Gall-inducing arthropods in a Neotropical savanna area in the EPA of Rio Pandeiros (Bonito de Minas, MG, Brazil): effects of plant species richness and super-host abundance

Several plant-related factors can influence the diversity of gall-inducing species communities. In the present study we performed an inventory of gall-inducing arthropods and we tested if the plant species richness and the abundance of superhost plants (Copaifera oblongifolia) influenced positively in the diversity of gall-inducing arthropod species. The study was realized in an area of Neotropical savanna (cerrado sensu stricto) in the Environmental Protection Area (EPA) of Rio Pandeiros, Minas Gerais, Brazil. Host-plant species and gall-inducing arthropods were sampled in 18 10  ×  10  m plots distributed in the vegetation. In total we found 40 arthropod gall morphotypes, distributed on 17 botanical families and 29 plant species. Cecidomyiidae (Diptera) induced the most arthropod galls (85%), and the plant family Fabaceae had the greatest richness of gall morphotypes (16). The plant species Copaifera oblongifolia and Andira humilis (Fabaceae) were the most important host species with five and three morphotypes, respectively. Galling species richness was not affected by none of explanatory variables (plant species richness and abundance of super-host plants). On the other hand, galling species per plant species was negatively affected by plant species richness and positively affected by abundance of super-host plants. This is the first study of arthropod-induced galls conducted in EPA of Rio Pandeiros, Brazil. Our results corroborate previous studies that highlight the importance of super-host plants for galling arthropod diversity on a local scale. Key-Words. Arthropod-plant interactions; Cecidomyiidae; Cerrado; Copaifera; Fabaceae.


INTRODUCTION
Plant-animal interactions between herbivorous arthropods and their host plants encompass the greatest diversity of terrestrial species (Price, 2002). Among herbivorous arthropods, gall-inducing species represent the most specialized guild (Araújo et al., 2019a), because they are the only arthropods capable of manipulating physiological and anatomical plant structures inducing the formation of galls (review in Miller III & Raman, 2019). Arthropod galls are structures formed by hyperplasia and hypertrophy of plant tissues, inside which the inductor development and feed (Fernandes et al., 2014). In addition, galls can provide shelter for galling arthropods against attack by natural enemies and environmental weathering (Miller III & Raman, 2019). Due to the intimate association between gall-inductors and their host plants, galling arthropod communities tend to be strongly influenced by host plant assemblages Altamirano et al., 2016;Costa & Araújo, 2019).
One of the main factors of plant assemblages that can influence galling communities is the richness of plant species (review in Araújo, 2013). Plant species richness represents a greater di-ISSN On-Line: 1807-0205 ISSN Printed: 0031-1049 ISNI: 0000-0004-0384-1825 versity of resources for the use of gall-inducing species (Cuevas-Reyes et al., 2004). In this sense, several evidences point that gall-inducing species richness increases as more potential host plant species are available (e.g., Wright & Samways, 1998;Gonçalves-Alvim & Fernandes, 2001;Cuevas-Reyes et al., 2004). On the other hand, there is also evidence that some galling communities are little influenced by richness of plant communities, because plant species vary greatly in their susceptibility to gall induction (Blanche, 2000;Araújo et al., 2013).
In tropical environments, many plant species have few or no gall-inducing species (Blanche, 2000), but some plant species can host many species of galling arthropods . The existence of super-host plants, i.e., plant species with a high number of gall-inducing species, indicates that the composition of plant assemblages also is important for the galling community . Therefore, the occurrence of super-host plants may increase the diversity of arthropod galls, regardless of the number of plant species (Veldtman & McGeoch, 2003;Araújo et al., 2014a). For Brazilian savannas, some super-host plants are recorded, such as Baccharis dracunculifolia DC. (Fernandes et al., 1996), Copaifera langsdorffii Desf. (Costa et al., 2010;Ribeiro et al., 2019), and Qualea parviflora Mart. , but few studies have investigated the effect of the presence of these taxa on the local diversity of galling species.
In the present study we performed an inventory of gall-inducing arthropods and their host plants in an area of Neotropical savanna (cerrado sensu stricto) located in the Environmental Protection Area (EPA) of Rio Pandeiros (Bonito de Minas, MG, Brazil). Additionally, we tested if the plant species richness and the abundance of super-host plants influenced positively in the diversity of gall-inducing arthropod species.

Study area
The study was performed in an area of Neotropical savanna (cerrado sensu stricto) in the Environmental Protection Area (EPA) of Rio Pandeiros (15°21′37.2″S and 44°54′45.9″W), which englobe the municipalities of Bonito de Minas, Cônego Marinho and Januária in the North of Minas Gerais State, Brazil. The reserve has an area of 431.401 hectares and was created in 1995 with objective to preserve the hydric resources and biodiversity in the valley of Pandeiros river. This area has been classified as of extreme biological importance, priority for scientific research and biodiversity conservation (Drummond et al., 2005). The EPA of Pandeiros river is located in an ecotone zone between the domains of Caatinga and Cerrado and has a remarkable floristic diversity, characteristic of this transitional effect (Bahia et al., 2009;Fagundes et al., 2019a). The cerrado sensu stricto is the predominant physiognomy of the EPA, but seasonally dry tropical forest and palm swamps are very common in the area (Bahia et al., 2009). Soils are classified as Quartzarenic Neosol, with sandy texture, Dystrophic Haplic Cambisol and Latosol with high water storage capacity, acid and low fertility (EMBRAPA, 2013). The climate of the region is tropical dry (Aw in the Köppen system), characterized by well-defined rainy periods, an average temperature of 24.2°C, and an average annual rainfall of 1,000 mm (Alvares et al., 2013).

Arthropod gall survey
Arthropod gall sampling was performed in a rapid ecological study realized in November of 2019. The sampling was realized in 18 plots of 10 × 10 m distributed in the study area . In each plot all woody plants with a circumference ≥ 10 cm at 1.3 m above ground were sampled. The botanical material was herborized and identified according to the botanical literature and/or the recommendation of experts. In the plots arthropod galls were sampled by active searches up to a height of 2.5 m in all plants. All sampled galls were classified into morphotypes using the host plant species and external morphology (organ of occurrence, shape, color, pubescence and size) (Carneiro et al., 2009). To designate gall morphotypes we use the terminology proposed by Isaias et al. (2013). Gall-inducing arthropods were determined from dissection of galls in field or laboratory and also using the arthropod gall literature from Neotropic and Brazil (e.g., Maia & Fernandes, 2004;Carneiro et al., 2009;Gagné, 2014;Araújo et al., 2013;Araújo et al., 2014b;Araújo et al., 2019b). In the area, we recorded the host-plant species Copaifera oblongifolia DC. (Fabaceae), which has been recently listed as a super-host plant harboring 15 different species of gall-inducing insects (Coutinho et al., 2019).

Statistical analyses
For the statistical analyses, galling species richness (total number of arthropod gall morphotypes in each plot) and galling species per plant species (mean number of arthropod gall morphotypes by host plant species in each plot) were used as response variable. We used the plant species richness and abundance of super-host plants as explanatory variables at plot level in generalized linear models (GLM's) built for each response variable. In the present study we used the abundance of Copaifera oblongifolia as a measure of abundance of super-host plants. All models were submitted to a residual analysis to determine the adequacy of error distribution and had a Gaussian error distribution assumed. All statistical analyses were performed in the software R version 3.4.1 (R Development Core Team, 2015).
Galling species richness was not affected by none of explanatory variables (Table 2), despite the tendency of a positive effect of abundance of super-hosts on the gall richness (p = 0.057). Already the galling species per plant species was significantly influenced both by plant species richness (p = 0.011) and abundance of super-host plants (p = 0.020) ( Table 2). We found that galling species per plant species was negatively affected by plant species richness (Fig. 6) and positively affected by abundance of super-host plants (Fig. 7).

DISCUSSION
The number of galling species observed in the area of EPA of Rio Pandeiros (40 morphotypes) is intermediary compared to other studies performed in Neotropical savannas (Table 3). For example, Urso-Guimarães et al. (2003) recorded only 22 gall morphotypes in cerrado fragments, rupestrian field and gallery forest in Delfinópolis, Minas Gerais State. In other study, Maia & Fernandes (2004) recorded 137 morphotypes of insect galls in an area of rupestrian fields and cerrado in the Serra de São José, Minas Gerais. These numbers extremely variable in the diversity of galling species can be explained by several factors, among which are different sampling efforts employed in the studies, as well as variations in the structural characteristics and diversity of the studied vegetation. The stan-   dardized measure of galling species per plant species obtained in the present study (1.37) can also be considered intermediate. In a recent review, Araújo et al. (2019b) founded that the mean number of insect gall morphotypes per plant species for Brazilian inventories was 1.72 (± 0.43), ranging between 1.16 and 3.50.
Gall-inducing arthropod fauna recorded in the EPA of Rio Pandeiros was composed of distinct and important galling groups such as gall-midges (Cecidomyiidae) and eriophyids (Eriophyidae), which induced 85% and 7.5% of the gall morphotypes, respectively. Gall-midges are the largest and most diverse group of galling insects in the world (Gagné, 2014). In gall surveys conducted in Brazil this family has shown to be quite diverse and abundant (review in Araújo et al., 2019b). Eriophyids are galling mites than can induce galls in more than 500 host plant species (Petanovíc & Kielkiewicz, 2010). In Brazil eriophyids are rarely include in gall inventories, but in other parts of globe some studies have highlighted the importance of the group for galling fauna (e.g., Nasareen & Ramani, 2014;Araújo et al., 2019a). According review of Maia (2006) gall-inducing insects of the orders Hemiptera and Lepidoptera are less common in the Neotropical region, which corroborates the results of the present study where these groups represented less of 8% of gall-inducers.
The plant family with higher gall richness was Fabaceae with 16 morphotypes, being this also the family with the largest number of host species (eight in total). The recent review about Brazilian gall inventories points that Fabaceae is the most important host family of Brazil, appearing as a super-host in 68.6% of the studies (Araújo et al., 2019b). The great intrinsic galling insect richness of this plant family may be due to the great diversity of species it presents (Mendonça, 2007). The host-plant species with higher diversity of gall morphotypes also were of family Fabaceae: Copaifera oblongifolia and Andira humilis. Recent studies have pointed Copaifera oblongifolia (Fabaceae) as a super-host plant of galling insects with 15 different gall morphotypes recorded (Coutinho et al., 2019;Fagundes et al., 2019b). Species of the genus Andira have also been listed as important hosts of galling insects in other areas of neotropical savannas, such as Andira paniculata Benth. (Santos et al., 2012).
Contrary to the expected the galling species richness was not affected by plant species richness and abundance of super host plants, despite the tendency of a positive effect of the abundance of Copaifera oblongifolia on the arthropod gall richness. A possible explanation for these results may be environmental factors, such as environmental stress and soil fertility, which were not measured in the present study, but may also influence the distribution of galling species (Gonçalves-Alvim & Fernandes, 2001;Ramos et al., 2019). On the other hand, we found that galling species per plant species was negatively affected by plant species richness and positively affected by abundance of super-host plants, corroborating our expectations. The negative relationship between plant species richness and galling species per plant species likely is a mathematical effect of increment in the number of host plant species (denominator) more accelerated than the number of galling species (numerator). However, the positive effect of the abundance of super-host plants on this measure may be indicative that each individual of Copaifera oblongifolia add new species of galling species to the community. Our results highlight the importance of super-host plants for galling arthropod diversity on a local scale, corroborating previous studies (e.g., Araújo et al., 2014a). This is the first study of arthropod-induced galls conducted in EPA of Rio Pandeiros, Minas Gerais, Brazil. Considering that our sampling was done through a rapid  ecological study, medium-and long-term sampling, considering different seasonal periods, will probably result in the recording of a greater diversity of arthropod galls. Nevertheless, compared to other areas of cerrado sensu stricto in Brazil the diversity recorded in the present study can be considered intermediate. Previous studies have shown that different factors can affect the diversity of gall-inducing arthropods in Neotropical savannas (Gonçalves-Alvim & Fernandes, 2001;Araújo et al., 2013;Araújo et al., 2014a). Contrary to expectations, we did not find effects of the plant species richness and abundance of super-host plants on the galling species richness but found that these explanatory variables affected significantly the number of galling species per plant species. Thus, we believe that future studies can investigate the extent to which the occurrence of other super-host species influences the local diversity of galling arthropods.