Open access journal: http://periodicos.uefs.br/ojs/index.php/sociobiology ISSN: 0361-6525 DOI: 10.13102/sociobiology.v67i2.4614Sociobiology 67(2): 261-267 (June, 2020) Introduction Studies on the nesting biology of wild bee species that have the potential to be managed as crop pollinators are an important step towards the development of management system for these bee species (Bosch & Kemp, 2002). Centris analis Fabricius is one of the wild bee species that have risen more interest as manageable pollinator in Brazil, given its abundance and great ability to colonize artificial nesting substrates (Jesus & Garófalo, 2000; Gazola & Garófalo, 2009; Oliveira & Schlindwein, 2009; Pina & Aguiar, 2011; Martins Abstract The ideal cavity dimensions for neotropical cavity-nesting bees with the potential to be managed as pollinators have not been getting proper attention. We investigated whether the occupancy of trap-nests by Centris analis Fabricius and Tetrapedia diversipes Klug, and other nesting aspects, are affected by trap-nest length. The used trap-nests were cardboard tubes measuring 5, 10, 15 and 20 cm in length, and Ø 8 mm. The percentage of occupied trap-nests of 10 cm by C. analis was higher than that of the 5 cm ones (χ2=11.17, gl=1, p<0.001). On the other hand, there was not difference between the occupation of 10 and 15 cm long trap-nests (χ2=0.51, gl=1, p=0.48), and between those measuring 15 and 20 cm long (χ2=1.36, gl=1, p=0.24). T. diversipes occupied a smaller number of 5 cm trap-nests than the 10 cm (χ2=1.52, gl=1, p=0.22), as well as that the 15 cm were more occupied than the 10 cm trap-nests (χ2=4.23, gl=1, p=0.04); moreover, there was not difference between the occupation of 15 and 20 cm trap-nests (χ2=0.28, gl=1, p=0.59). Both species showed higher percentages of dead immatures in nests set in the shortest trap-nests, whereas these mortality percentages were lower in the longest ones. By taking into consideration that there was not significant difference in many of the assessed parameters in comparison to values recorded for 15 and 20 cm long trap-nests, it seems likely to recommend the adoption of 10 cm long trap-nests for C. analis reproduction in agricultural sites that depend on the pollination service provided by this bee species. Sociobiology An international journal on social insects CO Santos1, PEC Peixoto2, CML Aguiar3 Article History Edited by Astrid Kleinert, USP, Brazil Received 18 July 2019 Initial acceptance 28 January 2020 Final acceptance 08 April 2020 Publication date 30 June 2020 Keywords Nesting biology, cavity-nesting bees, crop pollinator. Corresponding author Claudia Oliveira dos Santos Programa de Pós-Graduação em Ecologia e Evolução (PPGEcoEvol/UEFS) Universidade Estadual de Feira de Santana Av. Transnordestina s/nº, Novo Horizonte CEP 44036-900, Feira de Santana-BA, Brasil. E-Mail: cauoliver2@yahoo.com.br et al., 2012; Magalhães & Freitas, 2012). Knowledge about the biology of this species has been growing significantly; there are data about the local abundance and time distribution of the nesting activity of populations living in different landscapes, including crop areas (Oliveira & Schlindwein, 2009; Pina & Aguiar, 2011; Martins et al, 2012; Magalhães & Freitas 2012), as well as about nest architecture, nesting behavior (Jesus & Garófalo, 2000) and the attack of natural enemies to brood cells (Gazola & Garófalo, 2003). On the other hand, other nesting aspects, such as the ideal dimensions of nesting cavities, have been poorly investigated (Alonso et al., 2011). 1 - Pós-Graduação em Ecologia e Evolução, Universidade Estadual de Feira de Santana (UEFS), Bahia, Brazil 2 - Universidade Federal de Minas Gerais (UFMG), Belo Horizonte, Minas Gerais, Brazil 3 - Universidade Estadual de Feira de Santana (UEFS), Feira de Santana, Bahia, Brazil RESEARCH ARTICLE - BEES Cavity Length Affects the Occupation of Trap-Nests by Centris analis and Tetrapedia diversipes (Hymenoptera: Apidae) CO Santos, PEC Peixoto, CML Aguiar – Cavity length affects the occupation of trap-nests262 Many aspects of nesting biology can be influenced by the diameter or the length of the used cavities, such as the choice for the cavity (Bosch, 1994a; Bosch, 1994b; Rust, 1998), the number of brood cells produced per nest (Bosch, 1994a; Bosch, 1994b; Alonso et al., 2011), sex ratio (Stephen & Osgood, 1965; Torchio & Tepedino, 1980; Bosch, 1994b; O’Neill et al., 2010; Gruber et al. 2011; Alonso et al., 2011; Seildelmann et al., 2015) and the mortality ratio of immatures (Aguiar & Pina, 2012; Seildelmann et al., 2015). Alonso et al. (2011) have assessed the effect of subtle variations (5.5 to 7.0 cm) in trap-nest length on nesting-cavity selection by C. analis females, on the number of brood cells per nest and on sex ratio. Other studies focused on testing longer lengths are necessary in order to produce a more consolidated database capable of subsidizing the management of this crop pollinator. The relationship between sex ratio and nesting-cavity dimensions is quite important for Centris bees, since only females collect floral resources and provide pollination services to oil-flowers and pollen-flowers. C. analis females are bigger than males (Jesus & Garófalo, 2000); therefore, their production could be limited in small-diameter cavities, similar to what has been observed in other bee species that show male production bias in narrow cavities (Stephen & Osgood, 1965; Bosch, 1994; O’Neill et al., 2010; Gruber et al. 2011; Seildelmann et al., 2015). Accordingly, it is essential knowing the cavity dimensions capable of producing more females in order to be successful in managing Centris populations for crop pollination. The aim of the present study was to investigate whether the occupancy of trap-nests by two oil-bee species, C. analis and Tetrapedia diversipes Klug, and other nesting biology aspects, such as sex ratio and mortality of the offspring, are affected by different trap-nest lengths. Materials and Methods Study site Sampling was carried out at twoagricultural areas, DMQ (Maria Quitéria District) (12°16’00” S/ 38°58’00” W) and DHU (Humildes District) (12°20’08” S/ 38°51’17” W), Feira de Santana Municipality, Bahia State, Brazil. The prevailing soil use in DMQ is the cultivation of temporary crops (beans, guandu-beans, maize) and of Acerola trees (Malpighia emarginata) in small properties (family farming). The sampling procedure in DHU was carried out in Chácara Bocaiúvas (23.4 ha), which grows organic horticultural products. Sampling – A sampling point was installed in each site. Each sampling point consisted of 16 wooden blocks measuring varying thickness (from 4 cm to 19 cm); each wooden block had 60 cavities. The used trap-nests were cardboard straws, which were closed with cardboard caps in their rear tips. These straws were inserted into cavities drilled lengthwise in the wooden blocks (Camillo et al., 1995). The cardboard straws had 8-mm internal diameter, following Pina and Aguiar (2011), who reported a high nesting frequency of C. analis in trap-nests of this diameter, and 4 different lengths (5, 10, 15 and 20 cm). In all cases, the entire cardboard straw was sheltered inside the wooden block, except for 1 cm in the rear tip of it. Each site had four blocks with 240 trap-nests of each of the assessed lengths. The 16 blocks in each location were grouped on steel shelves protected by plastic tarpaulin, based on Aguiar et al. (2005). The trap-nests were inspected once a month, from October 2011 to September 2012, with an otoscope. Nests presenting concluded closing walls were removed and taken to the Entomology Laboratory of the Universidade Estadual de Feira de Santana (UEFS) to be daily observed until the emergence of adults, who were pinned, dry-mounted, separated by sex and taxonomically identified. Data analysis To evaluate if the occupation of trap-nests changed according to trap-nest length, we recorded for each trap-nest available (n=240) if it was occupied. Then, a Chi-square test was applied to assess whether the frequency of occupations (response variable) changed among trap-nestsof different lengths (explanatory variable). In case we found a significant difference among the four trap-nests length, we performed subsequent pairwise comparisons (using chi-square tests) between trap-nests of similar lengths (i.e. 5 and 10 cm, 10 and 15 cm and 15 and 20 cm). A generalized least squares model was performed to assess whether the number of provisioned brood cells changed due to trap-nest length. This model considered the number of brood cells in each trap-nest as the response variable and trap- nest length as explanatory variable. The variance of residues was adjusted according to each trap-nest length category, in order to achieve a better fit the model to the collected data. One analysis was carried out for each bee species. The same model structure described above was repeated to test whether the number of emerged adults changed due to trap- nest length, but we replaced the number of brood cells by the number of emerged adults. For both analyses considering the number of brood cells and emerged adults, only data of C. analis collected in DMQ were analyzed, given the low trap- nest occupation recorded in DHM. For T. diversipes only data collected in DHU and for traps nests with 10, 15 and 20 cm were analyzed, given the low trap-nest occupation recorded in DMQ and for nests with 5 cm. These analyses were carried out in the nlme package of the R software (R Development Core Team, 2004). A Chi-square test was applied to verify whether the number of males and females (sex ratio) of the offspring differed from an expected proportion of 1:1. To calculate the expected proportion for each sex, we divided the total number of offspring by two. To evaluate if sex ratio changed according to trap-nest length, we performed a general linear model. For this model we considered the number of males in relation to the total number of individuals in each trap-nest Sociobiology 67(2): 261-267 (June, 2020) 263 as the response variable and trap-nest length as the predictor variable. The general linear model was not carried for T. diversipes due to the low sample size for trap-nests of 5, 10 and 20 cm length. A generalized linear model with Poisson error distribution was performed to assess whether the number of brood cells with dead individuals changed due to trap-nest length. The number of dead individuals was considered as the response variable and trap-nest length was considered the explanatory variable. This analysis was only applied to C. analis in the DMQ site, given the low nesting frequency observed in the DHM site. Regarding T. diversipes, it was not possible to perform this analysis because offspring mortality was very low. Generalized linear models with Poisson error distribution were used to assess the association between the number of occupied trap-nests on every month of the year and the mean temperature, humidity and rainfall. Weather variables were obtained in the weather station of the Universidade Estadual de Feira de Santana, located in the same municipality as the sampling area. Nine models were built (Tables 4 and 5) and site and species were maintained as predictor variables in most of them (the exception occurred for the null model). We opted to keep site and species in most models because we were interested in evaluate if the number of occupied cells varied in response to climatic variables independent of site and species. The Akaike Information Criterion was used to select the most parsimonious model to explain the relationship between number of occupied cavities and the climatic variables. The significance of the selected model was calculated using a maximum likelihood ratio test. For this, the selected model was compared with a null model that retained only species and site as predictor variables. All analyses were performed in the R software (R Development Core Team, 2004). Results 1. Occupation of pre-existing cavities The number of nests established by bees in the DMQ site (Acerola orchard) was almost twice the number recorded for the site used for diversified crops (DHU) (Table 1). C. analis was the dominant species in the DMQ site, either when it comes to number of established nests (91% of all bee nests) or number of brood cells (n=460). On the other hand, this species recorded low frequency in the DHU site, whereas T. diversipes established 80% of all bee nests (202 brood cells) in it. Other cavity-nesting bee species (Centris tarsata Smith and Megachile spp) have established few nests in both sites (Table 1). The percentage of occupied trap-nests by C. analis differed between trap-nests of different lengths (χ2=17.85, df=3, p<0.001, Table 1). The proportion of occupied trap- nests recorded for the 10 cm trap-nests was higher than that of the 5 cm ones (χ2=11.17, df=1, p<0.001), but there was not difference in the proportion of occupied nests between 10 and 15 cm trap-nests (χ2=0.51, df=1, p=0.48) and between the 15 and 20 cm ones (χ2=1.36, df=1, p=0.24). The percentage of nests established by C. analis (n=84) in DMQ for nests with 5, 10, 15 and 20 cm were respectively 7%, 30%, 37% and 26%, indicating that the occupation of 5 cm trap-nests was the smallest, while the occupation was similar among trap-nests of 10, 15 and 20 cm. Bee species Trap-nest length 5cm 10cm 15cm 20cm Centris analis 6 25 31 22 Site DMQ Centris tarsata - 2 4 - Megachile (Acentron) sp. 4 - - 1 - Megachile (Sayapis) sp. 5 - - - 1 Total 6 27 36 23 % cavities occupied 2.5 11.2 15.0 9.5 Centris analis - 2 2 - Centris tarsata - 1 - 1 Site DHU Tetrapedia diversipes 1 5 15 19 Megachile (Tylomegachile) sp. 3 - - - 2 Megachile sp. 1 - - - 1 Megachile sp. 2 - - - 1 Total 1 8 17 24 % cavities occupied 0.4 3.3 7.1 10.4 Table 1. Number of nests established by solitary bees in trap-nests in two agricultural areas (DMQ and DHU), Feira de Santana, BA, Brazil. Most brood cells of C. analis in the DMQ site were found in the longest cavities, 15 cm (40%) and 20 cm (30%), whereas the smallest number of brood cells (3%) was observed in the shortest trap-nests (5cm long ones). The number of built brood cells changed according tonest length (Generalized least squares model: F(2,75)=18.54, p<0.001). The mean number of brood cells varied from 2.6 in 5 cm nests to 6.2 in 20 cm nests (Figure 1). Some cavities were not fully occupied by brood cells in trap-nests at all tested lengths. Variation in the space occupied by brood cells reached 44% of the cavity in 20 cm long trap-nests and 69% in the 5cm long trap-nests. For T. diversipes, the percentage of occupied trap-nests also differed among trap-nests of different lengths (χ2=22.12, df=3, p<0.001). There was no difference in occupation between 5 and 10 cm trap-nests (χ2=1.52, df=1, p=0.22), but the percentage of occupied trap-nest was higher for 15 cm trap-nests in comparison to the 10 cm ones (χ2=4.23, df=1, p=0.04). There was no difference in occupation between the 15 and 20 cm trap-nests (χ2=0.28, df=1, p=0.59). Percentage of nests established by T. diversipes (n=40) for nests with 5, 10, CO Santos, PEC Peixoto, CML Aguiar – Cavity length affects the occupation of trap-nests264 15 and 20 cm were respectively 2.5%, 12.5%, 37.5% and 47.5%. Therefore, the smallest occupation in T. diversipes occurred for trap-nests with 5 and 10 cm, while the greatest occupation occurred for trap-nests with 15 and 20 cm. Most brood cells of T. diversipes were built in longer cavities, 20 cm (55%) and 15 cm (33%), whereas the shortest trap-nests (5 cm) only sheltered 1.5% of the brood cells built by this bee species. The mean number of cells per nest established in cavities presenting different lengths varied from 3.0 in the only 5 cm nest that was occupied (although not included in the statistical analysis) to 5.8 in 20 cm nests (Figure 1). The occupied space by brood cells varied from 26- 27% of the trap-nest (in 20 and 15 cm long straws) to 60% of them (in 5 cm long straws). The mean number of brood cells provisioned by T. diversipes did not differ due to trap-nest length (Generalized least squares model: F(2,36)=2.36, p=0.11). For T. diversipes, the number of emerged adults did not differ due to trap-nest length (Generalized least squares model: F(2,36)=1.93, p=0.16). The incidence of attacks from natural enemies to the nests was low. Only 0.2% of C. analis brood cells in the DMQ site was lost, due to the attack by the cleptoparasitic bee, Mesocheira bicolor (Fabricius) (Hymenoptera, Apidae), whereas there was not any attack to C. analis nests in the DHU site. It was observed that 7% of T. diversipes brood cells were lost because of attack by the cleptoparasitic bee, Coelioxoides sp (Hymenoptera, Apidae). Fig 1. Mean number of brood cells provisioned in trap-nests of different lengths. Centris analis (= dark gray bars), Tetrapedia diversipes (= light gray bars). Bee species/ Site Number of dead immatures % of brood cells E L P A 5cm 10cm 15cm 20cm Centris analis/ DMQ 17 42 2 17 75 28 13 6 Centris analis/ DHU 2 3 0 0 0 11 50 0 Tetrapedia diversipes/DHU 4 12 1 0 100 14 3 8 2. Immature mortality, emerged adults and parasitism The mortality percentage recorded for C. analis offspring due to unknown causes reached 17% of the brood cells (n=460) assessed in the DMQ site and 29% in DHU (n=17). There was immature death in 8% of the brood cells (n=202) in T. diversipes nests. Mortality percentages were higher at the larval stage in both species (Table 2). Both species showed higher immature mortality percentage in nests established in the shortest trap-nests (5 cm long), whereas the mortality percentages were lower in the longest trap-nest (20 cm) (Table 2). The number of immature individuals of C. analis that died differed among trap-nests with different lengths in the DMQ site (χ2 = 20.64, df= 3, p < 0.001). The number of dead immatures gradually decreased from the 5 cm to the 20 cm long trap-nests (Figure 2). The number of emerged adults of C. analis was higher in the 15 and 20 cm nests than in the 10 cm ones (Generalized least squares model: F(2,75)=7.29, p=0.001; Figure 3). Table 2. Number of immatures of Centris analis and Tetrapedia diversipes dead in the nests and percentage of brood cells containing dead immatures in cavities of different lengths (5, 10, 15, 20 cm). Sites = DMQ, DHU. E = egg, L = larvae, P = pupae, A = adult Fig 2. Mean number of dead immatures of Centris analis in trap- nests of different lengths. 3. Sex ratio The sex ratio of C. analis offspring in the DMQ site was 0.9M:1F, which was similar to the expected proportion of 1:1(χ2=0.69, df=1, p=0.40). There was no difference in the sex ratio of C. analis offspring produced in trap-nests presenting different lengths (General linear model: F(2,47)=1.35, p=0.27) (Table 3). The sex ratio of T. diversipes offspring was 0.4 M:1F, which was significantly different from the expected proportion of 1:1 (χ2=5.45, df=1, p=0.02). Sociobiology 67(2): 261-267 (June, 2020) 265 4. Nesting activity and climatic factors There was nesting activity by solitary bees throughout most the year. C. analis mainly nested from September to March and recorded low nesting activity from April to August. T. diversipes kept its nesting activity from January to March and from June to September, except for July. Bee nesting activity was associated with climatic factors. According to the Akaike Information Criterion, the model based on temperature and rainfall effects was the most parsimonious (Table 4). According to this model, the number of occupied trap-nests in each month of the year increased as the mean Bee species/site Trap-nest Length Males Females Centris analis/ DMQ 5cm 02 03 10cm 22 21 15cm 48 46 20cm 24 34 Centris analis/ DHU 5cm 0 0 10cm 3 1 15cm 2 2 20cm 0 0 Tetrapedia diversipes/ DHU 5cm 0 0 10cm 4 7 15cm 13 17 20cm 13 14 Table 3. Number of males and females of Centris analis and Tetrapedia diversipes produced in trap-nests with different lengths. Sites = DMQ, DHU Models AICc df dAICc wi Site + species+ temperature + rainfall 162.7 7 0 0.83 Site + species + air humidity + temperature + rainfall 166.1 8 3.4 0.15 Site + species + air humidity + temperature 171.9 7 9.2 0.008 Site + species+ temperature 172.5 6 9.9 0.006 Site + species + rainfall 176.3 6 13.6 <0.001 Site + species 177.6 5 15.0 <0.001 Site + species +air humidity 179.4 6 16.7 <0.001 Site + species + air humidity + rainfall 179.4 7 16.8 <0.001 Null 233.1 1 70.4 <0.001 (dAICc represents the difference between the AICc value of model i and the AICc value of the most parsimonious model; wi is the weight of Akaike model i). AICc / bias corrected version of the Akaike information criterion. Table 4. Summary of generalized linear models with Poisson distribution, which describe the relationship between number of occupied nests, temperature and precipitation, using the study site and bee genera as covariates. The models are classified in ascending order of AICC values. Fig 3. Mean number of emerged adults in trap-nests of different lengths. Centris analis (= dark gray bars), Tetrapedia diversipes (= light gray bars). monthly temperature and rainfall also increased (χ2 = 21.22, df = 2, p = 0.001). This analysis indicated the presence of two outliers with extremely high number of occupied nests. After removing them, the most parsimonious model indicated that the number of occupied nests were related to temperature and air humidity variations (χ2 = 9.5, df = 2, p = 0.009; Table 5). The number of occupied trap-nests increased as temperature (b = 0.1) and humidity (b = 0.06) also increased. Models AICc df dAICc wi Site + species + air humidity + temperature 113.5 7 0.0 0.33893 Site + species + temperature 114.0 6 0.5 0.26567 Site + species + temperature + rainfall 114.8 7 1.3 0.17358 Site + species + air humidity + temperature + rainfall 116.5 8 3.0 0.07616 Site + species 116.5 5 3.0 0.07443 Site + species + air humidity 117.6 6 4.1 0.04314 Site + species + rainfall 119.2 6 5.7 0.01993 Site + species + air humidity + rainfall 121.0 7 7.5 0.00811 Null 131.6 1 18.1 <0.001 (dAICc represents the difference between the AICc value of model i and the AICc value of the most parsimonious model; wi is the weight of Akaike model i). AICc / bias corrected version of the Akaike information criterion. Table 5. Summary of generalized linear models with Poisson distribution, which describe the relationship between number of occupied nests, temperature and precipitation, using the study site and bee genera as covariates. In this case, two outliers were removed in relation to the results shown in table 4. The models are classified in ascending order of AICC values. CO Santos, PEC Peixoto, CML Aguiar – Cavity length affects the occupation of trap-nests266 Discussion The oil bee C. analis accepted a wide range of cavity lengths (from 5 to 20 cm) as nesting substrate; however, the shortest trap-nests (5 cm) were clearly less attractive than the longest ones (10 cm or longer). The present study did not show any preferential occupancy between the longest trap-nests (10, 15 and 20 cm), and this outcome suggests that they are all appropriate for C. analis nesting. Similarly, Alonso et al. (2011) did not find significant differences in occupancy by nesting females of C. analis in wooden blocks presenting small length differences (5.5, 6.0, 6.5 and 7.0 cm), although the shortest trap-nests recorded higher occupation in a third study site. In addition, the number of brood cells produced by C. analis was larger in longer cavities. When there is a low incidence of parasitism, as reported in this study, the occupation of longer cavities, where a female can produce a larger number of brood cells, seems to be an advantage, since it would lead to lower costs with the selection and establishment of new nests. On the other hand, in a different scenario, with a higher incidence of parasitism, could be advantageous to spread offspring over several nests, because some cleptoparasites can attack the same nest several times (Gazola & Garófalo, 2003; Aguiar, unpublished data). The oil bee T. diversipes also showed trend of preferring to nest in longer trap-nests. In these straws, females left a considerable portion of the cavity empty. This strategy can play a role, protecting the brood cells against natural enemies. The mortality percentage recorded for C. analis offspring due to unknown causes suggests that there are not considerable losses of such offspring in these sites, similar to previous observation in DMQ and in nesting areas close to it (Aguiar & Pina, 2012; Aguiar et al., 2013). On the other hand, Jesus and Garófalo (2000), Gazola and Garófalo (2003), as well as Couto and Camillo (2007) reported great brood cell losses for this species in an urban area. Factors causing the death of immature tropical solitary bees, who nest in trap-nests, remain poorly investigated. Some researchers suggest that high temperatures influence immature mortality (Frankie et al., 1988; Jesus & Garófalo, 2000; Gazola & Garófalo, 2003). However, Couto and Camillo (2007) did not find evidences about the influence of temperature on C. analis offspring mortality when they compared mortality percentages in nests established in shady vs. sunny areas. Aguiar et al. (2013) have suggested that nest management during its removal from the field, as well as during nest transportation to the laboratory – which, sometimes means long distances -, could also have some influence on mortality. Bosch and Kemp (2001) previously reported the importance of handling the nests to decrease immature mortality incidence, because the inappropriate handling of Osmia lignaria Say (Megachilidae) nests can take larvae away from their provisions and cause death due to starvation. Results have shown low mortality in the longest trap- nests, although T. diversipes and C. analis align the brood cells in the trap-nests (Jesus & Garófalo, 2000; Camillo et al., 1995), a fact that could result in difficulty for individuals reared at the bottom of the nest to reach the exit. This outcome suggests that longer cavities would not limit the survival of their offspring. On the other hand, both bee species showed higher mortality percentages in the shortest trap-nests (5 cm). The sex ratio of C. analis offspring produced in trap- nests of different lengths showed no differences; however, it is necessary to consider that the number of nests recorded in 5 cm long trap-nests, as well as the number of emerged adults, was too small, so they were not included in the analysis. On the other hand, Alonso et al. (2011) have concluded that cavity length has influenced offspring sex ratio, since there was male-biased sex ratio in the two shortest trap-nests (5.5 and 6.0 cm), whereas the sex ratio was not significantly different from 1:1 in the two longest ones (6.5 and 7.0 cm). The nesting activity of C. analis was intense in summer and decreased from May to August, as previously reported for DMQ (Pina & Aguiar, 2011). Based on results in the present study, temperature and rainfall have influenced the frequency of nesting in these cavity-nest bees. The low offspring production from mid-fall and throughout the winter must be considered when planning C. analis management for crop pollination, as it would result in low availability of female pollinators. Finally, the present study points out that longer trap- nests (10 cm or longer) are the best options to produce C. analis and T. diversipes offspring, since they have good acceptance by nesting females, and there is low offspring mortality in nests established in these longer cavities. It seems plausible recommending the adoption of 10 cm trap-nests for C. analis breeding in agricultural areas that depend on its pollination service, if one taken into account the higher costs and logistics impairments to produce wooden blocks at the appropriate thickness to shelter the longest trap-nests (longer than 10 cm), as well as the lack of significant difference in many of the assessed parameters in comparison to the parameters recorded for the 15 and 20 cm long cavities. Acknowledgments We thank the National Council for Scientific and Technological Development, Brazil (CNPq, proc. no. 475715/2008-0, no. 562518/2010-0), and the Fundação de Amparo à Pesquisa do Estado da Bahia (FAPESB, TO APP0042/2009) for financial support for this project. This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - Brasil (CAPES) - Finance Code 001. C.S. Oliveira received MSc scholarship from CAPES. References Aguiar, C.M.L., Garófalo, C.A., Almeida, G.F. (2005). Trap-nesting bees (Hymenopera, Apoidea) in areas of dry Sociobiology 67(2): 261-267 (June, 2020) 267 semideciduous forest and caatinga, Bahia, Brazil. Revista Brasileira de Zoologia, 22: 1030-1038. doi: 10.1590/S0101- 81752005000400031 Aguiar, C.M.L. & Pina, W. C. (2012). Mortalidade da prole de abelhas coletoras de óleo (Hymenoptera, Apidae) em áreas cultivadas com aceroleira. Magistra, 24: 136-142. Aguiar, C.M.L., Medeiros, R.L.S., Almeida, G.F. (2013). Mortalidade da prole em duas espécies de Centris (Hymnoptera, Apidae) em uma área urbana. Magistra, 25: 37-42. Alonso, J.D. S., Silva J.F. & Garofalo, C.A. (2011). The effects of cavity length on nest size, sex ratio and mortality of Centris (Heterocentris) analis (Hymenoptera, Apidae, Centridini). Apidologie, 43: 436-448. doi: 10.1007/s13592-011-0110-0 Bosch, J. (1994a). The nesting behaviour of the mason bee Osmia cornuta (Latr) with special reference to its pollinating potential (Hymenoptera, Megachilidae). Apidologie, 25: 84-93. Bosch, J. (1994b). Improvement of field management of Osmia cornuta (Latreille) (Hymenoptera, Megachilidae) to pollinate almond. Apidologie, 25: 71-83. Bosch, J. (1994). Osmia cornuta (Latreille) (Hym., Megachilidae) as a potencial pollinator in almond orchards. Journal of Applied Entomology, 117: 151-157. Bosch, J. & Kemp, W.P. (2001). How do manage the blue orchard bee as on orchard pollinator. Sustainable Agriculture Network. Beltsville, Maryland, 88p. Bosh, J. & W. P. Kemp (2002). Developing and establishing bee species as crop pollination: the example of Osmia spp. (Hymenoptera: Megachilidae) and fruit trees. Bulletin of Entomologial Research, 92: 3-16. doi: 10.1079/BER2001139 Camillo, E., Garofalo, C.A., Serrano, J.C. & Muccillo, G. (1995). Diversidade e abundância sazonal de abelhas e vespas solitárias em ninhos armadilhas (Hymenoptera: Apocrita: Aculeata). Revista Brasileira de Entomologia, 39: 459-470. Couto, R.M. & Camillo, E. (2007). Influência da temperatura na mortalidade de imaturos de Centris (Heterocentris) analis (Hymenoptera, Apidae, Centridini). Iheringia, Série Zoologia: 97: 51-55. doi: 10.1590/S0073-47212007000100008 Frankie, G.W., Vinson, S.B., Newstrom, L.E. & Barthell, J.F. (1988). Nest site and habitat preferences of Centris bees in the Costa Rican dry forest. Biotropica, 20: 301-310. Gruber, B., Eckel, K., Everaa, J. & Dormann, C.F. (2011). On managing the red mason bee (Osmia bicornis) in Apple orchards. Apidologie, 42: 564–576. doi: 10.1007/s13592-011-0059-z Gazola, A.L. & Garófalo, C.A. (2003). Parasitic behavior of Leucospis cayennensis Westwood (Hymenoptera, Chalcidoidea, Leucospidae) and rates of parasitism in populations of Centris (Heterocentris) analis Fabricius (Hymenoptera, Apidae, Centridini). Journal of the Kansas Entomological Society, 76: 131-142. Gazola, A.L. & Garófalo, C.A. (2009). Trap-nesting bees (Hymenoptera, Apoidea) in forest fragments of the State of São Paulo, Brazil. Genetics and Molecular Research, 8: 607- 622. doi: 10.4238/vol8-2kerr016 Jesus, B.M.V. & Garófalo, C.A. (2000). Nesting behaviour of Centris (Heterocentris) analis (Fabricius) in southeastern Brazil (Hymenoptera, Apidae, Centridini). Apidologie, 31: 503-515. doi: 10.1051/apido:2000142 Magalhães, C.B. & Freitas, B.M. (2012). Introducing nests of the oil-collecting bee Centris analis (Hymenoptera: Apidae: Centridini) for pollination of acerola (Malpighia emarginata) increases yield. Apidologie, 44: 234-239. doi: 10.1007/ s13592-012-0175-4 Martins, C.F., Ferreira, R.P. & Carneiro, L.T. (2012). Influence of the orientation of nest entrance, shading, and substrate on sampling trap-nesting bees and wasps. Neotropical Entomology, 41: 105-111. doi: 10.1007/s13744-012-0020-5 Oliveira, R. & Schlindwein, C. (2009). Searching for a manageable pollinator for acerola orchards: The solitary oil-collecting bee Centris analis (Hymenoptera: Apidae: Centridini). Journal of Economic Entomology, 102: 265-273. O’Neill, K.M., Pearce, A.M., O’Neill, R.P., Miller, R.S. (2010). Offspring size and sex ratio variation in a feral population of alfalfa leafcutting bees (Hymenoptera: Megachilidae). Annals of the Entomological Society of America, 103(5): 775–784. doi.org/10.1603/AN09183 Pina, W.C. & Aguiar, C.M.L. (2011). Trap-nesting bees (Hymenoptera: Apidae) in orchards of acerola (Malpighia emarginata DC) in a semiarid region in Brazil. Sociobiology, 58: 379-392. Rust, R.W. (1998). The effects of cavity diameter and length on the nesting biology of Osmia lignaria propinqua Cresson (Hym.: Megachilidae). Journal of Hymenoptera Research, 7: 84-93. Seidelmann K, Bienasch A, Pröhl, F. (2015). The impact of nest tube dimensions on reproduction parameters in a cavity nesting solitary bee, Osmia bicornis (Hymenoptera: Megachilidae). Apidologie 47: 114-122. doi: 10.1007/s13592- 015-0380-z Stephen, W.P. & Osgood, C.E. (1965). Influence of tunnel size and nesting medium on sex ratios in a leaf-cutter bee, Megachile rotundata. Journal of Economic Entomology, 58: 965-968 Torchio, P.F. & Tepedino, V.J. (1980). Sex ratio, body size and seasonality in a solitary bee, Osmia lignaria propinqua Cresson (Hymenoptera: Megachilidae). Evolution, 34: 993-1003.