Amazon Rainforest Ant-Fauna of Parque Estadual do Cristalino : Understory and Ground-Dwelling Ants

Ants are ecologically dominant, playing several ecosystem services in the environment (Del Toro et al., 2012) and occupying the most varied habitats (Bruhl et al., 1998; Dejean et al., 2015; Rocha et al. 2015; Pape et al., 2016). Ants strongly influence several terrestrial communities due to their abundance and potential relationships with various biological groups. The interactions of ants can be positive or negative, ranging from interactions with microorganisms (Hanshew et al., 2014; Sanders et al., 2014), plants (Izzo & Vasconcelos, 2009; Vicente et al., 2012, 2014; Koch et al., 2015) and other invertebrates (Dáttilo et al., 2012a; Freitas & Rossi, 2015; Puker et al., 2015), even other ants (Adams et al., 2007; Sanhudo et al., 2008; Gallego-Ropero & Feitosa, 2014). Therefore, ants have been used as a model to understand the various ecological patterns and as valuable bio-indicators of Abstract Ants are ecologically dominant and have been used as valuable bio-indicators of environmental change or disturbance being used in monitoring inventories. However, the majority of inventories have concentrated on ground-dwelling ant fauna disregarding arboreal fauna. This paper aimed to list the ant species collected both on the ground and in the vegetation of the Parque Estadual do Cristalino, an important protected site in the center of the southern Amazon. Moreover, we compared the composition of the ground dwelling and vegetation foraging ants. Two hundred and three (203) species distributed among 23 genera and eight subfamilies were sampled, wherein 34 species had not yet been reported in the literature for Mato Grosso State. As expected, the abundance and richness of ants was higher on the ground than in the understory. Also, the composition of the ant assemblages was different between these habitats (with only 20% occurring in both) indicating that complementary methods which include arboreal and terrestrial ants are indicated for efficient inventory. This study provides an inventory of the arboreal and ground ant fauna contributing to the knowledge and conservation of Amazonian ant fauna. Sociobiology An international journal on social insects


Introduction
Ants are ecologically dominant, playing several ecosystem services in the environment (Del Toro et al., 2012) and occupying the most varied habitats (Bruhl et al., 1998;Dejean et al., 2015;Rocha et al. 2015;Pape et al., 2016).Ants strongly influence several terrestrial communities due to their abundance and potential relationships with various biological groups.The interactions of ants can be positive or negative, ranging from interactions with microorganisms (Hanshew et al., 2014;Sanders et al., 2014), plants (Izzo & Vasconcelos, 2009;Vicente et al., 2012Vicente et al., , 2014;;Koch et al., 2015) and other invertebrates (Dáttilo et al., 2012a;Freitas & Rossi, 2015;Puker et al., 2015), even other ants (Adams et al., 2007;Sanhudo et al., 2008;Gallego-Ropero & Feitosa, 2014).Therefore, ants have been used as a model to understand the various ecological patterns and as valuable bio-indicators of Abstract Ants are ecologically dominant and have been used as valuable bio-indicators of environmental change or disturbance being used in monitoring inventories.However, the majority of inventories have concentrated on ground-dwelling ant fauna disregarding arboreal fauna.This paper aimed to list the ant species collected both on the ground and in the vegetation of the Parque Estadual do Cristalino, an important protected site in the center of the southern Amazon.Moreover, we compared the composition of the ground dwelling and vegetation foraging ants.Two hundred and three (203) species distributed among 23 genera and eight subfamilies were sampled, wherein 34 species had not yet been reported in the literature for Mato Grosso State.As expected, the abundance and richness of ants was higher on the ground than in the understory.Also, the composition of the ant assemblages was different between these habitats (with only 20% occurring in both) indicating that complementary methods which include arboreal and terrestrial ants are indicated for efficient inventory.This study provides an inventory of the arboreal and ground ant fauna contributing to the knowledge and conservation of Amazonian ant fauna.
Regardless of their ecological importance and diversity, studies on Brazilian ant fauna were usually concentrated in the Atlantic Forest (Feitosa & Ribeiro, 2005;Figueiredo et al., 2013;Silva et al., 2007), the Cerrado (Dáttilo et al., 2012b;Silva & Brandão, 2014;Camacho & Vasconcelos, 2015), and Pantanal (Ribas & Schoereder, 2007;Silva et al., 2013;Meurer et al., 2015).Although the Amazon is one of the largest and heterogeneous tropical forests remaining, the panorama is not different.The studies of Amazonian ants in Brazil are generally concentrated in small parts of the Central region (e.g.Vasconcelos et al., 2006Vasconcelos et al., , 2010;;Santos et al., 2007;Baccaro et al., 2013;Souza et al., 2012).Studies in the southern Amazon region are few, and because they focus on local ecological patterns, the list of species therein usually does not have confirmation of specialists and comprises a 1 -Universidade Federal de Mato Grosso, Cuiabá-MT, Brazil 2 -Museu de Zoologia da Universidade de São Paulo, São Paulo-SP, Brazil REsEARCh ARTICLE -AnTs large number of morphospecies (for example: Dáttilo et al., 2013;Falcão et al., 2015).The absence of surveys in this area is particularly problematic to the knowledge of the distribution of ant diversity in the tropics.More than being a frontier between the Amazonia and Cerrado biomes, the south of Amazonia is a frontier of fast agricultural expansion, being subject to enormous rates of deforestation (Fearnside, 2005).In such a scenario, several species could become, and probably some already have become extinct without being known by science.This is worrying in a region that has many species with unknown distribution (Vicente et al., 2011(Vicente et al., , 2012(Vicente et al., , 2015(Vicente et al., , 2016;;Prado et al., 2016).
Whereas inventories are fundamental to promote conservation of tropical forest remnants and the ants are important components of biodiversity, this work aims to contribute to the knowledge of the Amazonian ground-dwelling and arboreal ant fauna.The reserve is a terra firme Tropical Rainforest located in the Southern Amazon, but with little known diversity of various taxa.In this study, we list the fauna of ground-dwelling ants and the assemblages of ants foraging on lower vegetation in the same site.We compared the difference between the strata in order to demonstrate the increase in local biodiversity by using a simple additional method.

Study area
The surveys were conducted in an Amazonian Research Program Station in Biodiversity in the Brazilian Parque Estadual do Cristalino (henceforward PEC), with RAPELD methodology (see Magnusson et al., 2005).The park is a large protected area defined as "Priority for Conservation Site" (Ministério do Meio Ambiente [MMA], 2001) that hosts an area of 184,900 ha, situated in the middle of an area called "Arc of Deforestation" (Fearnside, 2005)

Ant sampling
The inventory of ants was conducted between November 2012 (begining of the rainy season with sporadic rains) and May 2013 (end of the rainy season with sporadic rains) in 11 trails 250m long, with a distance between sites of at least 1 km.In each trail, ants were collected every 25m, resulting in 20 samples per trail (10 samples of the ground-dwelling ants and 10 samples of the arboreal ants) and totaling 220 samples (110 samples of ground-dwelling ants and 110 samples of arboreal ants).The collection of the ground ant assemblages was made using a single pitfall trap that remains installed for 48 hours in each sample point.For sampling arboreal ants a beating-tray method was used in vegetation of understory.In each of these points, we selected the four treelets equidistant of about 2 m of the pitfall.Under each treelet, a white canvas was installed to prevent some ants from jumping and getting away from the sampling (Dáttilo et al., 2013), and about one meter from the ground to prevent ants from walking on the ground and climbing on the canvas.Then all vegetation in each treelet within 1m² with between 1-3 meters in height was steadily shaken and the ants that fell on the canvas were properly collected.The sampling of ground and arboreal ants was carried out concomitantly, so the sampling of arboreal ants was performed between 9 am and 3 pm on the same day that the pitfall was removed.

Ant identification
Ants collected initially were identified using dichotomous key for subfamilies and genera available in Baccaro et al. (2015).So as to separate into morphospecies and identify to a specific level we used several taxonomic keys (Brandão, 1990;Fernández, 2003;Longino, 2003a;MacKay & MacKay, 2010;Fernandes et al., 2014).Posteriorly we made comparisons with specimens deposited at the Laboratório de Ecologia de Comunidades from the Centro de Biodiversidade da Universidade Federal de Mato Grosso (UFMT) and the ant collection from the Laboratório de Sistemática, Evolução e Biologia de Hymenoptera from the Museu de Zoologia da Universidade de São Paulo (MZSP).We also consulted specialists to confirm species identification (see acknowledgments).Vouchers were deposited in the collections mentioned above.After the identification and confirmation by specialists we consulted the distribution of each species using the AntMaps (Janicki et al., 2016), a new interactive tool recently used by taxonomists and ecologists (Gérnard & Economo, 2015;Vicente et al., 2015;Santos-Silva et al., 2016;Wepfer et al., 2016).This tool comprises the geographic distributions of more than 15,000 species/subspecies in over 1.7 million records of about 8,650 publications, museum collections and specimen databases (Janicki et al., 2016).Furthermore, we researched in bibliographic references related to taxonomy of species or lists of ants from region.

Data analysis
We carried out a t-test to access the difference of ant abundance and richness patterns between ground and understory.Because ants are social insects, the sampled abundance of workers in pitfalls may be strongly related to the proximity to the nest (Gotelli et al., 2011) or workers' number in a colony, which varies greatly between species (Baccaro et al., 2015).To minimize this effect, we treated the abundance as sample-based occurrences (Gotelli et al., 2011) as is commonly done in studies with ants (Ryder-Wilkie et al., 2010;Dáttilo & Izzo, 2012;Baccaro et al., 2013).To access the distribution patterns of the ant assemblages we performed an ordenation with Principal Coordinate Analysis (PCoA) technique based on a matrix of Raup-Crick dissimilarity measures calculated in a binary matrix with presence and absence date.Raup-Crick was used because it is a robust index that calculates how different pairwise samples are than expected by chance implementing null models which consider the variation in the number of local species and alpha diversity (Chase et al., 2011).This index is frequently used to compare the dissimilarity of invertebrate communities between different habitats (Ryder-Wilkie et al., 2010;Ribas et al., 2012;Reis et al., 2013).We utilized the first two PCoA axis (which represented 66.76 % of explication) in a Multivariate Analysis of Variance (MANOVA) as dependent variables and vertical habitat (ground and understory) as a factor.All analysis were performed with R-software (R Core Team, 2015) using the Vegan-package (Oksanen et al., 2015).

Results
We recorded 1,581 occurrences of ants in the 220 samples.The ants collected belong to 203 species, 45 genera and eight subfamilies.Subfamilies with greater richness were Myrmicinae with 23 genera and 113 species, Ponerinae with eight genera and 23 species and Formicinae with five genera and 27 species.The genus with greater richness was Pheidole (37 species), Camponotus (17), Neoponera and Crematogaster (both 10 species).
Both abundance (p < 0.001) and richness (p < 0.001) of ants were different between the vertical strata (Table 1 -Fig 2), being greater on the ground than on the understory.We collected almost 63.4% of ant occurrences (total: 1,002, mean: 89.1, SD: ±26.3 per sample) and 65% of ant species (total: 143 species, mean: 47.91, SD: ±10.26) on the ground stratum.The understory showed a total of 579 occurrences (mean: 51±, SD: 15.1) and a total of 100 species (mean: 26.6, ±SD: 6.67).Of these 203 species, 164 species were restricted to a vertical stratum with 104 species collected only on the ground and 60 only on the understory, and just 39 species collected in both strata (Table 2).Although on average 41.09 species by trail (±SD: 9.07) were exclusive to the ground, there is a mean increase of 19.73 species by trail (±SD: 5.85) by adding beating tray in the ant collection.Only one subfamily (Amblyoponinae) was sampled exclusively on the ground.Consequently, the species composition was different between the habitats (MANOVA -Pillai-Trace: 0.852, F 1,20 : 54.873, p < 0.001 -Fig 3).Among the species sampled, 34 were recorded for the first time for Mato Grosso State, Brazil (Table 2).

Discussion
Of the inventories on ant assemblages conducted in the Amazon, the major part of the studies focused just on grounddwelling ants (Miranda et al., 2012;Souza et al., 2012Souza et al., , 2015;;Baccaro et al., 2013) and few comprised both terrestrial and arboreal ant fauna detailing the strata in which each species was collected (e.g.: Vasconcelos & Vilhena, 2006;Ryder-Wilkie et al., 2010).These few studies carried out in the Amazon embracing both soil and vegetation ants have demonstrated extremely different ant assemblages between these habitats (Vasconcelos & Vilhena, 2006;Ryder-Wilkie et al., 2010).The methods employed here are very different.The both methods capture active foraging ants, however the pitfall traps were active during day and night whereas the beating tray method was used only during the day.This can explain the small number of species collected on vegetation, preventing strong inferences towards the patterns of species richness.However we observed a strong turnover on ant composition, since only about 20% of all collected species were sampled both on the ground and on vegetation.Hence, even missing several nocturnal species on vegetation strata, the species collected just on vegetation were certainly absent on ground strata.Therefore, the vertical stratification in the ant fauna is a robust ecological pattern found in both the Amazon forest and in other biomes (Bruhl et al., 1998;Vasconcelos & Vilhena, 2006;Neves et al., 2013;Camacho & Vasconcelos, 2015).Additionally our results indicate that, using an additional simple method focusing on vegetation foraging ants, as beating-tray, one can increase the number of collected species.Therefore, besides the vegetation foraging ant community can show different ecological patterns they contribute to our understanding of the real ant biodiversity in an area and, as well as the knowledge of biogeographical patterns.
The Amblyoponinae subfamily, considered as a basal group in the evolution of ants (Saux et al., 2004), as well as the 18 other genera were sampled exclusively on the ground.Among these genera collected only in soil are the fungusgrowing ants Atta, Mycetarotes, Mycocepurus, Myrmicocrypta and Sericomyrmex.These ants nest directly on the ground or in rotten wood in contact with the soil (Baccaro et al., 2015).Except for the Atta species which climbs into vegetation to cut fresh vegetable matter, the other abovementioned species take advantage of organic matter found in litter such as fallen leaves, fruits, flowers, feces and corpses (Leal & Oliveira, 2000;Mayhé-Nunes & Brandão, 2006).Hylomyrma immanis, Octostruma balzani, Rogeria scobinata and three unidentified Hypoponera species also were collected exclusively in the soil.These genera have species with cryptic behavior (Lapolla  & Sosa-Calvo, 2006;Longino, 2013b).The sampling of only one Daceton armigerum worker on the ground should be considered as an accidental record, since species of this genus have adaptations such as hook-shaped claws which provide remarkable adhesion to the arboreal substrate (Billen et al., 2016) where they nest in tree branches (Azorsa & Sosa-Calvo, 2008;Vicente et al., 2011) and forage (Dejean et al., 2012).
Despite their hypogeal origin, ants diversify occupying other habitats (Lucky et al., 2013) and the occupation of the arboreal environment resulted in morphological and behavioral adaptations both to get around and feed on available resources in vegetation (Orivel & Dejean, 1999;Dejean et al., 2005).These adaptations were observed in the various Neoponera species sampled.Neoponera is composed mainly by arboreal behavior species (Schmidt & Shattuck, 2014) that live or forage on vegetation, and have morphological adaptations regarding modifications on claws, shape and adhesion (Orivel et al., 2001).Four other genera were restricted to vegetation: Azteca, Myrmelachista, Nesomyrmex and Tapinoma.Species of these genera usually nest on plants (Longino, 2007;Nakano et al., 2013;Longino, 2004).Cephalotes was another genus with a greater number of arboreal species, with just C. atratus being collected in both strata.The Cephalotes ant genus has as their main features a diet based largely on pollen and nectar and nesting in pre-existing plant cavities (Byk & Del-Claro, 2010), however, C. atratus can often be found foraging on the ground (Corn, 1980).Although some subfamilies as Pseudomyrmecinae and Dolichoderinae had a large proportion of species collected in the vegetation, no subfamily has been sampled exclusively to only this vertical strata.In Pseudomyrmecinae, only Pseudomyrmex tenuis was collected in both soil (one worker) and vegetation (seven workers).P. tenuis is a widespread ant species, frequently sampled foraging in all strata as epigaeic, arboreal and canopy stratum (Vasconcelos & Vilhena, 2006;Neves et al., 2013;Camacho & Vasconcelos, 2015;Souza et al., 2015).Also, in a Dolichoderinae subfamily with 18 ant species, only five species of Dolichoderus genus, occurred also on the ground.
Camponotus genus was particularly representative (17 species; 84 samples) in relation to the total number of collected species (5.6%).In two areas of Central Amazon, both in Amazonas state, Brazil, Camponotus species comprises 2.7% and 2.3% of total species collect on the ground (Souza et al., 2015).In addition, in the Central South Amazon this proportion ranges from 1.99% to 4.62% of ground ants in preserved areas of Rondônia state (Souza et al., 2015).On the other hand, in samples made in ground and understory of patches of forest in Southern Amazon in Rondonia state and Central East Amazon in Pará state, both in Brazil, there was a ratio of 9.33% and 7.09%, respectively (Santos-Silva et al. 2016;Vasconcelos et al., 2006).Therefore, this high proportion of Camponotus found in the Parque Estadual do Cristalino confirm that simple methodologies could sampled a greater number of considered arboreal ant species, such as Camponotus.
Mato Grosso State is the third largest state in Brazil, hosting three different Biomes and the many new records in an inventory (18 species collected in the soil, seven in vegetation and 10 in both strata) demonstrating a gap or shortcoming related mainly to the lack of surveys throughout the South American territory (but see : Kempf 1972;Brandão 1991;Silva et al., 2013;Meurer et al., 2015).Although there are a number of studies on the ecology and distribution of ants in the Amazon in northern Mato Grosso (Dáttilo et al., 2013;Falcão et al., 2015;Vicente et al., 2011Vicente et al., , 2012Vicente et al., , 2014Vicente et al., , 2015) ) this is the first list of ant species for the region, as well as being the most complete list published until now for the southern Amazon.In summary, this work demonstrates that inventories should consider ants of both ground and vegetation for a better sampling of local ant diversity.In addition, the number of new records of ant fauna (35 species) for Mato Grosso state and the number of unnamed species (97 morphospecies), show that the ant fauna of the Amazon in general are not common or are extremely unknown.The tropical forests are being increasingly threatened by anthropic activities and Mato Grosso state in particular because they are located in a region with a historically strong pressure for disturbance, being an agricultural frontier.Therefore, efforts should be made to understand ant diversity distribution patterns and propose conservation strategies in this region.
Appendix.List of species recorded at the Parque Estadual do Cristalino.Numbers represent the total number of occurrences in strata.*First record in literature from Mato Grosso state.

Fig 1 .
Fig 1. Location map of the Parque Estadual do Cristalino (filled circle).

Fig 2 .
Fig 2. (a) Abundance mean of ants and (b) richness of ant species in vertical strata (ground and understory).

Fig 3 .
Fig 3. Composition of ground (filled circle) and understory (filled circle) ant community of the Parque Estadual do Cristalino.Principal Coordinate Analysis axis represented 71.6% of total ordination.

Table 1 .
t-test measures for difference in abundance and richness between ground and arboreal ant assemblages.
Appendix.List of species recorded at the Parque Estadual do Cristalino.Numbers represent the total number of occurrences in strata.*First record in literature from Mato Grosso state.(Continuation) Appendix.List of species recorded at the Parque Estadual do Cristalino.Numbers represent the total number of occurrences in strata.*First record in literature from Mato Grosso state.(Continuation) Appendix.List of species recorded at the Parque Estadual do Cristalino.Numbers represent the total number of occurrences in strata.*First record in literature from Mato Grosso state.(Continuation) Appendix.List of species recorded at the Parque Estadual do Cristalino.Numbers represent the total number of occurrences in strata.*First record in literature from Mato Grosso state.(Continuation) Appendix.List of species recorded at the Parque Estadual do Cristalino.Numbers represent the total number of occurrences in strata.*First record in literature from Mato Grosso state.(Continuation)