two plots (Online Appendix Table 8). Considering most of the Italian Polyommatus are myrmecophilus, including P. ripartii, P. exuberans would likely present ant-associations. Thus, it is important to confirm this hypothesis on its myrmecophily and to study the ant genera involved (e.g. Tartally et al. 2019). Local adaptations to different ant species or genera could reinforce the differences with other populations of P. ripartii, as seen for myrmecophilous hoverflies (Schönrogge et al. 2002). Moreover, the use of different ant species would corroborate P. exuberans as a separate ESU of P. ripartii (Casacci et al. 2014). Different ant species required different microhabitats, determined by temperature and humidity at the soil level, soil texture and composition (Boulton et al. 2005;Casacci et al. 2011).

Conclusions
This is a pilot investigation that studied a small localized rare butterfly population in Mompantero. We demonstrated that the population is small in a restricted area, but with parameters comparable to other butterflies with similar characteristics. Considering that the species being studied is in decline, a general management of the area may help to reverse this trend. As a caveat, we recognize that other research has to be carried out to clarify the taxonomic status of P. exuberans, to identify the local host plant, to understand the possible myrmecophily and to study the connections between Susa Valley populations through genetic analysis (the latter has been carried out by Przybylowicz et al. (2014) for P. ripartii in Poland). Specifically on the unresolved taxonomy status, and given the results obtained from mitochondrial markers and enzyme electrophoresis studies, it is crucial that molecular analysis is broadened to include a set of nuclear markers (see Hinojosa et al. 2019).
Polyommatus exuberans is an oligophagous species whose larvae feed on a single plant genus (Onobrychis sp.). This may be an important factor determining the butterfly species' spatial distribution. It is crucial to verify the host plant of P. exuberans at the species level [either O. viciifolia Scop. and/or O. arenaria (Kit.) DC.].
Five out of the eleven genera of ants found within the study area have been reported to be used by the P. ripartii larvae. If myrmecophily is also confirmed for P. exuberans, the Susa population may be supported by the same taxa or different ones. The latter scenario may support P. exuberans as an ecologically distinct ESU of P. ripartii.
Considering that the populations are rare and far from each other, a metapopulation dynamic is difficult to expect. Information on the biology of this species is scarce, while its distribution is poorly understood. Genetic distances among populations could be important to study in order to understand connections and dynamics among populations.
The biggest threats to P. exuberans are over-collecting and reforestation (Sindaco et al. 2008;Bonelli et al. 2018). While the former seems mostly under control, the latter remains a problem, due to the abandonment of traditional agro-pastoral activities. One of the main forces opposing this phenomenon is wildfires, of which the last one occurred in October 2017. The critical situation in which the P. exuberans ESU is found necessitates a proper management of the area. Plants used by P. exuberans as nectar sources suggest that traditional management may be sufficient to maintain a suitable habitat for this species, in a landscape where host plants are mainly distributed along dirt road verges. Onobrychis sp. represents a feeding source both for the larval and the adult stages of P. exuberans. Thus, it is suggested to take into account both the butterfly phenology, the presence of Onobrychis sp. and the ant diversity when planning ordinary maintenance work of the road verges in the area. To ensure minimal anthropic disturbance, it is preferable to cut the grass on road verges and on the bordering grassland-where Onobrychis sp. is found-shortly before the flight period starts (early July). Grass cutting in the autumn appears more problematic, since the overwintering of taxa of the P. ripartii group remains under-investigated. In this way, road verges could be good habitats (Phillips et al. 2020) for both oviposition and feeding and, at the same time, they might favour dispersion as possible corridors (Modin and Öckinger 2020). Grass cutting causes temporary habitat deterioration (Bauerfeind et al. 2009;Halbritter et al. 2015), while heavy grazing limits the growth and expansion of Onobrychis, (Lafranchis et al. 2007). A combination of mowing and grazing is still to be encouraged, since these practices, while keeping reforestation at low levels, are known to contribute to support healthy lepidopteran communities (Mouquet et al. 2005;Bauerfeind et al. 2009;Skórka et al. 2013), including P. ripartii (Lafranchis et al. 2007).
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.