Contrasting patterns of habitat use in a threatened carabid (Carabus intricatus) and a sympatric congener in ancient temperate rainforest

There is increasing concern into the decline of insect populations, with corresponding calls for conservation action aimed towards threatened species in particular. However, there is a distinct paucity of knowledge surrounding habitat requirements, microhabitat selection and conservation actions that may be undertaken for the vast majority of invertebrate species. Carabus intricatus (Coleoptera: Carabidae) Linnaeus, 1761 is a threatened ancient woodland specialist in the United Kingdom with a highly restricted distribution and is listed as a section 41 species of principal importance in England. Despite this, no empirical systematic evidence exists for even basic habitat requirements. Here, we used a patch‐occupancy modelling framework to determine occupancy and detection parameters for C. intricatus and a sympatric generalist species, C. problematicus, which is a potential competitor. Our models showed that decayed deadwood availability and leaf litter covering (an indicator of the availability of mature broadleaved trees) leads to higher occupancy of C. intricatus, whereas decayed dead wood availability and slug abundance are primary determinants of overall abundance. Additionally, detection models showed that higher humidity substantially increased activity of C. intricatus. For C. problematicus, ground moss coverage and overall deadwood volume most strongly influenced occupancy and abundance, whereas there were no humidity related influences on activity. The results support our general predictions about the specialist–generalist nature of the two species in the United Kingdom and indicate that key characteristics of old‐growth temperate rain forests, particularly the availability of deadwood and high humidity, are central to the conservation of these charismatic carabids.


INTRODUCTION
The global decline of insect populations (Didham et al., 2020;Wagner et al., 2021) has led to renewed calls for studies focussing on invertebrate conservation (Harvey et al., 2020;Samways et al., 2020).Conservation of species requires a knowledge of their fundamental ecology (Christie et al., 2020;Hortal et al., 2015), particularly the selection and utilisation of habitat, as this directly influences survival and reproductive success of individuals (Van Moorter et al., 2016).
Habitat selection is the choosing of a specific habitat type among a set of available types (Van Moorter et al., 2016) and is intrinsically linked to movement (Boyce et al., 2016).Habitat utilisation is the more fine-scale interaction with particular resources and features within a habitat to meet life history needs (Jones, 2001).Key drivers of habitat selection and utilisation that transcend taxonomic boundaries include the need for (i) food, (ii) shelter and (iii) breeding opportunities (Allen & Singh, 2016;Van Moorter et al., 2013).Habitat selection and utilisation may be restricted by habitat fragmentation, as key resources (e.g., larger trees or decaying deadwood) become scarcer towards the edges of habitats (Ruete et al., 2016).In severe cases, when habitat patches no longer provide necessary resources, populations may go extinct (Banks-Leite et al., 2012;Öckinger & Nilsson, 2010).Thus, understanding the key drivers of habitat selection and utilisation in the sites available to species and populations is paramount to any efforts to conserve them.Species of large, flightless carabids (Coleoptera: Carabidae) typically have poor dispersal abilities and slow life histories, so are especially vulnerable to the effects of habitat loss and fragmentation (Hansen et al., 2018;Keller et al., 2005;Völler et al., 2018).Furthermore, evidence suggests carabids may be important indicator species of habitat quality in their respective environments (Koivula, 2011).
Many species of Carabus are forest dwelling and may depend on large deadwood objects to complete their life cycles (Negro et al., 2017;Seibold et al., 2014).Some indicators of habitat quality such as logged forests, for example, have reduced abundance of specialist carabids (Negro et al., 2017), and it can take up to 10 years for forest species to be restored to pre-logging levels (Koivula et al. 2019), while intact habitat mosaics and high humidity benefit carabid abundance (Volf et al., 2018;Negro et al. 2014).However, despite their importance in making conservation decisions, for the vast majority of Carabus species, especially forest-dwelling species, very little knowledge on finescale habitat selection exists.Around 60% of European forest edges are bordered by intensive land use types (Estreguil et al., 2013) that increase the hardness of edges and exacerbate impacts associated with small, fragmented habitats.Ancient broadleaved woodlands typically have more stable microclimates (Norris et al., 2012) and high amounts of mosses (Mölder et al., 2015) and deadwood (Bouget et al., 2014;Bujoczek et al., 2021) compared with other forest types.et al., 2013;Hopkins & Kirby, 2007) and a reduction in the size of forest interior habitats (Marrec et al., 2020).
The Blue ground beetle Carabus intricatus (Figure 1) Linnaeus, 1761 is a large carabid (elytron length size range: 11.00-19.75mm, weight range: 0.4-1.2g, females generally $40% larger than males) listed as 'Near Threatened' on the IUCN (International Union for the Conservation of Nature) red list with local populations classified as endangered or of unknown status throughout its temperate European range.C. intricatus on the continent extend across much of the temperate zone, excluding the Mediterranean and Scandinavia (although there are scattered records in isolated places in the southeastern Mediterranean and southeastern Sweden).In many regions, the beetle is already locally extinct (e.g., the majority of Belgium, Wouter Dekoninck pers.comm.)owing to dramatic reductions in habitat and widespread fragmentation over the past century.In the United Kingdom, C. intricatus are known to favour ancient woodlands and wood pasture habitats and are thought to primarily feed on slugs, particularly tree slugs (family Limacidae) (Walters, 2019) with the main period of activity and breeding period in April, May and June.In the United Kingdom, the species is listed as a priority species for conser-  (Ellis, 2016).Less than 1% of global land surface is suitable for temperate rainforest and 40% of the available European space is in the United Kingdom due to the oceanic climate (DellaSala, 2011).
However, these woodlands are patchily distributed in the landscape.
C. intricatus are considered to be good indicator species for healthy ancient woodland habitats but all populations on Dartmoor are apparently small and fragmented.Plans to increase the connectivity and extent of these fragmented habitats in the wider Dartmoor landscape may utilise C. intricatus as an effective umbrella species for wider woodland recovery.However, little is known about how this species, which appears to be an ancient woodland specialist, utilises these habitats or what features and resources are required to support healthy populations.Such information is essential to make effective conservation decisions.
In the present study, we aim to elucidate key habitat utilisation parameters for C. intricatus to inform understanding of their conservation requirements.We use plot-based detection surveys with repeated visits, within a single-season occupancy modelling framework.Ancient woodlands can have high levels of moss and lichen species as heterogeneity in age and size of trees, as well as abundance of dead wood, increases (Hofmeister et al., 2015;Ódor et al., 2013).Moss is thought to be a major habitat feature utilised by C. intricatus (Walters, 2019), and lichens may also be important as they are a favoured food of tree slugs.Previous work indicates slugs are key prey items for C. intricatus but did not find evidence of a preference for particular slug guilds (ground vs. tree slugs; unpublished data).Furthermore, large Carabus species are known to demonstrate a degree of niche partitioning to avoid intraguild competition (e.g., Wehnert & Wagner, 2019).C. problematicus, the only other abundant sympatric congener, is of a very similar size and also consumes slugs ( Šeri c Jelaska et al., 2014) but is found in a wider range of habitats (Chetcuti et al., 2019;Luff et al., 2007) so appears to be more of a generalist than C. intricatus.Based on previous unpublished work (e.g., Walters, 2019), the information outlined above and more general ecological knowledge on similar species, we predicted that within habitat occurrence and abundance of C. intricatus would be positively associated with moss, broadleaved woodland cover, the availability of deadwood (for shelter and as sites for oviposition) and the abundance of slugs, but negatively associated with the amount of understorey and open areas (i.e., herbaceous vegetation, grass) and the presence of C. problematicus.If C. intricatus presence and abundance is unrelated to the presence of C. problematicus, which is the only other abundant large carabid at the study site, we expect some degree of niche partitioning to be evident with C. intricatus having more specific, specialist requirements than the more generalist C. problematicus.

Site characteristics
The study was conducted at Hawns & Dendles (H&D) woodland complex National Nature Reserve, a small ($76 ha), isolated fragment of ancient upland oak-beech woodland (UK grid reference SX614625; see Figure 1) in Dartmoor National Park in southwest Devon.C. intricatus are found in eight other sites on Dartmoor, five in Cornwall and one in south Wales; all sites are small with the three largest areas being 200-300 ha (Walters, 2019;Buglife www.buglife.org.uk).H&D is bordered to the north, east and west by grazed moorland and to the south is predominantly farmland interspersed with small villages and towns.The River Yealm runs through the middle of H&D and the woodland complex covers the slopes of the valley around the river, ranging from 160-to 280-m elevation.Temperatures may drop as low as 1 C through winter months (December-February) and reach up to 18 C during summer (Met Office, 2024) with average monthly rainfall of 66.8 mm ranging from 48.5 mm in July to 95.8 mm in December.
H&D is subject to light to moderate grazing by sheep and deer and is considered the main stronghold for C. intricatus in the United Kingdom.

Survey design and data collection
Fieldwork was scheduled when adult C. intricatus come out of hibernation and are night active (Luff et al., 2007), from April 14th through June 25th 2022.We conducted 360 beetle presence/ absence surveys across this time period on 120 individual 25 Â 25 m Latin square plots (n = 3 surveys per plot), with a mean number of surveys per night of 9.2 (min = 6, max = 13) (see S1 for detailed information on survey plot criteria and block design).Very little is known about home range size in Carabus spp, but previous studies have published total movement distances of various species in the genus.Movement distances (which were highly non-linear) ranged between 3 and 806 m but were more typically 55-170 m in total over a three-week study period for C. hungaricus and C. olympiae (Bérces & Ru ˚žičková, 2019;Negro et al., 2017).We used these measures as a proxy for home range size to determine an appropriate survey plot size of 625 m 2 .

Plot-level microhabitat data collection
Plot-level habitat data were collected during the day by conducting four microhabitat assessments in the centres of 10 Â 10 m grids at the corners of each of the 120 plots.We used a line intercept method for our microhabitat assessments (Sutherland, 2006), whereby two 2.5-m ropes were placed over the top of one another in a cross shape and marked every 10 cm.Every 10 cm, the dominant microhabitat (plot-level covariates sampled can be found in Table 1) was assessed by the observer so that each microhabitat assessment yields 50 points (per corner; 10 Â 10-m plot) and 200 points in total (per 25 Â 25-m plot).
Two of the microhabitat assessment corners (alternating pairs of corners, e.g., north-east and south-east or north-west and southwest), were also used as the sites for deadwood surveys.Due to the sheer volume of deadwood and the time taken to conduct deadwood surveys, only two surveys were carried out rather than four (as in the microhabitat assessments).During a deadwood survey, all coarse woody debris (hereafter CWD; deadwood with >5-cm diameter, representing a potential shelter resource) was measured for circumference at each end, as well as length (cm) and decay class of the respective piece (ranking from 1 to 5) according to Motta et al., 2006 (see S2 for full classification of deadwood decay classes).

Beetle survey data collection
Night surveys assessing presence/absence and abundance of both C. intricatus and C. problematicus in plots typically took place between 2100 and 0300 and from 12th April to 24th June 2022.Each of the 120 plots were systematically surveyed in transects covering about 5 m width until the plot had been entirely sampled.We also surveyed for slugs during this time, taking a count of all slugs encountered to use as a plot-level covariate for prey abundance.C. intricatus and C. problematicus encountered on surveys were measured (left elytron in mm using dial callipers) and sexed via visual inspection of the first pair of tarsi (males have diagnostic enlarged and flattened tarsi for gripping females; Luff et al., 2007), and the left middle leg was clipped from the femur downwards for later genetic work (fitness impacts were negligible in a trial with C. problematicus; Gaublomme et al., 2003).
Microclimate data were measured using two EL-USB-2 data loggers from 'Lascar electronics' placed in separate halves of the forest at ground level for the duration of the study period.Conditions (temperature and humidity) were used for each individual survey as detection level covariates (see below).

Principal components
Proportional habitat variables (herbaceous vegetation, grass, leaf litter, moss, other; see Table 1) and coarse woody debris variables (volume and average decay class) were combined to independent principal component axes for further analyses.We opted to use principal components to reduce cross-co-linearity (Abdi & Williams, 2010).We further centred, scaled and ordinated the data using varimax rotation, ensuring each of the variables would load strongly on only one or two axes.Varimax rotation simplifies interpretation of results whilst retaining the high amount of explained variance (Abdi & Williams, 2010) and is commonly used in ecological studies (e.g., Lee & Carroll, 2014;Ramesh et al., 2016).We selected the first four principal components to use for statistical analysis as they explained 84% of variation in the T A B L E 1 Plot-level covariates measured during microhabitat assessments at the plot-level (see above for methodological details).

Generalised linear models
We used generalised linear models (GLMs) to estimate the effect size and significance level of covariates (independent variables) on Carabus abundance (both C. intricatus and C. problematicus abundance as separate response variables).We first created a 'global model', a GLM containing all covariates, and used a dredging approach to assess the most parsimonious combination of those models ranked by AIC (Grueber et al., 2011).Additionally, we constrained the maximum number of terms to be included to three covariates to avoid overparameterization (L opez-Angulo et al., 2020) as dredging may otherwise lead to model non-convergence (Grueber et al., 2011).

Software and packages
All statistical analyses were conducted in R version 3.5.2(R Core Team 2013).We applied varimax rotation using the 'fa' and 'target.rot' functions in the 'psych' package (Revelle, 2023).We fitted occupancy models using the 'occu' function in package 'unmarked' (Fiske & Chandler, 2011)

Occupancy
Coefficients derived from the global occupancy models for C. intricatus (Table 4; also see Figure 3) show the strongest relationships between C. intricatus probability of occupancy and H4, H1, and SLUG.
AB.In ecological terms, we find a positive relationship between C.
intricatus probability of occupancy and decay class of CWD, slug and leaf litter abundance, respectively, and a negative relationship with herbaceous vegetation and grass covering (Table 4).Note: further results on the stepwise occupancy modelling approach are detailed in S4.
The occupancy models (S4) and coefficients (Table 4) for C. problematicus show a strong positive influence of H2 (moss and the volume of CWD), and variable inconclusive relationships with other covariates.

Detection
Both temperature and humidity had positive relationships with C.
intricatus detection probability, with humidity having a stronger relationship than temperature (Figure 3a,b).C. problematicus detection probability was weakly positively related to temperature, but there was no relationship with humidity (Figure 3c,d).

C. intricatus final model set
Based on the initial assessment of covariates relating to both probability of occupancy (see Section 3.1.and S4) and detection (see

Abundance
C. intricatus abundance was primarily influenced by slug abundance (SLUG.AB), followed by decay class of CWD (H4), both of which were positively related (Table 6 and Figure 4).Weak effects included a negative relationship with H1 (which indicates an avoidance of herbaceous vegetation and grass but a positive association with leaf litter; see Table 3).C. intricatus abundance was also weakly positively associated to the amount of moss within plots and the volume of CWD available (H2).The three most parsimonious models all contained SLUG.AB and H4 (Table 6).Further details can be found in S6.
C. problematicus abundance was primarily related to moss covering and the volume of CWD in plots (H2), which was the only variable in all four of the well-supported models (Table 6 and Figure 4).There

General findings
The probability of occupancy by C. intricatus in plots was best predicted by the availability of well-decayed deadwood (CWD), but there was also a substantial positive association with the amount of leaf litter (i.e., broadleaved woodland cover) and a negative relationship with

Importance of dead wood
For C. intricatus, the significant positive relationship with decay class of deadwood (CWD) is likely related to utilising the resource as shelter or refugia habitat during the day and when overwintering as seen in many carabid communities (Parisi et al., 2018).Decayed woody debris in woodland interior habitats provide a microcosm of selfregulating microclimatic conditions, such as a constant humidity and substrate (Pouska et al., 2016).In addition to being refugia, these microhabitats may also be important in the context of oviposition, as demonstrated in many other large carabid beetles (Lövei & Sunderland, 1996;Turin et al., 2003; also see Martikainen et al., 2006).Additionally, the decay class of CWD objects signifies a preference for older forest patches in which there has been enough time for substantial decay to take place.C. intricatus conservation therefore relies upon the maintenance of older forest patches (i.e., ancient woodlands) with varying stages of well-decayed deadwood.Removal of wood for biofuel, which reduces deadwood, can therefore have impacts on landscape scale habitat availability for saproxylic organisms (Bouget et al., 2012) and can negatively impact deadwood-dependent species overall (Verkerk et al., 2011).Whilst fuelwood removal is not common on Dartmoor, it has previously taken place around the United Kingdom (Green & Peterken, 1997), and deadwood removal for aesthetic reasons around the country is estimated to remove 9,000,000 m 3 of 'underbark' wood from UK forests annually (Forest Research, 2023).Decay class of CWD was not a significant predictor of C. problematicus abundance and had no association with C. problematicus occupancy which supports the idea that, in contrast to C. intricatus, this species is more of a forest generalist than an old-growth forest specialist (Fuller et al., 2008).

C. intricatus abundance is associated with slug abundance
Slug abundance had the strongest relationship with C. intricatus abundance, even more so than deadwood decay class (CWD).Numerous studies link ground beetle abundance, movement and aggregations to that of slug prey (e.g., Bohan et al., 2000;Jouveau et al., 2022;Negro et al., 2008;Symondson et al., 2002;Turin et al., 2003) Kappes et al., 2009), so habitat fragmentation may impact the availability of favoured prey for C. intricatus.Furthermore, slug abundance and communities are also influenced by decay class of CWD as sites for refuge and breeding (Kappes, 2005).Interestingly, C. problematicus abundance was influenced to some degree by slugs, but the relationship was non-significant, and the effect size was small, despite appearing in the final top model, likely reflecting its more generalist diet.

C. intricatus prefer high woodland cover and avoid open areas
Leaf litter (a proxy for broadleaved woodland cover) had a positive relationship with C. intricatus occupancy, whereas understorey and open areas (i.e., herbaceous vegetation and grass) had a negative relationship with C. intricatus occupancy.Forest specialist carabid beetles are known to actively avoid understorey and grassy areas (Heliölä et al., 2001;Niemelä, 2001); this habitat may impede movement and reduce their ability to detect and track prey.(Turin et al., 2003), which may explain its largely invariant response to changes in humidity.Drier and hotter conditions are expected to increase dramatically through the 21st century in Europe.The Alps are a stronghold of C. intricatus on the continent (Turin et al., 2003;GBIF, 2024) and are expected to suffer from increased drought stress and reduced humidity through the 21st century (Gobiet et al., 2014).Furthermore, dramatic shifts in tree species composition in forests owing to climatic shift are expected in the coming century (Lindner et al., 2014).

Implications for conservation management
Preservation of existing deadwood at a site is likely to be of particular importance for maintenance of C. intricatus populations.Newer forest stands could be subject to limited girdling/felling, and manual dispersion of deadwood could be explored as a management technique (Vítková et al., 2018), as this has been shown to be beneficial for a range of species ranging from woodpeckers (Aszal os et al., 2020) to solitary wasps and bees (Westerfelt et al., 2015), although the benefits may take years to be apparent (Gaston et al., 2005).Some further interventions to speed the process of deadwood development include creating rips in higher up trunks, ringbarking, clean-cutting high stumps (1.5 -2 m) and artificial rot boxes filled with various substrates, all of which can all promote diversity and provide habitat for specialist saproxylic invertebrates (Hotchkiss, 2020).
Central to the efforts to conserve both C. intricatus and C. problematicus is the preservation of intact forest patches with typical oldgrowth forest characteristics.Intact and 'set-aside' forest patches provide a wealth of deadwood (Bouget et al., 2014) and humid environments necessary for both well-decayed deadwood and larger amounts of moss covering.These characteristics typically also promote forest slug abundances (De Smedt et al., 2019) and communities (Kappes et al., 2009).Furthermore, light grazing by large herbivores such as deer and sheep can have a suppressant effect on understorey vegetation (Öllerer et al., 2019) and is commonly cited as a positive management strategy for C. intricatus populations (Walters, 2019).

CONCLUSION
Our results indicated that decayed deadwood, leaf litter and wood- Thus, organisms specialised to ancient forest systems are at greater risk of extinction in the face of increased forest turnover (Estreguil F I G U R E 1 Top left: Wild mating pair of Carabus intricatus (photo credit: Nick Royle); bottom left: Devon and Cornwall boundary (light blue), Dartmoor National Park boundary (orange); right: Hawns and Dendles wood (orange).Maps made in QGIS v3.28.5 using ESRI (Environmental Systems Research Institute) satellite images as underlay.

F
I G U R E 2 Response plots of Carabus intricatus probability of occurrence in relation to selected covariates.Slug abundance and C. problematicus abundance have been standardised.Shaded areas represent the margin of SE.created univariate occupancy models to assess strength of relationships, then used a forward stepwise selection approach to arrive at a model set of four multivariate models, six univariate models and the null model for a grand total of 11 models.We constrained the maximum number of terms to be included to three covariates to avoid overparameterization (e.g., L opez-Angulo et al., 2020).We used a logit link function to estimate ß coefficients (effect size) from all covariates from the global occupancy model.All models were compared and ranked by ΔAIC (calculated by subtracting AIC (Akaike's Information Criterion) of model Â from the AIC of the top model; MacKenzie, 2006).Following the final model set, detection parameters were incorporated to create a separate four top-ranked model set that explained the maximum amount of variation in the dataset (detailed information on the occupancy modelling approach can be found in S3).
and compared them using the 'fitList' function of the same package.We assessed model fit between our top-ranked model and the reference distributions by comparing the Chisq statistic, using the 'MacKenzie and Bailey goodness of fit test'.The function 'mb.gof.test' in the package 'AICcmodavg' (Mazerolle, 2019), with 1000 simulations, showed that there was no indication of a lack of fit (p > 0.05).MacKenzie (2006) note that occupancy modelling without adequate tests of model fit risk weak inferences.We used the 'glm' function of the 'stats' package (R Core Team, 2013) to create our global model for abundance and the 'm.lim' and 'dredge' functions of the 'MuMIn' package (Barto n, 2023) to constrain the maximum number of terms and run each combination of suitable models thereof.

F
I G U R E 3 Response plots of Carabus intricatus (a,b) and C. problematicus (c,d) detection in relation to microclimate variables.Values are standardised.Obs -observed.were three additional weak effects, with greater numbers of C. problematicus found where there was greater slug abundance (SLUG.AB), but a negative relationship with H3 (i.e.areas with lots of bare ground and other uncategorized habitat variables and a marginally positive relationship with decay class of CWD).There were four wellsupported models for C. problematicus, with little between them in terms of overall model support (
grass and herbaceous vegetation (i.e., areas with extensive understorey or open, grassy areas).The abundance of C. intricatus was primarily predicted by the abundance of slugs and the availability of well-decayed deadwood.Consequently, we found support for most of our initial predictions.We also found clear evidence of niche separation with C. problematicus both in terms of their occupancy of habitat (spatial niche) and activity patterns (temporal niche), as C. intricatus activity was positively related to humidity but C. problematicus occupancy was independent of it.However, there was no evidence that either occupancy or abundance of C. intricatus was directly related to abundance of C. problematicus, and, although ground moss cover and the volume of available deadwood (H2) was weakly positively related to C. intricatus abundance, it was not related to the probability of C. intricatus occupancy.These results demonstrate the central importance of welldecayed deadwood for both occupancy and abundance of C. intricatus within forest habitats.The strong positive effect of broadleaved woodland cover, avoidance of more open areas and increased activity on humid nights show that C. intricatus is strongly associated with key features of old-growth temperate rainforest habitat.Fragmentation and associated edge effects (e.g., on microclimate) combined with reduction of availability of deadwood are therefore likely to have substantial negative impacts on populations of this large, flightless, slow life-history carabid.
land cover substantially improve the occupancy probability of C. intricatus, whereas slug abundance and availability of decayed deadwood influences population abundance and carrying capacity.There is niche separation, both spatially and temporally, with a generalist sympatric congener, C. problematicus.These findings offer important insights into factors governing both occupancy and abundance of C. intricatus in the United Kingdom, provide the first systematic study into the species globally and demonstrate the importance of temperate rainforest as a habitat for C. intricatus and other saproxylic species of invertebrates.
vation (Department for Environment Food & Rural Affairs, 2022), with the majority of known sites on Dartmoor in Devon.Dartmoor features a number of ancient woodlands that are examples of 'temperate Atlantic rainforest' habitat.Temperate rainforest is typified by epiphytes and generally contains distinct assemblages of lichens and bryophytes Loadings for the seven covariates included in the four H components used for further analysis.
Abdi & Williams, 2010ntly these were transformed into four meaningful varimaxed components (i.e., SS loading >1.00;Abdi & Williams, 2010; see Tables2 and 3for loadings) which we refer to as H1, H2, H3 and H4 (habitat components).H1 is characterised by high amounts of herbaceous vegetation and grass covering and low amounts of leaf litter.H2 is characterised by plots with high amounts of moss and large pieces of course woody debris.H3 represents 'other' substrate, such as bare ground, whereas H4 is characterised by well-decayed course woody debris.Occupancy modelsWe used single-season occupancy models described in MacKenzie (2006) to estimate occupancy (Ψ ) parameters and covariate effects on those parameters.Occupancy models also incorporate the possibility of false absences by using detection probabilities (p), which if unaccounted for can alter occupancy estimates to a large degree (MacKenzie, 2006; Figure 2).Relative humidity (%) and temperature ( C) at time of survey were our detection level covariates.We initially T A B L E 3 Note: (i) Minimum values are not presented for covariates as all are 0; (ii) units for the unspecified covariates are proportion of assessed area.Mean and maximum values for each covariate are included.
Estimates of ß coefficients from the global occupancy models for each of Carabus intricatus and Carabus problematicus.
Section 3.2.and S5), we constructed a final model set for C. intricatus with variations of the top-ranked models for each parameter related, and H1 herbaceous vegetation and grass coverage (negatively related) appear to be of primary importance in predicting C. intricatus occupancy, with slug abundance being secondary.Related to this, humidity is a more important predictor of detection probability (activity) than temperature, that is, occupancy is more likely to be detected when humidity is high.

Table 6
Best supported abundance GLMs from dredging, ranked by AIC for Carabus intricatus (uppermost model set) and C. problematicus.Bold values indicate well-supported models with ΔAIC < 2.
).Further details can be found in S6.T A B L E 5 Final model set for occupancy and detection parameters for Carabus intricatus.Bold values indicate well-supported models with ΔAIC < 2. Ψ (H1 + H4 + SLUG.AB) p (Humidity + Temperature) Note: Models with strong support in bold.Note: the fourth model was the final model with ΔAIC <2 for C. problematicus.Abbreviation: GLM, generalised linear model.
while gut content analysis has shown that a wide variety of large carabid species predate on slugs Šeri c Jelaska et al. (2014).However, guts of six C. intricatus from Croatia contained only Lepidoptera and earthworms in the diet ( Šeri c Jelaska et al., 2014).Gut content analysis was beyond the scope of this study, but we nevertheless found that C. intricatus . problematicus occupy microhabitats differently to C. intricatus For C. problematicus, ground moss covering and the overall volume of coarse woody debris (H2) were the strongest positive predictors of both occupancy and abundance.Our results support those of Kriegel et al. (2021), who also found C. problematicus to be most strongly associated with volume of deadwood in a study in Germany.How-The more restricted activity patterns of C.intricatus provides further evidence that they are 'forest specialists', and, more particularly, specialists of temperate rainforests (within the United Kingdom), with a requirement for wet, forest habitats, potentially to facilitate hunting of slug prey.In contrast, as the 'forest generalist', C. problematicus is known to have a longer period of seasonal activity than C. intricatus and be present in a wider variety of habitats across most of its European range C. intricatus populations, as leaf litter (and by extension broadleaved woodland cover) increases.Cabundance) with the other species.No significant effects were found.However, we did find support for niche partitioning and differential habitat use in C. intricatus and C. problematicus.This is not surprising but provides an important layer of ecological understanding to the fundamental niche of C. intricatus.Increased competition through range expansion of a generalist competitor (C.depressus) is predicted to impact the decline of the threatened specialist C. cychroides in the Cottian Alps (Anselmo & Rizzioli, 2022).Conversely, four Carabus species demonstrated clear niche separation in Brandenburg, Germany, with different levels of specificity towards various trees, or, in the case of C. arvensis, an entirely different functional ecology (an ability to climb tree stems and operate in vertical space) (Wehnert & Wagner, 2019), which may also apply to C. intricatus.Thus, due to the natural sympatry exhibited by C. intricatus and C. problematicus, it is tus, and a very strong relationship was found for moss covering and volume of CWD for C. problematicus.A preference for different elements of habitat and resources facilitates coexistence of these similar sized sympatric species.Interspecific competition is perhaps therefore unlikely to be of concern to C. intricatus populations in the United Kingdom, except in instances of novel competitors invading habitats through climatic shift.Novel competition through invasive insect generalists can have detrimental effects on natural range specialists (Crowder & Snyder, 2010).But, as found by Wehnert and Wagner (2019), naturally occurring sympatric congeners often have differing niches facilitating coexistence.C. intricatus have more restricted microclimatic niches than C. problematicus