Role of Seed Size, Phenology, Oogenesis and Host Distribution In

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Diposit Digital de Documents de la UAB

Integrative Zoology 2018; 13: 267–279 doi: 10.1111/1749-4877.12293

1 ORIGINAL ARTICLE 1 2 2 3 3 4 4 5 5 6 6 7 Role of seed size, phenology, oogenesis and host distribution in the 7 8 8 9 specificity and genetic structure of seed weevils Curculio( spp.) in 9 10 10 11 mixed forests 11 12 12 13 13 14 Harold ARIAS-LECLAIRE1,2 Raúl BONAL,3,4 Daniel GARCÍA-LÓPEZ5 and Josep Maria 14 15 1 15 16 ESPELTA 16 17 1CREAF, Centre for Ecological Research and Forestry Applications, Cerdanyola del Vallès, 08193, Spain, 2Exact and Natural 17 18 Sciences School, State Distance University of Costa Rica, UNED, Mercedes de Montes de Oca, San José, Costa Rica, 3Forest 18 19 19 Research Group, INDEHESA, University of Extremadura, Plasencia, Spain, 4DITEG Research Group, University of Castilla-La 20 20 Mancha, Toledo, Spain and 5Zoology Department, Autonomous University of Barcelona, Cerdanyola del Vallès, Spain 21 21 22 22 23 23 24 Abstract 24 25 25 26 Synchrony between seed growth and oogenesis is suggested to largely shape trophic breadth of seed-feeding in- 26 27 sects and ultimately to contribute to their co-existence by means of resource partitioning or in the time when in- 27 28 festation occurs. Here we investigated: (i) the role of seed phenology and sexual maturation of females in the 28 29 host specificity of seed-feeding weevils (Curculio spp.) predating in hazel and oak mixed forests; and (ii) the 29 30 consequences that trophic breadth and host distribution have in the genetic structure of the weevil populations. 30 31 DNA analyses were used to establish unequivocally host specificity and to determine the population genetic 31 32 structure. We identified 4 species with different specificity, namely Curculio nucum females matured earlier and 32 33 infested a unique host (hazelnuts, Corylus avellana) while 3 species (Curculio venosus, Curculio glandium and 33 34 Curculio elephas) predated upon the acorns of the 2 oaks (Quercus ilex and Quercus pubescens). The high spec- 34 35 ificity of C. nucum coupled with a more discontinuous distribution of hazel trees resulted in a significant genet- 35 36 ic structure among sites. In addition, the presence of an excess of local rare haplotypes indicated that C. nucum 36 37 populations went through genetic expansion after recent bottlenecks. Conversely, these effects were not ob- 37 38 served in the more generalist Curculio glandium predating upon oaks. Ultimately, co-existence of weevil spe- 38 39 cies in this multi-host-parasite system is influenced by both resource and time partitioning. To what extent the 39 40 restriction in gene flow among C. nucum populations may have negative consequences for their persistence in a 40 41 time of increasing disturbances (e.g. drought in Mediterranean areas) deserves further research. 41 42 42 Key words: Corylus avellana, Curculio spp., genetic structure, Quercus spp., trophic breadth 43 43 44 44 45 45 46 46 47 INTRODUCTION 47 48 48 49 Correspondence: Josep Maria Espelta, CREAF, Cerdanyola del Seed predation by insects may play a crucial role in 49 50 Vallès, Spain. plant population dynamics, by reducing the reproductive 50 51 Email: [email protected] output (Bonal et al. 2007; Espelta et al. 2008) and con- 51

© 2017 The Authors. Integrative Zoology published by International Society of Zoological Sciences, 267 Institute of Zoology/Chinese Academy of Sciences and John Wiley & Sons Australia, Ltd This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. H. Arias-LeClaire et al.

1 straining the regeneration process (Espelta et al. 2009b). seeds of different hosts in multi-specific systems has 1 2 Trophic breadth and specificity of seed-feeding insects been seldom explored. 2 3 is often explained by differences among plant hosts in The breadth of the trophic niche of seed-feeding in- 3 4 chemical or morphological traits (Bernays & Graham sects (specialist vs generalist) may influence the num- 4 5 1988; Forister et al. 2015). Differences in phytochemis- ber of species that predate upon different seeds and it 5 6 try (mainly nitrogen-based defensive compounds) have has consequences for the dynamics of the communi- 6 7 been observed to be highly relevant for the diversifi- ty of hosts (Espelta et al. 2009b). However, beyond the 7 8 cation of phytophagous insects and their diet breadth effects on plant fitness, differences in the trophic niche 8 9 (Kergoat et al. 2005). Concerning other seed features, may also influence the population dynamic of the seed 9 10 size is a trait claimed to influence ecomorphological di- consumers (Ylioja et al. 1999) depending on life-his- 10 11 versification in many endophytic insects (e.g. body size tory traits such as dispersal ability and landscape attri- 11 12 and rostrum shape), promoting differences in their tro- butes (i.e. abundance and spatial distribution of hosts). 12 13 phic niche, ecological adaptations and species radiation Spatial connectivity among plant-hosts has been shown 13 14 (Hughes & Voegler 2004a; Bonal et al. 2011; Pegue- to be especially relevant for insects with low disper- 14 15 ro et al. 2017). In addition to chemical and morpholog- sal ability (Thomas et al. 2001; Kruess 2003), resulting 15 16 ical differences, seeding phenology and stochasticity in a stronger genetic structure and reduced gene flow in 16 17 in the availability of this resource have also been sug- the insect populations located on more isolated hosts. In 17 18 gested as key factors influencing the guild of insect spe- the long run, host isolation may even result in coloni- 18 19 cies predating upon a particular plant host (Espelta et zation credits for some insect species, especially those 19 20 al. 2008, 2009b; Coyle et al. 2012; see also Pélisson et with a narrower diet breadth (Ruiz-Carbayo et al. 2016) 20 21 al. 2013a). As insects are short-living organisms, syn- and poor dispersal ability (Pélisson et al. 2013b; Hei- 21 22 chronization of their life-cycle with the resources upon neger et al. 2014). Conversely, generalist species may 22 23 which they depend is critical (Bale et al. 2002, 2007; show a more continuous distribution in the landscape, 23 24 Hood & Ott 2010). Therefore, processes such as adult benefiting from the spatial overlap of the different host 24 25 emergence (Espelta et al. 2017) and oogenesis (Trudel plants upon which they feed (Newman & Pilson 1997), 25 26 et al. 2002; Son & Lewis 2005) have to be tightly con- and show no genetic structure differences among popu- 26 27 nected with the presence of seeds for oviposition (Bonal lations owing to gene flow. Interestingly, for seed-feed- 27 28 et al. 2010). In particular, oogenesis (i.e. egg maturation ing insects a comparison of the genetic structure of their 28 29 in females) has been predicted to differ depending on populations and the spatial structure of their potential 29 30 the stochasticity of seeds availability. Thus, proovigen- hosts could provide strong evidence about differences 30 31 esis (i.e. females have already mature eggs at the onset of the trophic niche breadths. Moreover, the use of mo- 31 32 of their adult life) would be favored in species predating lecular techniques (DNA barcoding) may help to detect 32 33 upon hosts that regularly produce seeds while synovi- cryptic speciation and trophic niche segregation among 33 34 genesis (i.e. females start their adult life with imma- morphologically similar species (Peguero et al. 2017), 34 35 ture eggs) would be advantageous for species exposed and also to establish species specificity in an unequiv- 35 36 to more random fluctuations of seed production (Jervis ocal way in comparison to classifications based on the 36 37 et al. 2008; Richard & Casas 2009), as they can better presence or absence of a species on a particular plant, 37 38 adjust the amount of energy invested in reproduction to especially when the lack of morphological differences 38 39 the amount of seeds (but see Pélisson 2013b). Ultimate- at certain stages (e.g. larvae) make species identification 39 40 ly, the co-existence of the different seed consumers in impossible otherwise (Govindan et al. 2012). Unfortu- 40 41 a multi-host community could be mediated by resource nately, this combination of landscape ecology (i.e. host 41 42 partitioning (e.g. insects predate preferentially upon dif- connectivity) and population genetics when studying the 42 43 ferent species according to different seed traits; see Es- breadth of the trophic niche and dispersal ability of phy- 43 44 pelta et al. 2009a), time-partitioning (e.g. insects exhibit tophagous insects remains largely unexplored. 44 45 45 differences in life span and the timing of seed predation; The main aims of this study have been to investigate 46 46 see Pélisson et al. 2012) or the trade-off among disper- in a multi-host and multi-seed-predator system the role 47 47 sal versus dormancy ability to cope with resource scar- of seed size, seed phenology and ooegenesis in the host 48 48 city (Pélisson et al. 2012). Yet, the importance of the specificity of seed-parasite weevils (Curculio spp.) and 49 49 interplay among seed size, seeding phenology and oo- to analyze the consequences that potential differences 50 50 genesis in driving the guild of insects predating upon in trophic specialization and host distribution may have 51 51

Causes of host specificity in seed weevils

1 in the genetic structure of weevil populations. Curculio Study area and species 1 2 spp. (Coleoptera: Curculionidae) are seed parasites that 2 3 differ in their dispersal ability (Venner et al. 2011), dia- The study was carried out in mixed forests with the 3 4 pause duration (Pélisson et al. 2013a,b), oogenesis (with presence of oaks (Q. ilex and Q. pubescens and com- 4 5 both proovigenic and synovigenic species; Pélisson et mon hazel trees (Co. avellana) in Catalonia (north-east 5 6 al. 2013a) and the breadth of their trophic niche (Muñoz Spain, Fig. 1). The evergreen Q. ilex and the winter-de- 6 7 et al. 2014; Bonal et al. 2015; Peguero et al. 2017). We ciduous Q. pubescens are extensively distributed in pure 7 8 conducted this study in Catalonia (northeast Spain) in and mixed forests in all the western rim of the Mediter- 8 9 mixed forests including oaks (Quercus ilex, Quercus pu- ranean basin (Espelta et al. 2008), while the common 9 10 bescens) and common hazel trees (Corylus avellana) hazel (Co. avellana) often appears in scattered groups 10 11 with 4 different weevil species present (Curculio nucum in mixed deciduous forests or cultivated in monospe- 11 12 Linnaeus, Curculio glandium Marsham, Curculio veno- cific stands (AliNiazee 1998). Acorns in Quercus spp. 12 13 sus Gravenhorst and Curculio elephas Gyllenhaal). In- and hazelnuts in Co. avellana mature in 1 year and both 13 14 terestingly, in this region oaks show a much more con- are subjected to intense pre-dispersal seed predation 14 15 tinuous distribution and later seeding, while hazels often by weevils (Curculio spp.), a group of granivorous in- 15 16 appear in more discontinuous patches and have an earli- sects extensively distributed in the northern hemisphere 16 17 er production of fruits (Gracia et al. 2004). Concerning (Hughes & Voegler 2004a). In Catalonia, the most com- 17 18 weevils, the 4 species overwinter underground, but they mon weevil species predating upon acorns are C. glan- 18 19 differ in the duration of their diapause, the phenology of dium and C. elephas (Espelta et al. 2009b), the lat- 19 20 emergence, oogenesis and dispersal ability. Adults of C. ter also depredating upon chestnuts (Castanea spp.), 20 21 glandium, C. venosus and C. nucum emerge in spring 2 while in hazelnuts the unique species described up to 21 22 years after larvae buried into the soil, while C. elephas now has been C. nucum, a hypothesized highly specif- 22 23 exhibits variable diapause and adults emerge in early au- ic seed parasite (Guidone et al. 2007; Bel-Venner et al. 23 24 tumn (Bonal et al. 2010; Espelta et al. 2017) for up to 3 24 25 years (Pelisson et al. 2013b). Concerning oogenesis, in 25 26 C. glandium, C. venosus and C. nucum females are re- 26 27 productively immature (synovigenic) and ovarian devel- 27 28 opment is accomplished after 1 or 2 months of the feed- 28 29 ing period (Bel-Venner et al. 2009), while C. elephas 29 30 females are proovigenic and food intake is not required 30 31 for ovarian development (Pélisson et al. 2012). Regard- 31 32 ing host selection, previous studies have suggested that 32 33 C. nucum is highly specialized in hazelnuts (Bel-Ven- 33 34 ner et al. 2009), while the other weevils depredate upon 34 35 several oak species (Muñoz et al. 2014). However, these 35 36 results have not been confirmed by means of DNA anal- 36 37 yses as no study has been conducted in mixed hazel– 37 38 oak forests. Considering the traits of the species in- 38 39 volved in this multi-host and multi-predator system and 39 40 the spatial distribution of hosts, we hypothesize that: (i) 40 41 seed size and the synchronization of seeding phenology 41 42 and oogenesis will be responsible for the guild of wee- 42 43 vils predating upon the different plants; and (ii) the nar- 43 44 rower trophic breadth of C. nucum and the more patchy 44 45 distribution of hazels in comparison to the more gener- 45 46 alist habit of the other weevils and the continuous dis- Figure 1 (a) Location of study sites in Catalonia (north-east 46 47 tribution of oaks will result in differences in the genetic Spain). (b) Distribution of Corylus avellana, Quercus ilex and 47 48 structure of weevil populations of these species. Quercus pubescens according to the presence of this species in 48 49 plots inventoried in the Catalan Forest Inventory (Gracia et al. 49 50 MATERIALS AND METHODS 2004). RI, Ripoll (5 plots); OL, Olot (4 plots); MO, Montseny (4 50 51 plots); MA, Maresme (5 plots); PR, Prades (5 plots). 51

H. Arias-LeClaire et al.

1 2009). However, it must be highlighted that except for scars. We calculated the volume of both sound and in- 1 2 the weevil species predating upon oaks, ascription of fested seeds by measuring the length and width to the 2 3 weevil species to a plant host is based on the observa- nearest 0.01 mm with a digital caliper (see Espelta et 3 4 tion of adults in the foliage of that particular species, but al. 2009a). Infested seeds were placed individually in 4 5 no study has addressed this issue comprehensively (e.g. plastic trays for individual monitoring. Each seed was 5 6 identifying by means of molecular techniques the spe- checked daily to register the emergence of larvae, which 6 7 cies of the larvae inside chestnuts or hazelnuts). were immediately transferred to 2 mL Eppendorf with 7 8 During early summer on hazelnuts (AliNiazee 1998) 96% alcohol. Once larvae stop emerging (approximate- 8 9 9 and early autumn on acorns (see Bonal & Muñoz 2009) ly 3 weeks after seeds were collected) seeds were dis- 10 10 female weevils perforate the seed cover with their snout sected to check for the presence of non-emerged larvae. 11 11 and oviposit commonly a single egg so the larvae devel- From hazelnuts only Curculio larvae emerged, while for 12 12 op feeding on the seed kernel. At the middle of summer acorns the 6% of larvae corresponded to the Cydia spp. 13 13 in C. nucum (Bel-Venner et al. 2009) or late autumn in C. moth. 14 14 elephas and C. glandium (Espelta et al. 2009a), larvae In parallel, during the abovementioned field cam- 15 15 exit the seed and bury into the ground to overcome the paigns adult weevil were captured by shaking the can- 16 16 17 diapause period and undergo full metamorphosis. opy and collecting the fallen individuals in an inverted 17 umbrella held beneath the foliage for species identifi- 18 Sampling design 18 19 cation at the laboratory. To establish whether females 19 In 2013 we established a total of 23 sampling plots were sexually mature, they were dissected under a mi- 20 grouped into 5 geographical clusters (Sites) in a north 20 21 croscope to observe abdominal segments and ovary ma- 21 to south latitudinal gradient (see Fig. 1). This sampling turity. We considered the presence of eggs as a sign of 22 procedure was selected to account for the possible ef- 22 23 female ready for oviposition and the absence of eggs as 23 fects of latitude on the duration of the vegetative sea- 24 females that were still immature or had already ovipos- 24 son and, thus, on the seeding phenology of oaks and ha- 25 ited (Pélisson et al. 2013a). 25 zelnuts, their overlap and the overlap among these host 26 26 species and the weevils predating upon their seeds. Pre- DNA barcoding and larval species identification 27 27 sumably a tighter vegetative season in northern and A total of 1657 Curculio larvae emerged from ha- 28 28 colder sites would lead to more similar patterns of seed zelnuts and acorns. In order to establish unequivo- 29 29 production while these could be more relaxed and lon- cal trophic relationships between insects and their host 30 30 31 ger in southern and warmer places. Ultimately, this plants we used molecular techniques (DNA barcoding) 31 32 could lead to differences in the guild of weevils predat- as larvae cannot be determined according to morpho- 32 33 ing upon these plants. Plots were selected by searching logical characters. Therefore, from 342 larvae select- 33 34 for the presence of trees of Co. avellana and Q. ilex or Q. ed randomly among the ones emerged in the labora- 34 35 pubescens based on the Catalan Forest Inventory (Gracia tory from the 3 hosts we extracted DNA from a small 35 36 et al. 2004) and field observations of their reproductive piece of larval tissue (approximately 2-mm long) us- 36 37 status (i.e. presence of seeds). From late July (end of ha- ing the NucleoSpin-Tissue kit according to the manu- 37 38 zel seeding season) to early October (end of the acorn facturer’s instructions (MACHEREY-NAGEL GmbH, 38 39 crop) we carried out 3 sampling campaigns: (i) late July Düren, Germany; www.mn-net.com). We amplified a 39 40 to early August; (ii) late August to early September; and fragment (826 bp) of the mitochondrial cytochrome oxi- 40 41 (iii) late September to early October to account for pos- dase subunit 1 (cox1) using primers Pat and Jerry (please 41 42 sible differences in the phenology of seed infestation by see Hughes & Vogler [2004b] for details on primer se- 42 43 the different weevil species present. In every plot and quences and PCR protocols). We chose this fragment of 43 44 in each sampling period we randomly collected a mini- cox1 due to the availability of many reference sequenc- 44 45 mum of 100 seeds from each species (Co. avellana and es from correctly determined adults of European Cur- 45 46 Quercus spp.) under the canopies of several randomly culio spp. For comparison (Hughes & Vogler 2004b). 46 47 selected trees. Seeds were taken to the laboratory and Sequencing was performed using Big-Dye (Perkin-El- 47 48 classified as sound or infested to assess infestation rates mer) technology and an ABI3700 sequencer. Sequence 48 49 per species and sampling period. Infested seeds are eas- chromatograms were assembled and edited using Se- 49 50 ily recognizable by the presence of female oviposition quencher 4.6 (Gene Codes, Ann Arbor, MI, USA). For 50 51 51

Causes of host specificity in seed weevils

1 species identification we discarded those sequences that (Dupanloup et al. 2002). This method identifies the op- 1 2 after edition were shorter than 500 base pairs. Edited se- timal grouping option (K) that maximizes the among- 2 3 quences were aligned using CLUSTALW supplied via group component (FCT) of the overall genetic variance. 3 4 http://align.genome.jp, with default gap open and gap We defined the number of populations (K) and ran 100 4 5 extension penalties. The alignment sets were collapsed simulated annealing processes. We simulated different 5 6 into unique haplotypes and each of this compared to numbers of populations, ranging from K = 2 to K = 4, to 6 7 the Curculio spp. reference sequences available at Gen- determine the best population clustering option. 7 8 Bank. We applied the most conservative limit used in 8 9 DNA barcoding, which states a maximum genetic diver- RESULTS 9 10 gence (number of different nucleotides by the total num- 10 11 ber of compared nucleotides) of 1% with respect to the Molecular analyses allowed the identification of the 11 12 reference sequence for an unambiguous identification to larvae emerged from the seeds of the 3 host species (Co. 12 13 the species level (Ratnasingham & Herbert 2007). avellana, Q. ilex and Q. pubescens) as all sequences 13 14 showed a divergence below 1% with respect to Curcu- 14 15 Data analysis lio spp. reference sequences from GenBank. This diver- 15 16 To evaluate the occurrence of the different wee- gence was much lower than inter-specific differences, 16 17 vil species in the 3 potential hosts Co. avellana, Q. ilex which in all cases exceeded 8%. All larvae correspond- 17 18 and Q. pubescens) across the 5 study sites, we conducted to 4 species; namely, C. elephas, C. glandium, C. nu- 18 19 ed a Pearson’s χ2-test. Similarly, we used χ2-tests for the cum and C. venosus. As shown in Figure 2, weevil spe- 19 20 comparison of the presence of male and female wee- cies were not randomly distributed among hosts; that 20 21 vil proportion, as well as that of immature and mature is, C. nucum was exclusively present in hazelnuts while 21 2 22 females among sampling periods. A generalized linear the other 3 weevils emerged uniquely from acorns (χ 6 22 23 mixed model (GLMM), following a binomial distribu- = 263.9, P < 0.001). C. glandium and C. elephas were 23 2 24 tion, was used to test for the effects of the study site (RI, more abundant in Q. ilex (respectively, χ 14 = 91.8, P < 24 2 25 Ripoll; OL, Olot; MO, Montseny; MA, Maresme; PR, 0.001, and χ 8 = 23.3, P < 0.001) while there were not 25 26 Prades), sampling period (1, 2, 3) and host species (Co. significant differences in the presence of Curculio ve- 26 2 27 avellana, Q. ilex and Q. pubescens) on the seed preda- nosus between the 2 oak species (χ 5 = 7.47, P > 0.05). 27 28 tion rate by weevils. The factor “plot” was included as The different presence of larvae of the 4 weevil species 28 29 a random effect in the GLMM analyses to account for 29 30 the repeated nature of the measurements and other un- 30 31 explained variation. Analyses of deviance Type II Wald 31 32 χ2-tests were performed to establish the significance of 32 33 each different independent variable in the model. A gen- 33 34 eral linear mixed model was applied to test for the ef- 34 35 fects of host species, sampling period and seed condi- 35 36 tion (sound or infested) on seed size (volume in mm3) 36 37 with the factor “plot” included as a random effect. 37 38 For population genetic analyses we chose those spe- 38 39 cies in which there were a minimum of 10 individu- 39 40 als per population with sequences longer than 750 bp; 40 41 namely, Curculio glandium and Curculio nucum. We 41 42 42 used Arlequin software (Excoffier et al. 2005) to calcu- 43 late standard molecular diversity indices (gene diversi- 43 44 ty and nucleotide diversity) and to perform analyses of 44 45 the molecular variance (AMOVAs). Signatures of pop- 45 46 ulation demographic changes (bottlenecks or expan- 46 47 sions) were examined by Tajima’s D (Tajima 1989) and 47 48 48 Fu’s F (Fu 1997) as implemented in Arlequin software. Figure 2 Proportion of the different weevil species infesting 49 We also tested whether there was any geographic pat- the seeds of the 3 host species (Corylus avellana, Quercus ilex 49 50 50 tern in the population genetic structure using Samova 1.0 and Quercus pubescens) according to the DNA analyses of the 51 larvae emerging from the seeds. 51

H. Arias-LeClaire et al.

1 in the 3 hosts, especially among hazelnut and the 2 oaks, 1 2 was not due to the lack of a particular species in a given 2 3 site as we captured adult specimens of all weevil species 3 4 along the entire gradient. Moreover, as hazelnuts were 4 5 bigger than acorns during almost the entire seeding sea- 5 6 son (Table 1 and Fig. 3), the abovementioned differenc- 6 7 es in weevil specificity among these 2 groups of hosts 7 8 does not seem to be due to the exclusion of certain wee- 8 9 vil species from hazelnuts by a too small seed size. 9 10 Male and female weevils occurred in nearly the same 10 11 frequency with no significant variation along the sam- 11 12 2 12 pling periods (χ 2 = 2.28, P > 0.05). Yet the propor- 13 tion of females with presence of eggs and without eggs 13 14 2 14 showed significant differences through the season χ( 2 = 15 33.7, P < 0.001) and for the different weevil species. As 15 16 shown in Figure 4, through the season the presence of 16 17 females with eggs was earlier in C. nucum, followed by 17 18 C. glandium and C. elephas. In the 2 synovigenic spe- 18 19 2 19 cies, C. nucum had a decreasing pattern (χ 2 = 10.5, P 20 < 0.01) in the presence of females with eggs, while C. 20 21 2 21 glandium exhibited an increasing pattern (X 2 = 35.3, P 22 < 0.001). In the proovigenic C. elephas we did not find 22 23 females on the very first sampling period but as soon 23 24 as they appeared during the second and third sampling 24 25 2 25 dates they were already sexually mature (χ 2 = 16.5, P 26 < 0.001). Consistently with the seasonal patterns of the 26 27 presence of females ready to oviposit, we found that in- 27 28 festation rates showed significant variation among study 28 29 2 2 29 sites (χ 4 = 16.5, P < 0.001), sampling periods (χ 1 = 5.5, 30 2 30 P < 0.05) and host species (χ 2 = 6.4, P < 0.05). Over- 31 all, infestation was higher in northern localities and it 31 32 increased as the seeding season progressed (see coeffi- 32 33 cients for the different effects in Table 2). Concerning 33 34 host species, infestation rates showed contrasting tem- 34 35 poral patterns in hazelnut versus oaks (Table 2, Fig. 5), 35 36 in agreement with host seeding phenology and oogene- 36 37 37 38 38 39 39 40 40 Table 1 Estimates for the significant effects of tree host, sam- 41 41 pling period and seed condition (sound or infested) over seed 42 42 size (volume in mm3) according to the linear mixed model 43 43 44 Effects Estimate Standard error t-value 44 Intercept 3.199 0.0278 153.912*** 45 Figure 3 Proportion of females with eggs (black column) and 45 46 without eggs (white columns) for the 3 weevil species (Corylus Date 0.01695 0.004249 3.990*** 46 47 avellana, Quercus ilex and Quercus pubescens) captured in the Host, Q. pubescens −0.1876 0.008793 −21.336*** 47 48 3 sampling periods during the seeding season: Jul–Aug = from 48 Host, Q. ilex −0.4808 0.009984 −48.159*** 49 the end of July to the beginning of August; Aug–Sep = from 49 50 the end of August to the beginning of September; Sep–Oct = Seed condition, sound −0.03143 0.006962 −4.515*** 50 51 from the end of September to the beginning of October. *P < 0.05, **P < 0.01 and ***P < 0.001. 51

Causes of host specificity in seed weevils

1 a Table 2 Estimates for the significant effects of study site, tree 1 2 host and sampling period on weevil infestation rates according 2 3 to the generalized linear mixed model 3 4 Effects Estimate Standard error z-value 4 5 Intercept −2.8280 0.5462 −5.177*** 5 6 6 Site Olot 1.0830 0.4776 2.268* 7 7 8 Site Ripoll 1.5611 0.4788 3.260** 8 9 Host, Q. pubescens 0.9615 0.3810 2.523* 9 10 Date 0.4634 0.1973 2.349* 10 11 11 *P < 0.05, **P < 0.01 and ***P < 0.001. 12 12 13 13 14 14 15 b 15 16 16 17 17 18 18 19 19 20 20 21 21 22 22 23 23 24 24 25 25 26 26 27 27 28 28 29 c 29 30 30 31 31 32 32 33 33 Figure 5 Mean ± SE seed infestation rate of the 3 host species 34 34 (Corylus avellana, Quercus ilex and Quercus pubescens) along 35 35 the sampling dates. Jul–Aug = from the end of July to the be- 36 ginning of August; Aug–Sep= from the end of August to the 36 37 beginning of September; Sep–Oct = from the end of September 37 38 to the beginning of October. 38 39 39 40 40 41 41 42 sis in females; that is, in hazelnuts infestation occurred 42 43 earlier and slightly decreased through the season, while 43 44 3 it was absent during the first sampling date in the 2 oak 44 Figure 4 Mean ± SE volume (mm ) of sound (open dots) and species, and progressively increased towards the end of 45 infested (black dots) seeds of the 3 host species (Corylus avel- 45 the season (Fig. 5). 46 lana, Quercus ilex and Quercus pubescens) along the sampling 46 47 dates. Jul–Aug = from the end of July to the beginning of Au- The population genetic analyses showed marked dif- 47 48 gust; Aug–Sep = from the end of August to the beginning of ferences between C. nucum and C. glandium. Mean ge- 48 49 September; Sep–Oct = from the end of September to the begin- netic diversity was higher in C. nucum (Table 3), main- 49 50 ning of October. Notice the difference in the scale of the y-axis ly due to the higher number of distinct haplotypes; that 50 51 for the 3 host species. is, an ANOVA in which the population was included as 51

H. Arias-LeClaire et al.

1 Table 3 Values of gene diversity, nucleotide diversity, Tajima’s of our study, the first grouping the nearby populations 1 2 D and Fu’s F recorded at each population for Curculio nucum of Montseny (MO) and Maresme (MA) and another one 2 3 (a) and Curculio glandium (b) including the rest (see Fig. 1). No significant geograph- 3 4 (a) Curculio nucum ical pattern of molecular variance was found in C. glan- 4 5 5 Gene Nucleotide dium. 6 diversity diversity Tajima’s D Fu’s F 6 7 7 Ripoll 0.87 0.0015 −1.63* −4.54*** DISCUSSION 8 8 9 Olot 0.75 0.0024 −1.96** −8.34*** Seed infestation by weevils did not occur randomly 9 10 Montseny 0.89 0.0032 −1.40* −6.87*** but with 2 opposite breadths of host specificity; name- 10 11 Maresme 0.88 0.0034 −0.41 −1.20 ly, the highly specialized C. nucum infested a unique 11 host (hazelnuts), while up to 3 species (C. glandium, C. 12 Prades 0.59 0.0012 −1.69** −5.27*** 12 13 elephas and C. venosus) predated almost indistinctive- 13 (b) Curculio glandium 14 ly upon 2 oaks (Q. ilex and Q. pubescens). These dif- 14 15 Ripoll 0.71 0.0012 −1.10 −2.61* ferences in trophic specificity coupled with differenc- 15 16 Olot 0.57 0.0009 −0.46 −0.84 es in the geographical distribution of the hosts resulted 16 17 Montseny 0.69 0.0013 −1.22 −2.61* in 2 distinct patterns concerning the genetic character- 17 18 istics of weevils’ populations; that is, we only found a 18 Maresme 0.69 0.0013 −0.75 −1.95 19 significant genetic structure among the populations in 19 20 Prades 0.63 0.0034 −1.79 1.46 the highly specialist C. nucum. Ultimately, the results of 20 21 *P < 0.05, **P < 0.01 and ***P < 0.001. these genetic analyses confirmed the specialist or gener- 21 22 alist trophic breadth of the different weevil species ac- 22 23 cording to the identification of the larvae found in the 23 24 seeds and they stress how molecular techniques may 24 25 a random factor showed that mean gene diversity was help to establish unequivocal trophic relations for seed 25

26 significantly higher in C. nucum (F1,4 = 9.40; P = 0.03). feeding insects. 26 27 A total of 31 haplotypes were retrieved from the 118 se- Previous studies have suggested that seed size has 27 28 quences of C. nucum included in the analyses versus been a relevant trait promoting ecomorphological adap- 28 29 just 13 from 96 sequences in C. glandium (see Tables tations in the genus Curculio and driving species diver- 29 30 4 and 5). In C. nucum, 48% of the individuals had the sification (Hughes & Vogler 2004a; see also Peguero et 30 31 most common haplotype but there were many rare hap- al. 2017). In the end, a tight relationship between seed 31 32 lotypes sometimes found in just one individual and/or at and weevils’ body size would result in differences in 32 33 a single population (Table S1). In the case of C. glandi- the ability of larger and smaller weevils to infest seeds 33 34 um, there were much fewer rare haplotypes and the two of different size (differences in trophic breadth); that 34 35 most frequent ones were found in 80% of the individu- is, small species would be able to infest both small and 35 36 als (Table S2). The high proportion of rare haplotypes large seeds while species with a larger body size would 36 37 in C. nucum suggests population expansion after recent be restricted to larger seeds to obtain enough resourc- 37 38 bottlenecks and, accordingly, both Tajima’s D and Fu’s es to complete larvae development (see Bonal & Muñoz 38 39 F had negative and significant values in all populations 2008; Espelta et al. 2009a; Bonal & Muñoz 2011; Peg- 39 40 except one. In the case of C. glandium only for Fu’s uero et al. 2017). Yet, this does not seem to be the case 40 41 test were the values significant in two populations, thus in our study system where hazels, the species infested 41 42 showing that most populations were in equilibrium (Ta- by a single species (C. nucum), showed the largest seeds 42 43 jima 1989; Fu 1997). The AMOVA revealed a more re- during most of the season (Fig. 3) and experienced the 43 44 stricted gene flow between populations in the case of C. lower infestation rate (see Table S3). Instead of an in- 44 45 nucum, in which differentiation among populations ex- fluence of seed size, our results suggest that the exclu- 45 46 plained 5.02% of the total molecular variance (degrees sive infestation of hazelnuts by C. nucum could be more 46 47 of freedom = 4; P < 0.01), whereas in C. glandium in- related to a different pattern of sexual maturation of fe- 47 48 ter-population differences were not significant. The re- males among the 2 weevil species emerging from the 48 49 sults of the SAMOVA were marginally significant for C. soil in spring, specifically an earlier maturation in C. nu- 49 50 nucum (FCT = 0.08; degrees of freedom = 1; P = 0.08) cum in comparison to C. glandium (Fig. 4). These dif- 50 51 and defined two clusters within the geographical range 51

Causes of host specificity in seed weevils

1 ferences could be due to differences between the 2 wee- able diapause) to cope with seed scarcity. This seems to 1 2 vils in the requirements of resource acquisition as it has be the case for C. glandium and C. elephas; that is high 2 3 been demonstrated that sexual maturation in females of dispersal ability (up to 11 km) in the former species and 3 4 these synovigenic species critically requires some feed- an extended diapause (up to 3 years) in the later (see 4 5 ing at adulthood before reproductive development takes Venner et al. 2011; Pélisson et al. 2012). Yet, other fac- 5 6 place (Bel-Venner et al. 2009; Pélisson et al. 2012). The tors not covered in this study, such as the risk of parasit- 6 7 early maturation in C. nucum would be advantageous ism or survival of larvae during diapause, may also help 7 8 to oviposit in hazelnuts before the hardening of the nut- equalize their success to infest (Bonal et al. 2011). Sim- 8 9 shell, as this is a fast process occurring during seed ilarly, future studies with more intense and appropriate 9 10 growth and the main mechanism in hazels to avoid in- sampling schemes should address the relationship be- 10 11 festation (Guidone et al. 2007). Moreover, oviposition tween the number of adults of the different species and 11 12 of C. nucum would be expected to occur soon after mat- the number of larvae to disentangle the different preda- 12 13 ing as weevils do not adjust laying eggs to the moment tion rates upon each species and the influence of other 13 14 of highest seed availability, but they oviposit as soon as environmental factors. 14 15 females have mature eggs (Bonal et al. 2010). This be- Ultimately, differences in the trophic breadth leave a 15 16 havior is probably linked to the temporal unpredictabil- contrasting genetic signature in the populations of the 16 17 ity of seed crop size (Bonal & Muñoz 2008; Espelta et 2 species of weevils. A much higher number of local 17 18 al. 2008) and other constraints they have to cope with, rare haplotypes were found in the monophagus C. nu- 18 19 such as the need of rainfall episodes to soften the soil cum, along with a marginally significant genetic struc- 19 20 and allow the emergence of adults (Bonal et al. 2010, ture among populations, contrary to the more general- 20 21 2015; Espelta et al. 2017). Only during the 2 earlier ist C. glandium (see, for a similar example in aphids, 21 22 samplings, females of C. nucum seemed to preferential- Gaete-Eastman et al. 2004). Inter-specific differences 22 23 ly choose bigger seeds, a behavior related with the need in genetic characteristics of phytophagous insects could 23 24 to select a minimum seed size to ensure larvae develop- arise from differences in their dispersal ability or in the 24 25 ment and also owing to the availability of more seeds spatial distribution (isolated vs continuous) of the host 25 26 for oviposition (see Espelta et al. 2009a). (Peterson & Denno 1998; Kubish et al. 2014). Unfortu- 26 27 In comparison to the extreme host–parasite specificity nately, in comparison to the precise information about 27 28 of C. nucum, the other 3 weevils (C. venosus, C. glan- the dispersal ability of C. glandium (approximately 10 28 29 dium and C. elephas) predated indistinctively upon the km in Pélisson et al. 2012), we lack detailed knowledge 29 30 2 oaks with no evidence of a strategy in the partitioning about the dispersal range of C. nucum, except some ev- 30 31 of this resource according to the identity of the host spe- idence of weevils moving away from local sites to feed 31 32 cies or to seed size. The avoidance of competitive exclu- during adulthood and prior to mating (Bel-Venner et 32 33 sion among these species could be explained by several al. 2009). Yet the fact that C. nucum and C. glandium 33 34 mechanisms contributing to stabilize their coexistence; are sister species (Hughes & Voegler 2004a) and they 34 35 that is, time partitioning (Pélisson et al. 2012; see also share many similar life-history traits, such as ecomor- 35 36 Espelta et al. 2009a) and/or diversification of disper- phological adaptations and body size, adult emergence 36 37 sal-dormancy strategies (Pélisson et al. 2012). On the in spring, synovigenic females and a fixed diapause of 37 38 one hand, time partitioning in breeding activity can exist 2 years (see Hughes & Vogler 2004a; Bel-Venner et al. 38 39 when 2 competing species differ in the speed of energy 2009; Pélisson et al. 2012a,b), make us consider that 39 40 acquisition to be allocated to reproduction by females they may have a similar dispersal ability. Therefore, the 40 41 and the duration of their lifespan; that is, one species ac- differences we observed in their genetic characteristics 41 42 quires resources faster and it is able to oviposit earlier would be probably due to their different diet breadth and 42 43 on seeds, but it is exposed to a higher risk of seed abor- the more patchy and discontinuous distribution of hazels 43 44 tion, while the other oviposits later but has a longer life in comparison to the more abundant and constant pres- 44 45 span allowing it to lay eggs during a larger time frame ence of oaks (Gracia et al. 2004; see also Fig. 1), along 45 46 (see Pelisson et al. 2012 for C. pellitus and C. glandi- the geographical gradient sampled (approximately 225 46 47 um). On the other hand, stabilization can be reached by km from Ripoll to Prades). 47 48 48 means of different dispersal versus dormancy strategies Connectivity may be critical for population survival 49 49 with some species relying on a high dispersal ability (Fahrig & Merrian 1985; Fahrig & Paloheimo 1988) and 50 50 and others depending on dormancy strategies (e.g. vari- metapopulation dynamics (Levins 1970), especially in 51 51

H. Arias-LeClaire et al.

1 front of a disturbance: for example, the negative impact patchily distributed hosts (Co. avellana). To what extent 1 2 of severe drought episodes for the emergence of adult this restriction in gene flow (dispersal) may have nega- 2 3 weevils (Bonal et al. 2015; Espelta et al. 2017). In that tive consequences for the persistence of the populations 3 4 sense, our molecular data show that such disturbances of these highly specialized seed-feeding pests in a time 4 5 may have occurred and left their signature in C. nucum of increasing disturbances (e.g. drought events in Med- 5 6 population genetics. The significant negative values re- iterranean areas) is a fascinating question that deserves 6 7 trieved in the neutrality tests (Tajima’s D and Fu’s F) in- further research. 7 8 dicate that most of the C. nucum populations sampled 8 9 went through population expansion after recent bottle- ACKNOWLEDGMENTS 9 10 necks. Almost half of the individuals had the same hap- 10 11 lotype and there was an excess of rare haplotypes that This research was supported by the projects FOR- 11 12 differed little from the most common one. The lower ASSEMBLY (CGL2015-70558-P) and PLAGANADO 12 13 gene flow between populations (marginally significant (AGL2014-54739-R) of the Spanish Ministry of Econ- 13 14 genetic structure among populations) in C. nucum would omy, and the projects BEEMED (SGR913) (Generalitat 14 15 favor such bottlenecks as the patchy distribution of ha- de Catalunya) and PII1C09-0256-9052 (Regional Gov- 15 16 zel trees would complicate the arrival of immigrants. ernment of Castilla-La Mancha and the European So- 16 17 None of this happened in the case of C. glandium feed- cial Fund). R.B. was funded by a contract of the Pro- 17 18 ing on the widespread oak trees. Yet the interpretation of gram Atracción de Talento Investigador (Gobierno de 18 19 the results for C. nucum must be cautious as the shallow Extremadura). 19 20 genetic structure observed suggests that a fair amount of 20 21 gene flow still occurs, enough to overcome drift. More- REFERENCES 21 22 over, in addition to the current distribution of hazelnuts, 22 23 other abiotic environmental conditions could also be in- AliNiazee MT (1998). Ecology and management of ha- 23 24 volved in the genetic structure observed in C. nucum zelnuts pests. Annual Review of Entomology 43, 395– 24 25 (e.g. geological barriers or altitude for Trichobaris soror 41. 25 26 in De la Mora et al. 2015) Bale JS, Masters GJ, Hodkinson ID, Awmack et al. 26 27 (2002). Herbivory in global climate change research: 27 28 CONCLUSION Direct effects of rising temperature on insect herbi- 28 29 vores. Global Change Biology 8, 1–16. 29 30 The use of molecular analyses allowed us to precise- Barat M, Tarayre M, Atlan A (2007). Plant phenology 30 31 ly identify the weevil species depredating upon the vari- and seed predation: Interactions between gorses and 31 32 ous potential hosts in these mixed deciduous forests and weevils in Brittany (France). Entomologia Experi- 32 33 to unequivocally confirm the high specificity of the -ha mentalis et Applicata 124, 167–76. 33 34 zelnut C. nucum and the more flexible and wider trophic Bel-Venner MC, Mondy N, Arthaud F et al. (2009). 34 35 breadth of the rest of the weevils (C. venosus, C. glan- Ecophysiological attributes of adult overwintering in 35 36 dium and C. elephas) depredating upon acorns. In this insects: Insights from a field study of the nut weevil, 36 37 multi-host and multi-parasite system, co-existence of the Curculio nucum. Physiological Entomology 34, 61– 37 38 various weevil species seems to be mediated by a com- 70. 38 39 bination of extreme resource partitioning (i.e. among 39 Bernays EA, Graham M (1988). On the evolution of 40 C. nucum and the rest of species) and a combination of 40 host specificity in phytophagous arthropods. Ecology 41 time partitioning and differences in dispersal-dormancy 41 69, 886–92. 42 strategies among the 3 species depredating upon oaks. 42 43 Interestingly, although sometimes suggested, differences Bonal R, Muñoz A (2008). Seed growth suppression 43 44 in seed size did not have any effect in driving host spec- constrains the growth of seed parasites: Premature 44 45 ificity or the trophic breadth of the weevil species pres- acorn abscission reduces Curculio elephas larval size. 45 46 ent. Moreover, our results highlight that differences in Ecological Entomology 33, 31–6. 46 47 specificity in trophic breadth and in the spatial distribu- Bonal R, Muñoz A (2009). Seed weevils living on the 47 48 tion of hosts at a large geographical scale may result in edge: Pressures and conflicts over body size in the 48 49 the presence of genetic structure among the populations endoparasitic Curculio larvae. Ecological Entomolo- 49 50 of highly specific parasites C.( nucum) depredating upon gy 34, 304–9. 50 51 51

Causes of host specificity in seed weevils

1 Bonal R, Muñoz A, Díaz M (2007). Satiation of predis- genetics data analysis. Evolutionary Bioinformatics 1 2 persal seed predators: The importance of considering Online 1, 47–50. 2 3 both plant and seed levels. Evolutionary Ecology 21, Fahrig L, Merriam G (1985). Habitat patch connectivity 3 4 367–80. and population survival. Ecology 66, 1762–8. 4 5 5 Bonal R, Muñoz A, Espelta JM (2010). Mismatch be- Fahrig L, Paloheimo J (1988). Effect of spatial arrange- 6 6 tween the timing of oviposition and the seasonal opment of habitat patches on local population size. 7 7 timum. The stochastic phenology of Mediterranean Ecology 69, 468–75. 8 acorn weevils. Ecological Entomology 35, 270–78. 8 9 Forister ML, Novotny V, Panorska AK et al. (2015). 9 Bonal R, Espelta JM, Vogler AP (2011). Complex selec- 10 The global distribution of diet breadth in insect herbi- 10 tion on life-history traits and the maintenance of vari- 11 vores. PNAS 112, 442–7. 11 ation in exaggerated rostrum length in acorn weevils. 12 Fu YX (1997). Statistical tests of neutrality of mutations 12 Oecologia 167, 1053–61. 13 against population growth, hitchhiking and back- 13 14 Bonal R, Hernández M, Ortego J, Muñoz A, Espelta, JM ground selection. Genetics 147, 915–25. 14 (2012). Positive cascade effects of forest fragmenta- 15 Gaete-Eastman C, Figueroa CC, Olivares-Donoso R, 15 tion on acorn weevils mediated by seed size enlarge- 16 Niemeyer HM, Ramírez CC (2004). Diet breadth and 16 ment. Insect Conservation and Diversity 5, 381–8. 17 its relationship with genetic diversity and differentia- 17 Bonal R, Hernández M, Espelta JM, Muñoz A, Apari- 18 tion: The case of southern beech aphids (Hemiptera: 18 cio JM (2015). Unexpected consequences of a drier 19 Aphididae). Bulletin of Entomological Research 94, 19 world: Evidence that delay in late summer rains bias- 20 219–27. 20 21 es the population sex ratio of an insect. Royal Society 21 Govindan BN, Kéry M, Swihart RK (2012). Host selec- 22 Open Science 2, 150198. 22 tion and responses to forest fragmentation in acorn 23 Coyle DR, Mattson WJ Jr, Jordan MS, Raffa KF (2012). 23 weevils: inferences from dynamic occupancy models. 24 Variable host phenology does not pose a barrier to in- 24 Oikos 121, 623–33. 25 vasive weevils in a northern hardwood forest. Agri- 25 26 cultural and Forest Entomology 14, 276–85. Gracia C, Ibàñez JJ, Burriel JA, Mata T, Vayreda J 26 (2004). Inventari Ecològic i Forestal de Catalunya. 27 Dupanloup S, Schneider I, Excoffier L (2002). A simu- 27 CREAF, Bellaterra. 28 lated annealing approach to define the genetic struc- 28 29 ture of populations. Molecular Ecology 11, 2571–81. Guidone L, Valentini N, Rolle L, Me G, Tavella L (2007). 29 30 Espelta JM, Cortés P, Molowny-Horas R, Sánchez-Hu- Early nut development as a resistance factor to the at- 30 31 manes B, Retana J (2008). Masting mediated by sum- tacks of Curculio nucum (Coleoptera: Curculioni- 31 32 mer drought reduces acorn predation in Mediterra- dae). Annals of Applied Biology 150, 323–9. 32 33 nean oak forests. Ecology 89, 805–17. Jervis MA, Ellers J, Harvey JA (2008). Resource acqui- 33 34 34 Espelta JM, Bonal R, Sánchez-Humanes B (2009a). sition, allocation, and utilization in parasitoid repro- 35 35 Pre-dispersal acorn predation in mixed oak forests: ductive strategies. Annual Review of Entomology 53, 36 36 interspecific differences are driven by the interplay 361–85. 37 37 among seed phenology, seed size and predator size. Heiniger C, Barot S, Ponge JF et al. (2014). Effect of 38 38 Journal of Ecology 97, 1416–23. habitat spatiotemporal structure on collembolan di- 39 39 versity. Pedobiologia 57, 103–17. 40 Espelta JM, Cortés P, Molowny-Horas R, Retana J 40 41 (2009b). Acorn crop size and pre-dispersal predation Hood GR, Ott JR (2010). Developmental plasticity and 41 42 determine inter-specific differences in the recruitment reduced susceptibility to natural enemies follow- 42 43 of co-occurring oaks. Oecologia 161, 559–68. ing host plant defoliation in a specialized herbivore. 43 44 Espelta JM, Arias-LeClaire H, Fernández-Martínez M, Oecologia 162, 673–83. 44 45 Doblas-Miranda E, Muñoz A, Bonal R (2017). Be- Hughes J, Vogler AP (2004a). Ecomorphological adap- 45 46 yond predator satiation: Masting but also the effects tation of acorn weevils to their oviposition site. Evo- 46 47 of rainfall stochasticity on weevils drive acorn preda- lution 58, 1971–83. 47 48 tion. Ecosphere 8, e01836. Hughes J, Vogler AP (2004b). The phylogeny of acorn 48 49 Excoffier L, Laval G, Schneider S (2005). Arlequin ver. weevils (genus Curculio) from mitochondrial and nu- 49 50 3.0: An integrated software package for population clear DNA sequences: The problem of incomplete 50 51 51

H. Arias-LeClaire et al.

1 data. Molecular Phylogenetics and Evolution 32, Peterson MA, Denno RF (1998). The influence of dis- 1 2 601–15. persal and diet breadth on patterns of genetic isola- 2 3 Kergoat GJ, Delobel A, Fédière G, Le Rü B, Silvain JF tion by distance in phytophagous insects. The Ameri- 3 4 (2005). Both host–plant phylogeny and chemistry can Naturalist 152, 428–46. 4 5 5 have shaped the African seed-beetle radiation. Molec- Ratnasingham S, Hebert PDN (2007). Bold: The Bar- 6 6 ular Phylogenetics and Evolution 35, 602–11. code of Life Data System (www.barcodinglife.org). 7 Molecular Ecology Notes 7, 355–64. 7 8 Kruess A (2003). Effects of landscape structure and hab- 8 itat type on a plant–herbivore–parasitoid community. Richard R, Casas J (2009). Stochasticity and controlla- 9 bility of nutrient sources in foraging: Host–feeding 9 10 Ecography 26, 283–90. 10 and egg resorption in parasitoids. Ecological Mono- 11 11 Levins R (1970). Extinctions. Some mathematical ques- graphs 79, 465–83. 12 tions in biology. In: Gerstenhaber R, ed. Lectures on 12 Ruiz–Carbayo H, Bonal R, Espelta JM, Hernández M, 13 Mathematics in Life Sciences 2. The American Math- 13 Pino J (2017). Community assembly in time and 14 ematical Society, Providence, RI, pp. 77–107. 14 15 space: The case of Lepidoptera in a Quercus ilex L. 15 Kubisch A, Holt RD, Poethke HJ, Fronhofer EA (2014). 16 savannah–like landscape. Insect Conservation and 16 Where am I and why? Synthesizing range biology 17 Diversity 10, 21–31. 17 and the eco–evolutionary dynamics of dispersal. Oi- 18 Solar A, Stampar F (2011). Characterisation of select- 18 kos 123, 5–22. 19 ed hazelnut cultivars: Phenology, growing and yield- 19 20 Merville A, Venner S, Henri H et al. (2013). Endosym- ing capacity, market quality and nutraceutical value. 20 21 biont diversity among sibling weevil species compet- Journal of the Science of Food and Agriculture 91, 21 22 ing for the same resource. BMC Evolutionary Biolo- 1205–12. 22 23 gy, 13, 28. Son Y, Lewis EE (2005). Effects of temperature on the 23 24 Muñoz A, Bonal R, Espelta JM (2014). Acorn–weevil reproductive life history of the black vine weevil, 24 25 interactions in a mixed-oak forest: Outcomes for lar- Otiorhynchus sulcatus. Entomologia experimentalis 25 26 val growth and plant recruitment. Forest Ecology and et applicate 114, 15–24. 26 27 Management 322, 98–105. Tajima F (1989). Statistical method for testing the neu- 27 28 28 Newman D, Pilson D (1997). Increased probability of tral mutation hypothesis by DNA polymorphism. Ge- 29 netics 123, 585–95. 29 extinction due to decreased genetic effective popu- 30 Thomas CD, Bodsworth EJ, Wilson RJ et al. (2001). 30 31 lation size: experimental populations of Clarkia pul- 31 chella. Evolution 51, 354–62. Ecological and evolutionary processes at expanding 32 range margins. Nature 411, 577–81. 32 33 Peguero G, Bonal R, Sol D, Muñoz A, Sork VL, Espe- 33 Toju H, Fukatsu T (2011). Diversity and infection prev- 34 lta JM (2017). Tropical insect diversity: Evidence of 34 alence of endosymbionts in natural populations of the 35 greater host specialization in seed-feeding weevils. 35 chestnut weevil: Relevance of local climate and host 36 36 Ecology 98, 2180–90. plants. Molecular Ecology 20, 853–68. 37 37 Pélisson PF, Bel–Venner M, Rey B et al. (2012). Con- Trudel R, Lavallee R, Guertin C (2002). The effect of 38 trasted breeding strategies in four sympatric sibling 38 39 cold temperature exposure and long–day photoperi- 39 insect species: When a proovigenic and capital breed- 40 od on the termination of the reproductive diapause of 40 er copes with a stochastic environment. Functional 41 newly emerged female Pissodes strobi (Coleoptera: 41 Ecology 26, 198–206. 42 Curculionidae). Agricultural and Forest Entomology 42 43 Pélisson PF, Bel-Venner MC, Giron D, Menu F, Venner 4, 301–8. 43 44 S (2013a). From income to capital breeding: When Venner S, Pélisson PF, Bel-Venner MC, Débias F, Ra- 44 45 diversified strategies sustain species coexistence. jon E, Menu F (2011). Coexistence of insect species 45 46 PloS One 8, e76086. competing for a pulsed resource: toward a unified 46 47 Pélisson PF, Bernstein C, Francois D, Menu F, Venner theory of biodiversity in fluctuating environments. 47 48 S (2013b). Dispersal and dormancy strategies among PloS One 6, e18039. 48 49 insect species competing for a pulsed resource. Eco- Ylioja T, Roininen H, Ayres MP, Rousi M, Price PW 49 50 logical Entomology 38, 470–77. (1999). Host-driven population dynamics in an her- 50 51 bivorous insect. PNAS, 10735–40. 51

Causes of host specificity in seed weevils

1 SUPPLEMENTARY MATERIALS bearing each haplotype in the 5 study populations 1 2 Additional supporting information may be found in Table S3 Mean ± SE density of host plants and the per- 2 3 the online version of this article. centage of sound and infested seeds per location and 3 4 4 Table S1 Number of Curculio nucum individuals bear- host plant Density of host plants was calculated as 5 5 ing each haplotype in the 5 study populations the mean of the nearest inventoried plots included in 6 the Catalan Forest Inventory (Gracia et al. 2004). 6 7 Table S2 Number of Curculio glandium individuals 7 8 8 9 9 Cite this article as: 10 10 11 Arias-LeClaire H, Bonal R, García-López D, Espelta JM (2018). Role of seed size, phenology, oogenesis and host 11 12 distribution in the specificity and genetic structure of seed weevils Curculio( spp.) in mixed forests. Integrative 12 13 Zoology 13, 267–79. 13 14 14 15 15 16 16 17 17 18 18 19 19 20 20 21 21 22 22 23 23 24 24 25 25 26 26 27 27 28 28 29 29 30 30 31 31 32 32 33 33 34 34 35 35 36 36 37 37 38 38 39 39 40 40 41 41 42 42 43 43 44 44 45 45 46 46 47 47 48 48 49 49 50 50 51 51