Ant Community Structure ( Hymenoptera : Formicidae ) in Two Neighborhoods with Different Urban Profiles in the City of São Paulo , Brazil

Some ant species are highly abundant in cities, may form huge unicolonial populations with thousands of individuals able to displace native fauna, and impoverish ecological relationships in urban environments. In this work, we study the ant community in two neighborhoods with different urban profiles, one recently populated and another from the 1900s in the city of São Paulo, Brazil. Two hundred and ninety houses were sampled with baits for ant collections. Results show that the recent urbanized neighborhood with greater disturbance favors opportunistic and dominant species to colonize it, like Tapinoma melanocephalum. We also made a temporal analysis in the ancient neighborhood, collecting ants after ten years from a first survey. T. melanocephalum has a broader range than ten years ago, displaced other ant species, but confronts with Pheidole megacephala that was not found in the recent urbanized neighborhood.

Around the world, recordings on P. megacephala point this species as one of the most dominant and aggressive tramp ant species.In 1908, the big-headed ant invaded a Caribbean island, and surveys show that this is the only ant species found in that environment [6].The same data were found in two, from five islands surveyed in Florida [7].In Australia, P. megacephala displaced dominant Dolichoderinae species, Camponotini, and other Pheidole species in forested areas [8].The big-headed ant biomass in areas where it dominates is 18 times greater than the native ant species biomass where it does not occur [9].In a study conducted in urban areas from the Brazilian Cerrado, P. megacephala predominated in areas with low conservation efforts [5].
Tapinoma melanocephalum originated from the Indo-Pacific region is one of the ant species with broader distribution [10].As in other parts of the world, it was the most abundant species in a neighborhood in the city of São Paulo [3], in a hospital in the Central-Eastern Brazil [11], in residences in Ilhéus [2], Uberlândia [4], and Mogi das Cruzes [12].
Studies have focused on the biology of invading species and found out that they share some common characteristics like multiple queens, polidomic nests with a huge cooperation net among the nests, without any aggressive behavior [15], and also their clear association with disturbed environments [16].On the other hand, one might understand which are the patterns and characteristics in the urban environment to shelter invading ant species and use this knowledge to promote satisfactory management and control.In this way, Clarke et al. [17] examined soil characteristics and type of vegetation in urban parks in San Francisco, USA, and concluded that soil moisture positively affects ant species richness, and percent of natural area that is forested was negatively related.Forys and Allen [18] established a positive correlation between exotic and native ant species in Florida.The authors stated that the same biotic and abiotic characteristics may favor both species.But they highlighted that an increase on urban sprawl leads to an increase on the number of exotic ant species, once they are known to displace native ant fauna.
In New York, a study revealed that besides introduced species that are usually favored by disturbance, this process must be much more complex than it is known [19].In this survey, Tetramorium caespitum (L.), a generalist exotic species that nests in cracks on sidewalks, was the most collected ant, but in most streets, it occurred with other ant species, exotic or native to the USA.This same ant species was also collected in a survey conducted in the surroundings of Montreal close to a forested area [20].
For this purpose, two neighborhoods were surveyed, and ant presence was associated to recent or ancient urbanization and other abiotic factors furnished by man.Households were randomly chosen.We could repeat the ant collection in the same 25 houses in Vila Mariana in 2009-2011 after 10 years.The other 54 houses were surveyed for the first time in the recent years.Baits based on dehydrated liver, pineapple cake, and honey were set in 7 cm length straws which were placed in the houses, 15 baits in each house and three in each room: living room, bedroom, kitchen, bathroom, and outdoors (peridomiciliar area).In addition, manual collections were also performed using brushes and a bottle-type vacuum.The baits were left for 24 hours, and collected ants were taken to the laboratory for identification and counting.In addition to the collections, it was also filled in a form with data relating to the environment as house conservation, cleanness, number of children, and presence of pets.

Statistical Analysis.
We used descriptive statistics with graphics and tables.Categorical and numerical variables for neighborhood characterization were obtained from a form filled in for each ant collection (Table 1).We computed percentage values for categorical analysis (number of floors and children, pet and garden presence) and the mean number for numerical variables (number of rooms per house, number of residents and house age).
Association analysis was calculated for the five most collected ant species in the three samples.For that purpose, the Pearson's chi-square test was used or the Fisher's exact test (for expected values less than 5).
For numerical variables, the Mann-Whitney test was used to compare two categories, and the Kruskal-Wallis Test was used to compare three or more categories due to the absence of normal distribution from the variables [21].The level of 5% of significance was used (P < 0.05).The SAS system for Windows (statistical analysis system), version 8.02, was used for statistical analysis [22].
Relative frequency was calculated to analyze the most frequent ant species indoors and outdoors in each neighborhood, in each period of collection.
Shannon diversity index, equitability index, and Simpson's Index were calculated with the BioDap software [23] in order to record ant community in the neighborhoods.In Itaquera, 33 ant species were found, most native to Brazil, but 4 were exotic ones.In Vila Mariana, we collected 23 ant species in 1998/1999, 3 were exotic, and from 2009-2011 we collected 25 ant species and 4 were exotic.Tapinoma melanocephalum was the most abundant species in the three collections (VM 1998/1999, VM 2009-2011, and ITA 2009-2011).

Ant
In Vila Mariana, ten years later, it was found an increase on T. melanocephalum frequency and a decrease on the other ant species frequency as N. fulva, Brachymyrmex sp. 1, and Solenopsis sp., which are all native to Brazil.L. humile, also native to Brazil and found in 1998/1999, was not found in the surveys from 2009 to 2011.This is a species that could have had success in dominating the neighborhood.But the exotic big-headed ant, P. megacephala, expanded its foraging range ten years later.Besides its high frequency in Vila Mariana, P. megacephala was not found in Itaquera.
Ten years ago, N. fulva was the most frequent species, present in more than 40% of the sampled houses in Vila Mariana.Its occurrences were even indoors and outdoors.In 1998/1999, P. longicornis was also found outdoors with N. fulva.Our last survey from 2009 to 2011 showed that P. longicornis' frequency increased almost 3 times (Table 2), but its occurrence was limited to the peridomiciliar area.As N. fulva, P. longicornis is also an invasive ant species and can be displacing the native N. fulva.But the association analysis of N. fulva was negative with T. melanocephalum and P. megacephala (Figure 1) in the survey from 1998 to 1999.Thus, N. fulva may not succeed in colonizing Vila Mariana all these years, once it competed with three other invasive ants.As we calculated the association only among the five most frequent species, P. longicornis did not enter in 1998/1999 analysis.
From 2009 to 2011, few associations could be established among the ant species, but P. longicornis frequency increased and showed negative association with P. megacephala (Figure 2).
T. melanocephalum was mainly collected indoors (living rooms, bedrooms, kitchens, and bathrooms).When we compared samples from T. melanocephalum in 1998/1999 and ten years later, its relative frequency increased from 18% to 53% inside houses.P. megacephala indoor frequency was the same ten years later, but it expanded its foraging areas outdoors from 2009 to 2011.Brachymyrmex sp. 1 and Solenopsis sp.frequencies decreased in the outdoors in the last samples (Figure 3).
We analyzed the same 25 houses surveyed in 1998/1999 and 2009-2011, and data do not show significant differences from the 79 surveyed houses analyzed together (Figure 4).T. melanocephalum and P. megacephala are the most frequent species, and N. fulva decreased its frequency in ten years (Figure 4).
The number of ant species per house varied from 1 to 10 (Figure 5).The number of houses with only one ant species increased in the survey from 2009 to 2011 in Vila Mariana, due to T. melanocephalum increase.
In the opposite, Itaquera shows more houses with more ant species, compared to Vila Mariana.
The diversity indexes revealed that Vila Mariana has a more diverse and balanced ant community compared to Itaquera.The temporal analysis showed a decrease in the ant diversity in Vila Mariana (Table 3).

Neighborhood Profiles.
Itaquera is a neighborhood with small-and single-story houses, recently constructed, with more children and few areas of gardens and backyards.The worst conserved and with less cleanness houses were found there.In Vila Mariana, houses are bigger, more ancient than in Itaquera, with a large number of two-story houses, and with more gardens and backyards (Table 4).

Comparative Analysis between Ant Samples and Environment Characteristics.
The Mann-Whitney and Kruskal-Wallis tests showed that more ant species were found in  single-story houses (P = 0.007), mainly in their outdoors, and in both neighborhoods, in both periods of collections.T. melanocephalum was more frequent in houses with children (P = 0.001).On the other hand, P. megacephala was more Psyche frequent in houses without children (P = 0.007).Yet, according to the Mann-Whitney test, the highest frequencies of ant specimens outdoors (P < 0.001), as the number of ant species (P = 0.002), are higher in external areas in houses with gardens and backyards.P. megacephala is mainly found in houses with gardens and backyards (P < 0.001).
Houses without pets showed the highest species richness even indoors (P = 0.042), and outdoors (P = 0.016), the worst conserved houses showed species richness (P = 0.042) and pour cleanness houses showed higher ant abundance (P = 0.047).

Discussion
Tapinoma melanocephalum was the most common species in Vila Mariana in both studied periods and in Itaquera.
Other surveys in urban areas also pointed to this same species [4,14,24] as one of the most common ants in urban environments.T. melanocephalum is distributed throughout the world [10] and recruits a huge number of workers when a food resource is found [13], explaining its high abundance in our samples.As in São Paulo, in a survey conducted in New York in 2010, an exotic species was the most collected [19].
After ten years, it expanded its range in Vila Mariana as did P. megacephala and P. longicornis, contributing to the displacement of N. fulva, Brachymyrmex sp. 1, and Solenopsis sp.populations.P. megacephala is an aggressive and dominant ant species known to displace native ant fauna where it is introduced [7-9].P. longicornis was pointed by Forys and Allen [18] as the most common ant species in the region of Lower Florida Keys, and it was positively associated with the increase of urban development.
In a survey conducted in 1994 in a banana crop in southeastern Brazil, Fowler et al. [25] reported negative association between T. melanocephalum and P. longicornis.From 80 surveyed banana farms, 31 showed P. longicornis and 18 had T. melanocephalum presence, and in none of them both species were collected together.In our study, P. longicornis association was not analyzed with other ant species in Vila Mariana in 1998/1999 once its frequency and abundance were too low.But frequency data compared in both collected periods in Vila Mariana showed that this species is expanding its range.When association analysis was calculated to the most recent ant collections in Vila Mariana, P. longicornis only showed negative association with P. megacephala.This is explained by the fact that P. longicornis was not collected inside households, only outdoors where we also found P. megacephala.As T. melanocephalum occurrence is more frequent indoors than outdoors, they did not establish any ecological association.
In the central-eastern Brazil, surveys showed P. megacephala in ancient neighborhoods, with intense commercial activities and without expressive green areas [4], environments similar to Vila Mariana.Pacheco and Vasconcelos [5] also found this species in commercial areas and poorly preserved.In fact, commercial areas were also reported in a survey conducted by Menke et al. [26] with several invading ant species, as we have found in Vila Mariana.Therefore, it cannot be affirmed that this species prefers such urban environments.The establishment and dominance of invading species are related to several biotic and abiotic factors that must be better studied.
For T. melanocephalum to establish its nests, cleanness and resident habitats seem not to be relevant, but they can be decisive to their high frequency and abundance.In Itaquera, around 43% of the total sampled houses showed regular or very bad cleanness, and almost 25% were poorly preserved.Such conditions favor food offering and shelter [27] in order to promote suitable conditions for ant colonies development.Therefore, the best condition observed in Itaquera to be a good environment for T. melanocephalum was the absence of other ant species, as dominant as the ghost ant is, like P. megacephala.
Children presence may favor T. melanocephalum establishment, once data were similar in Vila Mariana after one decade and when the neighborhoods were separately analyzed.Children may offer more food on several rooms of the house favoring this opportunistic ant species.
A decrease on the ant diversity in Vila Mariana after 10 years was found when we calculated the Shannon-Wiever diversity index, but even though it is higher than in Itaquera.The low ant diversity in Itaquera must be attributed to T. melanocephalum abundance.Simpson's index corroborates this statement, once its value was higher in Itaquera, pointing to the dominance of one ant species.
An interesting data showed by the comparative analysis using the Mann-Whitney test between the collections from 2009 to 2011 in the two different neighborhoods is that T. melanocephalum was found in the peridomiciliar area in Itaquera and Vila Mariana equally.Peridomiciliar in these neighborhoods was different, with more gardens and backyards in Vila Mariana.But T. melanocephalum occurrence outdoors was lower than indoors, in both neighborhoods, probably due to the competition with other ant species and to abiotic factors such as temperature and moisture.In a survey conducted in 2009 by Wetterer [10], he reports that the ghost ant is restricted to indoors once the species can live wherever man is.He also emphasizes that it is found in tropical and subtropical zones in latitudes higher than 30 • .MacGown and Hill [28] confirm this statement and comment that, besides, it is found in the United States and in some states in Canada; the occurrences are in greenhouses and other warm places, strengthening that temperature influences its distribution, but it can be solved when occupies human dwellings with controlled temperature.In our work, T. melanocephalum was found mainly in the indoors even though there were differences among houses and human habits in both neighborhoods.Menke et al. [26] also highlight the role of cities as a refuge for generalist species that show more tolerance to dryer and hotter environments.
Differences in the environment outdoors are not significant for T. melanocephalum once this is not the most favorable environment for its establishment.
Itaquera has a recent urbanization process; new households, streets, and avenues have been continually built as other landscape transformation caused by such processes, what may favor opportunistic ant species with flexible adjustments to such disturbances, like T. melanocephalum.

Psyche
Many authors discuss this fact like Tschinkel [16] who suggested that S. invicta whose native environment is seasonally flooded and is a species that easily adapts to highly disturbed environments, even in different continents.When native ant populations were compared with invasive/exotic ant species in recent and ancient urban areas, the exotic species succeed in establishing in recent urbanized areas [29], while the native species remained in the ancient urban areas.Native and exotic species distribution in a city is still a not very well-explained question although the preference from exotic species for disturbed environments is known.Studies suggest that there is a series of relationships between biotic and abiotic factors that makes this question so complex [18,19].
This explains the species richness found in Itaquera compared to Vila Mariana and the high dominance by T. melanocephalum in the former neighborhood, a still unknown ant species but intimately linked to the São Paulo citizens.
Yet, the number of houses with several ant species in Itaquera is related to the green area that surrounds the neighborhood and many species that are not common to urban areas like Acromyrmex sp., C. crassus, C. rufipes, Cyphomyrmex sp., Linepithema spp., Octostruma sp., and 9 Pheidole species.The bad conservation of houses in Itaquera also may contribute to the establishment of ants inside and in the peridomiciliar areas of the houses.
Community.It was collected 7.249 ant specimens in the 132 collections in Vila Mariana in 1998/1999, 6.477 specimens in the 79 collections in Vila Mariana in 2009-2011, and 12.054 specimens in the 79 collections in Itaquera in 2009-2011 (Table 2).

Table 1 :
Data from surveyed households in Vila Mariana and Itaquera neighborhoods in the city of São Paulo, Brazil.

Table 3 :
Species richness, Shannon-Wiener diversity index (H ), equitability (E), and Simpson's diversity index (D) in Vila Mariana and Itaquera, in the city of São Paulo.

Table 4 :
Characteristics of houses sampled in Vila Mariana and Itaquera, in the city of São Paulo, Brazil.