Insect responses to host plant provision beyond natural boundaries: latitudinal and altitudinal variation in a Chinese fig wasp community

Many plants are grown outside their natural ranges. Plantings adjacent to native ranges provide an opportunity to monitor community assembly among associated insects and their parasitoids in novel environments, to determine whether gradients in species richness emerge and to examine their consequences for host plant reproductive success. We recorded the fig wasps (Chalcidoidea) associated with a single plant resource (ovules of Ficus microcarpa) along a 1200 km transect in southwest China that extended for 1000 km beyond the tree's natural northern range margin. The fig wasps included the tree's agaonid pollinator and other species that feed on the ovules or are their parasitoids. Phytophagous fig wasps (12 species) were more numerous than parasitoids (nine species). The proportion of figs occupied by fig wasps declined with increasing latitude, as did the proportion of utilized ovules in occupied figs. Species richness, diversity, and abundance of fig wasps also significantly changed along both latitudinal and altitudinal gradients. Parasitoids declined more steeply with latitude than phytophages. Seed production declined beyond the natural northern range margin, and at high elevation, because pollinator fig wasps became rare or absent. This suggests that pollinator climatic tolerances helped limit the tree's natural distribution, although competition with another species may have excluded pollinators at the highest altitude site. Isolation by distance may prevent colonization of northern sites by some fig wasps and act in combination with direct and host-mediated climatic effects to generate gradients in community composition, with parasitoids inherently more sensitive because of declines in the abundance of potential hosts.


Introduction
The spatial distributions of species reflect the net effects of numerous historical, geographic, biotic, and abiotic elements including speciation, migration, species interaction, resource availability, and climatic tolerances (Gaston 2000;He et al. 2005;Chen and He 2009). Phytophagous insects are the most species-rich components of terrestrial communities. The distributions of specialist phytophages are necessarily limited to within the ranges of their host plants, but they often only occupy a proportion of their host's range, suggesting that additional physical and biological variables such as climate and natural enemies also routinely play a part in determining range boundaries (Strong et al. 1984). Temperatures, precipitation, and the extent of seasonal fluctuations in climate all change with latitude and elevation (Hodkinson 2005;Deutsch et al. 2008;Benton 2009;Feeley et al. 2012;Yasuhara et al. 2012), and host plants can often tolerate a wider range of temperatures than their associated insects. The insect community associated with bracken provides an example, with some bracken-feeding species restricted to frost-free areas in Africa (Compton et al. 1989). This results in fewer insect associates at higher latitudes and altitudes and contributes to the all-pervading latitudinal gradient in species diversity (Gaston 2000;Willig et al. 2003;Witman et al. 2004;Buckley et al. 2010). The distributions of insects may nonetheless respond differently and more rapidly to changes in climate than their hosts (Sch€ onrogge et al. 2012;Nooten et al. 2014), resulting in the proportion of a host plant's range that is utilized changing over time (Hodkinson and Bird 1998;Schweiger et al. 2008).
Gradients in altitude display some environmental changes that are similar to latitudinal gradients, and can generate similar selection pressures on phytophagous insects (Pellissier et al. 2012), but there are also major differences that can influence how insects respond (Hodkinson 2005). Altitudinal effects are typically displayed over much shorter distances, making dispersal between sites with contrasting temperature regimes easier than between equivalent climatic changes generated by latitude. Seasonal variation in temperatures and day length are also lower at high elevations at lower latitudes, compared with lower elevation, higher latitude sites. Challenges posed to phytophagous insects and their host plants at higher latitudes are therefore similar, but not the same as those generated at higher altitudes (Rasmann et al. 2014).
Phytophagous insects often support diverse communities of parasitoids, many of which also have restricted host ranges. The distributions of carnivores and others that feed at higher trophic levels often decline rapidly with latitude (Hillebrand 2004;Freestone et al. 2011;Santos and Quicke 2011) and suitable hosts may be less abundant or entirely absent at higher latitudes for parasitoids with highly specific host requirements (Condamine et al. 2012;Cruaud et al. 2012). Similarly, individuals of a phytophagous species that develop at higher altitudes can be subject to attack by fewer species of parasitoids, and suffer lower mortality rates (Randall 1982).
Phylogeographic history and interactions within complex large-scale communities often make the drivers of insect distributions and responses to environmental gradients difficult to distinguish (Buckley et al. 2010;Romdal et al. 2013). Relatively simple spatially defined insect communities associated with a single plant resource provide good systems to try to tease apart drivers of species distribution Bannerman et al. 2012). Large-scale manipulations of these communi-ties result when the distributions of their host plants are extended beyond their natural range margins, especially where the expanded range extends into areas subject to more extreme environmental conditions. Here, we focus on a community of plant-feeding and parasitoid fig wasps (Hymenoptera, Chalcidoidea) associated with a single plant resourcethe figs produced by a species of fig tree that is widely planted within and beyond its natural distribution in China.
The  (Wiebes 1979;Herre et al. 2008;Chen et al. 2012). Related fig trees tend to be pollinated by related fig wasps, suggesting a long history of co-evolution, although host switching between lineages has also taken place (Cook and Segar 2010;Cruaud et al. 2012). The mutualism is also of broader ecological significance, because so many vertebrates eat ripe figs, resulting in fig trees and fig wasps often being keystone species (Shanahan et al. 2001;Herre et al. 2008). Most fig trees have tropical or subtropical distributions, and few species are exclusively temperate. Factors influencing their range margins may be linked to the trees themselves or reflect limitations imposed by the environmental tolerances of their pollinators (Liu et al. 2014;Zhang et al. 2014).
Figs are also utilized by other groups of fig wasps belonging to families of Chalcidoidea other than Agaonidae. More than 30 nonpollinating fig wasps (NPFW) species have been recorded from a single Ficus species, although most support less than half this number Cook and Rasplus 2003;Wang et al. in press). Like the pollinators, NPFW generally develop in galled ovules. Figs lacking pollinators usually abort, but some NPFW are capable of developing in unpollinated figs, which allows them to be independent of the pollinators. The host ranges of most NPFW are poorly known. Some lineages appear highly host plant specific, but others contain species that utilize more than one host plant (Cook and Segar 2010). Detailed knowledge of the larval feeding behavior of NPFW is only available for a small number of species, but it is becoming increasingly apparent that NPFW display a diverse range of feeding behaviors which includes seed predators, ovule and fig wall primary gallers, secondary gallers that enlarge the galls of primary gallers, primary parasitoids (most of which also feed on some plant tissue), and specialist hyperparasitoids (Pereira et al. 2007;Compton et al. 2009;Segar and Cook 2012;Chen et al. 2013). The specific insect hosts attacked within figs by parasitoid NPFW are rarely documented (Cook and Segar 2010;Segar et al. 2013), but niche conservatism induced by morphological characters such as ovipositor length and body size generates some specific matching of parasitoids and gallers, and there is also evidence for strict sense co-evolution between some gall-formers and their specific parasitoids (Compton 1993;West et al. 1996;Dunn et al. 2008). Because of this limited knowledge, it is usually only possible to characterize the species within a particular fig wasp community as being either exclusively phytophages (most or all of which are ovule gall-formers) and parasitoids in a broad sense, which kill larvae of other fig wasps and develop in galls that other species had initiated.
Fig wasp communities display convergence and relatively homogeneous structures across continents (Segar et al. 2013), but also display between-site variation in species richness. Latitudinal gradients in the species richness and composition of fig wasp faunas in southern Africa have been investigated along a gradient extending from six degrees north to 34 degrees south Hawkins and Compton 1992). These studies failed to detect significant latitudinal trends in the species richness of galler NPFW, whereas species richness among parasitoid fig wasps was generally slightly lower at lower latitudes, but for most species, only a small number of crops were available for analysis. Here we focus on   Fig. 1). As a result of its popularity as a street and ornamental tree, and the widespread introduction of its pollinators, F. microcarpa populations have become established in many tropical and subtropical areas, including the Mediter-ranean and Caribbean, mainland USA and Hawaii and Brazil (Nadel et al. 1992;de Figueiredo et al. 1995;Kobbi et al. 1996;Beardsley 1998;Burrows and Burrows 2003;Starr et al. 2003;van Noort et al. 2013;Wang et al. 2015). In urban environments, small plants can cause damage to buildings, but the plant can also become a serious invader of natural habitats (Mckey 1989;Beardsley 1998;Starr et al. 2003;Corlett 2006;Caughlin et al. 2012). The plant's success in seasonal climates may be related to the ability of its pollinators' populations to rapidly recover from winter shortages of figs . F. microcarpa has small seeds that are mainly dispersed by frugivorous birds, and ants also serve as secondary dispersal agents (Kaufmann et al. 1991;Shanahan et al. 2001;Caughlin et al. 2012).

Materials and Methods
In China, F. microcarpa is indigenous to south Fujian, Guangdong, Guangxi, Hainan, the south of Yunnan Province and Taiwan. It is also one of the most widely planted ornamental and street trees in southern China, both within its natural range and extending to around 1000 km further north. Within its natural range, F. microcarpa is generally an uncommon component in natural forests. It is present at much higher densities in urban or suburban areas and we mainly focused on the fig wasp communities found on planted trees. F. microcarpa produces crops of up to several thousand figs from among the axils of leaves. Fruiting is asynchronous at the population level, with figs present on the trees throughout the year. On individual trees, the crops may develop synchronously, but when few fig wasps are present, the synchrony tends to break down. Low winter temperatures in the introduced range of the plant restricts the emergence of adult fig wasps from their natal figs to the warmer months of the year, whereas emergence is all year at our southern-most sites, with a peak during the cool dry    Peng, pers. comm.). Numerous varieties and forms of F. microcarpa have been described within its extensive range (Berg and Corner 2005). The trees in southwest China are uniform in appearance, but could not be assigned to a particular variety.
At  (Galil and Copland 1981). As with the agaonids, a single larva develops inside each ovule. Parasitoid NPFW may or may not consume some plant tissue, but they always destroy the gall causers. Some gallers of F. microcarpa ovules can develop in figs that are not entered by pollinators. They have associated parasitoids that do not attack pollinator larvae (S.G. Compton & R. Wang, unpubl.). Generally, species from subfamily Epichrysomallinae (family Pteromalidae) are the hosts of species from family Eurytomidae, and species from subfamily Sycoryctinae (family Pteromalidae) are parasitoids of both agaonids and species from subfamily Otitesellinae (family Pteromalidae). Philotrypesis taiwanensis Chen (Sycoryctinae) is the only obligate seed predator (Wang et al. 2014). In our analyses, we grouped the species associated with F. microcarpa into two trophic levels based on their feeding behavior: "phytophages" with larvae that feed exclusively on plant ovules or seeds and "parasitoids" with larvae that kill other fig wasp species.

Study sites
Ficus microcarpa fig crops were sampled in Mianyang, Chengdu, Xichang, and Panzhihua (Sichuan Province), and Kunming and Xishuangbanna (Yunnan Province) ( Fig. 1). They formed a north-south transect across southwestern China, covering about 1200 km and 9.5 degrees of latitude. Xishuangbanna is located on the border between subtropical and tropical China, with hot and humid summers and mild, dry winters. It is the only study site believed to be within the native range of F. microcarpa (Table 1). At the other sites, F. microcarpa is not present in local natural forests, but has been widely planted in urban areas. Winter and summer temperatures at the sites generally decline with increasing latitude, but Kunming has a cooler climate than the other sites, because of its higher elevation (Table 1). Variation in annual precipitation among the study sites is slight, ranging from 850 to 1100 mm. The trees in Xishuangbanna were growing in a botanic garden. Elsewhere, they were planted along roadsides and in public amenity areas.

Collecting methods
We haphazardly collected mature figs without fig wasp exit holes (late C/early D) phase sensu Galil and Eisikowitch (1968) from at least six F. microcarpa trees at each study site (Table 1) and stored them in 70% ethanol. Sampling was concentrated in the periods when most trees locally had mature figs. Figs that are not colonized by fig wasps are retained on the trees for long periods before they abort. They continue to grow and could only be reliably distinguished from figs entered by fig wasps after dissection. To record the contents of the figs, they were cut into quarters and soaked in water for at least 10 min to soften the galls before the figs were examined under a binocular dissecting microscope. Each flower was checked and recorded in one of the following categories: male flowers, seeds, unfertilized and ungalled female flowers, galls containing wasps, and failed galls.

Statistical analyses
All statistical analyses except species accumulation and estimated species richness curves were carried out using R version 2.14.2 (R Development Core Team 2012). Likelihood ratio tests were carried out to assess the significance of fixed effects.  We tested whether we had detected most or all of fig wasp species in their regional species pools by delineating curves of accumulated species richness with increasing sample size using a first-order jackknife algorithm (Burnham and Overton 1978;Heltshe and Forrester 1983), in SDR version 4.1.2 (Seaby and Henderson 2006).
The Meteorological data for the period 2004-2013 were obtained from the website of Weather Underground (http://www.wunderground.com). For each study site, except Panzhihua, we collated eight climate factors: annual average temperature and precipitation, summer (3 months from June to August) average and average high temperatures, summer average rainfall, winter (3 months from December to February) average and average low temperatures, and winter average rainfall. Panzhihua was excluded from the climate analyses due to a lack of local meteorological data. We then ran a principal component analysis (PCA) and selected the first principal component (PC1, which explained 61.1% of the total variance) as the factor representing "local climate" (Table S1 in supplementary materials).

Results
The cheater Eupristina species was found only in the figs from Kunming and Xishuangbanna, where it occupied 12.1% and 82.6% of the figs, respectively (Table 4). In a very small number of figs, it was recorded as an "accidental" pollinator, with 50.0% (2 of 4 figs in Kunming) and 19.6% (9 of 46 figs in Xishuangbanna) of the figs that contained offspring of this species (but not E. verticillata) also containing small numbers of seeds.

Latitudinal and altitudinal effects on fig wasp community
We recorded 21 fig wasp morphospecies from the figs of F. microcarpa in southwest China (Table 5), but no more than 13 species were recorded from any individual site (Tables 2 and 5). Xishuangbanna had several species that were not recorded elsewhere, but there were also other species that were only recorded at other sites. Although several species of phytophages and their associated parasitoids were only recorded at intermediate latitudes, no clearly northern species were present (Table 5). Both direct accumulation and first-order jackknife methods suggest that regional species richness almost reached asymptotes within our range of sample sizes at every site (Fig. 4) and estimates of the size of the regional pools from which our samples were drawn suggest that we had recorded most but not all of the species predicted to be present at each site (Table 2).  wasps are included). Note that we did not find any significant effect of the interaction between latitude and altitude (Table S3), and therefore, it was deleted from all statistical models and the two factors were analyzed separately. LR = likelihood ratio. Response variables are as follows: (1)   NS, not significant, *P < 0.05; ***P < 0.001.    (Table 2). There was a latitudinal shift in the character of the fig wasp communities, with a northwards decline in the abundance of the agaonids and their associated parasitoids and an increasing preponderance of fig wasps that make larger galls such as Meselatus, Odontofroggatia, and Walkerella species, together with their associated parasitoids (mainly Sycophila maculafacies and Philotrypesis okinavensis sensu Chen et al. (1999); Table S2). Mean species richness per fig also declined significantly with latitude (Tables 2 and 3; Fig. S1), but did not exceed three species per fig, even at the southerly sites, despite many more species in total being recorded there. Diversity, as measured by the Shannon-Wiener index, was highest at intermediate latitudes. In the two most northerly sites, Mianyang and Chengdu, this reflected the low species richness, while in the south, at Xishuangbanna, species richness was high, but many species were rare and offspring of the two Eupristina agaonids predominated, occupying over 97% of the figs and comprising over 91% of all fig wasps (Tables 4 and S2). Despite this pattern, there was a significant decline in the Shannon-Wiener index with increasing latitude (Table 3; Fig. 3C).
Altitude was negatively correlated with ovule occupancy rate, fig wasp abundance, and species richness, and was positively related to Shannon-Wiener index values, suggesting altitude is also playing an important role in shaping the fig wasp fauna (Table 3; Fig. 3D-F). We failed to detect any significant effects of the interaction between latitude and altitude (Table S3).

Comparisons between trophic levels
The overall fig wasp community from the six sites combined included 12 phytophages and nine parasitoids ( Table 5). Generally, offspring of phytophagous species were far more abundant than those of parasitoid species, comprising 90.5% of the total number of fig wasp individuals in the figs (Table S4). There were no parasitoids present in the figs from the two most northerly sites. Figs from higher latitude sites contained significantly fewer species at both trophic levels, but parasitoid species declined significantly more rapidly with latitude than phytophagous species (Tables 3 and S4; Fig. 5). In contrast, altitude only had a significant influence on species richness of phytophagous species (Table 3).

Discussion
Apparent exceptions to global-scale declines in parasitoid species richness at higher latitudes may often be a result of sampling bias Sime and Brower 1998;Quicke 2012). Within more local communities, contrary species richness patterns have nonetheless been reported among gall wasp parasitoid assemblages in Canada (Bannerman et al. 2012) (Warren et al. 2010). Which stages of the insects' life cycles are particularly climate sensitive are unclear, but low temperatures will influence larval development times, the ability of the adult offspring wasps to emerge from the figs, and their ability to migrate between trees to look for oviposition sites ). Between-species variation in flexibility of development rates may be critical, because low winter temperatures at the more northern ( (Peng et al. 2010;Yang et al. 2013;Liu et al. 2014;Zhang et al. 2014).
Local climatic variables within the geographic range encompassed by our study sites were correlated more strongly with altitude than latitude, due to the large range in elevations. At Kunming, the higher altitude site located toward the southern edge of our study area, this is reflected in lower average minimum temperatures than the other sites, but also with relatively low mean maxima and less seasonal variation temperatures than at sites further north.
The dispersal abilities of different groups of fig wasps are unknown, but long-distance dispersal events (extending to 100 km or more) have been recorded in some species (Ahmed et al. 2009;Wang et al. 2009;Chen et al. 2012). Nevertheless, given that F. microcarpa is an introduced species at most of our study sites, and we know of no examples of the plant establishing itself outside of urban areas, there must be large gaps between the plant's introduced populations, some of which may be beyond the dispersal limits of most fig wasp species and result in isolation by distance effects contributing to the northerly decline in community complexity.
Some postulated reasons for global declines in parasitoid species richness with latitude, such as a lack of alternative hosts at more northerly sites, can be rejected because most fig wasps associated with F. microcarpa appear to be host plant specific (R. Wang & S.G. Compton, unpubl.). The more pronounced northerly decline in species richness and abundance among parasitoid NPFW appears to be linked to a shortage of hosts, rather than their complete absence. At higher latitudes, potential hosts were present at lower densities both within individual figs and in terms of the proportion of the figs that contained any fig wasps. Differences in the dispersal ability of phytophagous and parasitoid NPFW could also contribute if some species go extinct locally each winter at higher latitude sites and there are annual rescue effects from populations further south (Bannerman et al. 2012). The relative dispersal ability of different groups of fig wasps is unknown, but it is possible that dispersal among parasitoid species is hampered by the long ovipositors possessed by many species. These allow the parasitoids to lay their eggs through the walls of figs, but seem likely to reduce flight efficiency. Some parasitoid fig wasps with long ovipositors are nonetheless able to colonize even very isolated desert fig trees in Africa (Ahmed et al. 2009).
Planted F. microcarpa trees are capable of surviving beyond the natural range limit of the species, suggesting that there are germination and establishment issues that limited the tree's distribution in the past. From the tree's perspective, individuals planted further north also increasingly produced figs that were of no reproductive value, because they were seldom or never colonized by pollinators and were only colonized by gall-forming NPFW. Seed production was therefore limited or absent. The monoecious fig tree with a natural distribution that extends furthest north in China, F. virens, also struggles to support populations of its pollinator fig wasp through the winter at its northern range limit, but seed production is supported by the seasonal migration of pollinators from further south . F. virens is pollinated by a species of Platyscapa, a genus where long-distance pollinator dispersal may be the norm (Burnham and Overton 1978). The E. verticillata pollinator of F. microcarpa may be less mobile and also more sensitive to temperature effects than Platyscapa sp.
Eupristina verticillata was absent from the southerly, but high altitude, Kunming study site, which experiences low mean monthly temperatures. It was replaced there by an abundant second ("cheater") Eupristina species, which had a distribution limited to our two most southern study sites. Why this species should be restricted to the south, but apparently thrive at high altitude, is unclear. The two Eupristina species coexist at the lowland Xishuangbanna study site, but at Kunming Eupristina sp.

Supporting Information
Additional Supporting Information may be found in the online version of this article: