Butterflies are widely recognized as effective indicators of environmental conditions, owing to their close ecological associations with vegetation, their sensitivity to habitat quality, and their rapid demographic responses to environmental changes (Horn 2003; Poniatoswki et al. 2018; Pallottini et al. 2023). Their dependence on specific resources at different life stages, combined with relatively short life cycles, makes them particularly suitable for detecting both natural and anthropogenic alterations in ecosystems (Chowdhury et al. 2023). Consequently, long-term trends in butterfly populations are increasingly used to infer changes in biodiversity, ecosystem functioning, and landscape integrity at different spatial scales ranging from site-specific to population, country and continent-wide extents (Gerlach et al. 2013; Ghazanfar et al. 2016; van Swaay et al. 2025).

Across Europe, widespread declines in butterfly populations have been documented over recent decades, also involving widespread and once abundant species, raising concern among conservation scientists (Warren et al. 2021; Habel et al. 2022). Habitat loss and fragmentation, climate change, agricultural intensification, urban expansion, and the abandonment of traditional land-use practices have been identified as major drivers of these declines (Thomas 1995; van Swaay et al. 2010; Balletto et al. 2015). Species with narrow ecological requirements are particularly vulnerable, as even subtle changes in habitat structure or resource availability can have disproportionate effects on their population viability (Nakonieczny et al. 2007). For this reason, several butterfly species are listed in Annex IV or II of the Habitats Directive (HD, Council Directive 92/43/EEC), i.e. they are recognized as taxa of European-wide conservation relevance, thus requiring strict protection and timely monitoring (Warren et al. 2021). Within this framework, assessing, monitoring and reporting the conservation status of listed species is a legal obligation for all the EU member states (HD, Art.11 and 17), and an imperative challenge for the scientific community (Bonelli et al. 2018, 2021). Guidelines clearly indicate that effective monitoring should not be limited to presence–absence data but should incorporate demographic and reproductive indicators capable of revealing early signs of population decline (van Swaay et al. 2012). Parameters such as adult abundance, oviposition rates, larval occurrence, and early survival provide essential information on habitat suitability and population performance (e.g., Scherer et al. 2025), especially in fragmented and human-modified landscapes. However, despite their importance, such demographic data remain scarce for many strictly protected butterfly species due to limited resources that impose efficient, timesaving, and yet consistent indicators to be monitored (Bonelli et al. 2018).

Butterfly demography is shaped by multiple environmental drivers whose effects are often species-specific and mediated by life-history traits (Zografou et al. 2021; Martin et al. 2025). Host plant availability is a fundamental constraint on oviposition and larval development, while nectar resource abundance primarily affects adult foraging and persistence (Melero et al. 2016; Essens et al. 2017). Habitat structure and microclimatic conditions influence behaviour, survival, and reproductive success, whereas landscape composition and configuration may affect dispersal, connectivity, and population stability (Alves et al. 2024). In addition to this, female specific egg-laying site selection can also be influenced by an energetic assessment of the plant’s nutritional value, as well as the potential for both intraspecific (within-species, e.g., due to the occurrence of other female eggs or larvae) and interspecific (between-species, e.g., active behavioural exclusion by territorial males) competitive interactions at the site, or by the presence of predators and parasitoids (Shiojiri et al. 2015). Importantly, these drivers may simultaneously act at different spatial scales and influence distinct life stages, leading to complex and sometimes decoupled responses across the butterfly life cycle (Dover and Settele 2009; Hill et al. 2021). As an example, very local-scale conditions, such as microclimate, host-plant abundance, or duration of direct sunlight illumination, may significantly vary in very few meters, and their effects widely differ in comparison to landscape-scale conditions (acting at hundreds or thousands of meters scales), such as the overall habitat type in which a population occurs, including agricultural and urban areas. Given this, both scales may significantly shape butterfly demography by altering different phases of e.g., reproduction, such as oviposition and larval survival (local scale; Rabasa et al. 2005) or dispersal and colonisation (landscape scale; Dover and Settele 2009).

Despite the acknowledged importance of these factors, their relative contributions to population performance remain poorly understood for many protected species (Cini et al. 2020; Warren et al. 2021). In particular, the role of disturbance regimes and habitat heterogeneity is often overlooked or treated as uniformly negative, despite growing evidence that intermediate levels of disturbance can enhance habitat suitability for certain specialist insects (Stefanescu et al. 2011; Bubovà et al. 2015). Similarly, the extent to which butterflies, and particularly highly specialized and protected species, can persist and reproduce in anthropogenic or urbanized landscapes remains a key open question for conservation biology and green space management (Wood and Pullin 2002; Ramírez-Restrepo and MacGregor-Fors 2017). Addressing these knowledge gaps requires field studies designed to disentangle the effects of multiple, interacting environmental drivers across life stages and along ecological gradients (Konvicka and Kadlec 2011; Kaiser et al. 2016; Dennis et al. 2017).

In this study, we adopted a field approach to investigate the environmental drivers shaping the demography of a strictly protected butterfly species across heterogeneous landscapes. By combining data from natural, semi-natural, and anthropogenic environments, and by simultaneously assessing adult abundance, reproductive output, and early developmental stages, we aim to identify the key factors influencing population indicators and evaluate their conservation implications. Among species potentially suitable for tackling the presented topic, we selected the Italian festoon Zerynthia cassandra as a model. This species is in fact an excellent representative of highly specialized butterfly species that though can occur also in anthropogenic habitats, intertwining subtle ecological connections (e.g., Labadessa and Ancillotto 2023) and thus being disproportionately threatened by environmental human-induced alterations (Bonelli et al. 2018).

Specifically, we tested the following hypotheses and associated predictions: (1) larger habitat patches may provide larger amounts of key resources and support greater population performance, thus we predict that demography indicators will increase along with the surface of surrounding natural areas (Blomfield 2021); (2) resource availability affects different life stages in distinct ways, so that we predict that nectar (flower) availability positively influences adult abundance but has limited effects on reproductive indicators, while host plant abundance will positively affect oviposition and larval presence, but not necessarily adult abundance (see Fowler et al. 2016); (3) moderate levels of disturbance enhance habitat suitability to a butterfly specialized upon margins and ecotones, so that disturbed sites will show higher values of demographic indicators at intermediate values of either local and landscape disturbance pressures (Cardoso et al. 2013; Attiwilli et al. 2022).

By linking environmental drivers to demographic responses across life stages, we ultimately aim to provide a mechanistic understanding of habitat suitability for a protected butterfly species. In doing so, we seek to generate evidence-based insights to inform adaptive conservation and management strategies under the Habitats Directive, particularly in landscapes increasingly shaped by human activities and environmental change such as agricultural and urban areas.

Study species

We selected the Italian festoon (Zerynthia cassandra Geyer, 1828) as model species, as it is a EU-wide protected papilionid butterfly species; the species was only recently recognized after a taxonomic split from the protected Z. polyxena (Dapporto 2010), thus resulting protected under the Habitats Directive with a unique identification code (Z. cassandra: 6943; Z. polyxena: 1053), as also reported by the V Report (2019–2024) ex Art.17 Habitats Directive (ISPRA, 2025). As such, specific and standardised protocols have been proposed for monitoring the status of its populations (Trizzino et al. 2013; Bonelli et al. 2016), including indicators of adult, egg and larvae abundances, as well as availability of oviposition habitat. Zerynthia cassandra is considered a highly specialized species due to its close and strict dependency upon herbaceous, annual and non-palatable plant species belonging to the Aristolochia genus L. which represent its only larval host plants (Labadessa and Ancillotto 2023). This given, and due to its relatively low mobility, it is deemed as vulnerable to habitat loss and fragmentation (Vovlas et al. 2014). It is an early flying univoltine species that actively selects suitable habitat patches for oviposition, according to a set of environmental factors, favouring sunny patches of host plants (Cini et al. 2021) and being apparently favoured by local-scale soil disturbance, e.g. wild boar rooting (Labadessa and Ancillotto 2023). The species, endemic to central and southern Italy, is still widespread and relatively abundant, so that it is listed as Least Concern in the Italian butterflies IUCN Red Lists (Balletto et al. 2015; Bonelli et al. 2018), even though its populations are often small and restricted to small patches of suitable habitats, and several of them have declined or gone locally extinct in the last decades (Bonelli et al. 2018). Despite being a microhabitat specialist and featuring biological traits that make it highly susceptible to habitat changes, e.g., univoltinism and host plant specialisation (Essens et al. 2017), Z. cassandra is able to persist even within deeply modified landscapes such as agroecosystems and urban areas, provided that such populations are ecologically connected (Ancillotto et al. 2024). All these factors make the timely monitoring of Z. cassandra populations a key conservation activity; besides, connecting habitat characteristics to population performance is a highly needed information to foster active conservation actions (Ghesini et al. 2019).

Study areas and sampling protocol

We conducted our sampling campaign in spring (March to June) 2024 and 2025 at 10 study areas in central Italy (Fig. 1a-b); areas were selected based on previous knowledge on the species’ presence due to past research (e.g., Ancillotto et al. 2024) and monitoring activities by protected areas. Overall, the areas selected cover most of the ecological contexts where the Z. cassandra is usually observed i.e., mosaic agricultural landscapes, broadleaf woodlands, and grassland-forest margins, both in natural and urban areas, thus providing an excellent overview of the ecological needs and plasticity of the species (Stoch and Genovesi 2016). Within each area, 1 or 2 transects on pre-existing trails or path were identified by applying the following criteria: (i) crossing potentially profitable habitat types based on our knowledge on the species’ needs (Cini et al. 2021), (ii) representative of the general landscape present in the area, (iii) in proximity to previous records of Z. cassandra by colleagues or area management staff. This given, we overall surveyed 13 transects, totalling 112 transect sections.

Adults were sampled along 200, 500–1000 m long transects, subdivided in sections of 100 m, as this is a widely used and standardized method for carrying out repeatable quantitative sampling of butterflies (Barkmann et al. 2023). Data was registered following the same protocol during all study years, as described by Pollard and Yates (1993), crossing each section in approximately 3 min, recording flying or resting adults sighted 5 m in front and 2.5 m to the sides of the observer and reversing the direction of travel at each visit. The adult surveys were conducted from the first week of March to the second week of May, covering each transect between one and three times a week, during the hottest hours of the day (i.e., approximately between 10:00 and 15:00). Even though weekly visits varied along the season, sampling effort across transects remained constant, resulting in the same numbers of visits for all transects. The time window during which the visits were conducted was selected based upon observations during the previous years in order to cover both the onset and the end of the species local flying period. At each visit, the following environmental variables were recorded: temperature and humidity, cloud cover (categorized into the ranges 0%-25% cloud cover, 26%-50%, 51%-75%, 76%-100%), and wind speed according to the Beaufort scale (Mather 2005).

The abundance of host plants, eggs, and larvae was also recorded between March and June, to identify the seasonal appearance and disappearance of the Aristolochia plants and the peak count period for each category. In particular, host plants were recorded once every two visits, while eggs and larvae weekly, from the second week of March to the second week of June (when all sites featured N larvae = 0). Plant abundance was measured by counting the numbers of visible stems/shoots within each section buffer area (as in Labadessa and Ancillotto 2023), thus excluding all smaller stems (i.e., height usually < 10 cm from the ground) covered by larger ones. Eggs and larvae were visually counted by checking on all the Aristolochia plants through gently manipulation of stems to inspect apical parts, the underside of larger leaves, and flowers, within the buffer zone surrounding each transect section. All eggs and larvae were counted, regardless of their maturity stage. The number of plants occupied by eggs and larvae was also recorded at each egg/larval count.

Environmental features were also recorded either during sampling or a posteriori, and namely:

1. 1.

   Size of available natural and seminatural habitat (as absolute area values), approximated as the surface (in hectares) of uninterrupted natural and semi-natural vegetation surrounding a transect section, as assessed from photointerpretation of satellite images (from Google Earth Pro); this procedure excluded all impervious surfaces, agricultural areas, and bare soil surfaces, as these are all considered unsuitable and poorly permeable to dispersal by our target species (Ancillotto et al. 2024).

2. 2.

   Host plant abundance, as the numbers of Aristolochia stems in a 2-meter buffers surrounding the transect line of each section (as in Bonelli et al. 2016).

3. 3.

   Flower Index, as a proxy of nectar resources availability, visually estimated once per transect, between mid-April and mid-May (i.e., peak flight period of Z. cassandra), along a 2 m buffer surrounding each transect section; ranked as follows: 0 = no flowers visible, 1 = 1–10% of cover by flowering plants, 2 = 11–20%, 3 = 21–30%, 4 = 31–40%, 5 = 41–100% of visible vegetation cover). Only entomophilous flowering plants were considered.

4. 4.

   Rooting index, as the percent soil with visible wild boar rooting activity surrounding each transect section, assessed by following Ferretti et al. (2021)

5. 5.

   Landscape class, classified as either natural/seminatural and urban environment surrounding the transect; classification was done a priori, with urban transects (N = 6) were those included within dense urban matrix, while natural/semi-natural were those inside protected areas outside of any urban settlement (N = 7).

6. 6.

   Habitat type, classified as either woodland, margin, or open grassland, according to the main habitat type present at the given transect section; more specifically, sections at < 10 m from habitat edge were considered as margin.

The two latter variables were meant to describe butterfly habitat at a wider scale (i.e., landscape), by defining the ecological context in which each transect occurred.

Table 1 summarizes the general characteristics of all study areas, including main habitat type, altitude, and whether they are legally protected.

Statistical analyses

We first assessed whether Z. cassandra population indicators (maximum and total numbers of adults, eggs, and larvae) correlated with each other, retaining only independent indicators for adult and eggs/larval stages (Green 1979; Sokal and Rohlf 2012). This resulted in a strong correlation between total and maximum numbers of adult butterflies observed per transect section (r: 0.93, p < 0.001), between the total and maximum numbers of eggs (r: 0.97, p < 0.001), and between maximum numbers of eggs and total numbers of larvae (r: 0.84, p < 0.01). As such, we used the total numbers of adult butterflies observed, the maximum numbers of eggs, and the maximum numbers of larvae as species’ indicators. Moreover, the short maximum time interval between consecutive visits (1 week) may have led to double counts of eggs and larvae, so that maximum numbers recorded at a given visit likely provide a more genuine indicator. Additionally, we also calculated host plant occupation ratio per section, as the numbers of stems featuring eggs/larvae of Z. cassandra upon the total numbers of available stems, and the apparent egg survival as the maximum numbers of larvae upon the maximum numbers of eggs. We checked for collinearity among all the numerical environmental variables measured along transects, at the transect-section scale, resulting in no significant relationship and thus the retaining of all the variables. We then checked for spatial autocorrelation among variable values across transect sections by computing a Moran I test using the DHARMA package for R (Hartig and Hartig 2017); all tests resulted as non-significant (p > 0.05), thus confirming the absence of spatial pattern in value distribution. All numerical variables were preliminary rescaled by using the function rescale as built in the scales package for R (Wickham et al. 2016), to range in value between 0 and 100.

For each response variable (i.e., demographic indicator of Z. cassandra), we run a generalized additive mixed model (GAMM; Zuur et al. 2009) including all the selected explaining variables, and adopting a smoothing spline value of 6 for variables expected to have non-linear effects upon butterfly indicators e.g., due to local-scale responses of butterflies such as saturation or behavioural interactions (Oostermejer and Van Swaay 1998), and namely for host plant abundance, such as Flower and Rooting indexes. GAMMs were chosen as an approach that may be better suited than others (e.g., GLMMs), for testing the effects of environmental variables upon biological processes when there is no previous information upon their relationships, by allowing to test for data-driven curve estimations. In addition, since we analysed data at transect section scale, we included transect ID as random factor, to take into account potential differences due to transect-specific environmental conditions. From each full model, we inspected the significance of each variable by assessing p-value (< 0.05), and whether the 95% confidence interval around the estimate encompassed 0 or not (Zuur et al. 2009). For categorical variables with > 2 classes (N = 1), post-hoc contrasts were assessed by using Tukey’s post hoc test with Bonferroni correction for multiple comparisons, to identify significant differences among classes. All analyses were performed with R 4.5.2 (R Core Team 2025).

A total of 122 observations of adult butterflies were performed during the study (ranging in numbers per visit per section from 1 to 6); we counted 4,168 eggs (range per visit: N = 1-706), and 523 larvae (range per visit: N = 1–51). The main host plant species at all sites was Aristolochia rotunda, yet 13 (14.6%) sections featured A. lutea alone, or in syntopy with A. rotunda, and one (1.1%) site featured A. clematitis in syntopy with A. rotunda. In all cases, eggs and larvae of Z. cassandra were observed upon all the Aristolochia species found at each site. Most sites featured the occurrence of wild boars to some extent, and 43.8% of transect sections were subjected to rooting (between 5 and 35 of Rooting Index value). We counted 7,477 Aristolochia stems across all transect sections, and host plant abundance per 100 m section ranged between 0 (11.4% of sections) and 1825 stems (on average: 176.3).

Overall, our models proved moderately to highly effective in explaining the variance of the considered response variables (R-squared ranging between 0.64 and 0.84; Table 2). Flower index, as expected, significantly influenced only the numbers of observed adult Z. cassandra. Wild boar rooting activity had a significant and consistently positive effect on the numbers of observed adults and maximum numbers of larvae and increased apparent egg success. The local abundance of Aristolochia stems influenced the total numbers of adults, eggs and of larvae per transect section, in all cases showing a positive effect. Habitat type also showed significant effects on some of the considered indicators, with transect sections at open and margin habitats featuring higher numbers of adults and higher occupation rates. Lastly, landscape class and area size had no effect upon any of the demographic indicators we measured (Table 2; Fig. 2). Both host plant abundance and rooting index showed the same response pattern by adult butterflies (both variables) and by larvae and egg success (rooting only), showing a bimodal relationship: namely, adult butterflies peaked at low and high values of host plant abundance, quickly reaching a low plateau at intermediate values (Fig. 2), while all indicators responded to rooting with an initial peak (at values > 0) and again at highest values of rooted soil, a pattern especially evident in the case of egg counts.

We identified which local habitat features, disturbance regimes, and landscape contexts shape the demography of a strictly protected butterfly species. By combining indicators of adult activity, reproductive output, and early survival, our results provide a life-stage–integrated perspective that is particularly relevant for conserving Z. cassandra, and to monitor the species under the obligations of the Habitats Directive (Council Directive 92/43/EEC), in both natural and anthropogenic landscapes (see Stoch and Genovesi 2016).

Contrary to our first hypothesis, larger areas of profitable natural and semi-natural habitats did not necessarily translate in higher demographic indicator values for Z. cassandra. The lack of relationships between Z. cassandra demographic indicators and size of the natural and semi-natural areas surrounding our sampling sites is a crucial finding, indicating that even smallest patches of suitable habitats surrounded by inhospitable matrix, may sustain relatively large or at least vital populations (Angold et al. 2006). Our sites had sections surrounded by highly variable amounts of potentially profitable habitats (range: 0.5-5,00 ha), thus consistently picturing the variability of conditions where the species may be found, ranging from isolated small patches of remnant habitat to extensive and relatively pristine natural areas. The comparability of population performance among different levels of habitat availability adds to the increasingly recognized importance of small natural patches in anthropogenic landscapes (Tulloch et al. 2016; Lindenmayer 2019), and highlights that such small populations should be carefully considered and monitored in conservation assessments, especially when falling outside of any network of protected areas, e.g., Natura2000 sites (Kajzer-Bonk and Nowicki 2022). As an example, the three sites we monitored in the outskirts of Firenze (Tuscany) are all small (between 0.4 and 0.9 ha) remnants of natural vegetation close to wetlands within a densely built-up and intensively cultivated landscape (Ancillotto et al. 2025), none of which is legally protected yet. As such, these limited patchy sites, despite featuring population values of Z. cassandra comparable to populations in more natural and larger areas, are at disproportionately higher risk of being affected by human activities like agricultural expansion and urban development, thus representing a conservation emergency for the species (Ancillotto et al. 2025). It is likely that conditions such as small populations in agricultural landscapes and urban or suburban parks are more common than expected for Z. cassandra (Ghesini et al. 2018), and possibly other butterfly species, making our results even more relevant for a general re-evaluation of these habitats and their importance to butterfly conservation. In line with our second hypothesis, environmental characteristics influenced different life stages of our target butterfly species, highlighting a synergistic effect of different drivers upon specific phases of butterfly life cycle. Flower presence, representing nectar availability, positively influenced adult abundance, confirming its role in sustaining adult activity and persistence independently from oviposition opportunities, particularly in heterogeneous or anthropogenic landscapes (Cole et al. 2017). Availability of nectar resources is key not only to sustain flight of adult butterflies, but also to boost individual fertility (King and Schultz 2024). Nonetheless, host plant abundance strongly influenced numbers of observed adults too, besides increasing egg and larval numbers, reinforcing its central role in regulating reproductive output (Dennis et al. 2004); in fact, frequenting areas rich in Aristolochia spp. could allow adults of a species with a short flight period, such as Z. cassandra, to optimize the timing of the entire reproductive process (e.g., courtship, mating, searching for an egg-laying site, etc.: Wickman 2009). As reported in other studies, A. rotunda proved to be the primary host plant species for Z. cassandra in our study sites, followed by A. lutea, with one site also featuring the use of A. clematitis, which is usually avoided (see also Ghesini et al. 2019); the species is also known to use A. clusii and A. pallida, which though are not present in the area (Bonelli et al. 2016; Labadessa and Ancillotto 2023). The lack of any effect of host plant abundance on occupation rates and on egg apparent success suggests a partial decoupling between adult oviposition preferences and reproductive outputs, emphasizing the need to incorporate life-stage–specific and population fitness indicators into conservation assessments. Namely, large patches of Aristolochia spp. may prove attractive to adult Z. cassandra, and yet not sustain high numbers of eggs and larvae to their final stages, not offering suitable microenvironmental conditions (Griese et al. 2020), i.e. potentially representing ecological traps (Ghesini et al. 2019). As an example, highly accessible patches of host plants in restored wetlands of Oregon are extremely attractive to the locally rare Lycaena xanthoides (Boisduval, 1852), despite leading to very low egg and larval survival (Severns 2011). As such, habitat quality for Z. cassandra should be evaluated not only on the abundance of Aristolochia plants, but also on the availability of nectar resources to adults. Moreover, the plateau rapidly reached by numbers of adults at increasing abundances of Aristolochia stems may also be explained by behavioural constraints. In fact, males of territorial species (e.g., Z. cassandra, pers. obs.), may displace competitors or even mated females from profitable host plant patches, thus limiting the numbers of flying individuals, as observed in other species that feature territorial behaviour (Bergman et al. 2007).

Finally, our results also add to the increasing evidence that disturbance, at least at moderate levels, can enhance habitat suitability across multiple life stages of some butterfly species (Swart et al. 2016), thus supporting our third hypothesis. Specifically, wild boar rooting showed a consistently positive effect on adult abundance, egg and larval numbers, and apparent egg survival (as in Scherer et al. 2025; and de Schaetzen et al. 2018). These findings align with previous studies that indicate higher oviposition probability by Z. cassandra at boar-rooted patches (Labadessa and Ancillotto 2023) and yet challenge the widespread perception of wild boar activity as uniformly detrimental to biodiversity (Barrios-Garcia and Ballari 2012). Intermediate levels of soil disturbance may instead promote conditions favourable to specialist butterflies, albeit the mechanisms underlying the relationship between the demography of Z. cassandra and wild boar rooting are still to be clarified. A potential explanation to this pattern is that rooting clears relatively limited soil patches from vegetation (Sims et al. 2014) during the excavation by boars searching for food (invertebrates and underground plant parts); as such, toxic plants like Aristolochia are generally avoided by rooting boars. Consequently, Aristolochia plants are left untouched by wild boar (Labadessa and Ancillotto 2023), while competitor plant species are cleared from the surroundings. This given, host plants in rooted areas are actually less cluttered by vegetation and thus more detectable and accessible to butterflies, while also receiving more direct and prolonged sunlight and thus providing better conditions for egg and larval development, as also seen in other case studies (Cini et al. 2020; Nagaya et al. 2021). From a conservation perspective, these results support the concept of disturbance-dependent habitat management, where certain forms of physical disturbance can maintain early-successional conditions that would otherwise be lost through vegetation closure (Sandoval et al. 2019). Encroachment by shrubs and trees is a key driver of habitat loss and local extinction for butterfly species dependent upon margins and open habitats, even in protected areas; as an example, the protected Parnassius mnemosyne (Linnaeus, 1758) has dramatically declined in a National park of the central Apennines in recent years, likely due to forest recolonization of former pastures and meadows (Cini et al. 2021). It is important to emphasise that the positive effects observed in this case can hardly be generalised to contexts with chronic or high-intensity disturbances, and that our transects featured relatively low maximum levels of wild boar rooting. Conservation strategies should therefore aim to maintain habitat disturbance within a functional range, avoiding both excessive degradation and complete suppression. In landscapes where traditional practices such as grazing or mowing have declined, wild boar activity may partially substitute for these processes, although careful monitoring is required to prevent negative cascading effects. As an example, Scherer et al. (2025) found clear positive effects of wild boar rooting activity to the rare Melitaea aurelia and Euphydryas aurinia in abandoned calcareous grasslands, by disturbing soil and re-establishing a mosaic of early-successional plant assemblages, including vital host plants. In line with these considerations, habitat type and landscape context emerged as key constraints on reproductive success, even where adult presence was high. Open habitats such as meadows and dry grassland patches supported higher adult abundance and higher occupation rates, likely due to favourable thermal conditions and greater accessibility to - and higher abundance of - nectar resources. These habitats may function as important dispersal areas, facilitating movement across the landscape, especially for poorly mobile species such as Z. cassandra (Vovlas et al. 2014). In contrast, inner woodland sections showed lower apparent egg survival despite relatively high occupation ratios. This pattern suggests that such habitats may indeed be used by adults but offer suboptimal conditions for early development of eggs and larvae, potentially acting as demographic sinks (Camerini et al. 2018). Whether such processes are due to microclimatic unfavourable or suboptimal conditions at host plants, or due to stronger biotic pressures acting on early life stages (e.g., predation risk or competition and parasitism levels for eggs and larvae), is still under study (Shiojiri et al. 2015), yet the key role of microclimate in shaping oviposition selection (Cini et al. 2021) suggests a prime role in explaining our results too. From a conservation standpoint, this highlights the risk of overestimating habitat quality based solely on adult occurrence or on the abundance of host plants and underscores the importance of managing habitat structure to enhance reproductive success, for example by maintaining small openings or ecotones within woodland matrices (Kalarus and Nowicki 2025; van Halder et al. 2015). Landscape context further shaped population indices at even coarser scales, with natural and semi-natural landscapes supporting higher egg numbers than urban ones (Pla-Narbona et al. 2022). Although adults were observed in patches within highly urbanized contexts, reproductive investment was reduced, indicating that urban habitats may need interventions to increase population fitness although apparent egg success did not differ between natural and urban sites. Nonetheless, some of the monitored populations within the urban area of Rome have been documented for several decades now, suggesting that highly fragmented populations like these may still thrive for long, provided a network of ecologically connected sites is still available (Ahon-Vasquez and Capuano 2025). This finding reinforces the need for landscape-scale conservation approaches which preserve connectivity and prioritize semi-natural habitats as core reproductive areas.

By integrating habitat extent and resources, disturbance, and habitat structure and landscape context, our study highlights the multi-scale nature of habitat suitability for protected butterflies (Turlure et al. 2019). Effective conservation strategies should prioritize the maintenance of profitable habitat patches, by ensuring the persistence of host plant populations, allow moderate levels of physical disturbance to vegetation in order to foster high nectar resources to adult butterflies, and promote structurally heterogeneous habitats within connected landscapes, with a special attention upon populations occurring at very small habitat patches (Dennis et al. 2006). Moreover, maintaining moderate levels of local disturbance by e.g., prescribed livestock grazing, may also ensure profitable habitat heterogeneity. Quantifying the effects of different grazing/disturbance intensities and timing should be a key research topic for effective conservation management in the future. Our results also showed that adult presence alone may be a partial indicator of conservation success, and that a set of different demographic measurements derived from different life stages may provide a deeper and more nuanced understanding of habitat suitability at a local scale or e.g., of the success of conservation actions. We also highlight that our study overlooked the pupal stage of our target species, since Z cassandra spends this phase in the litter or in secluded sites at ground level, challenging the actual monitoring of pupae. Nonetheless, the long time spent as pupae make this stage a critical one for the conservation of this and other butterfly species, highlighting the need to integrate it in conservation assessments. As an example, forestry management implying the use of heavy vehicles may damage pupae or make soil less suitable to butterflies (Scherer and Fartmann 2024). Moreover, disentangling the ecological relationships between environmental conditions and the demography of EU-wide protected species is a key asset for more complex procedures, such as the definition of favourable reference values (FRV), needed for assessing the conservation status of species listed in the Annexes of the Habitats Directive (Bonelli et al. 2021). Conservation assessments and monitoring programs of Z. cassandra, as well as other legally protected butterflies, should aim to also incorporate reproductive and early survival metrics to accurately evaluate population viability, possibly building up long term data series for properly assessing demographic trends. In a rapidly changing world, where both land-use practices and disturbance regimes constantly reshape our landscapes, such integrative and evidence-based approaches are essential to ensure the long-term persistence of strictly protected butterfly species.

Authors

1. Leonardo Ancillotto
2. Martina Chiara Recchilungo
3. Giulia Bacco
4. Giacomo Bruni
5. Camilla Bongini
6. Alice Lenzi
7. Arianna Giannini
8. Sandra Rosselli
9. Alessandro Campanaro
10. Emanuela Maurizi
11. Emiliano Mori
12. Fabio Mosconi

Cite this article

Ancillotto, L., Recchilungo, M.C., Bacco, G. et al. Identifying key demographic drivers of a protected butterfly to inform conservation in natural and anthropogenic landscapes. Biodivers Conserv 35, 198 (2026). https://doi.org/10.1007/s10531-026-03403-y

Download citation

• Received: 09 March 2026

• Revised: 16 June 2026

• Accepted: 18 June 2026

• Published: 29 June 2026

• Version of record: 29 June 2026

• DOI: https://doi.org/10.1007/s10531-026-03403-y

Keywords