Is the social wasp fauna in the tree canopy different from the understory ? Study of a particular area in the Brazilian Amazon Rainforest

The Polistinae social wasps are comprised of 26 genera and 958 species widely distributed in the Neotropical region (Pickett & Carpenter, 2010). These wasps are important components of Neotropical ecosystems due to their ubiquity and diversity, as well as their complex interactions with other organisms (Silveira et al., 2012). The Amazon Rainforest is the largest biome in Brazil, covering an area of 4,196,943 Km2 within the Brazilian territory, and is one of the most biodiverse places in the world, presenting the highest diversity of social wasps (Silveira, 2002; Somavilla et al., 2014; Barbosa et al., 2016). Silveira (2002) highlighted that 200 species have been recorded in the Brazilian Amazon, which represents about two-thirds of all Brazilian Polistinae fauna. However, Somavilla and Oliveira (2013) emphasized the need for more studies focusing on wasp distribution in the region. Furthermore, the richness of social wasps in the Brazilian Amazon Forest is still underestimated (Somavilla & Oliveira, 2017). Abstract Most studies about the insect community in rainforests only focus on the forest understory, and even though the rainforest canopy is one of the most fascinating and diverse environments, it remains poorly explored. Therefore, we analyzed the difference between the social wasp composition in these two strata at the ZF-2 Station in the Brazilian Amazon Rainforest, Amazonas, Manaus, using flight interception Malaise traps, in the rainforest understory and canopy. We collected ninety-two species belonging to 18 genera; Polybia was the richest genera (22 species), followed by Mischocyttarus (14) and Agelaia (13). Fortyfour species were exclusively collected in the understory, twenty exclusively collected in the canopy, and twenty-eight in both strata. The understory was distinctly more diverse and more abundant than the canopy, while some rare or poorly collected species were only found in the canopy. We found a relationship between the species composition at the ZF-2 Station and the Ducke Reserve, Manaus. Therefore, we suggest using traps in canopy in the Amazon biome as an effective method for collecting a higher diversity of social wasps. Sociobiology An international journal on social insects

Despite the contributions of these works, Somavilla et al. (2014) stated "there are many sample gaps in the Amazon region and distribution and occurrence studies are necessary for improving this prior knowledge".This may be due to limited exploration in the rainforest understory (Somavilla et al., 2014).Inventories from the Amazon show that some Polistinae species have been represented by one or a few specimens with all collections carried out in understories only, while tree canopies have been poorly explored by researchers.The previously mentioned studies from understories, and only two studies used the suspended trap method in tree canopies (Roraima State and Ducke Reserve), but only as a complementary method and reaching low height, reaching at most 20 meters (Barroso et al., 2017;Somavilla & Oliveira, 2017).
The rainforest canopy is one of the most fascinating environments to develop studies about arthropod communities (Erwin, 2013;Nakamura et al., 2017).Although the tropics hold the highest species diversity (Erwin, 2013), many interactions in tropical forests are made in the canopies, due to the high number of species that inhabit canopies and the large amount of biomass that is generated in this stratum (Nakamura et al., 2017).Compared to other rainforest strata, the rainforest canopy presents higher illumination, temperatures, wind speeds, and relative humidity (Nakamura et al., 2017).
In this way, we asked: Is the social wasp fauna different between the tree canopy and the understory?Which environment/strata has the highest richness of social wasps?Herein, we present a list of species and discuss the use of interception traps, Malaise traps mounted in the canopy and understory, for collecting social wasps (Polistinae) at the ZF-2 Station in the Brazilian Amazon Rainforest.

Study area
The study site is located inside one of the National Institute for Amazonian Research (INPA) reserves -Tropical Silviculture Experimental Station, at Km 934 of BR 174 on the road ZF-2 (2°35'21"S -60 o 06'55"W), herein called ZF-2 station.This area is about 50 km North of Manaus in Amazonas, Brazil and is managed by the project: The Large-Scale Biosphere-Atmosphere Experiment in Amazonia (LBA-INPA), which is the largest study about climate change in the Amazon region.
According to the Köppen (1948) climate classification, the climate at the station is Am type, with low annual thermic range and average monthly rainfall over 60 mm.The average annual temperature is 26.7 ºC, ranging between 23.3 ºC and 31.4 ºC, while average annual rainfall is 2286 mm and relative humidity is around 80%.The rainy season is from December to May, and the dry season from June to November (Barbosa et al., 2015).The topography in this area is undulating with elevation ranging from 50 to 140 meters above sea level.

Wasp collection
Currently, Brazil holds one of the largest research programs that uses towers as a standard structure to reach the tree canopy.Such sampling tower is metallic and 40 m high, located inside a typical Amazonian landscape, which is comprised of an ombrophilous dense forest with a canopy reaching 40 m, or even 50 m due to emergent trees.In this type of forest, it is difficult to distinguish between median and lower strata, but the mean canopy height in the Amazon is 28.6 m (Higuchi et al., 2009).A description of the flora at the research station can be found in Martins et al. (2006).In the forest understory and canopy, we used a modified Malaise trap Gressit and Gressit model (1972) that was 6-meter long with two collector vials (Figure 1).In the canopy, the trap was mounted 32 meters from the ground in a metallic tower (6m wide x 6m length) to collect wasps and in understory on the forest floor.Both traps were active for fifteen consecutive days each month, for a period of eight months between July 2016 to March 2017.
The Polistinae specimens were sorted and identified at the Hymenoptera Laboratory of the National Institute of Amazonian Research (INPA).The vouchers were deposited into the INPA's Invertebrate Collection.Specimens were identified using the keys proposed by Richards (1978),  Carpenter andMarques (2001), andCarpenter (2004) and were compared to previously identified species from the INPA Invertebrate Collection.

Data analysis
We used Nonmetric Multidimensional Scaling (NMDS) (Minchin, 1987) to ordinate the inventories carried out in the Brazilian Amazon (see Introduction topic) according to their species composition and to verify whether species displayed a preference for a particular area and used dissimilarity coefficient for the species composition from the ZF-2 Station compared to other inventories in the Brazilian Amazon.This analysis was conducted in R version 3.3.3.(R Core Team, 2017) using vegan package 2.4-0 (Oksanen et al., 2016).For this analysis, we excluded the specimens identified as "morphospecies" or "varieties" and we used only the species collected in Malaise or suspended traps, and we exclude the species of active search, light trap or attractive trap.

Results
The survey from ZF-2 station revealed a rich fauna of social wasps.We found 1.548 specimens in ninety-two species belonging to genera (Table 1).Polybia Lepeletier was the richest genera (22 species), followed by Mischocyttarus de Saussure ( 14) and Agelaia Lepeletier (13).Together, these three genera comprised about 55% of the species listed.The only Brazilian Polistinae genera we did not collect from in the area were Asteloeca Raw, Chartergus Lepeletier, and Protonectarina Ducke, the first two not specious genera with three described species and Protonectarina with occurrence only for the Atlantic Forest.
In terms of specimens' numbers collected, 870 were from understory and 678 from canopy.Forty-four species were exclusively collected in the trap mounted in understory, twenty species were exclusively collected in the trap mounted in the tree canopy, and twenty-eight species were found in both traps (Table 1).All species of Agelaia, Apoica, and Pseudopolybia were collected in the understory.Species of Metapolybia, Nectarinella, and Synoeca were collected in the canopy.Polybia species were collected in both strata.
Several species are new records for Brazil or Amazonas state: Agelaia pallidiventris (new record for Brazil), Nectarinella manauara and Protopolybia rugulosa (second time both species were collected for Amazonian Rainforest), A. flavipennis, Polybia affinis, Po. minarum, Pr. nitida, Pr. rotundata and Pr.sedula (new records for Amazonas State).The following specimens could not be determined to the species level: one species of Agelaia, one species of Polybia (Myrapetra), and two species of Mischocyttarus (Haplometrobius).
According to the dissimilarity analysis between the social wasps' composition from different areas in the Amazon, we found a relationship between the species composition of ZF-2 Station andDucke Reserve (Figures 02, 03).The same relationship was found for Alvarães and Mamirauá Reserve (AM) and Caxuanã Reserve (PA).However, the composition of wasp fauna from the Rio Branco region (AC), Serra do Divisor National Park (AC), Maracá Ecological Station (RR), and Lagos Region (AP) were dissimilar from the ZF-2 Station.

Discussion
In surveys conducted in the Amazon rainforest, the diversity of social wasps is generally higher than in other biomes.Surprisingly, we collected 92 species from the ZF-2 Station in only eight months, using two passive collection trap sin two strata (understory and canopy 32 meters high).The 92 species from the ZF-2 station were distributed into 18 genera, mainly Polybia (22 species), Mischocyttarus (14), and Agelaia (13).
In other areas of Amazonas State, as the Ducke Reserve, 103 social wasp species were collected from the understory stratum only, however, six types of collection methods were used during several years of collection and different methods.At Ducke Reserve, 19 genera were collected, mainly Polybia (28 species), Agelaia (12 species) and Mischocyttarus (12 species).Therefore, the results of species composition from both the ZF-2 Station and the Ducke Reserve were very similar.Silva and Silveira (2009) and Somavilla et al. (2014) showed that fast inventories were efficient for sampling the most abundant species, recording three genera: Agelaia, Polybia and Mischocyttarus.Herein, we found, the same most abundant genera, which constituted more than 50% of the specimens collected specimens.Agelaia species usually form large colonies with millions of individuals (Zucchi et al., 1995), and, consequently, are more likely to be captured (Silva & Silveira, 2009).Mischocyttarus is the genus with more described species in social wasps, that can support the high diversity in this study.(Silveira, 2002) and Polybia has a very active foraging behavior, which facilitates the collection of these species in traps (Carpenter & Marques, 2001).
Most species of Epiponini, belonging to Apoica, Brachygastra, Clypearia, Leipomeles, Polybia, and Protopolybia, were only collected in Malaise traps.Conversely, this method was less efficient for capturing Polistes, a pattern that has also been reported by other researchers (Silveira, 2002;Silva & Silveira, 2009).Polistes nests at lower height and has usually small nests with few individuals, they are more difficult to be seen in forest (Carpenter & Marques, 2001).Therefore, using only Malaise traps to collect social wasps may underestimate the actual richness of Polistes.
Conversely, there was a strong relationship of nesting in the forest canopy, as the rare species Clypearia duckei, Epipona tatua, and Polybia depressa were only recorded in the canopy stratum at the ZF-2 Station and the Ducke Reserve (Somavilla & Oliveira, 2017).Furthermore, Richards (1978) recorded an E. tatua nest that was over 12 meters above ground in Mato Grosso, and Somavilla (2013) photographed a nest above 20 meters at the Ducke Reserve.Synoeca, and some Brachygastra and Polybia nests are also common in canopy, Richards (1978)  Insects inhabiting the canopy of the Amazon rainforest strata are poorly represented in collections, with many unknown specimens in these habitats (Rafael & Gorayeb, 1982).From inventories in the Amazon, some Polistinae species are represented by only one or a few specimens in the collections and occur in the tree canopy, a habitat that has been poorly explored by collectors.For example, in the two works they explored the canopy, 25 species in Ducke Reserve and seven in Roraima State, manly Agelaia, Clypearia, Metapolybia and Polybia (Barroso et al., 2017;Somavilla & Oliveira, 2017).Likewise, there are few methods for collecting insects in these strata, that are often times costly and impractical.Based on this observation and the need for collections in these environments, traps mounted in the canopy are efficient ways of collecting and improving knowledge about social wasp species, mainly in the Amazon rainforest.Furthermore, such methods should be used in conjunction with Malaise traps mounted on the forest floor and active search.Finally, our study shows that some species were only collected in the canopy (Table 1).
There are different methods for sampling social wasps, however, few studies have attempted to standardize these methods or establish comparable and adequate protocols to survey the fauna at a given site.One important factor to consider when implementing new social wasp sampling protocols is the distribution pattern of these organisms (Silveira, 2002;Somavilla et al., 2014).Similarly, it is also important to determine the most efficient traps to collect the target group and dispose of them in a standardized manner (Noll & Gomes, 2009).Through this study, we verified, that using traps in the upper forest canopy is very important for collecting social wasp species.Furthermore, such method should be included in the collection protocol for the Amazon biome.The similarities of social wasp species composition between ZF-2 Station and Ducke Reserve was not surprising since they are geographically close (50 km apart) and both areas have the same phytophysiognomy characteristics (ombrophilous dense and humid forest).In addition, these two sites, as well as some localities in Roraima State, are the only places where canopy fauna was collected.The species composition between Alvarães and Mamirauá in Amazonas and Caxiuanã in Pará were very similar, with both reserves located in lowland areas.
The sites with different phytophysiognomies and that were geographically distant were the most dissimilar sites.The Maracá Ecological Station with savanna formations, Serra do Divisor with more open and elevated vegetation, Rio Branco Region and the Amapá Lake Region, were the most dissimilar from the ZF-2 Station.Collection efforts were discontinuous in these places.
Unfortunately, different methods and collection efforts in surveys are complicated, as Silveira (2002) pointed out: "Comparisons of local fauna by the use of information in collections always confront obstacles arising from unsystematic collecting methodologies.In general, appropriate information about the effort spent in finding a certain number of species is not available, and data about the relative abundance of the species are hardly recoverable".We hope that this work stimulates other standardized surveys that explore the canopy fauna in the Amazon Region.
Finally, in this study we found that the social wasp fauna in the canopy is slightly different from the fauna of the understory, with many similar species in both forest strata.In addition, when comparing the strata, the understory had more diverse fauna than the canopy, while the canopy had some rare and poorly collected species.In this way, we suggest using traps suspended in the Amazon biome to collect a high diversity of social wasps.In addition, continuous collection efforts are important for better sampling in a particular area.

Fig 1 .
Fig 1. Sampling method at the ZF-2 Station (LBA/INPA): (A) Malaise trap in the understory and (B) Malaise trap in the canopy at 32 m high.
recorded a B. bilineolata nest above 13 meters above ground and Somavilla et al. (2012) recorded a S. surinama and P. liliacea nest about 20 meters above ground.