Depauperation and divergence of plant‐specialist herbivore assemblages in a fragmented tropical landscape

1. Our understanding of the structure and spatial organisation of biological assemblages in human‐modified tropical landscapes has critical importance to improve conservation actions. Investigations on this topic have focused on local (α) diversity patterns, overlooking the changes in species turnover (β diversity) between sites, and its consequences on total (γ) diversity. 2. This study assessed the differences in α, β and γ diversities of galling insects and their host plants (saplings) in a fragmented Atlantic forest landscape in northeast Brazil. Both assemblages were recorded in 30 plots (total of 0.1 ha for each forest type) located in the interior and on the edges of a large fragment and small forest fragments (10 plots per forest type). 3. α diversity of host plants and galling insect assemblages was significantly higher in interior (reference) plots than in edge and fragment plots. Yet, both assemblages showed higher β diversity in fragment and edge plots than in reference plots – a finding potentially associated with the hyperdynamism of fragmented forests and consistent with the landscape divergence hypothesis. 4. However, biotic differentiation of host plant and galling insects was not great enough to compensate the loss of α diversity, and thus γ diversity, because most host plant and galling insect species in forest fragments were also registered in reference plots. Our findings indicate that, despite each small forest fragment being very dissimilar from each other, they have low importance for the conservation of plant assemblages and their specialized herbivores at landscape scale.     

Galling insects in a fragmented forest: incidence of habitat loss,edge effects and plant availability

Galling insects tend to be highly sensitive to changes in their host plants or their environment. Here we analyze the effects of Chaco Serrano forest fragmentation on gall inducing species associated with four native plants species, simultaneously examining area and edge effects as well as the role of host plant availability on such effects. At edge and interior locations in each of nine forest sites in an area gradient in Central Argentina, we estimated herbivory as (1) the proportion of galled plants and (2) the number of galls per plant. Herbivory variations in relation to forest area and edge/interior locations were analyzed with generalized linear models, whereas the influence of plant availability in mediating area and edge effects was assessed by Structural Equation Models. Different responses to fragmentation were observed, depending on the insect species and also on the chosen herbivory indicator. Significant edge and area effects were detected in four and two out of ten performed models, respectively. When significant, edge effects were mostly positive and consistently direct rather than mediated by plant availability; instead, area effects varied from positive and led by plant abundance, to negative and independent of plant availability. Our study provides new evidence of a tendency for galling insects to benefit from edge conditions, while showing less consistent effects regarding forest size. Our results also suggest a very limited role for plant availability as a mechanism mediating fragmentation effects on herbivory by galling insects.     

Edge effect and species–area relationships in the gall-forming insect fauna of natural forest patches in the Brazilian Pantanal

The aim of this study was to (i) measure differences in species richness between edge habitats versus interior habitats, or more precisely the edge effect, and (ii) test the species–area relationship for gall-forming insects in natural forest patches in a Brazilian floodplain (Pantanal of Mato Grosso do Sul). These patches are regionally known as capões, basically composed of woody vegetation. Twenty-seven patches were surveyed. In each patch two transects were conducted for gall sampling. One transect encircled the patches while the other was conducted in the interior of the patch, totaling 54h of sampling. Host plant and galling insect species composition differed quite characteristically between the edge and the interior of patches, but galling insect richness did not. When insect gall richness was expressed as the ratio between insect gall and host plant richness (gall per plant ratio), a weak species–area relationship was found. Our results suggest that the number of galling insects per individual plant is not affected by the size of the patch. Despite these results, the natural forest patches found in this region seem well suited for long-term studies addressing species–area relationships. With regard to herbivorous insects, these studies should be combined with research on host plant dynamics during flooding and dry seasons.     

Habitat conversion and galling insect richness in tropical rainforests under mining effect

Human-induced habitat change is the main cause of species loss and can have severe effects on plant communities and the associated herbivore fauna. In this study, we investigated the effects of habitat conversion due to mining on communities of galling insects in areas of tropical rainforest in the Brazilian Amazon. We sampled galling insects in the Floresta Nacional de Saracá Taquera, Pará, Brazil, where forest plateaus are used by the Mineração Rio do Norte Group to extract bauxite. Our results show that human-induced habitat change via mining activities increased the local species richness of galling insects. We also found that after impact there was greater species richness of galling insects closer to the forest edge than in the forest interior. Changes in plant physiology and in the diversity of natural enemies in human-modified habitats, along with the endophagous life-form, might account for the high incidence of galling in human-disturbed habitats. This result highlights the importance of understanding how different insect groups respond to human activities, since such idiosyncrasies might have profound effects on the species’ patterns of ecological interactions and in the outcomes of those interactions.     

Edge effect benefits galling insects in the Brazilian Amazon

We investigated the influence of edge effects on communities of galling insects in the Brazilian Amazon. For this, were performed data analyses comparing the galling richness between disturbed and undisturbed areas, and subsequently evaluated the richness of galling at different distances from the forest edge. We registered 194 galling insect morphotypes in 116 host plant species. Galling insect richness was more pronounced in disturbed areas and closer to the forest edge, both to community in general as to specific host plant taxa. These results indicate positive responses of galling insect richness to edge effects in forest environments. Edge effect could influence the galling distribution in modified habitats because increases the stress conditions of the plants and decreases the attack pressure of natural enemies.     

Interactions among three trophic levels and diversity of parasitoids: a case of top-down processes in Mexican tropical dry forest

The objective of this study was to analyze the relationship between plant hosts, galling insects, and their parasitoids in a tropical dry forest at Chamela-Cuixmala Biosphere Reserve in western Mexico. In 120 transects of 30 by 5 m (60 in deciduous forest and 60 in riparian habitats), 29 galling insects species were found and represented in the following order: Diptera (Cecidomyiidae, which induced the greatest abundance of galls with 22 species; 76%), Homoptera (Psylloidea, 6.9%; Psyllidae, 6.9%; Triozidae, 3.4%), Hymenoptera (Tanaostigmatidae, 3.4%; which were rare), and one unidentified morphospecies (3.4%). In all cases, there was a great specificity between galling insect species and their host plant species; one galling insect species was associated with one specific plant species. In contrast, there was no specificity between parasitoid species and their host galling insect species. Only 11 species of parasitoids were associated with 29 galling insect species represented in the following families: Torymidae (18.2%), Eurytomidae (18.2%), Eulophidae (18.2%), Eupelmidae (9.1%), Pteromalidae (9.1%), family Braconidae (9.1%), Platygastridae (9.1%), and one unidentified (9.1%). Most parasitoid species parasitized several gall species (Torymus sp.: 51.1%, Eurytoma sp.: 49.7%, Torymoides sp.: 46.9%). Therefore, the effects of variation in plant defenses do not extend to the third trophic level, because a few species of parasitoids can determine the community structure and composition of galling insect species in tropical plants, and instead, top-down processes seem to be regulating trophic interactions of galling insect species in tropical gall communities.     

Gall‐forming insect species richness along a non‐scleromorphic vegetation rainfall gradient in South Africa: The importance of plant community composition

Abstract Currently there is no single accepted hypothesis to explain gall‐forming insect species richness at a particular locality. Hygrothermal stress, soil nutrient availability, plant species richness, plant structural complexity, plant family or genus size, and host plant geographical range size have all been implicated in the determination of gall‐forming insect species richness. Previous studies of such richness at xeric sites have included predominantly scleromorphic vegetation, usually on nutrient‐poor soils. This study is the first to investigate gall‐forming insect species richness of xeric, non‐scleromorphic vegetation. Two habitat types were sampled at each of five localities across a rainfall gradient in the savanna biome of South Africa. The habitat types differed with respect to plant species composition and topography. Gall‐forming insect species richness did not increase with increasing hygrothermal stress or decreasing soil fertility. Rather, gall‐forming insect species richness was largely dependent on the presence of Terminalia sericea as well as other members of the Combretaceae and Mimosaceae. Plots where all these taxa were present had the highest gall‐forming insect species richness, up to 15 species, whereas plots with none of these taxa had a maximum of four galling‐insect species. Despite herb, shrub and tree strata not differing in gall‐forming insect species richness, insect galls were more common on woody than non‐woody plants. Also, stem galls were more frequent than apical or leaf galls. An alternative hypothesis to explain local gall‐forming insect species richness is suggested: galling insects may preferentially select those plant species with characteristics such as chemical toxicity, mechanical strength, degree of lignification or longevity that can be manipulated to benefit the galler. Thus plant community composition should be considered when attempting to explain gall‐forming insect species richness patterns.     

Direct and indirect interactions involving ants, insect herbivores, parasitoids, and the host plant Baccharis dracunculifolia (Asteraceae)

Abstract.  1. The relative importance of direct and indirect interactions in controlling organism abundance is still an unresolved question. This study investigated the role of the direct and indirect interactions involving ants, aphids, parasitoids, and the host plant Baccharis dracunculifolia (Asteraceae) on a galling herbivore Baccharopelma dracunculifoliae (Homoptera: Psyllidae).
2. The effects of these interactions on the galling herbivore's performance were evaluated by an exclusion experiment during two consecutive generations of the galling insect.
3. Ants had a direct negative effect on the performance of the galling herbivore by reducing the number of nymphs per gall. In contrast, ants had no indirect effects on gall mortality through the associated parasitoids.
4. Aphids negatively affected gall development, suggesting that galls and aphids might be partitioning photoassimilates and nutrients moving throughout host-plant tissues.
5. In addition, galls that developed during the rainy season were heavier, indicating that variation in the host plant, due to weather changes, can affect the development of B. dracunculifoliae galls. However, variation in the development of B. dracunculifoliae galls due to presence of aphids or the weather changes did not affect parasitoid attack.
6. These results suggest that direct interactions between ants and galls influenced galling insect abundance, whereas numerical indirect effects involving galling insects, ants, aphids, and host plants were less conspicuous.     

Size, age and composition: characteristics of plant taxa as diversity predictors of gall-midges (Diptera: Cecidomyiidae)

Many hypotheses have been proposed to explain the diversity of gall-midge insects (Diptera: Cecidomyiidae), some of them taking into account plant diversity. This study aims to test the importance of size, age and composition of host plant taxa in the diversity of Cecidomyiidae. For this we used inventories data on the diversity of galling and host plants in Brazil. We found that Asterales, Myrtales and Malpighiales, were the most important orders, with 34, 33 and 25, gall morphotypes, respectively. The most representative host families were Asteraceae (34 morphotypes), Myrtaceae (23) and Fabaceae (22). In general, the order size and the plant family were good predictors of the galling diversity, but not the taxon age. The most diverse host genera for gall-midges were Mikania, Eugenia and Styrax, with 15, 13 and nine galler species, respectively. The size of plant genera showed no significant relationship with the richness of Cecidomyiidae, contrary to the prediction of the plant taxon size hypothesis. The plant genera with the greatest diversity of galling insects are not necessarily those with the greatest number of species. These results indicate that some plant taxa have a high intrinsic richness of galling insects, suggesting that the plant species composition may be equally or more important for the diversity of gall-midges than the size or age of the host taxon.     

Comparing galling insect richness among Neotropical savannas: effects of plant richness, vegetation structure and super-host presence

Plant species diversity maintains the stability of ecosystems and the diversity of consumer species such as insect herbivores. Considering that gall-inducing insects are highly specialized on their host plants and dependent on the occurrence, abundance and distribution of plants, we evaluated the diversity patterns of gall-inducing insect along Brazilian Neotropical savannas and the potential role of plant species richness, vegetation structure and super-host presence on determining these patterns. We found 1,882 individual plants that belonged to 64 different host plant species grouped in 31 families, associated to 112 galling insect species. The galling richness was positively influenced by plant species richness and the presence of the super-host genus Qualea (Vochysiaceae). Plant species richness explained 48 % of the galling richness and areas with presence of super-hosts had more than twice of galling species than areas where they were absent. On the other hand, we found no evidence that larger plants hosted more species of galling insects. We observed that for the diversity of galling insects in the Brazilian Cerrado, vegetation structure explained almost the same portion as plant richness, because structural variables did not have an effect on residuals of galling richness and plant richness regression. Our findings suggests that plant richness has a more important role on the mitigation of natural enemies and adaptive radiation of galling species, while structural aspects of the vegetation does not seem to have that effect. Furthermore, we show that the super-host taxa provide an increment in local galling richness because they present a great diversity of local number of gall morphospecies (i.e. alpha diversity) and the high turnover of morphospecies among different localities (i.e. beta diversity). Therefore we argue that the quality of resources (richness and super host presence) appears to be a most important factor for the diversity of galling insects in Neotropical systems, than the amount of resources.     

Diversity of gall wasps (Hymenoptera: Cynipidae) associated with oak trees (Fagaceae: <Emphasis Type="Italic">Quercus</Emphasis>) in a fragmented landscape in Mexico

Habitat fragmentation reduces the available habitat area and increases both the distance between fragments and the amount of fragment edges. Therefore, there are more probabilities of plant population size reduction and species extinction. In the same way, biotic and abiotic changes associated with forest fragmentation can dramatically alter plant growth and phenological patterns. We conducted a 3-year study to analyze effects of habitat fragmentation and seasonal variation on host plant quality (quantity of leaves, diameter at breast height, tree height), gall abundance and species richness in a temperate oak forest. Our results show that host plant quality was significantly higher in isolated oaks and small fragments, increasing the abundance and species richness of oak gall wasp species in most fragmented habitats. Oak canopy cover is altered by forest fragmentation, there being higher production of leaves on trees that are more exposed to fragmentation, and can provide important resources for maintaining gall wasp species diversity in a fragmented landscape. We found higher gall wasp richness and abundance in autumn than in the spring, which matches with the higher quantity of leaves in this season.     

Hypersensitivity of Fagus sylvatica L. against leaf galling insects

Hypersensitivity is known as a localized resistance of plants against pathogens. It also can be detected in response to galling insects, i.e., in the area immediately adjacent to the site of oviposition and attempted penetration by the galling larva. This host response includes morphological and histological changes that cause the death of the attacked tissue. It is observed as a rounded dark brown halo around the gall induction site. We provide the first observation on the occurrence and possible relevance of this induced mechanism by which one of the most common tree species in Germany, Fagus sylvatica L., resists attack by two of its most common galling insects, Mikiola fagi and Hartigiola annulipes (Diptera: Cecidomyiidae). Galls induced by these cecidomyiids were extremely common in the studied area in beech forests around Darmstadt, Germany. The availability of resources (leaves on a stem) was a poor predictor of attack by the galling insects as well as for gall abundance (galls successfully formed). Hypersensitive reaction was the most important factor acting against the galling population studied. More than 77% of the attempts of the insects to induce galls on F. sylvatica resulted in failure and consequently the death of the galling larvae. Therefore, few live galls remained to be found and destroyed by natural enemies. This corroborates the view that in galling insect–host plant system interactions plant-driven factors may play a major role in determining herbivore failure and success, and perhaps the resulting community structure.     

Subtropical Interactions: Comparing Galling Insect and Host Plant Diversity in Southern Brazil and Florida

Gall-inducing insects seem to have a diversity pattern distinct from the usual latitudinal decrease in species, with more species occurring in xeric environments instead. Many questions regarding galler diversity over geographical scales remain unanswered: for example, little is known about beta diversity, and the role super host plants play in local/regional richness. Our aim was to compare galling insect and host plant diversity in different biogeographical regions, but under similar environmental conditions. We sampled short stature coastal woodlands on sandy soils of the Atlantic coast in both USA (Florida) and Brazil (Rio Grande do Sul, RS), between 25° and 30° latitude. Little-used 200-m long trails were searched during 90 min for galls; there were four trails in USA and five in Brazil. Gall functional traits (galled plant organ, gall shape and colour) proportions were not different between Florida and RS. Local galling and host plant species richness also did not differ, and neither did regional galling diversity. The beta diversity pattern, however, was distinct: sites in Florida have more similar galling faunas than sites in RS. Common diversity patterns indicate common environmental biotic (plant diversity, vegetation structure) and abiotic (climate, soil) factors might be contributing to these similar responses. As Brazilian sites are in the Atlantic forest hotspot, a high galling insect beta diversity might be caused by a higher heterogeneity at larger scales—sample-based rarefaction curves were ascending for Brazil, but not for USA. Myrtaceans were super hosts in Brazil, but not in Florida, where oaks take up this role.     

Diversity of galling arthropods and host plants in a subtropical forest of Porto Alegre, southern Brazil

Many hypotheses have been proposed to explain diversity patterns of galling insects. However, there are contradictory evidences on the evolutionary and ecological factors responsible for the trends. Furthermore, questions such as arthropod seasonality, sampling sufficiency and sampling team experience have been almost ignored. This study records galling arthropod diversity while paying attention to these questions. Seasonal sampling of galling arthropods and host plants were conducted in a humid subtropical forest of southern Brazil. Four transects were sampled twice per season, with two persons searching the vegetation for galls during 1h30min. After 96h.persons of sampling, 130 gall morphotypes on 84 species of host plants were recorded. An analysis of the numbers of galls and gall morphotypes found per transect along time showed that sampling team experience influences galler richness results and the interpretation of galler seasonality patterns. Different species had distinct seasonal patterns. Galling arthropod richness was bound to plant richness. Our results suggest that sampling team experience is an important factor that must be explicitly considered, as well as seasonality patterns of different galling species, at least for tropical/subtropical areas. Although sampling sufficiency was not reached, fauna heterogeneity at small spatial scales seems substantial: despite the proximity of the sampled transects (500 m), they harboured significantly specific faunas. This work adds to the literature records suggesting that both plant richness and specific composition of the vegetation have a strong influence on galler richness at least for local scales.     

Synchronization of gallers with host plant phenology

In addition to various bottom-up effects, the synchronization of herbivores with their host plant phenology determines quality and quantity of food resources and affects the preference–performance linkage and abundance of herbivores. The synchronization has a more critical meaning for such short-lived galling insects as cecidomyiid adults and young aphid stem mothers than for other insects. This review, first, presents general information about gall midges and gall aphids, together with their life history patterns and some ecological attributes. Second, some important topics of galling insect–host plant relation are briefly reviewed. Then, synchronization patterns between gall midge emergence and host plant phenology are analyzed to discuss the adaptive strategies of gall midges and to show how the amount of available food resources is affected by the time lag in synchronization. The spatial distribution pattern and the preference–performance linkage of aphid stem mothers is also discussed in relation to synchronization. Received: October 2, 1998 / Accepted: July 3, 2000     

The interplay between plant traits and herbivore attack: a study of a stem galling midge in the neotropics

1. The relationship between plant traits and the frequency of attack by a stem galling midge, Neolasioptera sp. (Diptera: Cecidomyiidae), on Eremanthus erythropappus (Asteraceae) was studied. The morphological changes of the host after a galler attack and the potential effects of these changes on attacks by the next generation of gallers were analysed. The study was conducted in the Serra do Japi, São Paulo, south-eastern Brazil. 2. Galled branches were significantly longer, thicker, and had more leaves than ungalled branches. Accordingly, gall establishment was higher in the longer and more foliose branches. Hence, it is suggested that ovipositing females are maximizing their performance by selecting larger branches. 3. Galled branches were larger than ungalled branches of the same age. Two hypotheses, not necessarily exclusive, can explain this pattern: (1) the plant vigour hypothesis that the females are choosing the more vigorous, fast-growing branches, which still remain more vigorous after galling; or (2) the resource regulation hypothesis that galling increases branch growth rates, thus increasing resource quality for forthcoming conspecifics. 4. Co-occurrence frequencies of current and past generation galls showed that the likelihood of a branch being galled increased when it, or the branch from which it stemmed, had been galled before. The data indicated that this preference was conditioned by the number of previous attacks. Heavier attack intensities, such as one gall in the same branch and another in the branch from which it stemmed, decreased the probability of further galling. 5. The suggested links between herbivore attack and plant traits indicate that studies on host selection by phytophagous insects must take into account that herbivory itself may change the plant traits that are postulated to be selected by the insects.     

The applicability of the plant vigor and resource regulation hypotheses in explaining Epiblema gall moth–Parthenium weed interactions

A basic question in insect–plant interactions is whether the insects respond to, or regulate plant traits, or a complex mixture of the two. The relative importance of the directions of the influence in insect–plant interactions has therefore been articulated through both the plant vigor hypothesis (PVH) and the resource regulation hypothesis (RRH). This study tested the applicability of these hypotheses in explaining the interactions between Parthenium hysterophorus L. (Asteraceae) and its stem‐galling moth, Epiblema strenuana Walker (Lepidoptera: Tortricidae). Parthenium plants exposed to galling were sampled at three sites in north Queensland, Australia, over a 2‐year period, and the relationship between gall abundance and plant vigor (plant height, biomass, flowers per plant, and branches per plant) was studied. To test the predictions of PVH and RRH, the vigor of parthenium plants protected from galling using insecticides was compared to galled plants and plants that escaped from galling. The vigor of ungalled plants was less than the vigor of galled plants. The higher plant vigor in galled plants was not due to galling, as was evident from insecticide exclusion trials. The insect seemed to preferentially gall the more vigorous plants. These findings support the predictions of the PVH and are contrary to those of RRH. Since gall abundance is linked to plant vigor, galling may have only a limited impact on the vigor of parthenium. This has implications for weed biological control. If the objective of biological control is to regulate the population of a plant by a galling insect, a preference for more vigorous plants by the insect is likely to limit its ability to regulate plant populations. This may explain the paucity of successes against biocontrol of annual weeds using gall insects.     

Effects of host plant architecture on colonization by galling insects

Abstract: To study the abundance and occurrence of herbivore insects on plants it is important to consider plant characteristics that may control the insects. In this study the following hypotheses were tested: (i) an increase of plant architecture increases species richness and abundance of gall‐inducing insects and (ii) plant architecture increases gall survival and decreases parasitism. Two hundred and forty plants of Baccharis pseudomyriocephala Teodoro (Asteraceae) were sampled, estimating the number of shoots, branches and their biomass. Species richness and abundance of galling insects were estimated per module, and mortality of the galls was assessed. Plant architecture influenced positively species richness, abundance and survival of galls. However, mortality of galling insects by parasitoids was low (13.26%) and was not affected by plant architecture, thus suggesting that other plant characteristics (a bottom‐up pressure) might influence gall‐inducing insect communities more than parasitism (a top‐down pressure). The opposite effect of herbivore insects on plant characteristics must also be considered, and such effects may only be assessed through manipulative experiments.     

Different relationships between galling and non-galling herbivore richness and plant species richness: a meta-analysis

The plant richness hypothesis (PRH) is used to explain herbivorous insect richness based on the number of plant species, predicting a positive relationship. However, the influence of plant richness on insect distribution can become stronger with greater levels of specialization of herbivores. In this meta-analysis, I tested whether there is any difference in the correlation force recorded between studies that investigated endophagous versus exophagous herbivores, and galling versus non-galling guilds, in order to determine whether more specialized groups have a stronger relationship. Furthermore, I calculated whether effect sizes were homogeneous between galling studies carried out at local and regional scales, and between tropical and temperate regions. A total of 52 correlations were analyzed between plant species richness and herbivore species richness, with 18 correlations derived from galling herbivores and 34 from non-galling herbivores. The effect sizes were significant and positive in all studies, being higher for endophages than for exophages, and for galling than for non-galling studies. These results provide evidence that groups of insects with a higher level of host specialization and specificity (e.g., endophagous and galling) exhibit a greater dependence on plant richness. There was no difference in effect sizes for galling studies between the local and regional level or between tropical and temperate groups. Despite the large variability found for galling studies, effect sizes were consistent independently of climatic region and latitudinal variation. These results suggest that the PRH for galling insects can be generalized to most ecosystem and vegetation types.     

Plant architecture and meristem dynamics as the mechanisms determining the diversity of gall-inducing insects

Plant architecture is considered to affect herbivory intensity, but it is one of the least studied factors in plant–insect interactions, especially for gall-inducing insects. This study aimed to investigate the influence of plant architecture on the speciose fauna of gall-inducing insects associated with 17 species of Baccharis. Five architectural variables were evaluated: plant height, number of fourth-level shoots, biomass, average level and number of ramifications. The number of galling species associated with each host plant species was also determined. To test the effects of plant architecture on gall richness at the individual level, we used another data set where the number of fourth-level shoots and gall richness were determined for B. concinna, B. dracunculifolia, and B. ramosissima every 3 weeks during 1 year. The average similarity between host species based on gall fauna was low (9%), but plants with the same architectural pattern tended to support similar gall communities. The most important architectural trait influencing gall richness at the species level was the number of fourth-level shoots, which is indicative of the availability of plant meristems, a fundamental tissue for gall induction and development. This variable also showed a positive correlation with gall richness at the individual level. We propose that variations in gall richness among host species are driven by interspecific differences in plant architecture via availability of young, undifferentiated tissue, which is genetically controlled by the strength of the apical dominance. Plant architecture should have evolutionary consequences for gall communities, promoting insect radiation among architecturally similar plants through host shift and sympatric speciation. We also discuss the role of plant architecture in the global biogeography of gall-inducing insects. Electronic supplementary material The online version of this article (doi:) contains supplementary material, which is available to authorized users.