A Quantitative Baseline of Ants and Orchid Bees in Human-Modified Amazonian Landscapes in Paragominas , Pará , Brazil

Tropical forest ecosystems are the richest ecosystems on Earth (Pimm and Raven 2000), harboring up to two thirds of the planet biodiversity (Gardner et al., 2009). Yet, the region has been suffering intense human-impacts and is the among the most actives frontiers of land-cover changes in the world (Malhi et al., 2008). In Brazil, government efforts were yielding positive results, and deforestation rates have decreased from 2004 to 2014, although these rates have increased recently (PRODES-INPE 2015). Given this loss Abstract The lack of effective biodiversity baselines is a major impairment to implement conservation plans. Hence, constructing and updating species lists provides vital information about species distribution records. The Sustainable Amazon Network (in Portuguese Rede Amazônia Sustentável; RAS) is an interdisciplinary research initiative that aims to evaluate land-cover changes effects in eastern Brazilian Amazonia. Within the scope of this project, we sampled ants and orchid bees and herein present a list of species collected in Paragominas, PA, Brazil; the most complete lists of species published to date of these groups for the eastern Amazon. We sampled these insects across several land-cover types, from undisturbed forested habitats, through varyingly disturbed forested habitats and secondary forests to production areas (silviculture, pastures and arable fields). In total we recorded 285 species of ants and 36 species of orchid bees. Species richness was higher in primary forests for both groups, followed by production areas. Orchid bees reached their highest richness in secondary forests. For orchid bees, production areas were dominated by a few hyper-dominant species, such as Eulaema nigrita. For future assessments if the aim is to make a complete inventory, we recommend the use of additional sampling methods. Finally, we expect this study can be used as a baseline for understanding the effectiveness of ongoing changes in forest conservation and land management practices and in determining conservation status for several taxa described here. Sociobiology An international journal on social insects


Introduction
Tropical forest ecosystems are the richest ecosystems on Earth (Pimm and Raven 2000), harboring up to two thirds of the planet biodiversity (Gardner et al., 2009).Yet, the region has been suffering intense human-impacts and is the space and time, as well as across human-modified landscapes is a valuable tool for studying population ecology and biodiversity responses to human impacts in order to judge which human activities are most affecting biodiversity (Lach et al., 2010).
The Sustainable Amazon Network (in Portuguese, Rede Amazônia Sustentável, RAS, Gardner et al., 2013) assessed the biodiversity, ecosystem services and social aspects of human-modified forest landscapes in two frontier regions (Paragominas and Santarém-Belterra) of the eastern Brazilian Amazon, in the state of Pará.These regions have suffered from intense deforestation since the 1970's (Lindenmayer et al., 2004), although several governmental and social initiatives have been contributing to minimize and revert this process (Viana et al., 2016).Within the municipality of Paragominas, RAS fieldwork sampled trees and lianas (Berenguer et al., 2014), birds (Lees et al., 2012) and here we present data on a comprehensive survey of two terrestrial invertebrate groups: ants (Hymenoptera: Formicidae) and orchid bees (Hymenoptera: Apidae: Euglossina) selected for their ecological importance.
Ants (Hymenoptera: Formicidae) are a ubiquitous group, being numerically and ecologically dominant in tropical forests (Hölldobler & Wilson 2009;Lach et al., 2010).They play roles as seed dispersers (Christianini et al., 2007), in moving nutrients across soil horizons (Sousa-Souto et al., 2007) and in regulating populations of prey species (Folgarait, 1998).Ants are also easy to sample, have a relatively well established taxonomy and are numerically dominant nearly everywhere in the Neotropics throughout the year (Underwood & Fisher, 2006).
Orchid bees encompasses around 250 species endemic to the Neotropics (Nemesio & Rasmussen, 2011).One of the striking characteristics of this group is pollination of tightly associated plant species (Janzen, 1971).Given their sensitivity to environmental change (Nemesio & Vasconcelos, 2013) and ease with which they can be sampled makes them a costeffective ecological disturbance indicator group (Gardner et al., 2008).
Here we present an annotated checklist of ants and orchid bees collected in Paragominas, PA, Brazil; the most complete list species list produced to date for the western Amazon that, together with other efforts already conducted in the region (e.g.Kempf, 1970;Kalif et al., 2001;Santos et al., 2008) establishes a crucial biodiversity baseline for ongoing environmental monitoring.

Study site
We sampled the insects in Paragominas, a 2 million ha Amazonian municipality in Pará state, northeastern Brazil (Fig 1).The region was originally almost entirely covered with evergreen forests and regional climate is Am according to the Köppen classification (Alvares et al., 2013) with an average annual rainfall of 1800mm (Andrade, 2011) and mean annual temperatures of 26.3ºC (Pinto et al., 2009).We conducted all fieldwork between January-June 2011, during the rainy season.
Sampling was undertaken in 18 ca.5.000 ha.catchments covering the entire municipality (2 million hectares), distributed along a gradient of deforestation from undisturbed primary forest through varyingly disturbed primary forests, secondary forests, pastures and mechanized agriculture, where we established from 8-12 transects (300m) in each catchment, in a density of 1 transect/400ha (Fig 1).In total, we sampled 192 transects across the major land-use classes present in the region including undisturbed primary forests, varyingly disturbed (from logging and fire) primary and secondary forests and production areas (silviculture -Eucalyptus and Schyzolobium amazonicum, cattle pastures and arable fields).

Insect sampling
Within each transect we sampled both insect groups concurrently.Ants were sampled with epigaeic pitfall traps, composed of plastic containers (8 cm diameter) half filled with a solution of water, salt (5%) and soap (5%) and baited with sardines and honey, both inaccessible to the ants.In each transect we installed six pitfall traps separated 50 m from one another.
To sample orchid bees, we used four plastic bottles per transect, with one scent bait each (2L, 10 cm diameter, 35 cm height), tied to a tree trunk, 1.5 m above the ground.Male orchid bees were attracted to four types of scent baits (eugenol, methyl salicylate, vanilla and eucalyptol), separated by 50 m from each other.In both cases traps remained in the field for 48h prior to removal, see Fig 1 for a graphical representation of our sampling design.
We processed and identified the ants to the most precise taxonomic level possible using genera taxonomic keys (Fernández, 2003, Baccaro et al., 2015), and the most upto-date taxonomic revisions of each taxa, directly comparing with the available type images on AntWeb (Available from http://www.antweb.org.Accessed in April/2016), and checked against the reference collection of the Community Ecology Lab, affiliated with the Museu Regional de Entomologia da Universidade Federal de Viçosa (UFVB).We checked all the species in Ant Cat (Bolton; http://antcat.org, accessed in April/2016) to confirm the validity of the nomenclature we used.
We processed and identified orchid bees at EMBRAPA -Amazônia Oriental, adapting available taxonomic keys (Nemesio, 2009) and using the reference collection of EMBRAPA -Amazônia Oriental.A taxonomist, Dr André Nemésio, checked species identifications.We deposited voucher specimens of ants in the reference collection of the Community Ecology Lab, Universidade Federal de Viçosa.Orchid bees are deposited on the reference collection in EMBRAPA -Amazônia Oriental.

Statistical analyses
To assess our sampling sufficiency, we built sitebased species accumulation curves (Colwell et al., 2004) and estimated the total number of species sampled in each taxon using the first order Jackknife richness estimator.All analyses were performed using the R v.3.1.2(R Core Team 2015), using the package vegan (Oksanen et al., 2015).

Results
For both groups, species richness was higher in forested areas (average ± standard deviation; Ants 25.2 ± 7.3; Bees 6.45 ± 3.8) and lower in production areas (average ± standard deviation; Ants 16.9 ± 6.0; Bees 3.8 ± 2.5).Worth of highlighting, the richest assemblage of orchid bees was found in secondary forests.In Table 1, we present how species richness of both groups was distributed in transects across the studied land-use classes.

Ants
We sampled a total of 285 species of ants, in 60 genera, belonging to nine subfamilies.We assigned a name to all genera and among them, 132 were identified to specieslevel or very close to a given taxa, and 11 were assigned to Map of the sampling region within the Brazilian contexts.Circles represent each sampled transect and circle sizes represent relative species richness for each taxa.In colours are represented micro-catchments, where samplings were located and greener colours represent higher primary forest cover, while redder colours represent lower primary forest cover.
species groups or complexes where exact species identification was impossible.The remaining 142 are identified and were assigned a morphospecies code that applies only to this study.A list of the species and morphospecies is given in Table 2.At least two new species of the genera Xenomyrmex (Xenomyrmex PGM-01) and Hylomyrma (Hylomyrma PGM-01) were sampled (L.do Prado and M. Ulysséa, personal communications, respectively) and these specimens are deposited at the Museu de Zoologia da Universidade de São Paulo.By comparing with the type photos (AntWeb Available from http://www.antweb.org.Accessed in March 2015) and/or original descriptions, we considered some species in the list of ants to be slightly different from the original species (e.g.Strumigenys aff.perparva; Pachycondyla aff.purpurascens) and are therefore treated as morphospecies very close to that particular species.It remains to be investigated whether each of them represent still undescribed sister species or a case of intraspecific variation.Others were tentatively assigned to a given taxa but precise identification was not possible (e.g.Sericomyrmex cf.parvulus).The regional species accumulation curve is not asymptotic but does flatten towards the end (Fig 2a), and the 1 st order Jackknife estimator suggests that we sampled 77.5% of the total species richness.

Discussion
Our study is the most comprehensive sampling to date of ants and orchid bees for any area of the eastern Amazon and we hope it both fosters future biodiversity studies in the region, and can be used as an evidence-baseline for future Red Listing classification exercises for invertebrates.We consider our sampling effort sufficient for both taxa at the regional scale, with at least 77% of the estimated diversity sampled for all taxa.For ants, the only previous study we are aware of in Paragominas yielded only 74 species belonging to 30 genera (Kalif et al., 2001).Even so, by using a different sampling method (Winkler extractors), this study sampled species not represented in our study, demonstrating the importance of implementing complementary methods of sampling to survey this region.Exploring seldom-studied habitats such as the forest canopy (Basset et al., 2012) or underground soil layers (Wilkie et al., 2007;Schmidt & Solar, 2010;Schmidt et al., 2014) also offers significant potential to increase the number of species described for the region.
Orchid-bees are poorly known in the Amazonian region, our total of 36 species is of a similar magnitude as other studies in current and former terra-firme forest habitats in the region (Oliveira & Campos, 1996;Rasmussen, 2009;Storck-Tonon et al., 2009;Abrahamczyk et al., 2011).We sampled in a very diverse range of habitats and in a large area, however additional species are likely to be encountered using a greater range of bait types (Nemesio & Vasconcelos, 2013).For both taxa, we acknowledge that by sampling only during the rainy season, we might have missed some species that prefer dry climates.We believe this was not the case for ants, as they are colonial organisms and are know to be far more active during the rainy season (Hölldobler & Wilson, 2009), improving capture by pitfall traps.Orchid bees do have marked seasonality (Abrahamczyk et al., 2012;Abrahamczyk et al., 2014), however the highest species richness for this group is found in the rainy season (Abrahamczyk et al., 2011).In order to cover a larger area as possible, while capturing the maximum diversity, we opted to sample during the rainy season.
Unsurprisingly we found forests to be more species rich than non-forest habitats, as was the case for other taxonomic

Orchid bees
We sampled 3,769 orchid bees of 36 species, belonging to four of the five known genera in the group.Out of the total, 34 species could be identified to species level.Only one species of Eufriesea and one Eulaema were assigned to morphospecies.The complete list of species is available in Table 3. Species accumulation curves were near asymptotic (Fig 2b), and we sampled 87% of the total species richness estimated by 1 st the order Jackknife estimator.
groups in the same study region (e.g.Moura et al., 2013).However, it is worth highlighting here that in a detailed analysis of ant responses to land-use changes and forest disturbance, we observed more subtle patterns of diversity within and between the major land-use types (Solar et al., 2016).Studies on ants and other groups have been showing that the recovery of species diversity in disturbed forests is not guaranteed, even when considering samples conducted relatively mature secondary forests on average (Wilkie et al., 2009;Barlow et al., 2016;Solar et al., 2016).
By contrast the orchid bees exhibited similar levels of richness in all forest types, with highest richness in secondary forests.This is an expected result, considering orchid bees have high vagility, being able to fly several kilometers a day (Janzen, 1971), which in turn could cause high rates of spillover.On the other hand, they may rapidly colonize new habitats, and may persist in small forest patches (Nemesio & Silveira, 2007, 2010).Nevertheless, orchid bees are seriously affected by deforestation and forest fragmentation (Nemesio & Silveira, 2010) and forest-dependent species are seriously threatened; see Nemesio (2013) for case study in the Atlantic forest.
Open areas are often the least hospitable environments (Gascon et al., 1999), and are commonly dominated by generalist species.This is the case of the orchid bee Eulaema nigrita Lepeletier de Saint Fargeau, 1841.This species is an example of a non-forest species, which can be rarely recorded in forest fragments but is highly abundant in open areas.Compared with Eu. nigrita, all other species in pasture transects had a relative occurrence frequency of less than 1%.

Conservation implications
Enhanced documentation of local diversity patterns of insects and other organisms are invaluable in helping to assess conservation priorities and assess management effectiveness.Indeed, it would be highly desirable to develop conservation strategies or conclusions taking into account a more comprehensive understanding of diversity and distribution of the major groups of organisms inhabiting a given locality.We hope this assessment provides the baseline for new community and population studies on these groups of insects in the region.Paragominas is the flagship municipality in the state of Para for the Green Municipalities Program (in Portuguese, Programa Municípios Verdes -http://municipiosverdes.com.br/), an initiative aiming to stop deforestation and promote secondary forest recovery and sustainable land-use practices in the region (Viana et al. 2016).We suggest therefore this study and the patterns of species distributions can be used as baselines for future studies of forest changes in that region to assess the conservation success of the program.for their suggestions in the first draft of this manuscript.Alexander C. Lees revised English grammar and spelling.We received financial support from Instituto Nacional de Ciência e Tecnologia -Biodiversidade e Uso da Terra na Amazônia (CNPq 574008/2008-0), Empresa Brasileira de Pesquisa Agropecuária -Embrapa (SEG:02.08.06.005.00), the UK government Darwin Initiative (17-023), The Nature Conservancy, and Natural Environment Research Council (NERC) (NE/F01614X/1 and NE/G000816/1).RRCS is funded by CAPES/PNPD.JHS is funded by CNPq.This is the contribution number 51 of the Sustainable Amazon Network (http://www.redeamazoniasustentavel.org).

Fig
Fig 1.Map of the sampling region within the Brazilian contexts.Circles represent each sampled transect and circle sizes represent relative species richness for each taxa.In colours are represented micro-catchments, where samplings were located and greener colours represent higher primary forest cover, while redder colours represent lower primary forest cover.

Fig 2 .
Fig 2. Species accumulation curves for both studied taxonomic groups.Each curve was drawn after 10.000 randomisations of original data and the shaded area represents the standard deviation.In the x-axis we have number of sampled transects, in the y-axis, accumulated species richness.

Table 1 .
Average species richness ± confidence intervals for species richness of each taxon per transect in each land-use type.

Table 2 .
List of ant species collected in this study.Values represent number of records per pitfall traps of each species in each land-use type: PFU -primary forest undisturbed, PFL -primary forest logged, PFLB -primary forest logged and burnt, SEF -secondary forest, REFreforestation with commercial species, PAS -pasture, AGR -agricultural areas.

Table 2 .
List of ant species collected in this study.Values represent number of records per pitfall traps of each species in each land-use type: PFU -primary forest undisturbed, PFL -primary forest logged, PFLB -primary forest logged and burnt, SEF -secondary forest, REFreforestation with commercial species, PAS -pasture, AGR -agricultural areas.(Continuation)

Table 2 .
List of ant species collected in this study.Values represent number of records per pitfall traps of each species in each land-use type: PFU -primary forest undisturbed, PFL -primary forest logged, PFLB -primary forest logged and burnt, SEF -secondary forest, REFreforestation with commercial species, PAS -pasture, AGR -agricultural areas.(Continuation)

Table 2 .
List of ant species collected in this study.Values represent number of records per pitfall traps of each species in each land-use type: PFU -primary forest undisturbed, PFL -primary forest logged, PFLB -primary forest logged and burnt, SEF -secondary forest, REFreforestation with commercial species, PAS -pasture, AGR -agricultural areas.(Continuation)

Table 2 .
List of ant species collected in this study.Values represent number of records per pitfall traps of each species in each land-use type: PFU -primary forest undisturbed, PFL -primary forest logged, PFLB -primary forest logged and burnt, SEF -secondary forest, REFreforestation with commercial species, PAS -pasture, AGR -agricultural areas.(Continuation)

Table 2 .
List of ant species collected in this study.Values represent number of records per pitfall traps of each species in each land-use type: PFU -primary forest undisturbed, PFL -primary forest logged, PFLB -primary forest logged and burnt, SEF -secondary forest, REFreforestation with commercial species, PAS -pasture, AGR -agricultural areas.(Continuation)

Table 2 .
List of ant species collected in this study.Values represent number of records per pitfall traps of each species in each land-use type: PFU -primary forest undisturbed, PFL -primary forest logged, PFLB -primary forest logged and burnt, SEF -secondary forest, REFreforestation with commercial species, PAS -pasture, AGR -agricultural areas.(Continuation)

Table 2 .
List of ant species collected in this study.Values represent number of records per pitfall traps of each species in each land-use type: PFU -primary forest undisturbed, PFL -primary forest logged, PFLB -primary forest logged and burnt, SEF -secondary forest, REFreforestation with commercial species, PAS -pasture, AGR -agricultural areas.(Continuation)

Table 3 .
List of orchid bees species collected in this study.Values represent the number of individuals of each species in each land-use type: PFU -primary forest undisturbed, PFL -primary forest logged, PFLB -primary forest logged and burnt, SEF -secondary forest, REFreforestation with commercial species, PAS -pasture, AGR -agricultural areas.