Different ecological demands shape differences in population structure and behaviour among the two generations of the small pearl-bordered fritillary

Study species Boloria selene

The small pearl-bordered fritillary Boloria selene (Denis & Schiffermüller, 1775) belongs to the family Nymphalidae, subfamily Heliconiinae. Boloria selene is widely distributed in Eurasia at elevations from 0 to 2,200 m asl (Tuzov & Bozano, 2006; Kudrna et al., 2011). The species inhabits nutrient-poor open landscapes, damp to wet meadows, fens and fen edges with abundance of flowering plants but also forest paths and clearings (Settele et al., 2008; Settele et al., 2015). The host plants of the caterpillars are different violet species (Viola spec.), in Germany mostly the marsh violet (Viola palustris) (Ebert & Rennwald, 1991; Gelbrecht et al., 2016) but also V. canina, V. riviniana and V. hirta (Tolman & Lewington, 1998). In the temperate zone, the species is mesophilic (Gelbrecht et al., 2016) and bivoltine with the second generation sometimes just being a partial one (Settele et al., 2008); however, it can be univoltine at higher altitudes, e.g., above 1,000 m asl. in the Alps, and in northern Europe (Schweizerischer Bund für Naturschutz, 1987). The two subsequent generations are temporally separated; the flight period of the first generation is mid-May to mid-June; the second generation is on the wing from late July to late August (Ebert & Rennwald, 1991; Gelbrecht et al., 2016). The adults feed on a wide spectrum of flowering plants (Schweizerischer Bund für Naturschutz, 1987; Ebert & Rennwald, 1991). In a comparison with other Boloria species, Boloria selene is graded as a generalist species regarding the use of multiple nectar resources as an adult and no known specialisation in other resources and microhabitat structures (Turlure et al., 2010; Turlure et al., 2019). On the German Red List, Boloria selene is listed on the pre-warning list (Reinhardt & Bolz, 2011), while it is categorised as endangered on the Red List of Brandenburg (Gelbrecht et al., 2001; Settele et al., 2015).

Study site

The study was carried out in eastern Brandenburg (North-East Germany) in the administrative district of Märkisch-Oderland, near the city of Müncheberg (Fig. 1). The annual average temperature is 9.0 °C and the annual average rainfall 563 mm (1981–2010; DWD weather station Müncheberg (Deutscher Wetterdienst (DWD), 2010). The study site, the so-called “Gumnitzwiesen” (52°30′N, 14°04′E), is part of the nature reserve “Gumnitz und Großer Schlagenthinsee” and the Natura 2000 site of the same name, which are completely embedded in the nature park “Märkische Schweiz”. The study area of approximately 10.5 ha consists mainly of nutrient-poor to nutrient-rich wet meadows of the plant community Molinion caeruleae and smaller parts of dry, sandy, calcareous grassland (Landesamt für Umwelt, 2019). The area is delimited by a large pine forest and a smaller fen woodland area, a lake at the eastern border and a forest path used for hiking alongside the forest edge (Landesamt für Umwelt, 2019). The next known habitats of Boloria selene are several km away from our study site so that exchange of individuals should be rare or completely missing.

Data collection

We applied the mark-release-recapture (MRR) method (Ehrlich & Davidson, 1960). Butterfly individuals were captured with an insect net, marked with a continuous, unique ID on the undersides of fore- and hindwings with a permanent felt tip pen (“OHpen universal” in black by Stabilo^®) and released immediately after capture. The ID consists of a letter that represents the capture day and a running number, starting each day with 1. We took the GPS position of every capture and recapture event using smartphones with the app TourCount© (Stein, 2019). For every capture event, we noted the individual’s sex, its wing condition and activity prior to capture, the plant species or genus when the individual was feeding, temperature, wind speed (scale: 1–4) and cloudiness (percentage of sky coverage). The wing condition was classified on a scale from 1 to 4 (1: fresh with undamaged wing fringe, 2: missing wing fringe, 3: slightly damaged, 4: heavily damaged) (Thomas, 1983; Zimmermann et al., 2005).

Data were collected 24 May–20 June 2021 (first generation) and 11 July–13 August 2021 (second generation). Hence, our analyses all rely on data collected during one single year. However, weather conditions were favourable during the entire field work period and the population densities were well manageable so that our data should be well suitable for representing our study species. Nevertheless, it would be nice to have data from additional years to cross-check our results. The daily number of collectors varied from one to six people, i.e., three people regularly participating, who were sometimes accompanied by up to three students. The whole study area was surveyed using transects. Every cluster of Viola palustris was recorded with its GPS position using the app TourCount©.

Ethics declaration

No butterflies were harmed or killed for this study. Field work was authorised by Landesamt für Umwelt, Gesundheit und Verbraucherschutz, Frankfurt (Oder).

Data analyses

Population demography

The best model for the first generation presumed an additive effect of sex and quadratic time on the survival rate, an additive effect of sex and factorial time on the capture probability, an interactive effect of sex and squared time on the proportional recruitment pent, and an effect of sex on the number of individuals (Table 1(A)). This model estimated a population size of 155 males (± 6 SE; i.e., 15/ha ± 0.6 SE) and 155 females (± 10 SE; i.e., 15/ha ± 1.0 SE), which corresponds to a sex-ratio of 1:1.

During the first three capture days of the second generation, we only recorded one male specimen per day; these initial days were inappropriate for the applied algorithm and therefore were excluded from analyses due to high standard errors. The differences in the Akaike Information Criterion for the first four models were <2, and therefore a weighted average of the estimated population sizes over the best four models was used. The best four models share an effect of sex and factorial time on the capture probability and an effect of factorial time on the recruitment. The two best models assume a constant population size and an effect of sex and linear/squared time on the survival rate. In the third and fourth models, the number of individuals differs by sex and the survival probability is affected by time (squared or linear) (Table 1(B)). The calculated ĉ factor for under-dispersion was 0.41. The weighted model estimated a population size of 382 males (± 18 SE; i.e., 37/ha ± 4 SE) and 276 females (± 19 SE; i.e., 26/ha ± 3 SE), which corresponds to a sex ratio of 1.4 males per female.

The first generation displayed protandry, which was absent in the second generation (Fig. 2). The weather conditions were sufficient for flight activity throughout both flight periods and started to deteriorate at the end of the second flight period.

Behaviour

Males were highly flight active in both generations (Fig. 3). The proportion of basking individuals compared to resting ones was significantly higher in the first than in the second generation (Chi-squared test of homogeneity: χ² = 11.87, df = 4, p = 0.018). Females of the first generation were significantly more flight active than those of the second generation during which flight activity decreased while nectaring and resting increased (Chi-squared test of homogeneity: χ² = 15.47, df = 5, p = 0.009). There were no significant differences in activity between sexes of the first generation (Chi-square test of homogeneity: χ² = 8.78, df = 4, p = 0.07; N males = 123; N females = 110), but the difference was significant in the second generation (Chi-square test of homogeneity: χ² = 29,5, df = 5, p = 0.0001; N males = 191; N females = 105).

Plant genera used for nectaring

The first generation used nine different plant genera for nectaring, with Knautia as the most frequently used (males: 45.5%; females: 58.6%), followed by Cardamine for males (33.3%) and Lychnis for females (17.2%) (Fig. 4A). Three plant genera were used for nectaring only by males (Arabidopsis, Veronica, Hieracium), and another three only by females (Myosotis, Ranunculus, Cirsium), resulting in significant sex-specific preferences (χ² = 17.8, df = 8, p = 0.03) (N males = 33; N females = 29). Removing all individuals recorded more than once; however, the difference is only marginally significant (χ² = 13.6, df = 7, p = 0.06) (N males = 28; N females = 23).

Seventy-four nectaring events were recorded in the second generation (27 males, 47 females). Lythrum (males: 44.4%; females: 46.8%) and Cirsium (males: 29.6%; females: 21.3%) (Fig. 4B) were visited most by both sexes. Despite both sexes using some genera exclusively, sex-specific preferences were not significant (χ² = 9.1, df = 10, p = 0.52) (N males = 27; N females = 47).

Wing condition

In the first generation, mean wing wear is linearly correlated with time, for males and females (LM males: df = 11, R² = 0.91, p < 0.0001, N = 13; LM females: df = 12, R² = 0.85, p < 0.0001, N = 14) (Fig. 5A). Mean wing wear did not differ significantly between sexes (U test: p = 0.56). The slope of the correlation also did not differ significantly between sexes (males: 0.09, females: 0.086), tested with a linear model with an interaction between the two predictors time and sex (F statistics: df = 23, p = 0.79).

In the second generation, the mean value of wing condition was linearly correlated with time for males (LM: R² = 0.54, df = 13, p = 0.002) but not for females (LM: R² = 0.02, df = 11, p = 0.63) (Fig. 5B). Means did not differ between sexes (U test: p = 0.43), but the correlation slopes differed significantly (males: 0.03, females: 0.004; F-statistics: df = 24, p = 0.037). Comparing the slopes of wing decay of first and second generation, they were significantly steeper in the first (F statistics – males: df = 24, p < 0.0001; females: df = 23, p < 0.0001).

Mobility and movement pattern

In the first generation, mean flight distance was significantly higher for females than for males (means – males: 99 m ± 112 SD, N = 101, females: 158 m ± 173 SD, N = 71) (U-test: p = 0.005) (Table 2). The mean lifespan movements were significantly higher in females as well (U test: p = 0.015) (Table S1). A linear model with the number of days between the capture events as predictor and the flight distance as response variable did not reveal a correlation for neither sex (males: df = 99, R² = 0.0002, p = 0.9; females: df = 69, R² = 0.006, p = 0.5) (Fig. S1A).

In the second generation, mean flight distance was again higher for females (142 m ± 190 SD, N = 39) than for males (97 m ± 128 SD, N = 100) (Table 2), but these did not differ significantly (U test: p = 0.9), nor did the lifespan movements (U test: p = 0.4) (Table S1). There was no significant correlation between passed time since capture and flight distance (males: df = 98, R² = 0.03, p = 0.08; females: df = 37, R² = 0.006, p = 0.17) (Fig. S1B).

The travelled distances did not differ significantly between generations, neither for males (U test: p = 0.9), nor for females (U test: p = 0.1).

In the extrapolation of dispersal potential (Fig. 6), the best fits of NEF and IPF were obtained for 50 m classes; only NEF for first-generation males had its best fit with 20 m interval. In the first generation, NEF always had a better fit than IPF, while it was the opposite in the second generation (Table 3). In both generations, the predicted dispersal probabilities were always higher for IPF than for NEF, and for females than for males. Comparing generations, the modelled proportions of long-distance movements were considerably higher in the second generation; except for the IPF of first-generation males (Table 4).

The flight paths in both generations showed long and short flights for both sexes and no sex-specific movement pattern (Figs. S3 & S4). Accumulations of butterfly individuals were in the majority of cases in close proximity to mapped stands of the larval food plant Viola palustris but accumulations also existed without the confirmed presence of this plant (Fig. S5).

Population structure and protandry

The observed population density of both generations had intermediate levels if compared to other studies (Cozzi, Müller & Krauss, 2008; Konvicka et al., 2011) ; this indicates suitable habitat quality. The higher density of the second generation might be explained by an increased survival rate of the non-overwintering larvae which is a common feature in bi- and multivoltine species (Altermatt, 2010; Plazio & Nowicki, 2021). The overall moderate population size of both generations facilitated high recapture rates (Fischer & Fiedler, 2001b; Fric, Klimova & Konvicka, 2006). Sex-specific differences in the recapture rates between generations might be explained by the decreased flight activity of second-generation females (Weyer & Schmitt, 2013; Jugovic, Crne & Luznik, 2017).

In the majority of cases, protandry is assumed to be beneficial for the reproductive success of both sexes (Fagerström & Wiklund, 1982; Zonneveld, 1992) and is achieved by faster development and earlier eclosion of males (Wiklund, Nylin & Forsberg, 1991; Fischer & Fiedler, 2000). However, this beneficial feature was only observed for the first and not for the second generation. This raises the question why this apparently advantageous trait is not always expressed. We assume that the first generation of Boloria selene is synchronized when winter diapause ends in early spring. During the flight period of the first generation, the eggs are laid along a two-week time window; additionally, the second generation apparently is lacking a synchronizing environmental cue. This results in asynchronous eclosion of the second generation, blurring protandry (or at least weakening it so strongly that it could not be detected by our model) and making the second-generation flight period longer than the one of the first, apparently also affecting the time period of egg-laying (Komata & Sota, 2017). A prolongation of the second generation due to bad weather conditions that might have limited the daily flight periods (Wiklund, Nylin & Forsberg, 1991; Junker & Schmitt, 2010; Larsdotter Mellström & Wiklund, 2010), is rather unlikely because weather conditions were rather favourable all along the second flight period of Boloria selene.

Behaviour and use of nectar plants

First-generation males and females had similar activity patterns. This is a rather unusual pattern for butterflies (but see Parile, Piccini & Bonelli, 2021) because males in general fly more whereas females rest and feed more (Zimmermann et al., 2005; Ehl et al., 2019). Interestingly, this pattern changed during the second generation. While male flight activity remained unchanged, it decreased for females with an increase of nectaring, thus reflecting the commonly observed pattern. The analysis of wing condition decay supports these flight activity patterns, with sex-specific differences and a slower decay in females only for the second generation. The generally high male flight activity is explained by their search for mates; females on the other hand search for nectar sources and oviposition sites, activities not demanding such high investments in flight time (Ebert & Rennwald, 1991; Fischer, Beinlich & Plachter, 1999; Kuras et al., 2003; Weyer & Schmitt, 2013).

These sex- and generation-specific differences in behaviour also coincide with differences in nectar plant choice, with first-generation individuals being more selective about plant genera and exhibiting sex-dependent selection, while the second generation was mostly opportunistic. A general decrease in the availability of plants for nectaring in the late season might be one possible explanation for our results, supported by Rusterholz & Erhardt (2000).

Flower selection by butterflies also depends on the specific nectar composition as males rely more on sucrose and sugar because of their higher, energy-demanding flight activity, while females have a higher demand for amino-acids for egg production (Rusterholz & Erhardt, 2000; Mevi-Schütz & Erhardt, 2002). Thus, the nectar sources available during the first generation might be more efficient to satisfy butterflies’ nutritional needs, for males and females, so that individuals need less time for feeding. In particular, Knautia is a highly efficient nutritional source for both sexes because its nectar contains similar proportions of sucrose, glucose and fructose and high amounts of essential and non-essential amino-acids (Venjakob, Leonhardt & Klein, 2020). When such high-quality resources are missing during the second generation, individuals are forced to use a mix of different plants and to visit more inflorescences to get the same amount of nutrition, probably explaining the larger time investment of second-generation females in nectaring. Interestingly, females broadened their nectaring behaviour and their time investment considerably more than males. This might underline the higher difficulty of females fulfilling their nutritional needs compared with males.

Furthermore, Rusterholz & Erhardt (2000) observed that females of Polyommatus bellargus were more selective for nectar sources with high amino-acid content in the spring than in the autumn generation. They hypothesised that overwintering negatively affects the nutritional conditions of the larvae and consequently enhances the requirements for amino-acids of spring-generation females. This might also explain the broader range of nectar sources used in our second generations and the smaller numbers in the first generation because individuals of the first generations might be forced to be more selective for the most nutrient-rich nectar plants. The negative effect of overwintering on the nutritional conditions of larvae might also in males lead to more specialised nectar and nutrient requirements after eclosion as males have a generally higher weight loss caused by metamorphosis, as shown for Lycaena tityrus (Fischer & Fiedler, 2000). The hypothesis of a seasonal change in nectar plant selectivity might therefore not only be true for females, as supposed by Rusterholz & Erhardt (2000), but also for males and well corresponds with our observations.

Movement patterns and dispersal

First-generation males had a slightly higher likelihood of emigrating than those of the second generation if considering IPF extrapolations as suggested by Baguette (2003). For females, we observed the opposite with considerably higher extrapolations for long-distance dispersal in the second than in the first generation. While males are motivated by the search for females and thus respond with emigration to lower population densities (Kuussaari et al., 1998), female mobility is affected by the availability of suitable oviposition host plants after mating (Timus et al., 2017). Consequently, high population densities increase intraspecific competition for host plants, hereby enhancing dispersal. Additionally, male harassment is augmenting with higher population densities and therefore may also lead to increased emigration pressure after mating (Baguette et al., 1998).

This is well in concordance with our findings for Boloria selene. The species’ sex- and generation-specific differences in dispersal probabilities can therefore be explained by the different densities of the two generations and the different evolutionary functions of both sexes, maybe even mirrored in their morphology as suggested by work of Fric, Klimova & Konvicka (2006) obtaining significant differences in body design between generations. The overall higher dispersal probability of females compared to males (in particular in the second generation) can be explained by the sex-specific advantage of potential founder effects: After mating, one individual female can establish a new population in a so-far empty habitat, whereas the reproductive success of males always depends on the presence of females (Weyer & Schmitt, 2013).