Is the “ Centro de Endemismo Pernambuco ” a biodiversity hotspot for orchid bees ?

The orchid-bee faunas (Hymenoptera: Apidae: Euglossina) of the three largest forest remnants in the “Centro de Endemismo Pernambuco”, northeastern Brazil, namely Estação Ecológica de Murici (ESEC Murici), RPPN Frei Caneca, and a forest preserve belonging to Usina Serra Grande, in the states of Alagoas and Pernambuco, were surveyed using seventeen different scents as baits to attract orchid-bee males. Eight sites were established in the three preserves, where samplings were carried out using two protocols: insect netting and bait trapping. We collected 3,479 orchid-bee males belonging to 29 species during 160 hours in early October, 2012. Seven species were collected in the “Centro de Endemismo Pernambuco” for the first time. Richness proved to be one of the highest of the entire Atlantic Forest domain, and diversity in some sites, especially at ESEC Murici, revealed to be one of the highest in the Neotropics. Eulaema felipei Nemésio, 2010, a species previously recorded only at ESEC Murici, was found in no other preserve in the region and its conservation status is discussed.

Despite this recent "boom" of studies in the Atlantic Forest domain, there still are many areas that could be considered "data deficient" concerning our knowledge on their orchid-bee fauna.Areas in northeastern Brazil, particularly at the northern margin of the São Francisco river, are a good example.Only a few areas have been effectively sampled in Paraíba (Bezerra and Martins, 2001;Farias et al., 2007;Farias et al., 2008), Pernambuco (Milet-Pinheiro and Schlindwein, 2005;Darrault et al., 2006), and Alagoas (Darrault et al., 2006;Moura and Schlindwein, 2009;Nemésio 2010b).Almost nothing is known from the states of Ceará, Sergipe and Rio Grande do Norte, although a new orchid bee has recently been described from Ceará (Nemésio and Ferrari, 2012).
The Atlantic Forest situated to the north of the São Francisco river is often called the "Centro de Endemismo Pernambuco" (hereafter CEPE).Compared to other areas in the Atlantic Forest domain, CEPE is the most devastated, the least known and the least protected (e.g.Hayer, 1988;Coimbra-Filho and Câmara, 1996;da Silva and Tabarelli, 2001).Originally the forested area ranged from the state of Alagoas to the state of Rio Grande do Norte covering a total area of almost 60,000 km 2 (Brown, 1982;Prance, 1982).Only ca.2,000 km 2 (less than 5% of its original cover) remains, scattered in small fragments (Cardoso da Silva and Tabarelli, 2000;Ribeiro et al., 2009).Deforestation of CEPE began early in Brazilian history (Coimbra-Filho and Câmara, 1996), but reached catastrophic levels during the 1970's, when forests in Alagoas and Pernambuco gave way to sugar cane plantations, largely due to fiscal incentives provided by the Brazilian government under a development plan known as "Pró-Álcool"; this program consisted of sugar cane cultivation to produce ethanol fuel for motor vehicles (Goerck and Wege, 2005).Nevertheless, many plant and animal species considered endemic in this region (e.g.Teixeira andGonzaga, 1983a, b, 1985;Prance, 1987;Teixeira, 1987, Pennington, 1990;Siqueira Filho, 1998;Olmos, 2005;Nemésio, 2010a) have been resilient enough to survive in the remaining forest patches, although some species have most probably vanished from these areas (Asfora and Pontes, 2009).Besides the endemic species, CEPE holds ca.50% of all bird species of the entire Atlantic Forest (Roda and Pereira, 2006) and at least 8% of all tree species (Uchoa Neto and Tabarelli, 2002).
Currently, the largest forest preserves in CEPE are located in the states of Alagoas and Pernambuco, namely Estação Ecológica de Murici, in the municipality of Murici, and Mata do Coimbra, in the municipality of São José da Laje, both in the state of Alagoas; and Reserva Particular do Patrimônio Natural (RPPN) Frei Caneca, in the municipality of Jaqueira, southern Pernambuco (Figure 1).The orchid-bee fauna of the former area was sampled in 2009 (see Nemésio, 2010b) and the two latter areas were sampled in 2003 by Darrault et al. (2006).Nonetheless, sampling protocols used in both studies were different and Darrault et al. (2006) did not present data on abundance of each species.More importantly, many species were not identified at species level by Darrault et al. (2006).Nemésio (2010b) recognised 25 species for the region, but the absence of species-level identification for some taxa in Darrault et al. (2006) suggests that richness in the region may be higher.
The goals of the present study were, thus, to investigate the actual richness and diversity of orchid bees in CEPE using the same sampling protocol simultaneously in the three largest forest remnants in the region.If we assume that most orchid-bee species are forest dependent (Dressler, 1982a;Roubik and Hanson, 2004;Nemésio, 2009), then the best chances of recording the highest richness and diversity of these insects would be in the largest forest patches, theoretically the best preserved ones.Besides, we also used simultaneously different protocols (handnet collecting versus bait trapping) in order to investigate whether differences in sampling protocols would influence the results, a current topic in orchid-bee studies (Nemésio andMorato, 2004, 2006;Mattozo et al., 2011;Nemésio, 2012b).

Study sites
This study was conducted in three forest remnants in CEPE, two in the state of Alagoas and one in the state of Pernambuco (Figure 1).In the state of Pernambuco, samplings were performed at RPPN Frei Caneca (RFC), a private protected forest remnant with a total area of about 1,000 ha, split into several forest fragments.Three sampling sites were selected at RFC: the largest (RFC-1; 500 ha) fragment, locally known as "Serra do Quengo"; a 50-ha fragment, known as "Mata do Ageró" (RFC-2); and another 50-ha fragment, known as "Mata do Espelho" (RFC-3).At the two former sites orchid-bee males were actively collected with insect nets, whereas in the latter site bees were bait trapped (see below).All three fragments are connected by corridors, although the small ones present more disturbed vegetation.In the state of Alagoas, two areas were sampled: Mata do Coimbra, a private area belonging to Usina Serra Grande (USG), with a continuous block of forest of about 3,500 ha, and (ii) Estação Ecológica de Murici, a protected federal area of about 6,000 ha split into two larger fragments (about 2,500 ha each) and several smaller ones (see Nemésio, 2010a, b).Four sites were sampled at USG, one at the forest edge and three in the interior of the forest, one of the latter ones with bait traps, all the other ones with insect nets.Only one sampling site was selected at ESEC Murici, and sampled with insect nets since three sites had been sampled in this area three years before under the same protocol employed here.All these areas are immersed in a matrix of sugar cane plantations and pastures.Samplings were carried out from the 2 nd to the 11 th of October, 2012.

Data analysis
Diversity was estimated through Shannon-Wiener diversity index (H'), as H' = -Σ p i ln (p i ), where p i is the proportion of total number of species made up of the ith species (Pielou, 1975).Evenness (E) was estimated through the formula E = H'/ ln (S), where S is the species richness.The similarity in faunistic composition among sites was estimated by the Renkonen similarity coefficient, recommended by Wolda (1981) for small samples.Similarity indexes that take into account not only presence of species, but also their relative abundances, are strongly recommended by Balmer (2002) since theoretically they better reflect natural processes.For the similarity analysis we only included the six sites where bees were collected with insect nets, since the sampling protocol was identical and, thus, comparable.Based on those similarities, the areas were grouped using UPGMA (Sneath and Sokal, 1973).
The ordination of the sites according to the similarity of their faunas showed a medium to high overall (55% to 80%) similarity among the three forest areas (see Figure 2).The sites situated at the interior of the forest at Usina Serra Grande showed the highest similarity (80%) and grouped with the site at ESEC Murici (69.9%), also in the interior of the forest.The site situated at the edge of the forest at Usina Serra Grande grouped first with the small site at RPPN Frei Caneca (69.4%) and, then, to the first group (67.8%).The site situated at the interior of the forest in the largest fragment at RPPN Frei Caneca was the most dissimilar of all six sites, sharing only 55.2% of  2).This latter site was the only one where specimens belonging to Eulaema Lepeletier, 1841 together represented more than 50% of the orchid-bee community (Table 1).Samplings carried out with bait traps captured substantially less bees than samplings performed with insect nets (Table 2), even considering that four traps were used for each scent.Besides the lowest abundance, richness, diversity and evenness were also among the lowest when compared to sites sampled with insect nets (Table 2).Larger bees belonging to Eulaema were more commonly captured with bait traps than insect nets.At Usina Serra Grande, 18.5% of the trapped bees belonged to the genus Eulaema, whereas 14.3% of the bees captured with insect nets were Eulaema spp. at the other three sites.At RPPN Frei Caneca, 61.7% of the trapped bees were Eulaema spp., whereas 36.4% of the bees collected with insect nets were Eulaema at the other two sites (Table 2).Sixteen of the 17 scents used in the present study were attractive to orchid-bee males (Tables 3 and 4).No specimen was collected on linalool.More than 94% of the specimens and all species were collected on seven scents: skatole (25%), cineole (21%), β-ionone (18%), vanillin (12%), eugenol (7%), methyl-salicylate (6.5%) and benzyl acetate (5.3%) (Table 3).Cineole attracted the highest number of species (19).None of the ten lesser scents attracted exclusive species (Table 4).

Sampling protocol
The lack of standardisation in orchid-bee inventories has been pointed out as a main concern for comparative studies for over a decade (Morato, 1998;Nemésio and Silveira, 2007b;Nemésio, 2012b).Recently, Nemésio (2010b) proposed a 20-hour sampling methodology that consists in collecting all bees in a given site over 20 hours Table 3. Diversity, evenness, species richness and number of specimens of each orchid-bee species collected with insect nets attracted to the seven most efficient scents in six sampling sites in "Centro de Endemismo Pernambuco", northeastern Brazil, after 20 hours of sampling in each site.See text for location of each site.(in consecutive days or not) in any period from 07:00h to 17:00h, using as many scents as possible, during the season when orchid bees are most active.Roubik (2004a) had already suggested that surveys of orchid-bee males with scent baits during a single day have great utility, and may reveal almost as much about local community structure as studies lasting a full year.The protocol proposed by Nemésio (2010b) has already been used in more than twenty areas in northeastern Brazil (Nemésio, 2010b; this study), southern Bahia (Nemésio, 2011a(Nemésio, , 2013a, c, d), c, d), northern Espírito Santo (Nemésio, 2011b(Nemésio, , 2013b)), Minas Gerais (Nemésio, 2012c;Nemésio and Paula, 2013), and the Peruvian Amazon (Nemésio et al., 2014), always with a high number of collected specimens during a relatively short period of time.With more than 3,000 bees collected in only ten days, the present study revealed one of the highest abundances ever recorded with this 20-hour sampling strategy, only rivalled by the regions of Monte Pascoal, in southern Bahia (Nemésio, 2013d), and Linhares, in northern Espírito Santo (Nemésio, 2013b).On the other hand, it must be emphasised that this protocol potentially fails to record the actual richness of species   belonging to Eufriesea Cockerell, 1908, since most species in this genus are univoltine, highly seasonal (Kimsey, 1982), and it becomes a matter of chance reaching the sampling areas when one or more species of this genus are active.It is particularly true in northeastern Brazil, since the rainy season, when most bees are active (see Bezerra and Martins, 2001), occurs during winter, with no active species of Eufriesea.

Insect nets versus bait traps
Although some authors (Aguiar andGaglianone, 2008, 2011) insist that bait traps may be an efficient sampling method, there is now a growing body of evidence to contradict it in different parts of the Neotropics (see Nemésio andMorato, 2004, 2006;Mattozo et al., 2011;this study).When both methodologies are used simultaneously, the commonly used bait traps are always less efficient than active collecting with insect nets (Nemésio andMorato, 2004, 2006;Mattozo et al., 2011;this study).Specimens belonging to Eulaema are more commonly found trapped than specimens belonging to Euglossa Latreille, 1802, suggesting that small bees belonging to Euglossa escape more easily than the larger Eulaema spp.Results obtained with the sole use of bait traps should, thus, be considered with great care.If, as pointed out by Nemésio (2012b), our main objective is to understand the actual relative abundance of each species in a given orchid-bee community, then the most efficient sampling methodology should always be preferred.The sole use of bait traps should be avoided whenever possible, since it is not only the lower number of captured bees that counts here (which, theoretically, could be counterbalanced through a larger number of traps), but the serious bias introduced by this methodology, given that it favours trapping the largest bees.

Scent preferences
Although sixteen scents were attractive to orchidbee males, our data strongly suggest that almost the same results could be reached with the use of only the seven most efficient scents, since more than 94% of the specimens and all 29 species were attracted to them (see Tables 3 and 4).The most efficient scents may vary from region to region, depending on the species composition of each community, since some species present particular scent preferences (e.g.Ackerman, 1989).For example, β-ionone was an important scent in the presente study, but only three specimens were attracted to it during 120 hours of active sampling in Peruvian Amazon only two months before this study (Nemésio et al., 2014).Cineole, methyl salicylate and vanillin, among the most powerful attractants in the present study, are widely recognised as strong attractants.On the other hand, skatole has been neglected by most researchers, who argue that it stinks.Although it is true, it is a powerful attractant.In the present study it attracted the largest number of specimens and it was an important scent for collecting Euglossa marianae, a species only known to visit skatole and cineole baits.In a recent survey in Peru, skatole attracted the highest number of species (Nemésio et al., 2014).Other scents that were not particularly strong attractants in the present study, such as p-tolyl acetate and trans-methyl cinnamate, have shown to be important baits in southern Bahia (A.Nemésio, unpubublished data) and in the Amazon (Nemésio et al., 2014).Based on the results here presented and those presented by Nemésio et al. (2014) for the Amazon, we encourage researchers to use a wider array (more than the usual five or six) of scents in future orchid-bee studies to better sample those odd little-responsive species that usually escape our attention, especially in highly speciesrich areas (see also Nemésio, 2012b).

Faunistics, richness and diversity
Twenty-nine species were recorded in the present study, the highest number ever recorded for this part of the Atlantic Forest.Nemésio (2010b), retrieving data from all published inventories in the "Centro de Endemismo Pernambuco", assumed that 25 orchid-bee species were known to occur in the region.That study, however, recorded three species of Eufriesea not collected in the present study since no species of Eufriesea was apparently active during field samplings.On the other hand, seven species of Euglossa are here recorded for the first time in the "Centro de Endemismo Pernambuco" and are dealt with below.

Euglossa adiastola Hinojosa-Díaz, Nemésio and Engel, 2012
This species was considered in previous studies in the Atlantic Forest as Euglosa augaspis Dressler, 1982c.It is a common species in "Hileia Baiana" (the Atlantic Forest of southern Bahia and northern Espírito Santo), occurring in most forest remnants (Nemésio, 2011a(Nemésio, , b, 2013a-d)-d), and reaching the northeastern portion of the state of Minas Gerais (Nemésio, 2012c).The records of this species in ESEC Murici and Usina Serra Grande extend its known geographic distribution over 650 km northwards.

Euglossa aratingae Nemésio, 2009
This species is apparently widespread in southeastern Brazil (Nemésio, 2009) and its previous northernmost record in the Atlantic Forest was in southern Bahia.The records presented here extend its geographic distribution over 700 km northwards.There is also a male recently collected in the state of Paraíba currently deposited in the Entomological Collection of the Universidade Federal de Minas Gerais (UFMG), which would represent the northernmost record of this species in the Atlantic Forest.

Euglossa bembei Nemésio, 2011d
This species had already been recorded for the state of Pernambuco (at RPPN Frei Caneca) by Darrault et al. (2006) (as Eg. ioprosopa Dressler, 1982b).It is here recorded for Alagoas for the first time, at ESEC Murici.Its discontinuous distribution (it was not collected in the intermediately located Usina Serra Grande) may be partially explained since this species apparently prefers higher elevations (see Nemésio, 2011dNemésio, , 2012c;;Nemésio and Vasconcelos, 2013) and sites at Usina Serra Grande were those situated at the lowest elevations.

Euglossa carinilabris Dressler, 1982c
This is another species considered to be found only in "Hileia Baiana" by Nemésio (2009).Nevertheless, it seems to be rarely attracted to scent baits.During a recent inventory throughout the "Hileia Baiana", over 15,000 orchid-bee males were collected (see locations in Nemésio et al., 2012) and only one specimen of Eg. carinilabris was collected.The record of this species at RPPN Frei Caneca extended its known geographic distribution over 700 km northwards.

Euglossa hemichlora Cockerell, 1917
This species is not restricted to the Atlantic Forest, being also present in the Amazon Basin (in most studies it is treated as Eg.gaianii Dressler, 1982d, a junior synonym).
In the Atlantic Forest it had been recorded in Rio de Janeiro (Tonhasca Junior et al., 2002), Minas Gerais (Nemésio and Silveira, 2006), Espírito Santo (Nemésio, 2011b) and Bahia (Nemésio, 2013d).The present record represents a range extension of over 700 km northwards, but it must be pointed out that specimens belonging to this species have been also collected in the states of Ceará and Piauí, and currently deposited at the Entomological Collection of Universidade Federal de Uberlândia.

Euglossa monnei Nemésio, 2012a
This species was recently described from "Hileia Baiana", where it was recorded in low abundance (only 14 males collected among over 15,000 orchid bees) from Linhares, in Espírito Santo, to Igrapiúna, in Bahia (see Nemésio, 2012a).The present record extends its known geographic distribution over 650 km northwards from Igrapiúna.

Euglossa pepei Nemésio and Engel, 2012
This small Euglossa was described from Parque Nacional do Pau Brasil (Porto Seguro, Bahia) based on only four specimens.It has been subsequently found in Linhares (Nemésio, 2013b) and, now, in ESEC Murici, greatly extending its geographic distribution.

Euglossa viridis (Perty, 1833)
This species seems to present a wide distributional range in South America, occurring in the Amazon Basin and the Atlantic Forest.In the latter biome, it was recorded from the state of Paraná, in the south (Giangarelli and Sofia, 2011) to Bahia, in the north (Nemésio, 2011a).Since this species is weakly attracted to common scents in orchid-bee inventories, its rareness seems to be an artifact.The record of this species at RPPN Frei Caneca is the northernmost record for this species in the Atlantic Forest.
At least four species should be added to the 29 species collected in the present study: Eufriesea atlantica Nemésio, 2008, Ef. mussitans Fabricius, 1787, Ef. nordestina Moure, 1999and Ef. pyrrhopyga Faria and Melo, 2011 were already recorded for the "Centro de Endemismo Pernambuco".Thus, orchid-bee richness in the region reaches at least 33 species, more than 50% of all Atlantic Forest species.Some species weakly attracted to synthetic scents, as Exaerete dentata (Linnaeus, 1758), are expected to occur in the region.This latter species, for example, occurs, in eastern Brazil, from Piauí (Perty, 1833, as Chrysantheda nitida), in the north, to São Paulo, in the south.According to GAR Melo (personal communication), a specimen of Ex. salsai Nemésio, 2011c from Pernambuco is also deposited in the Entomological Collection of the Universidade Federal do Paraná.Given the number of species formerly believed to be endemic in "Hileia Baiana" now found to also occur in the "Centro de Endemismo Pernambuco", it would be no surprise if Eulaema seabrai Moure, 1960 and Euglossa cyanochlora Moure, 1996, two species weakly attracted to synthetic scents (the latter one considered to be endemic in "Hileia Baiana"), were collected in the region in the future.
The above considerations show that the "Centro de Endemismo Pernambuco" holds one of the richest orchidbee faunas of the entire Atlantic Forest, only surpassed by that recorded in "Hileia Baiana", with more than 40 species ever recorded (reviewed by Nemésio, 2013d).Diversity is also extremely high in the region, with the diversity recorded for ESEC Murici (H' = 2.55) being one of the highest ever recorded for the entire Neotropical region (see Storck-Tonon et al., 2009, p. 700-701 for a review).On the other hand, the number of endemic species in the "Centro de Endemismo Pernambuco" is particularly low: only three species (Eufriesea pyrrhopyga, Euglossa perpulchra, and Eulaema felipei) were exclusively collected in this region to date.The new records presented in this study, however, show that the number of endemic species formerly believed to occur in "Hileia Baiana" was overestimated, since many endemics to the latter region were now found to occur in the "Centro de Endemismo Pernambuco", with the sole exception of Euglossa cyanochlora.This finding strongly suggests that both subunits of the Atlantic Forest, well characterised by vegetation and other animal groups as strongly distinct, are apparently a continuous and indistinct biome for orchid bees, the rarest species of Euglossa and Eulaema in "Hileia Baiana" are also the rarest species in the "Centro de Endemismo Pernambuco".The absence of some species currently found in "Hileia Baiana" from the "Centro de Endemismo Pernambuco", such as Euglossa cognata Moure, 1970, may be just a consequence of the more dramatic devastation that took place in the Atlantic Forest of Alagoas, Pernambuco and Paraíba.

Conservation
Only a few forest fragments remain in the "Centro de Endemismo Pernambuco" and none exceeds 10,000 ha (Ribeiro et al., 2009).The few forest areas left are, thus, extremely threatened, although still holding a huge diversity of life forms (Cardoso da Silva and Tabarelli, 2000;da Silva and Tabarelli, 2001;Uchoa Neto and Tabarelli, 2002;Goerck and Wege, 2005;Olmos, 2005;Asfora and Pontes, 2006;Nemésio, 2010b).The situation of these few areas is dramatic because strong anthropogenic pressures still exist in the region (reviewed by Nemésio, 2010a, b) and predicted climatic changes can heavily impact the forest remnants (Williams et al., 2007), changing their suitability for many species.Concerning orchid bees, two species deserve particular attention: Euglossa marianae and Eulaema felipei.
Euglossa marianae is perhaps the most sensitive forest-dependent of all orchid-bee species in the Atlantic Forest.Nemésio (2011b) realized that this species was only recorded in 13 forest remnants along the Atlantic Forest, the smallest of them with ca.3,000 ha (ESEC Murici, see Nemésio, 2010b).The record of this species in Alagoas by Nemésio (2010b) was mysteriously ignored by Faria and Melo (2012), although these latter authors followed Nemésio's (2010b) data on the close ally Euglossa calycina for the region.Nine specimens were collected previously in ESEC Murici (Nemésio, 2010b) and in the present study we demonstrate that this species also occur at Usina Serra Grande, where it is one of the most abundant species (Table 1), especially at sites in the interior of the forest (as previously suggested by Tonhasca Junior et al., 2002 andNemésio andSilveira, 2006, as Eg. analis).The presence of this species at Usina Serra Grande is outstanding, because it apparently remained unnoticed in a previous sampling in that area (Darrault et al., 2006).This species was not found at RPPN Frei Caneca, a smaller forest patch, supporting Nemésio's (2011b) point of view that ESEC Murici, and now Usina Serra Grande, present the smallest area possible to maintain viable populations of this sensitive species.Since many small fragments persist between both preserves, which are separated by ca.40 km, there is a remarkable opportunity to test if they function as corridors for this species or if both populations are isolated.Anyway, any further deforestation in both areas may add a potential risk of extirpating this species from the only two areas where this species is known to occur in the "Centro de Endemismo Pernambuco".
Eulaema felipei presents a still more dramatic situation if compared to Euglossa marianae.Although apparently as sensitive as the latter species, El. felipei is only known from ESEC Murici.This species would be expected to occur at both Usina Serra Grande and RPPN Frei Caneca, close to ESEC Murici.Nevertheless, our field work carried out simultaneously in all three areas revealed that it is absent from the latter two.We can only speculate about the reasons why it is restricted to ESEC Murici.Pristine vegetation can be found more abundantly in ESEC Murici than in the other two areas.Moreover, ESEC Murici is closer to the coast, presenting the highest humidity levels.Whatever the reasons, the fact is that El. felipei was never recorded outside ESEC Murici in any previous field study in the "Centro de Endemismo Pernambuco" (Bezerra and Martins, 2001;Martins and Souza, 2005;Milet-Pinheiro and Schlindwein, 2005;Darrault et al., 2006;Farias et al., 2007Farias et al., , 2008;;Moura and Schlindwein, 2009;this study).If it is true, we hypothesise that a high level of endogamy may be present in this species, one additional factor, besides (and most probably a consequence of) its restricted geographic distribution, to threaten this species.Further samplings in nearby forest remnants are needed to search for other populations of this species and urgent measures are needed to protect other forest patches in the area.Except for ESEC Murici, almost all other forest remnants in the region are situated in private areas and most of them are completely unprotected.

Figure 1 .
Figure 1.Map of northeastern Brazil showing the location of RPPN Frei Caneca, Usina Serra Grande and ESEC Murici.See Material and Methods for exact coordinates.

Figure 2 .
Figure 2. Clustering of the six sites where sampling was conducted with insect nets and 17 scent baits according to their similarity.MUR: ESEC Murici; RFC: RPPN Frei Caneca; USG: Usina Serra Grande.See Material and Methods for exact coordinates of each sampled site.

Table 1 .
Diversity, evenness, species richness and number of specimens of each orchid-bee species collected with insect nets with 17 different scents (see Material and Methods) in six sampling sites in "Centro de Endemismo Pernambuco", northeastern Brazil, after 20 hours of sampling in each site.See text for location of each site.

Table 2 .
Diversity, evenness, species richness and number of specimens of each orchid-bee species collected with insect nets and bait traps with six different scents (see Material and Methods) in eight sampling sites in "Centro de Endemismo Pernambuco", northeastern Brazil, after 20 hours of sampling in each site.See text for location of each site.

Table 4 .
Diversity, evenness, species richness and number of specimens of each orchid-bee species collected with insect nets attracted to the nine less efficient scents in six sampling sites in "Centro de Endemismo Pernambuco", northeastern Brazil, after 20 hours of sampling in each site.See text for location of each site.