Deceptive signals and behaviors of a cleptoparasitic show local adaptation to different host species

Leslie Saul-Gershenza,1, Jocelyn G. Millarb, J. Steven McElfreshb, and Neal M. Williamsa

aDepartment of Entomology and Nematology, University of California, Davis, CA 95616; and bDepartment of Entomology, University of California, Riverside, CA 92521

Edited by May R. Berenbaum, University of Illinois at Urbana–Champaign, Urbana, IL, and approved August 2, 2018 (received for review October 26, 2017) Chemosensory signals play a key role in species recognition and pallida in the Mojave Desert in south-central California. It must be mate location in both invertebrate and vertebrate species. Closely emphasized that these M. franciscanus populations have been related species often produce similar but distinct signals by determined to be the same species using both morphological (10) varying the ratios or components in pheromone blends to avoid and molecular analyses (11). interference in their communication channels and minimize cross- Female M. franciscanus lay large egg masses averaging 761 attraction among congeners. However, exploitation of reproduc- eggs (SD 533, n = 8) (4). The eggs hatch synchronously, and the tive signals by predators and parasites also may provide strong newly hatched triungulins climb nearby vegetation to form a selective pressure on signal phenotypes. For example, bolas single aggregation (4, 12). These aggregations remain together spiders mimic the pheromones of several moth species to attract their prey, and parasitic larvae, known as triungulins, for up to 14 d, during which time they produce a signal that cooperatively produce an olfactory signal that mimics the sex mimics the sex pheromone produced by females of their bee pheromone of their female host to attract male bees, as the hosts (4, 12). It is essential that they act collectively to produce a first step in being transported by their hosts to their nests. In both signal of sufficient intensity. Their olfactory signal attracts male

cases, there is strong selection pressure on the host to discriminate bees to pseudocopulate, and the instant that a male bee contacts EVOLUTION real mates from aggressive mimics and, conversely, on the predator, an aggregation, the entire larval mass attaches to the male (Fig. parasite, or parasitoid to track and locally adapt to the evolving 1). The male then flies off with his unwanted passengers, which signals of its hosts. Here we show local adaptation of a beetle, transfer to a female during mating or mating attempts (4). The Meloe franciscanus (Coleoptera: Meloidae), to the pheromone chem- infested female then carries the triungulins to their ultimate istry and mate location behavior of its hosts, two species of solitary destination, the female bee’s nest, where they disembark to feed bees in the genus Habropoda.WereportthatM. franciscanus’ de- on the nest provisions and host offspring, completing their de- ceptive signal is locally host-adapted in its chemical composition and ratio of components, with host bees from each allopatric population velopment and emerging from the nest as adults the following preferring the deceptive signals of their sympatric parasite popula- spring (8, 9). tion. Furthermore, in different locales, the triungulin aggregations have adapted their perching height to the height at which local Significance male bees typically patrol for females. This study provides strong evidence for two different but local adaptation | deceptive signals | –parasite interactions | complementary types of local adaptation in geographically behavioral adaptation | mimicry isolated populations of a parasitic insect. Specifically, we report that different populations of a blister beetle, Meloe francis- pecies recognition and mate location chemosensory signals canus, a nest parasite of bees, locally adapt their deceptive Splay key roles in both invertebrate and vertebrate species (1, chemical signals, which mimic the sex pheromones of their host 2). Predators such as bolas spiders, parasites (3), and parasitoids bees, to the differing pheromone blends of their local host such as blister beetle larvae (4) can exploit these communication species. We show that local nest parasites are significantly signals. Parasites are expected to exhibit local adaptations to more attractive to male bees than nonlocal parasites, using their host populations because, in general, parasites exhibit transplant experiments. We report the identification of at- larger population sizes and shorter generation times than their tractant blends for the two host species and the compounds hosts (5, 6) (but also see ref. 7). The beetle Meloe franciscanus that the beetle larvae produce to attract their hosts. Further- (Coleoptera: Meloidae), the larvae of which parasitize ground- more, we show that the two parasite populations have nesting bees, has neither of these stereotypical characteristics. evolved divergent host-matching behaviors. First, M. franciscanus is univoltine, with a similar generation time Author contributions: L.S.-G. and J.G.M. designed research; L.S.-G., J.G.M., and J.S.M. as its host, and second, its population sizes are small relative to performed research; J.G.M. contributed new reagents/analytic tools; L.S.-G., J.G.M., those of its hosts. The adult are flightless and have lim- J.S.M., and N.M.W. analyzed data; and L.S.-G., J.G.M., and N.M.W. wrote the paper. ited capacity to disperse (8), but beetles can disperse phoretically The authors declare no conflict of interest. as larvae attached to host bees (9). In addition, the various extant This article is a PNAS Direct Submission. populations of M. franciscanus are geographically isolated, with Published under the PNAS license. minimal or no possibility of gene flow between populations. Here M. franciscanus Data deposition: The sequences reported in this paper have been deposited in GenBank we report that different populations of exhibit database (accession nos. MH487729.1–MH487733.1, MH489082.1–MH489086.1, MH491842.1– local adaptations that mimic both the behaviors and the chemical MH491845.1, MH511214.1–MH511218.1,andMH511209.1–MH511213.1) and are available in composition of the sex pheromones of locally available bee SI Appendix, Table S1. Experimental data are available in SI Appendix, Tables S1–S6. species. Specifically, we compare a population of M. franciscanus 1To whom correspondence should be addressed. Email: [email protected]. larvae, known as triungulins, parasitizing nests of Habropoda This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. miserabilis (Hymenoptera: Apidae) from the coastal sand dunes 1073/pnas.1718682115/-/DCSupplemental. of Oregon with a population parasitizing the congener Habropoda

www.pnas.org/cgi/doi/10.1073/pnas.1718682115 PNAS Latest Articles | 1of5 Downloaded by guest on September 24, 2021 Fig. 1. Cleptoparasite M. franciscanus adapts to two allopatric bee hosts in the genus Habropoda.(A)AmaleH. pallida from the Mojave Desert with M. franciscanus triungulins attached. (B) A male H. miserabilis from Oregon with M. franciscanus triungulins covering the entire dorsal surface of the abdomen. (C)AnM. franciscanus aggregation on grass stems. [C is reprinted with permission from ref. 4. Copyright (2006) National Academy of Sciences, USA.] (All images, L.S.-G.)

In a previous study, we showed that the sex attractant phero- populations become isolated in space and/or time, the different mone of female H. pallida was produced in the head, and consists localized forces of natural selection result in divergence of the of a group of odd-numbered, C21-to-C31 monounsaturated hy- signals from the ancestral state. Furthermore, in systems where drocarbons (4). Of these components, those with double bonds parasites or predators exploit the signals of their hosts, exploiting at the 9 position were found from both males and females, organisms must track the evolving signals of their local hosts or whereas only females produced alkenes with double bonds in multiple hosts to persist. Here we provide evidence for local positions other than 9, and it was these additional isomers that adaptation by M. franciscanus, demonstrating how populations made the signal sex-specific. We also showed that the triungulins have matched both the pheromone chemistry and the reproductive parasitizing this population mimicked a subset of the female- behaviors of their local hosts, two solitary bee species in the genus specific alkenes, with the subset being both necessary and suffi- Habropoda. cient to attract male bees effectively. A recent meta-analysis of hymenopteran sex pheromones Results suggests that both saturated and unsaturated hydrocarbons are Cafeteria Experiments. In a first set of experiments, we showed strongly correlated with evolutionary relationships within the that H. miserabilis females from Oregon produced a sex phero- order, and alkenes with a (Z)-9 double bond were suggested as mone that attracted males, and that aggregations of sympatric M. the ancestral state (13). Divergence from the ancestral state has franciscanus larvae also attracted male bees. Thus, male bees resulted in double bonds in other positions, greatly increasing the were highly attracted to cages containing live female bees, but number of different signals possible, particularly as compounds not to caged live male bees or to control, empty cages (Fig. 2A can occur as blends. Selection favors receivers’ ability to detect and SI Appendix, Table S2). Furthermore, male bees were divergent ratios or components (14), providing a mechanism by attracted to caged aggregations of M. franciscanus triungulins, which signals can evolve in response to selection pressures such suggesting that they too produced a volatile attractant. A follow- as interspecific competition for a pheromone channel or ex- up experiment testing different body parts (female and male bee ploitation of the signal by predators and parasites. Over time, as head, thorax, abdomen) showed that males were most attracted

Fig. 2. (A) Male H. miserabilis were significantly more attracted to caged female bees and M. franciscanus larval aggregations than to caged male bees (Friedman test: χ2 = 20.52, P < 0.001, k = 4, n = 9). Female bees were significantly more attractive than male bees (P = 0.014). Female bee and Meloe ag- gregation attraction were not significantly different from each other (P = 0.998), and both were significantly more attractive than controls (females, P = 0.014; Meloe aggregations, P = 0.024). However, male bees were not significantly different from controls (P = 1.00) (SI Appendix, Table S2). (B) Body regions of the two sexes differed in attractiveness (Friedman test: F = 4.55, χ2 = 42.98, k = 9, P < 0.001, n = 7). Female bee heads were significantly more attractive to male bees than female thoraxes, all male body segments, and controls (P = 0.001). Male body regions did not differ significantly from controls (P = 1.00; male thorax, P = 0.879) (male data are in SI Appendix, Table S3). (C) Synthetic lures (syn lure OR) reconstructed from analyses of local Oregon M. franciscanus triungulins were more attractive to male H. miserabilis bees than were control lures (Poisson GLIMMIX trial: F = 4.31, df = 3, P = 0.131; treatment: F = 24.58, df = 1, P = 0.016; n = 4, control 0 ± 0.0 SEM) (SI Appendix, Table S4). Figures show means ± SEM of male inspections and contacts of test subjects in cafeteria experiments. Blue bars represent bees or bee body parts; gold bars represent Meloe larval aggregations or Meloe syn lure. The different lowercase letters in A–C indicate that the inspections + contacts were significantly different from each other.

2of5 | www.pnas.org/cgi/doi/10.1073/pnas.1718682115 Saul-Gershenz et al. Downloaded by guest on September 24, 2021 to the heads of females (Fig. 2B and SI Appendix, Table S3). were absent in the H. pallida Mojave females. Conversely, H. SI Appendix Body parts of males were not attractive ( , Table S3), pallida Mojave females had three (Z)-9-alkenes [(Z)-9-C21,(Z)- demonstrating the sexual specificity of the signal. We then ana- 9-C23, and (Z)-9-C25] that were absent in H. miserabilis Oregon H. miserabilis lyzed extracts of heads of bees of both sexes and females, as well as a group of higher molecular mass C31 alkenes. whole-body extracts of sympatric triungulin aggregations. A The triungulin extracts from the two sites were also markedly subset of compounds found in the extracts of the triungulins was different. The triungulins from both sites had (Z)-9-C23 and (Z)- H. miserabilis also found in extracts of the heads of female bees. 9-C25, whereas only the Oregon triungulins had (Z)-7-C23 and The extracts consisted almost exclusively of straight-chain al- (Z)-7-C25, each matching local host female bees. The propor- kanes and alkenes. The extract of heads of male bees was the tions of the individual alkenes in the beetle extracts also differed least complex, consisting primarily of tricosane (C23) and pen- between the Mojave and Oregon populations. tacosane (C25) and three alkenes with (Z)-9 double bonds [(Z)- Finally, a blend of compounds found in extracts of both the 9-C23,(Z)-9-C25, and (Z)-9-C31], with (Z)-9-C25 being the most female H. miserabilis bees and the Oregon triungulins, recon- abundant (Fig. 3). In contrast, extracts of the heads of female structed from synthetic compounds, was attractive to male bees were characterized by 5 alkanes with chain lengths of C21 to H. miserabilis in a field bioassay (Fig. 2C and SI Appendix, Table C29 and 12 alkenes with chain lengths of C23 to C29 and with S4), confirming that the Oregon triungulins have adapted their double bonds in positions 7, 10, 11, 12, 13, and 14 but not po- mimicry of their host’s sex pheromone to the local host species. sition 9 (Fig. 3). The whole-body extracts of the triungulins were We had previously documented the attraction of Mojave male intermediate, with 5 alkanes between C21 and C29 and 8 alkenes H. pallida to the synthetic blend of Mojave M. franciscanus with double bonds in positions 7, 9, 10, 11, and 12 (Fig. 3). triungulins (4). In comparing the chemistries of the H. miserabilis bees and triungulins from the cool Oregon site and the previously studied Transplant Experiments. To determine whether the triungulin H. pallida bees from the Mojave Desert site, the extracts were populations parasitizing the different allopatric bee species have superficially similar, consisting of complex mixtures of alkanes locally adapted to the chemical signals of their respective hosts, and alkenes. However, closer examination revealed major dif- we tested responses of male bees to sympatric and allopatric ferences. For example, the Oregon H. miserabilis females had triungulin aggregations. We found that male bees were signifi- three (Z)-7-alkenes [(Z)-7-C23,(Z)-7-C25, and (Z)-7-C27] that cantly more attracted to triungulin aggregations from their own EVOLUTION

Fig. 3. M. franciscanus triungulins from different populations produce locally adapted blends of semiochemicals to mimic their specific host bees. Chemical profiles of (A) the Oregon population of host bee H. miserabilis and its locally adapted nest parasite M. franciscanus, and (B) the Mojave Desert population of host bee H. pallida and its locally adapted nest parasite M. franciscanus. Bars in blue are alkenes, which are present in the extracts of female and male bees

and/or locally adapted triungulins. Note the presence of (Z)-7-C23 in female H. miserabilis and Oregon M. franciscanus. The blue-shaded sections highlight the areas of mimicry between local M. franciscanus and Habropoda females and the area of difference between the two Meloe populations and the two Habropoda species. The asterisk indicates the most abundant compound differed between extracts. [B is reprinted with permission from ref. 4. Copyright (2006) National Academy of Sciences, USA.]

Saul-Gershenz et al. PNAS Latest Articles | 3of5 Downloaded by guest on September 24, 2021 local adaptation is fundamental to understanding parasite speci- ation and the maintenance of enduring parasite–host relationships. Methods Collection of and Field Experiments. Specimens were collected from and field bioassays were conducted at Hidden Lake Dunes north of Waldport, Lincoln County, Oregon (N44.45666667, W-124.0794444, 14 m) and Kelso Dunes in the Mojave National Preserve, San Bernardino County, California (N34.88968, W-115.72147, 760 m), from March through May 2010 to 2015. Sand dunes at the Oregon site are stabilized with European beach grass, Ammophila arenaria (L.) Link, surrounded by Scotch broom Cystisus sco- parius (L.) Link. A forest composed of Pinus contorta Douglas ex Loudon var. contorta and California huckleberry Vaccinium ovatum Pursh, the local major host plant of adult H. miserabilis, surrounds the nest site. Kelso Dunes are active dunes which support ∼184 species of plants (16), including Astragalus lentiginosus var. borreganus, a nectar plant of H. pallida and the exclusive host plant of adult M. franciscanus at Kelso Dunes; Larrea tridentata, which is the major pollen host plant of H. pallida; Petalonyx thurberi;and several native grasses including Hilaria rigida, Panicum urvilleanum,andStipa Fig. 4. Reciprocal transplant experiments showed that male Habropoda hymenoides, which serve as triungulin aggregation perching sites. Astragalus l. bees were most attracted to their local M. franciscanus triungulin aggrega- var. borreganus blooms every year, providing a reliable nectar source for H. tions. M. franciscanus larval aggregations transplanted from the Mojave De- pallida, which emerges in synchrony with its bloom at Kelso Dunes. H. pallida sert to the Oregon site were significantly less attractive to male H. miserabilis nesting begins a few days after the onset of Larrea bloom. than were local M. franciscanus aggregations from Oregon (GLmixedM: P < Voucher specimens of M. franciscanus and host bees H. miserabilis and 0.0006, n = 5). Similarly, M. franciscanus larval aggregations transplanted from H. pallida from the multiple study localities in Oregon and California have Oregon to the Mojave Desert study site were less attractive to male H. pallida been deposited in the Bohart Museum, University of California, Davis. Our bees than local Mojave M. franciscanus aggregations (GLmixedM: P < 0.001, sequence data for M. franciscanus, for the CO1, 16S, 28S, ITS-2, and EF1-α n = 3) (SI Appendix, Table S5). Figures show means ± SEM of male inspections genes, are available in GenBank (SI Appendix, Table S1). The data from the and contacts of test subjects in cafeteria experiments. experiments are provided in SI Appendix, Tables S2–S6. To test for local adaptation, we conducted transplant experiments in two habitats (Mojave and Oregon) with local and transplanted Meloe aggrega- locale than to allopatric aggregations, for both bee species (Fig. 4 tions, and used the trait male bee attraction as a measure of parasite per- and SI Appendix, Table S5), confirming the parasites’ ability to formance (17). Cafeteria and transplant experiments conducted in the track the pheromones of their coevolved hosts. Oregon coastal dunes were performed from 10:00 to 17:00 hours during February through April 2010 to 2015. For the experiments testing perching Behavioral Adaptation. We observed that H. miserabilis males at heights, triungulins were released in the bottom of a vial containing a the Oregon sand dunes site typically flew 2 to 12 cm above the vertical 60-cm-long wooden dowel, and allowed to climb to their preferred height inside a screen cage. Each cafeteria experiment presented all test ground while patrolling for emerging females, whereas H. pallida males in the Mojave Desert flew 10 to 40 cm above the ground while patrolling for emerging females. Consequently, we tested whether the two populations of triungulins varied in their pre- ferred aggregation heights. We found that Oregon triungulins aggregated at a significantly lower height in their native Oregon range (n = 53) and maintained this height when transplanted as egg masses and allowed to emerge at the Mojave Desert site (Fig. 5 and SI Appendix, Fig. S1 and Table S6). Similarly, Mojave triungulin aggregations (n = 53) studied in the Mojave Desert showed an innate preference for significantly higher aggregation heights than Oregon triungulins, even when transplanted as egg masses to the Oregon site (n = 23) (Fig. 5). Thus, local stimuli are not needed to elicit this response (15), and selection of high or low aggregation sites is not due to the heights of available perches or temperature. Instead, perching height appears to be genetically determined, as an adaptation to the heights at which local male bees typically patrol (SI Appendix, Fig. S1). Discussion We identified two separate but complementary local adaptations by two geographically isolated populations of a cleptoparasitic Fig. 5. Triungulin aggregations show local adaptation in perching heights. The aggregation heights of M. franciscanus larvae from the Mojave Desert species, which allow these parasites to exploit two different host (34.94 ± 2.35 cm SEM) and tested in the Mojave Desert and in the Oregon species in different parts of their range. Our reciprocal trans- coastal sand dunes were not significantly different (gold bars). Similarly, the plant experiments confirmed that these traits reflect innate ge- aggregation heights of M. franciscanus from coastal Oregon (8.07 ± 0.70 cm netic differences, rather than phenotypic plasticity, in response SEM) and tested in the Mojave Desert and in coastal sand dunes in Oregon to differences in environmental conditions between coastal sand were not significantly different (blue bars). The aggregation heights of the ’ Mojave Desert and the Oregon populations are significantly different at dunes in Oregon and California s Mojave Desert. Further studies – < = are needed to investigate the selection pressures that the para- both sites (Kruskal Wallis: P 0.001, n 53) (SI Appendix, Table S6). The different lowercase letters (a and b) indicate that the perching heights of the sites exert on the chemical signaling and reproductive behaviors Meloe larvae from different populations (Mojave, OR) were significantly of their hosts and, in turn, the reciprocal selection on the parasites different from each other when tested in different locations (transplant themselves. Understanding the evolutionary forces that shape experiments).

4of5 | www.pnas.org/cgi/doi/10.1073/pnas.1718682115 Saul-Gershenz et al. Downloaded by guest on September 24, 2021 subjects simultaneously within each trial, along with a control. After each 280 °C, hold time 30 min. Sample aliquots (1 μL) were injected in trial was performed, positions of test treatments were randomly changed splitless mode. for each subsequent trial. To determine the positions and stereochemistries of the alkene double For bioassays of attractants, lures consisted of 11-mm gray rubber serum bonds, 100-μL aliquots of extracts were concentrated to dryness under a stoppers (The West Company) or 1-cm squares of dark brown felt cloth stream of nitrogen and treated with five drops of a solution of meta- treated with hexane extracts of heads, thoraxes, and abdomens of male and chloroperbenzoic acid in methylene chloride (2 mg/mL) for 2 h at room female H. pallida or H. miserabilis or local M. franciscanus triungulins, or temperature. The solutions were concentrated to dryness and the residue hexane solutions of test compounds (typical dose 0.5 mg of the major was treated with 200 μL 1 M aqueous NaOH and 0.5 mL hexane, and vor- component, with amounts of the minor components corresponding to ratios texed. The hexane layer was removed and dried by passage through a plug found in the extracts). Controls consisted of identical materials (stoppers, felt of anhydrous Na2SO4 in a pipette, concentrated, and analyzed by GC-MS. cloth, cages, pins, or wood dowels) treated with hexane rather than extracts The positions of the derivatized double bonds were determined from the or compounds. Bioassays tested attraction to (i) live bees or triungulins in enhanced fragments from cleavage on either side of the epoxide, and the cages, versus empty cage controls; (ii) attraction to bee body parts, versus geometries of the former double bonds were determined by epoxidation of controls; and (iii) triungulin aggregations from Oregon, versus triungulin authentic standards and matching of retention times. Epoxides from the (Z)- aggregations from Kelso Dunes. At Kelso Dunes, lures, controls (hexane- and (E)-alkenes separated to baseline, with the (E)-isomers eluting first. treated septa or empty screen cubes), and cages were placed at ∼60-cm height above the sand surface on the dowels within the Meloe aggrega- Straight-chain alkanes were obtained from commercial sources. (Z)-9- tion height range in patches of plants located in dune depressions that were heneicosene and (Z)-9-tricosene were purchased from Sigma Chemical and regularly visited by bees of both sexes. At the Oregon coastal dunes site, Lancaster Synthesis, respectively. Other (Z)-alkenes were synthesized by (Z)- lures, controls, or cages were placed on insect pins on the ground. For all selective Wittig reactions, and purified by recrystallization as previously bioassays, the number of male bee inspection hovers and contacts to cages described (19). The synthetic reconstruction of compounds identified in ex- containing live bees or triungulins, lures, or controls, and the behaviors of tracts of the Oregon triungulins and the H. miserabilis females consisted of bees around the various treatments, were recorded by visual observations. tricosane (0.5 mg), (Z)-7-tricosene (0.5 mg), (Z)-9-tricosene (1 mg), and (Z)- These behaviors were defined as male bees approaching females, triungulin 11-tricosene (0.5 mg), loaded onto gray rubber septa as a hexane solution. aggregations, a lure, or cage, within 10 cm for ≥2 s (11, 18). The bees are sexually dimorphic, so sex of responding could be determined vi- Statistical Analysis. To analyze our transplant experiments, we used R (20) and sually at close range. Real-time notes were used for analysis. Subsets of lme4 (21) to perform a linear mixed effects analysis of the relationship be- responding animals were caught with a sweep net to verify that they were tween the mean inspections and contacts of male bees to the local versus males, and bioassays were recorded by digital video as vouchers. Heights of transplanted Meloe aggregations. The random effects specified were trial natural aggregations were measured at both study sites. Air temperature location and type of aggregation (origin) [glm(formula = meanIC ∼ trialloc + EVOLUTION and wind speed for each experiment were recorded with a Kestrel 4500 type, family = poisson, data = choice.dat)]. To analyze our cafeteria exper- anemometer. iments with live bees and Meloe aggregations, as well as female and male bee heads, abdomens, and thoraxes, we used the Friedman test (two-way Chemical Analysis. Individual male and female H. pallida and H. miserabilis nonparametric ANOVA) followed by a posteriori multiple comparisons using bees were collected with a sweep net and held in glass vials. After brief the Nemenyi multiple comparison test for unreplicated blocked data with chilling, each bee was dissected into head, thorax, and abdomen, and each the PMCMR package (version 4.0.0, 2018) in R (22). The synthetic lure ex- body part was extracted with 1 mL hexane for 1 h. Triungulin aggregations periments were analyzed with a Poisson GLIMMIX procedure in SAS (23). The were extracted in 1 mL hexane for 1 h. The resulting extracts from bee body P values for the Friedman tests used to analyze Habropoda and Meloe ag- parts or triungulins were transferred to clean vials for use in bioassays, or gregation and different body part attractiveness to male H. miserabilis bees stored at −20 °C until analyzed by coupled gas chromatography–mass were derived from tables (24). spectrometry (GC-MS). Extracts of triungulin aggregations, and of heads, thoraxes, and abdomens ACKNOWLEDGMENTS. We thank the following for their generous support of male and female H. miserabilis, were analyzed by GC-MS, using a Hewlett- of this research: The Community Foundation’s Desert Legacy Fund, California Packard 6890 gas chromatograph interfaced to an HP 5973 mass selective Desert Research, Disney Wildlife Conservation Fund, Sean and Anne Duffey detector (Hewlett-Packard) operated in electron impact ionization mode and Hugh and Geraldine Dingle Research Fellowship, and Department of (70 eV). The gas chromatograph was equipped with a DB-5MS column Entomology and Nematology, University of California, Davis fellowships. (20 m × 0.2 mm i.d., 0.25-μm film thickness; J&W Scientific), with helium We thank S. Nadler, J. Rosenheim, J. D. Pinto, R. Westcott, N. Gershenz, carrier gas with a flow rate of 37 cm/s, injector and transfer line tempera- and C. Pagan and our reviewers for helpful suggestions and comments that tures of 280 °C, and a temperature program of 100 °C/0 min, 15 °C/min to greatly improved this manuscript.

1. Smadja C, Butlin RK (2009) On the scent of speciation: The chemosensory system and 14. Roelofs WL, Rooney AP (2003) Molecular genetics and evolution of pheromone bio- its role in premating isolation. Heredity (Edinb) 102:77–97. synthesis in Lepidoptera. Proc Natl Acad Sci USA 100:9179–9184. 2. Johansson BG, Jones TM (2007) The role of chemical communication in mate choice. 15. Pigliucci M, Murren CJ (2003) Perspective: Genetic assimilation and a possible evolu- Biol Rev Camb Philos Soc 82:265–289. tionary paradox: Can macroevolution sometimes be so fast as to pass us by? Evolution 3. Haynes KF, Gemeno C, Yeargan KV, Millar JG, Johnson KM (2002) Aggressive chemical 57:1455–1464. mimicry of moth pheromones by a bolas spider: How does this specialist predator 16. Calflora (2014) Information on California Plants for Education, Research and Con- attract more than one species of prey? Chemoecology 12:99–105. servation (The Calflora Database, Berkeley, CA). Available at www.calflora.org/. Ac- 4. Saul-Gershenz LS, Millar JG (2006) Phoretic nest parasites use sexual deception to cessed July 17, 2018. obtain transport to their host’s nest. Proc Natl Acad Sci USA 103:14039–14044. 17. Kawecki TJ, Ebert D (2004) Conceptual issues in local adaptation. Ecol Lett 7: 5. Hamilton WD, Axelrod R, Tanese R (1990) Sexual reproduction as an adaptation to 1225–1241. resist parasites (a review). Proc Natl Acad Sci USA 87:3566–3573. 18. Kullenberg B (1973) Field experiments with chemical sexual attractants on aculeate 6. Gandon S, Michalakis Y (2002) Local adaptation, evolutionary potential and host– Hymenoptera males II. Zoon (Suppl 1):31–41. parasite coevolution: Interactions between migration, mutation, population size and 19. Ginzel MD, Blomquist GJ, Millar JG, Hanks LM (2003) Role of contact pheromones in generation time. J Evol Biol 15:451–462. mate recognition in Xylotrechus colonus. J Chem Ecol 29:533–545. 7. Lively CM (1999) Migration, virulence, and the geographic mosaic of adaptation by 20. R Core Team (2013) R: A Language and Environment for Statistical Computing parasites. Am Nat 153:S34–S47. (R Found Stat Comput, Vienna). Available at www.R-project.org/. Accessed May 8. Pinto JD, Selander RB (1970) The bionomics of blister beetles of the genus Meloe and 17, 2018. a classification of the New World species. Illinois Biol Mon 42:1–222. 21. Bates D, Maechler M, Bolker B, Walker S (2015) Fitting linear mixed-effects models 9. Siebold CT (1841) Ueber die Larven der Meloiden. Stettiner Entomol Ztg 9:130–136. using lme4. J Stat Softw 67:1–48. 10. Pinto JD (1982) The larva of Meloe franciscanus Van Dyke and MacSwain’s Meloe sp. a 22. Pohlert T (2018) PMCMRplus: Calculate Pairwise Multiple Comparisons of Mean Rank (Coleoptera: Meloidae). Pan-Pac Entomol 58:367–368. Sums Extended, Version 4.0.0. Available at https://rdrr.io/cran/PMCMRplus/. Accessed 11. Saul-Gershenz L (2017) Host range evolution of cleptoparasite Meloe franciscanus May 20, 2018. (Coleoptera: Meloidae). PhD dissertation (University of California, Davis). 23. SAS (2013) SAS/STAT 9.3 User’s Guide (SAS Institute Inc., Cary, NC). 12. Hafernik J, Saul-Gershenz L (2000) Beetle larvae cooperate to mimic bees. Nature 405: 24. Lopez-Vazquez C, Hochsztain E (2017) Extended and updated tables for the Friedman 35–36. rank test. Commun Stat Theory Methods. Available at https://www.tandfonline.com/ 13. Kather R, Martin SJ (2015) Evolution of cuticular hydrocarbons in the Hymenoptera: A action/doSearch?AllField=10.1080%2F03610926.2017.1408829. Accessed June meta-analysis. J Chem Ecol 41:871–883. 28, 2018.

Saul-Gershenz et al. PNAS Latest Articles | 5of5 Downloaded by guest on September 24, 2021