Received: 28 August 2018 | Accepted: 5 November 2018 DOI: 10.1111/1365-2656.12933 RESEARCH ARTICLE Phenotypic biomarkers of climatic impacts on declining insect populations: A key role for decadal drought, thermal buffering and amplification effects and host plant dynamics Jofre Carnicer1,2 | Constantí Stefanescu2,3 | Maria Vives-Ingla1 | Carlos López2 | Sofia Cortizas1 | Christopher Wheat4 | Roger Vila5 | Joan Llusià2 | Josep Peñuelas2 1Department of Evolutionary Biology, Ecology and Environmental Abstract Sciences, University of Barcelona, Barcelona, 1. Widespread population declines have been reported for diverse Mediterranean Spain butterflies over the last three decades, and have been significantly associated 2CREAF, E08193 Bellaterra (Cerdanyola del Vallès), Catalonia, Spain with increased global change impacts. The specific landscape and climatic drivers 3Natural History Museum of Granollers, of these declines remain uncertain for most declining species. Granollers, Spain 2. Here, we analyse whether plastic phenotypic traits of a model butterfly species 4Department of Zoology (Population Genetics), University of Stockholm, (Pieris napi) perform as reliable biomarkers of vulnerability to extreme tempera- Stockholm, Sweden ture impacts in natural populations, showing contrasting trends in thermally ex- 5 Institute of Evolutionary Biology (CSIC- posed and thermally buffered populations. UPF), Barcelona, Spain 3. We also examine whether improved descriptions of thermal exposure of insect Correspondence populations can be achieved by combining multiple information sources (i.e., inte- Jofre Carnicer Email: [email protected] grating measurements of habitat thermal buffering, habitat thermal amplification, host plant transpiration, and experimental assessments of thermal death time Funding information VENI-NWO, Grant/Award Number: (TDT), thermal avoidance behaviour (TAB) and thermally induced trait plasticity). 863.11.021; Spanish Government, Grant/ These integrative analyses are conducted in two demographically declining and Award Number: CGL2016-78093-R, CGL2013-48074-P and CGL2013-48277-P; two non-declining populations of P. napi. Catalan Government, Grant/Award Number: 4. The results show that plastic phenotypic traits (butterfly body mass and wing size) SGR 2014-274; European Research Council Synergy Grant, Grant/Award Number: are reliable biomarkers of population vulnerability to extreme thermal conditions. ERC-2013-SyG 610028; Knut and Alice Butterfly wing size is strongly reduced only in thermally exposed populations dur- Wallenberg Foundation, Grant/Award Number: KAW 2012.0058; Swedish ing summer drought periods. Laboratory rearing of these populations documented Research Council, Grant/Award Number: reduced wing size due to significant negative effects of increased temperatures VR-2012-4001 affecting larval growth. We conclude that these thermal biomarkers are indicative Handling Editor: Albert Phillimore of the population vulnerability to increasing global warming impacts, showing contrasting trends in thermally exposed and buffered populations. 5. Thermal effects in host plant microsites significantly differ between populations, with stressful thermal conditions only effectively ameliorated in mid-elevation populations. In lowland populations, we observe a sixfold reduction in vegetation thermal buffering effects, and larval growth occurs in these populations at signifi- cantly higher temperatures. Lowland populations show reduced host plant quality (C/N ratio), reduced leaf transpiration rates and complete above-ground plant J Anim Ecol. 2019;1–16. wileyonlinelibrary.com/journal/jane © 2018 The Authors. Journal of Animal Ecology | 1 © 2018 British Ecological Society 2 | Journal of Animal Ecology CARNICER ET AL. senescence during the peak of summer drought. Amplified host plant tempera- tures are observed in open microsites, reaching thermal thresholds that can affect larval survival. 6. Overall, our results suggest that butterfly population vulnerability to long-term drought periods is associated with multiple co-occurring and interrelated ecologi- cal factors, including limited vegetation thermal buffering effects at lowland sites, significant drought impacts on host plant transpiration and amplified leaf surface temperature, as well as reduced leaf quality linked to the seasonal advance of plant phenology. Our results also identify multiannual summer droughts affecting larval growing periods as a key driver of the recently reported butterfly population de- clines in the Mediterranean biome. KEYWORDS butterflies, climate change, host plant, multiannual drought, phenotypic biomarker, Pieris napi, thermal buffering 1 | INTRODUCTION global warming (De Frenne et al., 2013; Nieto- Sánchez, Gutiérrez, & Wilson, 2015). Declines in butterfly populations across diverse species over the Populations inhabiting sites lacking effective habitat thermal last three decades have been described in the Mediterranean basin buffering could experience increased negative impacts of extreme (Stefanescu, Herrando, & Páramo, 2004; Stefanescu, Carnicer, temperatures, resulting in substantial long- term demographic de- & Peñuelas, 2011; Stefanescu, Torre, Jubany, & Páramo, 2011; clines (Parmesan, Root, & Willig, 2000). In addition to the effects Wilson et al., 2005; Wilson, Gutiérrez, Gutiérrez, & Monserrat, of habitat thermal buffering, the thermal exposure of butterfly pop- 2007; Carnicer, Brotons, Stefanescu, & Peñuelas, 2012; Carnicer, ulations can be crucially determined by other key processes, such Stefanescu, et al., 2013; Zografou et al., 2014; Melero, Stefanescu, as the seasonal variation of host plant transpiration and leaf water & Pino, 2016). Negative effects of land use changes and global content during summer drought, the operation of thermal amplifica- warming have been proposed as the main drivers of the observed tion processes in microhabitats or the display of thermal avoidance declining trends (Stefanescu et al., 2004; Wilson et al., 2005, 2007; behaviours in the insect larvae allowing the selection of cool micro- Stefanescu, Carnicer, et al., 2011; Stefanescu, Torre, et al., 2011). sites at the host plant (Carnicer et al., 2017). These key co- acting These negative demographic trends affect both habitat generalist processes are often not measured, and their complex interactions and specialist butterfly species in the Mediterranean biome. The remain poorly described. To understand the relative importance of spatial diversity of most functional groups is negatively associated all these processes, integrative studies combining multiple informa- with increased temperatures and aridity (e.g., host plant use, dis- tion sources in intensively studied populations are warranted. persal capacity, habitat specialisation and thermal niche groups; Here, we provide an integrative study of the thermal exposure Stefanescu, Carnicer, et al., 2011; Stefanescu, Torre, et al., 2011). in four populations of Pieris napi, combining multiple sources of in- Furthermore, the available evidence suggests that global warm- formation (demographic and climatic data, phenotypic trait data, ing-induced population responses are intimately linked to com- measurements of habitat thermal buffering, host plant traits and plex interactions with habitat features and host plant dynamics experimental assessments of thermal responses). Furthermore, we (Bennett, Severns, Parmesan, & Singer, 2015; Carnicer et al., 2017; explore whether temperature- responsive phenotypic traits can Carnicer, Barbeta, Sperlich, Coll, & Peñuelas, 2013; De Frenne et al., be applied as reliable biomarkers of the different vulnerability to 2013; Merrill et al., 2008; Oliver, Stefanescu, Páramo, Brereton, increased temperatures in these intensively studied populations. & Roy, 2014; Oliver et al., 2015; Suggitt et al., 2012). In line with Ample experimental evidence supports that diverse life- history and this idea, it has been suggested that specific habitat attributes can functional traits of butterflies are highly responsive to temperature modify global warming impacts on butterfly populations, triggering variation and show predictable responses to extreme temperature both positive and negative demographic responses. For example, it treatments (Bauerfeind & Fischer, 2013a,b, 2014; Jones, Hart, & has been shown that the densification of forest habitats associated Bull, 1982; Nail, Batalden, & Oberhauser, 2015; Sheridan & Bickford, with land abandonment can cool local microclimates, buffering 2011). In particular, wing and body size measures have been identi- the impacts of global warming in some plant and insect popula- fied as traits highly responsive to temperature variation and climate tions and resulting in positive or neutral demographic responses to change impacts (Atkinson, 1994; Atkinson & Sibly, 1997; Forster, CARNICER ET AL. Journal of Animal Ecolog y | 3 Hirst, & Atkinson, 2012; Kingsolver, 2009; Nygren, Bergström, & (October–early November), with overwintering in the pupal stage. Nylin, 2008; Sheridan & Bickford, 2011; Talloen, Van Dongen, Van Maximum abundance is typically recorded in early summer, in co- Dyck, & Lens, 2009). Therefore, it is likely that an extensive quanti- incidence with the peak of the third generation. This peak is fol- fication of plastic phenotypic traits in declining and non- declining lowed by a period of 1–2 months when abundance is much reduced, natural populations could indicate their different vulnerability to in coincidence with summer