Global Change Biology (2011) 17, 194–205, doi: 10.1111/j.1365-2486.2010.02233.x Global warming will affect the genetic diversity and uniqueness of Lycaena helle populations JAN CHRISTIAN HABEL*w z,DENNISRO¨ DDERz,THOMASSCHMITTz and GABRIEL NE` VE§ *Institute of Ecology and Environmental Chemistry, Leuphana University Lu¨neburg, D-21335 Lu¨neburg, Germany, wMuse´e National d’Histoire Naturelle, Section Zoologie des Inverte´bre´s, L-2160 Luxembourg, zDepartment of Biogeography, Trier University, D-54296 Trier, Germany, §Aix-Marseille Universite´, CNRS IRD UMR 6116-IMEP, F-13331 Marseille Cedex 3, France Abstract The climate warming of the postglacial has strongly reduced the distribution of cold-adapted species over most of Central Europe. Such taxa have therefore become extinct over most of the lowlands and shifted to higher altitudes where they have survived to the present day. The lycaenid butterfly Lycaena helle follows this pattern of former widespread distribution and later restriction to mountain areas such as the European middle mountains. We sampled 203 individuals from 10 populations representing six mountain ranges (Pyrenees, Jura, Massif Central, Morvan, Vosges and Ardennes) over the species’ western distribution. Allozyme and microsatellite polymorphisms were analysed to study the genetic status of these highly fragmented populations. Both molecular marker systems revealed a strong genetic differentiation among the analysed populations, coinciding with the orographic structure and highly restricted gene flow among them. The large-scale genetic differentiation is more pronounced in allozymes (FCT: 0.326) than in microsatellites (RCT: 0.113), but microsatellites show a higher resolution on the regional scale (RSC: 0.082) compared with allozymes (FSC: n.s.). For both analytical tools, we found private alleles occurring exclusively in a single mountain area. The highly fragmented and isolated occurrence of populations is supported by the distribution pattern of potentially suitable climate suggested by species distribution models. Model projections under two climate warming scenarios predict a decline of climatically suitable areas, which will result in the extinction of most of the populations showing unique genetic characteristics. Keywords: allozymes, climate change, climate envelope, ecological niche modelling, fragmentation, Lycaenidae, microsatellites, mountain regions Received 25 August 2009; revised version received 18 December 2009 and accepted 10 March 2010 nowski, 2000). For more than four decades, allozyme Introduction electrophoresis has been applied to assess effects of The number of studies addressing the genetic structure of ecosystem changes (Hedrick, 1999), for example, in populations has increased considerably over the last studies on past climate oscillations (see Schmitt, 2007, decade (Hedrick, 2001; Storfer et al., 2007; see Schmitt, 2009 for recent reviews). However, more recently devel- 2007, 2009 for a recent review). This process has been oped analytical techniques using hypervariable, non- accelerated by the development of powerful molecular coding DNA have become more widely used in techniques and new statistical methods, which offer a landscape genetics allowing high levels of resolution variety of sophisticated analyses (e.g. Haig, 1998; Zane of genetic structures even over small spatial and short et al., 2002; Excoffier & Heckel, 2006). Meanwhile, neutral temporal scales (Storfer et al., 2007). For example, micro- and nonneutral, dominant and codominant markers offer satellites have currently become a popular tool in a broad choice of ways of analysing different temporal population and conservation genetics (e.g. Collevatti and spatial scales in the fields of phylogeography, ecology et al., 2001; Gao et al., 2002; Gaiotto et al., 2003; Dutech and conservation (Allendorf & Luikart, 2007). et al., 2004), despite the difficulty of identifying work- Codominant marker systems, such as allozymes or able loci in some taxonomic groups such as Lepidoptera microsatellites, represent suitable tools for estimating (Ne`ve & Megle´cz, 2000; Ne`ve, 2009). This marker is recent and past population structures (Hedrick & Kali- particularly useful for the estimation of the effective population size (Ne), genetic structures – even among Correspondence: J. C. Habel, Muse´e National d’Histoire Naturelle neighbouring populations – and recent genetic effects of Luxembourg, 25, Rue Mu¨nster, L-2160 Luxembourg, e-mail: environmental changes (cf. Conte et al., 2003; Keller & [email protected] Largiade`r, 2003; Keller et al., 2005). However, data 194 r 2010 Blackwell Publishing Ltd GLOBAL WARMING AND GENETIC UNIQUENESS 195 obtained from microsatellite analyses should be inter- lar marker systems and species distribution modelling. preted with caution, as several severe shortcomings of Our selected model organism has suffered severely this marker system, such as the frequent presence of from the advancing temperature increase combined null alleles, have been noted (Chapuis et al., 2008). with land-use changes; today most of the populations Therefore, the most important observable differences of central Europe therefore occur as isolated remnants between allozymes and microsatellites are (i) the much (Finger et al., 2009). This may lead to strong genetic drift higher level of polymorphisms in microsatellites, (ii) effects (cf. Lande, 1995; Hedrick & Kalinowski, 2000), their significantly higher level of heterozygosity and often even combined with reductions in population size (iii) a two to four orders of magnitude higher mutation and individual fitness (Bouzat et al., 1998; Madsen et al., rate in microsatellites (Estoup et al., 1998; Streiff et al., 2000; Chapman et al., 2009). 1998; Gao et al., 2002). Both methods, then, have their We combine the analysis of allozymes and microsa- specific strengths and therefore should be applied in tellites to study the population genetic constitution at combination. Nevertheless, in most cases our knowl- the westernmost distribution of this butterfly species edge about population genetic structures of species is with species distribution modelling for its past, present based on a single genetic marker system, although some and possible future distribution patterns. We analysed studies even showed that various molecular markers 203 individuals from 10 populations scattered from may result in conflicting patterns (Vandewoestijne & the Pyrenees to the Ardennes and representing six Baguette, 2002; Garcia-Paris et al., 2003; Veith et al., mountain systems and modelled distributions based 2008). on 458 presence localities. The population genetic ana- In addition to genetic analyses, species distribution lyses performed assess (i) the recent genetic status of models (SDMs) allow spatial assessments of areas these highly isolated populations and (ii) potential potentially suitable for a species and the connectivity differences between these two molecular marker patterns of these areas (Guisan & Zimmermann, 2000; systems. Via species distribution modelling, we evalu- Jeschke & Strayer, 2008; Ro¨dder et al., 2010). SDMs rely ate (iii) the consequences of our genetic results for on the assumption that environmental conditions are conservation. the primary drivers of the target species’ distribution and that the range of this species fits with those condi- tions (Arau´ jo & Pearson, 2005). These models have Materials and Methods frequently been applied to predict changes in species’ potential distributions under current, past and Study species future climate scenarios (Arau´ jo et al., 2004; Arau´ jo The Violet Copper L. helle (Denis & Schiffermu¨ ller, 1775) is a & Guisan, 2006; Heikkinen et al., 2006; Hijmans & boreo-montane butterfly of the Palearctic. This lycaenid spe- Graham, 2006; Pearman et al., 2008). cies was widely distributed over Central Europe during the The application of SDMs is of particular interest in late glacial and early postglacial period and declined in the combination with genetic analyses, preferably those wake of rising temperatures during the postglacial warming based on more than one single marker system for the over major parts of the European lowlands (Habel et al., 2010). Isolated remnant populations have remained restricted to understanding of past, present and future changes in mountain areas like the lower mountains of Western and ranges. As a recent altitudinal and latitudinal range Central Europe (Nunner, 2006; Bachelard & Fournier, 2008), shift due to global warming is a generally accepted but some populations also occur in lowland wet meadows in phenomenon (Parmesan & Yohe, 2003), predictive tools some countries of Central Europe (Skorka et al.,2007). The are highly relevant, but so far do not take into account habitats of L. helle are cool and damp grasslands with sheltered the intraspecific diversity. However, the preservation of stands, marshes, clearings in forests and along streams, genetic diversity and unique lineages within species in springs and bogs with sheltered warm pockets (Re´al, 1962; a rapidly changing world is considered as a key aspect Bachelard & Descimon, 1999) and specific vegetation struc- in conservation biology (Allendorf & Luikart, 2006). tures (Turlure et al., 2009); another requirement is the existence Therefore, the combination of SDMs and genetic ana- of the exclusive larval food plant, Polygonum bistorta. For these lyses allows the delimitation of particularly endangered reasons, the habitats of L. helle are rather scattered