Popul Ecol (2005) 47:193–202 DOI 10.1007/s10144-005-0222-3
O R I GI N A L A R T IC L E
Piotr Nowicki Æ Madgalena Witek Æ Piotr Sko´rka Josef Settele Æ Michal Woyciechowski
Population ecology of the endangered butterflies Maculinea teleius and M. nausithous and the implications for conservation
Received: 16 February 2005 / Accepted: 23 May 2005 / Published online: 11 August 2005 The Society of Population Ecology and Springer-Verlag Tokyo 2005
Abstract Butterflies of the genus Maculinea are highly endangered in Europe. The cuckoo species M. rebeli has been thoroughly investigated through both empirical and modelling studies, but less is known about the population ecology of predatory Maculinea. We present the findings of a 2-year research study on sympatric populations of two endangered butterflies: Maculinea teleius and M. nausithous in the Krako´w region, southern Poland. The study comprised mark–release–recapture sampling and laboratory rearing of butterflies from larvae collected in the field. For both species the sex ratio was slightly, but consistently, female-biased and there was little year-to-year change in the seasonal population sizes. Daily numbers showed greater variation between the 2 years of the study due to the differences in daily survival rate. The average life span of laboratory-raised butterflies kept in ideal conditions was more than 6 days, compared to only 2–3 days in the field. The recruitment of both males and females consistently followed a bimodal pattern. A small proportion of individuals (maximum 25%) changed sites, in spite of relatively short distances of ca. 100 m separating them. The results indicate that populations of both species are typically stable within their sites, possibly due to larval polymorphism, but there is little inter-site mobility and thus landscape corridors seem necessary to enhance metapopulation viability. A further problem to be considered in the conservation of Maculinea butterflies is the fact that their very short life span in relation to flightperiod length reduces the effective population size.
P. Nowicki (&) Æ M. Witek Æ P. Sko´rka Æ M. Woyciechowski Institute of Environmental Sciences, Jagiellonian University, Gronostajowa 7, 30-387 Krako´w, Poland E-mail:
[email protected] Tel.: +48-12-6645125 Fax: +48-12-6646912 J. Settele Department of Community Ecology, UFZ Centre for Environmental Research, Halle, Germany
Keywords Dispersal Æ Mark–release–recapture Æ Population fragmentation Æ Population size Æ Robust Design
Introduction Butterflies of the genus Maculinea (Lycaenidae) are regarded as the ‘flagships’ of nature conservation in Europe (Thomas 1995; Thomas and Settele 2004). The scientific community has long been intrigued by their specialised myrmecophilous lifestyles: the larvae feed for a short time on specific foodplants and then spent most of their life in Myrmica ant nests, either preying on ant broods (‘predatory species’: M. arion, M. nausithous and M. teleius) or being fed by nurse ants (‘cuckoo species’: M. alcon and M. rebeli) (Thomas et al. 1991). Furthermore, all five European species of Maculinea have been declining in numbers and are highly endangered with local extinctions in many countries (Wynhoff 1998; Munguira and Martin 1999). Biotope loss or deterioration is undoubtedly the principal threat to Maculinea butterflies. Many former habitats have been destroyed by the recent development of urban, industrial and recreational areas, while the intensification or abandonment of traditional agriculture has led to the deterioration of many other sites (Thomas 1995; Wynhoff 1998; Munguira and Martin 1999). At present most European populations exist in highly fragmented landscapes, frequently disturbed by human activities; the populations are often small and isolated. Understanding the mechanisms shaping the population dynamics of the species and implementing this knowledge into conservation programmes is crucial to ensuring the long-term persistence of the species under such conditions (Thomas 1995; Settele et al. 1996; Maes et al. 2004). The population ecology of M. rebeli has been extremely well documented by field research and investigated theoretically in numerous modelling studies (Hochberg et al. 1992, 1994; Elmes et al. 1996; Clarke
194
et al. 1997; Thomas et al. 1998a). It is likely that the knowledge acquired in this way may also be applicable to the second cuckoo species, M. alcon, which resembles M. rebeli in many aspects of its ecology to the extent that for a long time both were considered conspecific (Kaaber 1964; Wells et al. 1983). On the other hand, less is known about the population ecology of the predatory Maculinea, especially the wet grassland species M. nausithous and M. teleius. While the population patterns of the cuckoo species can be investigated with relative ease by counting their conspicuous eggs, no such method could be effectively implemented for the predatory species that oviposit their eggs deep within the inflorescences of their foodplants (Thomas et al. 1991). Instead, they have to be studied with mark–release–recapture (MRR) techniques, which are very laborious and time-consuming. However, the advantage of MRR studies is that, if properly designed, they can provide information on various aspects of population ecology (New 1991; Warren 1992). We decided to apply this method for a comprehensive study of the populations of the two predatory species, M. teleius and M. nausithous, occurring sympatrically in wet meadows of southern Poland, with the aim of gathering knowledge of importance for their conservation. In addition, we raised butterflies in the laboratory from larvae and pupae collected in the field to obtain independent estimates of population parameters.
Materials and methods Study sites The study was conducted in a meadow complex located in southern Poland (50 01¢N; 19 54¢E), 4 km southwest of Krako´w’s city centre. The meadows occupy the flat bed of the Vistula River valley, at an altitude of 200– 240 m above sea level. They comprise 63 sites covered with Sanguisorba officinalis, which is the foodplant of both investigated species (Thomas 1984, 1995). All of the S. officinalis sites are inhabited by M. teleius, with M. nausithous also present on most of them. The metapopulations (sensu Harrison 1994) of both species in the Fig. 1 Map of the investigated Sanguisorba officinalis sites and their surroundings. White areas represent fragments of meadows without S. officinalis. The location of the study area in the map of Poland is shown in the top left corner
region can be roughly estimated at tens of thousands of individuals and have been relatively stable in recent years (authors’ unpublished data). At present, the Krako´w metapopulations are threatened with habitat loss due to the rapid expansion of the city. Major urban and industrial development is planned for the area up to the year 2009 (Bo¨hm 2000; Pepkowska 2002). The sites have no special geographic names, thus for research purposes we developed a nomenclature system in which they are assigned consecutive numbers. The MRR study was performed on three sites, which form the easternmost tip of the wet meadow complex. The largest of the three sites (Krako´w 17) comprises 8.77 ha and is isolated from the others by forest patches and residential areas (Fig. 1). The site was affected by the construction of an underground central-heating main pipeline in October 2002. About 10% of its surface was destroyed by the engineering works, and it is most likely that all the ant nests in this area were lost as well. However, the vegetation including S. officinalis fully recovered by the following summer. The two other sites, Krako´w 18 and Krako´w 9, are an order of magnitude smaller (0.94 and 0.63 ha, respectively), with no apparent environmental barrier but only an 80-m-wide fragment of meadow without S. officinalis between them (Fig. 1). Further S. officinalis sites located to the west and south are separated by strips of 3-m-high reeds, which create a major obstacle for Maculinea movements. This was indicated by the fact that except for a single M. teleius from site 18, no butterflies marked on the investigated sites were observed further to the west or south during an intensive 2-day survey of all the sites in the region in mid-July 2003. Larvae and pupae for the laboratory rearing of M. teleius were collected from several large sites located in the core area of the wet-meadow complex, approximately 2–4 km west of the sites on which MRR sampling was conducted. Mark–release–recapture field sampling The MRR study was conducted from 3 July to 13 August 2002, and from 25 June to 18 August 2003.
195
Admittedly, the very beginning of the flight period in 2002 was missed as a few adults were already observed flying on 1 July. Nevertheless, the effect of this on the seasonal population size estimates on sites 17 and 18 should be negligible, because the lack of the recruitment estimate for the interval of 1–3 July was compensated for by relatively high initial population size estimate for 3 July. Site 9 was only surveyed in 2003. Sampling took place every third day, except for the beginning of July 2002 when a 2-day interval occurred once. Each capture day consisted of four capture sessions, each 100 min long, conducted between 09:00 and 17:00. To allow butterflies to mingle between capture sessions, the sessions were separated with 20–40 min intervals. A fifth capture session was not possible, even though this might considerably improve the precision of daily population size estimates (Otis et al. 1978). With shorter capture sessions the numbers of butterflies captured per session would be too low for a reliable estimation, and the duration of daily sampling could not be extended either, because very few individuals flew before 9:00 and after 17:00. Moreover, sampling before 9:00 would inevitably break the assumption of population closure as Maculinea butterflies eclose in morning hours (Elferrich 1963, 1998; Thomas 1984; Thomas and Lewington 1991). Four to six people participated in capturing butterflies on site 17, two to three people on site 18 and a single person on site 9. Butterflies were captured in nets, marked with fine-tipped waterproof Staedtler Lumocolor 313 pens on the underside of the right hind wing, and immediately released at the place of capture. Consecutive numbers in four different colours (red, black, green and blue) were used as marks. Mark–release–recapture data analysis The Robust Design model (Pollock 1982; Pollock et al. 1990) was applied for the data analysis with capture days constituting primary periods and capture sessions being secondary periods. The analysis was carried out with the software MARK 2.1 (White and Burnham 1999), including the programs CAPTURE (Otis et al. 1978) and RELEASE (Burnham et al. 1987) that have been incorporated as independent modules into MARK 2.1. Prior to the analysis proper the assumptions of the model were evaluated. The inter-sexual differences in residence rate and capture probability were tested jointly with the overall test 1 of the program RELEASE. However, as it has relatively little power (Burnham et al. 1987) we also developed two additional testing procedures. We used Kolmogorov-Smirnov tests (Sokal and Rohlf 1981) to compare observed residences for recaptured individuals of both sexes, defined as the time between the first and the last capture of an individual. The chi-square tests, or Fisher exact tests in the case of small sample sizes, were applied to detect differences in the sex ratios of butterflies captured and those recaptured at
least once on the day of their initial capture (so as to avoid the effect of survival) over the entire flight period. It should be noted that both of the above procedures are modified versions of Tabashnik’s (1980) joint residencecatchability test. As in the Tabashnik’s test we used the pattern that the ratio of recaptured proportions of males and females should be constant regardless of the time interval (t, expressed in days) if the survival rate is equal for both sexes, and that this ratio should be 1 on the day of initial capture (i.e. t=0) if the capture probability is equal for both sexes. We believe that the applied procedures are more sensitive than linear regression fitting of the Tabashnik’s test, especially in the case of infrequent sampling days, which leads to a small number of possible time intervals and thus a small number of points to which the regression line is fitted. The assumption of age-independent survival was evaluated by analysing the observed residence curves (Watt et al. 1977; Brakefield 1982). With survival being independent of butterfly age, the proportions of individuals reaching a certain age should logarithmically decrease with age. Therefore, we tested the fit of a regression line against logarithmically transformed proportions observed in the populations surveyed—the lack of significance would indicate that survival rate differed between age classes. Demographic closure of the investigated populations within primary periods (i.e. between 09:00 and 17:00 on each capture day) was assessed by testing whether the estimates of survival between secondary periods of the same primary periods were significantly lower than 1—significant differences would indicate the lack of population closure. The population parameters estimated included daily (for capture days) and seasonal population size, recruitment and survival between capture days, average life span and sex ratio. It has to be noted that the sample size was large enough to estimate all the above parameters from the model only in the case of the populations on site 18 as well as that of M. teleius on site 17. In contrast, the number of recaptures in the M. nausithous population on this site made it impossible to obtain any estimates except for the sex ratio. Following the results of the model selection routine of the program CAPTURE, the heterogeneity model (Mh) was applied for the estimation of ^ i Þ: The recruitment ðB ^ 0 Þ; and the daily population sizes, ðN i thereforeP also seasonal population size ^ 0 ¼ k1 B ^0 ¼ N ^ 1 Þ; were adjusted for the ^ 0 ; where B ðN 0 i¼0 i total individuals emerging and dying within a single interval between consecutive primary periods. The sex ratio was corrected for inter-sexual differences in capture probability, even if the differences were insignificant. The sur^ Þ were scaled with vival rates between capture periods ð/ i respect to the length of intervals between primary periods ^ ¼ / ^ 1=ti Þ: Subsequently, the to give daily survival rates ð/ i i ^ was calculated as the average daily survival rate / weighted mean, weights being the numbers of individuals captured with respect to different numbers of individuals
196
captured within primary periods. The average life span was derived from the daily survival rate as ^ Þ1 0:5: The above equation, in reverse ^e ¼ ð1 / form and with average observed residence replacing life ^ for M. nausithous on span, was used to approximate / site 17, for which model estimates were not available. ^ may be regarded as the de For the isolated site 17 / facto survival rate, as there were extremely few individuals emigrating from this site. On the other hand on the non-isolated sites 9 and 18 it should rather be termed ‘residence rate’ (though we stick to the ‘survival rate’ for the sake of continuity with cited references), because it was affected by both mortality and emigration. Having assumed equal mortality on all sites, which seems justified because they experience the same weather conditions and have similar communities of potential predators, the difference between the average daily survival rate for site 17 and that for sites 9 or 18 could be ^ Þ used as an estimate of average daily emigration ðE ^ total Þ from the two latter sites. Seasonal emigration ðE P1 ^ n ^ from the two sites was calculated as n¼0 E / (where n ^ defines the fraction of individuals reaching the age of / ^ Þ1 : Applying ^ ð1 / n days), which can be solved as E an alternative approach, we also expressed seasonal emigration as the proportion of individuals that were recaptured on a site other than the site of their initial capture as in Sutcliffe et al. (1997). Laboratory rearing of butterflies Pre-pupating larvae and pupae of M. teleius were sought in Myrmica ant nests from June to early July 2002 and
2003. The material was collected with the permission of the Polish Ministry of Environment. Larvae and pupae found were placed in plastic containers with a wet sponge to provide adequate humidity and transported to the laboratory, where they were weighed and then kept separately in a thermostatic chamber with a constant temperature of 20C. The date of eclosion (and whenever possible its precise time) and the sex were recorded for each adult butterfly emerging. Butterflies obtained in 2002 were kept outside in net enclosures until natural death. Two types of enclosures were used: large ones (2·2·2 m) containing parts of grassland with S. officinalis and natural invertebrate communities, and small ones (40·40·40 cm), in which butterflies lived in a predator-free space and fed on sugar solution. Living conditions in the large enclosures were presumably similar to natural ones, the only exception being the lack of bird predators. On the other hand, those in small enclosures were almost ideal (the weather was generally fine during this part of the experiment), so the butterfly life span recorded there can be assumed as a rough approximation of the physiological life span. In 2003 M. teleius adults were used for genetic research and had to be killed soon after eclosion, thus no information on life span is available for that year.
Results Mark–release–recapture model assumptions Tests conducted revealed no significant inter-sexual differences in survival rate among the populations investigated (Table 1). However, the proportions of males and females recaptured on the day of their initial
Table 1 Results from testing the Robust Design assumptions Population
Year
Inter-sexual differences
Age-independent survival—fit of linear regression line
RELEASE Test 1a,b
KolmogorovSmirnov testa
Chi-square/Fisher testb
P
D
P
P
R2
df
P
M. teleius Krako´w 17 Krako´w 17 Krako´w 18 Krako´w 18 Krako´w 9
2002 2003 2002 2003 2003
0.5994 0.8841 0.8718 0.5728 0.4563
0.073 0.034 0.095 0.038 0.411
>0.5 >0.5 >0.5 >0.5 0.2–0.5
0.2780 0.0347* 0.1654 0.1849 0.2533c
0.9904 0.9686 0.9972 0.9931 0.9819
3 4 4 4 2
0.0004*** 0.0004*** 0.5 0.1–0.2 ca. 1.0 >0.5
0.0389*c 0.2655c 0.8315 0.0741 0.2303
0.9978 0.9643 0.9885 0.9701 0.9605
1 1 3 3 2
0.0297* 0.1210 0.0005*** 0.0022** 0.0200*
a
Differences in survival Differences in capture probability c P values of the Fisher exact test, used when sample size was too small to apply chi-square test * P
