Global Change Biology (2010) 16, 2923–2929, doi: 10.1111/j.1365-2486.2010.02177.x Adaptation to host plants may prevent rapid insect responses to climate change SHANNON L. PELINI1 , JESSICA A. KEPPEL2 , ANN E. KELLEY3 and JESSICA. J. HELLMANN Department of Biological Sciences, University of Notre Dame, Notre Dame, IN 46556, USA Abstract We must consider the role of multitrophic interactions when examining species’ responses to climate change. Many plant species, particularly trees, are limited in their ability to shift their geographic ranges quickly under climate change. Consequently, for herbivorous insects, geographic mosaics of host plant specialization could prohibit range shifts and adaptation when insects become separated from suitable host plants. In this study, we examined larval growth and survival of an oak specialist butterfly (Erynnis propertius) on different oaks (Quercus spp.) that occur across its range to determine if individuals can switch host plants if they move into new areas under climate change. Individuals from Oregon and northern California, USA that feed on Q. garryana and Q. kelloggii in the field experienced increased mortality on Q. agrifolia, a southern species with low nutrient content. In contrast, populations from southern California that normally feed on Q. agrifolia performed well on Q. agrifolia and Q. garryana and poorly on the northern, high elevation Q. kelloggii. Therefore, colonization of southern E. propertius in higher elevations and some northern locales may be prohibited under climate change but latitudinal shifts to Q. garryana may be possible. Where shifts are precluded due to maladaptation to hosts, populations may not accrue warm-adapted genotypes. Our study suggests that, when interacting species experience asynchronous range shifts, historical local adaptation may preclude populations from colonizing new locales under climate change. Keywords: climate change, geographic mosaic, local adaptation, plant-insect interactions, range shifts, resource specialization Received 27 September 2009 and accepted 13 December 2009 Insects have been particularly responsive to climate Introduction change because much of their life history is influenced Species have shifted their geographic distributions in by temperature, but they also are strongly affected by response to past and recent climate changes, and these the quality and availability of plants as food resources shifts are expected to continue with future climate (Pelini et al., 2009b). Because insects and plants disperse change (Parmesan & Yohe, 2003). These distributional at different rates, their geographic range shifts during shifts occur as habitats at poleward latitudes or higher periods of climate change could differ substantially elevations become suitable and individuals disperse to (Schweiger et al., 2008). For example, a number of and establish in new locales. The increased frequency of butterfly species have recently shifted their distribu- dispersal from equatorial populations and/or strong tions poleward or to higher elevations (e.g., 240 km over selection against poleward genotypes under a changing 30 years) (Parmesan et al., 1999) while the expected climate has resulted in increases in equatorial geno- average rate that trees can track a changing climate is types toward the poles and could increase the climatic only 20–40 km per 100 years (Davis & Shaw, 2001). This tolerance of poleward populations (Rodrı´guez-Trelles & host plant lag could limit the establishment of insect Rodrı´guez, 1998; Umina et al., 2005; Balanya´ et al., 2006). populations in new regions. Interactions between insects and host plants often vary Correspondence: Jessica. J. Hellmann, tel. 1 1 574 631 0296, fax throughout species’ ranges, and it is important that we 1 1 574 631 7413, e-mail: [email protected] think about this in the context of range shifts under climate change. Geographic turnover in host plant suitability and 1Present address: S. L. Pelini, Harvard Forest, Harvard University, Petersham, MA 01366, USA. availability often results in populations of herbivorous insects that are locally adapted to different host plants 2 Present address: J. A. Keppel, Department of Biology, University throughout the insect’s geographic range, creating a geo- of North Carolina, Chapel Hill, NC 27599, USA. graphic mosaic of host plant specialization (Thompson, 3Present address: A. E. Kelley, School of Natural Resources and 2005). Strong adaptation to local host plants can lead to Environment, University of Michigan, Ann Arbor, MI 48109, USA. reduced fitness on alternate host plants (Boecklen & r 2010 Blackwell Publishing Ltd 2923 2924 S. L. PELINI et al. Mopper, 1998). Rapid adaptive evolution of novel host (red vs. white oak groups) (Pavlik et al., 2002). Popula- plant use could relax this barrier, but this process may be tions of E. propertius in southern and central California constrained by the rapid pace of modern climate change are multivoltine and feed on the leaves of Quercus (see Etterson & Shaw, 2001; Hellmann & Pineda-Krch, agrifolia (coast live oak) (Fig. 1), an evergreen red oak. 2007). Therefore, equatorial genotypes could be reduced E. propertius populations at higher elevations (600– in frequency or lost if equatorial individuals disperse 1800 m) between southern California and southwestern poleward but fail to colonize because they encounter Oregon feed on Quercus kelloggii (California black oak) unsuitable or less suitable host plants. This could compro- (Fig. 1), a deciduous red oak. Populations north of south- mise poleward or upland populations by limiting the western Oregon are univoltine and feed exclusively on arrival of genotypes adapted to warmer conditions. the only available oak species, Quercus garryana (Oregon In this study, we examined the geographic structure white oak or Garry oak) (Fig. 1), a deciduous white oak. of host plant specialization within the geographic dis- Other Quercus species also may be used by E. propertius tribution of Erynnis propertius, the Propertius duskyw- in California, but Q. agrifolia and Q. kelloggii are the only ing skipper (Lepidoptera: Hesperiidae) (Guppy & recorded hosts and the most common in sites where Shepard, 2001). E. propertius occurs along the western E. propertius has been collected (personal observations). coast of North America from Baja California, Mexico to Lab experiments revealed that E. propertius has complete British Columbia, Canada where oaks (Quercus) occur mortality on alternate Quercus species, including the (Scott, 1986; Opler, 1999) (Fig. 1). widespread Quercus chysolepis (canyon live oak), how- The three primary Quercus host plant species used by ever (S. L. Pelini, unpublished results). E. propertius offer interesting contrasts in geography The orientation of the transitions in dominance of these (northern vs. southern and low vs. high elevation), host plant species across latitude and elevation is impor- phenology (evergreen vs. deciduous) and phylogeny tant when assessing the capacity for range shifts in E. propertius under climate change, particularly because these Quercus species overlap at their range boundaries. Q. kelloggii co-occurs with Q. garryana at its northern range edge and co-occurs and hybridizes with Q. agrifolia in southern California (Pavlik et al., 2002) (Fig. 1). There areafewplaceswhereQ. garryana and Q. agrifolia overlap in Q. garryana-dominated woodlands (Pavlik et al., 2002), butwedidnotsampleE. propertius from those locations due to small population sizes. In this study, we conducted a fully crossed feeding experiment to determine if E. propertius populations are adapted to local host plants. We assume that the rate of current climate change constrains rapid adaptive evolu- tion of host plant use in E. propertius and that the Quercus host plants will disperse slower than E. propertius.His- torical adaptation to local Quercus host plants could prevent range expansion if southern E. propertius popula- tions have reduced fitness on northern or high elevation host plants. Given an unchanging host plant landscape, host plant limitations could occur for E. propertius in two geographic areas. First, individuals dispersing from south (i.e., those that occur with and use Q. agrifolia) (Fig. 1) to north will have to switch to Q. garryana. Second, individuals dispersing to higher elevations and other northern locales will have to switch to Q. kelloggii. Materials and methods Fig. 1 Distribution map of Quercus host plants with Erynnis Feeding trials propertius sampling sites. Quercus distributions are from Little (1971). Triangles represent ‘northern’ sites and squares represent We collected E. propertius individuals from four sites ‘southern’ sites. dominated by Q. agrifolia in southern and central California r 2010 Blackwell Publishing Ltd, Global Change Biology, 16, 2923–2929 HOST PLANTS LIMIT INSECT RANGE SHIFTS 2925 (Qa1–Qa4) (Fig. 1) and from two sites dominated by ous insects that specialize on Quercus are insensitive to, or even Q. garryana in northern California and southwestern Oregon benefit from, tannins (Roslin & Salminen, 2008). (Qg1 and Qg2) (Fig. 1). We also collected individuals from one Leaves were collected from greenhouse Q. agrifolia, Q. kelloggii-dominated locale with Q. garryana in the Sierra Q. kelloggii and Q. garryana plants used in the E. propertius Nevada Mountains of central California (Qk1) (Fig. 1). Eggs feeding trials during July of 2007 and 2008. The leaves were were collected from Qk1, Qg1 and Qg2 (‘northern’) in April–