Euglossine bees ( Hymenoptera , Apidae ) in a remnant of Atlantic Forest in Paraná State , Brazil

Species composition, relative abundance, seasonal changes in the species abundance and scent association of male Euglossini collected in a semi-deciduous forest fragment in the north of the State of Paraná, southern Brazil, were recorded. Euglossine males were collected twice a month, for twelve months, from 8:00 am to 3:00 pm. The scents eucalyptol, eugenol, vanillin, methyl salicylate and benzyl acetate were used as baits. A total of 434 males distributed among 3 genera and 9 species were attracted to the chemical baits. Eufriesea violacea (Blanchard, 1840) (49.8%), Eulaema nigrita Lepeletier, 1841 (23.0%) and Euglossa pleosticta Dressler, 1982 (13.8%) were dominant in number of individuals. Among the non-dominant species, Euglossa fimbriata Rebêlo & Moure, 1995 was more common (9.0%), followed by E. cordata (L., 1758) (1.8%), E. truncata Rebêlo & Moure, 1995 (1.4%), E. melanotricha Moure, 1967 (0.7%), E. townsendi Cockerell, 1904 (0.23%) and Eufriesea auriceps Friese, 1899 (0.23%). In general, bees were more abundant in warm-wet season (September-March). Eufriesea violacea was the most seasonal species, showing activity through the warm-wet season, from October to February. Eucalyptol was the most attractive fragrance, which was responsible for 92.6% of all visits by euglossine bees.

Purified components of floral fragrances and some synthetic chemicals can be used to attract male euglossines and thereby census them (DODSON et al., 1969;WILLIAMS & DODSON, 1972;DRESSLER, 1982).The number of males found at baits seems to reflect patterns of emergence of these bees and is correlated with frequency of visits to natural sources, so censuses may indicate actual bee abundance (ACKERMAN, 1983b).The use of chemical baits as attractants has also made possible to estimate geographical distribution of many species (DODSON et al., 1969;REBÊLO & SILVA, 1999).Chemical baiting techniques have been widely used to study different aspects of euglossine communities in Costa Rica (JANZEN et al., 1982;ROUBIK, 2001), Panama (RICKLEFS et al., 1969) and Peru (PEARSON & DRESSLER, 1985).In Brazil, euglossine bees studies have been developed primarily in Amazonian Forest (BECKER et al., 1991;OLIVEIRA, 1999;SILVA & REBÊLO, 2002) and in Atlantic Rainforest remnants (PERUQUETTI et al., 1999;BEZERRA & MARTINS, 2001;SANTOS & SOFIA, 2002).
The Atlantic Rainforest is one the most endangered forests, which has been severely destroyed after centuries of deforestation.Nowadays only about 5 to 12% of the original forest remains as isolated forest remnants scattered throughout a landscape dominated by agricultural uses (BROWN & BROWN, 1992;CHIARELLO, 1999).Recently the Parque Estadual Mata dos Godoy, an Atlantic Forest remnant, in northern region of Paraná State, was included by the Brazilian government as a priority area for surveys of invertebrate fauna, considering its extreme importance to conservation and management of this biome (BRASIL, 2000).The purpose of this work was to study the community of euglossine fauna of this State Park in southern Brazil.

MATERIAL AND METHODS
The Parque Estadual Mata dos Godoy (MG) is a forest fragment of 680 ha, surrounded by cropped area, located in the district of Londrina city (23 o 27'S and 51 o 15'W), north of Paraná State, southern Brazil (fig.1).The vegetation of the Park is classified as semi-deciduous subtropical forest (CHAGAS-E-SILVA & SOARES-SILVA, 2000; TORREZAN, 2002), which represents an associated ecosystem of Atlantic Rainforest (FERNANDES & BEZERRA, 1990).
The local climate is classified as Cfa, humid subtropical with warm summer (IAPAR, 1994;MENDONÇA, 2000), rain is distributed in all seasons, with a decline in rainfall in the winter (June, July and August).From 1977 through 2000, mean annual rainfall, temperature and relative humidity registered for the region were: 1626.0 mm, 20.9 o C and 70.9%, respectively (data provided by IAPAR, Instituto Agronômico do Paraná).
The sampling method was based on REBÊLO & GARÓFALO (1991), with some modifications.From November 1998 through October 1999, twice a month, male euglossine bees were collected with an entomological net when they arriving at chemical baits.Baits consisted of little balls (diameter = 5 cm) of absorbent paper each saturated with one of five different chemicals (benzyl acetate, eugenol, eucalyptol, methyl salicylate and vanillin), placed at the edge of forest, about 2.0 m apart and 1.5 m above the ground.Baits were available from 8:00 am until 3:00 pm, and were constantly monitored by the collector.To make the census more effective, clear days were chosen whenever possible (JANZEN et al., 1982).All chemicals were replenished on the baits every hour to prevent losses due to their volatility.
In laboratory, the specimens were identified and voucher specimens of all species of bees are deposited at the Museu de Zoologia, Universidade Estadual de Londrina (UEL).
Air temperature and relative humidity were measured every hour during the collections using a hygrometer with dry and wet bulbs hung 1.2 m above the ground.Rainfall, temperature and relative humidity for the study area were obtained from IAPAR.
Abundance was calculated from the total number of individuals attracted to all chemical baits throughout the year.Seasonal abundance was estimated as the number of bees collected each month.The dominant species was determined as described by KATÔ et al. (1952).Faunal diversity was estimated by means of the Shannon's index (H') and species evenness by means of Pielou's index (J') (LUDWIG & REYNOLDS, 1988).Sorensen's similarity coefficient (LAROCA, 1995) was used to compare the results of this study with data reported in other works (REBÊLO & GARÓFALO, 1991, 1997;GARÓFALO et al., 1998;JESUS & GARÓFALO, 2000) concerning euglossine bees in similar semi-deciduous forests.Chi-square test (2 x 2) was employed to analyze bee preference for fragrances.Correlations between biotic (number of individuals and species of bees captured monthly) and abiotic (temperature, humidity and rainfall) data were tested for significance using Spearman's rank correlation coefficient (r s ) (LUDWIG & REYNOLDS, 1988;ZAR, 1996).Values were considered different where P < 0.05.
The highest rainfall (699.3 mm) occurred during December, January and February.June had a high rainfall (135.7mm) and August was the driest month (fig.2).The highest relative humidity occurred in January, February and June, and varied from 79.2 to 81.4%.
The highest mean temperatures occurred from November to March, and the lowest from May to July (fig.3).The bees were most abundant during the warmest and rainiest period (November to March) when 69.1% of the total males were captured (fig.2).A clear reduction in bee activity occurred during the colder months (April to August) when the mean temperature was 18.2°C.No bees were attracted to baits in May or June, the coldest months of the period (figs.2, 3).There was a significant correlation between bee abundance and mean temperature (r s = 0.668, n = 12, P < 0.05).
The number of species ranged from zero (May and June) to eight (October).Monthly bee abundance and species richness were highly correlated (r s = 0.86, n = 12, P < 0.0001).During the period from September to March (warm-wet season) and April to August (cold-dry season) it was found all nine species were active during the warmwet season, but only five in the cold-dry season (fig.3).Among the four most abundant species (tab.I), only E. violacea was restricted to the warm-wet season, while Eulaema nigrita, Euglossa pleosticta and E. fimbriata Rebêlo & Moure, 1995 were active in both seasons (fig.4).Of these species, Eufriesea violacea was both the most abundant and the most seasonal species (fig.4).
The present results suggest that some euglossine species in semi-deciduous forest ecosystems have a more constant pattern of abundance and dominance than others.Few studies in neotropical semi-deciduous forests such as MG reported the remarkable abundance of E. violacea males found in the MG community (GARÓFALO et al., 1998;JESUS & GARÓFALO, 2000), while three abundant species in MG (Eulaema nigrita, Euglossa pleosticta and E. fimbriata) were also frequent in all studies performed in semi-deciduous forest fragments in southeastern Brazil (REBÊLO & GARÓFALO, 1991, 1997;GARÓFALO et al., 1998;JESUS & GARÓFALO, 2000).The values of species diversity (H' = 1.37) and evenness (J' = 0.62) found at MG reflected the dominance of Eufriesea violacea.
Plant communities that are widely separated geographically, but show similar physiognomic characteristics, tend to have similar bee communities (MOLDENKE, 1975).All the species attracted to baits at MG have been found in different semi-deciduous forest remnants in southeastern Brazil (REBÊLO & GARÓFALO, 1991, 1997;GARÓFALO et al., 1998;JESUS & GARÓFALO, 2000), suggesting that the euglossine fauna of Brazilian semideciduous forests are very similar.The comparison between euglossine fauna of MG and others remnants in southeastern Brazil, using Sorensen's coefficient, showed values of similarity ranging from 63.2% to 94.7%, supporting Moldenke's idea.
The effectiveness of cineole, a relative scent of eucalyptol, in attracting euglossine males is well known (DRESSLER, 1982;ACKERMAN, 1989;REBÊLO & GARÓFALO, 1991;MORATO et al., 1992;SILVA & REBÊLO, 2002).In contrast, eugenol and methyl salycilate, which were very attractive to male euglossine in other sites, were poor attractants at Godoy Forest.Geographical differences in fragrance preference are common (PEARSON & DRESSLER, 1985;ACKERMAN, 1989).At two sites in Panama, ACKERMAN (1989) found that seven out of eleven euglossine species showed seasonal differences in fragrance preference.He suggested that geographic variation in fragrance choice may be due to age or genetic differences among populations, or differences in fragrance resource availability.
Climatic parameters certainly determine the occurrence of some euglossine bees species in the subtropics, while in the tropics, rainfall plays a marked role in the incidence of all euglossine bees species (SILVA & REBÊLO, 2002).In this study it was observed that euglossine bees of MG showed a similar pattern of activity to other semi-deciduous forest (REBÊLO & GARÓFALO, 1991, 1997;GARÓFALO et al., 1998;JESUS & GARÓFALO, 2000), that is, bees are more active during the rainy-wet season.
As mentioned, sampling was performed preferentially on sunny days, when weather conditions were more suited to bee activity.Hence, the measured local temperature was frequently higher than the mean monthly temperature provided by IAPAR, except in May and September (fig.2).The high mean temperature during the sampling days (27.5 o C) may explain the increased bee abundance in August (figs.2, 3).Conversely, the low number of males captured in February might be due to the unfavorable weather conditions during the sampling days or to the rainy period that immediately preceded the samplings in that month (fig.2).
Several species of Eufriesea Cockerell, 1908 exhibit a seasonal pattern of activity with a very short active period (from 2 to 4 months) during the year, usually in the rainy season (DRESSLER, 1982;JANZEN et al., 1982;ACKERMAN, 1983b;ROUBIK & ACKERMAN, 1987).Such bees spend most of the annual cycle as diapausing prepupae (DRESSLER, 1982).The high seasonal variation observed for E. violacea in MG has previously been described (WITTMANN et al., 1989;REBÊLO & GARÓFALO, 1991;PERUQUETTI & CAMPOS, 1997;REBÊLO & GARÓFALO, 1997;GARÓFALO et al., 1998;JESUS & GARÓFALO, 2000).This species has one generation a year (WITTMANN et al., 1989), and males usually emerge in September and females after October (PERUQUETTI & CAMPOS, 1997).Then, a high abundance of these bees is expected from October to December.Besides, the high frequency of males might be related to a high reproduction investment to increase mating success, given that this species shows a remarkable seasonal behavior, with a short period of activity throughout the year.
Populations of some euglossine bees species may remain active but reduced in size and locally restricted at Euglossine bees (Hymenoptera, Apidae) in a remnant of Atlantic Forest in Paraná State, Brazil certain times of the year.Such seasonal fluctuations in species abundance are probably caused by local nesting and emergence patterns of the bees (ACKERMAN, 1983a;REBÊLO & GARÓFALO, 1991).The long active period of E. nigrita during the year (fig.4) confirms the pattern of activity not strongly seasonal of these bees in semideciduous forests described elsewhere (REBÊLO & GARÓFALO, 1991, 1997;GARÓFALO et al., 1998).
Population studies of bees allow graphic understanding of trends in pollinator abundance and what kinds of abundance variability can be expected, but long-term and continuous studies about bee populations are still scant.However, they are needed for meaningful insights on pollinator population shifts in nature (ROUBIK, 2001).This is the first characterization of a euglossine bees community in a remnant of Atlantic Forest in northern Paraná State.It is now necessary long-term studies to better understand their seasonal patterns, annual fluctuations and population stability (MORATO, 1998;ROUBIK , 2001).

Table I .
Number and percentage of males of Euglossinae bees captured at each chemical bait visited by bees, in the Parque Estadual Mata dos Godoy, Londrina, from November/1998 to October/1999 (N, number of individuals). (ºC)