The introduction of the honey bee, Apis mellifera L. (Hymenoptera: Apidae), into the arid environments of Mexico has affected the behavioral ecology of native bees. We described the behavioral repertoire and interactions between A. mellifera and the native bee Lithurgus littoralis Cockerell (Hymenoptera: Megachilidae) on Opuntia huajuapensis Bravo (Cactaceae) flowers in a semiarid environment. We filmed the bees in 150 cactus flowers to obtain the diversity of behaviors and their durations and thereby quantify the interactions. The behavior accumulation curve (Clench model) showed differences in the behavioral repertoire between the 2 bee species and between the sexes of L. littoralis. We found that A. mellifera and L. littoralis females invested more time in feeding behavior than L. littoralis males and recorded a wider repertoire of agonistic behaviors in male compared with female bees. Native male bees often perched in flowers and were inactive for long periods. The results indicate a possible interference competition between native and non-native bee species that are visiting the flowers of O. huajuapensis.
The directed or unintentional introduction of non-native bees in different environments has affected native pollinators mainly because they compete for floral resources (Goulson 2003; Paini 2004). The honey bee, Apis mellifera L. (Hymenoptera: Apidae), has been considered beneficial because of its importance as a generalist pollinator of many economically important plants and has, therefore, been introduced in many ecosystems. However, its introduction has had negative consequences for native biota, as A. mellifera can reduce the diversity of native bees due to increased competition for floral resources (Huryn 1997; Badano & Vergara 2011), and influence plant-pollinator networks (Campos-Navarrete et al. 2013). As a consequence, the introduction of non-native bee species has modified the behavior of both native and non-native bees (Goulson 2003; Thomson 2004).
The consequences are worse when non-native and native bees converge on the same floral resource, which affects their behavioral responses. Honey bees affect native solitary bees at the peak of the blooming season (Thomson 2016). However, Shavit et al. (2009) indicated that evidence supporting a negative effect is circumstantial and can be more severe during the dry season or during droughts when floral resources are limited. Identifying and measuring the duration and frequency of non-native and native bee behaviors and their intra and interspecific interactions allows us to understand the likely ecological and evolutionary consequences of the introduction of non-native bees (Ishii et al. 2008). This is a particularly critical subject because research suggests that successful pollination is enhanced by increased diversity of native pollinators (Peso & Richards 2010; Brittain et al. 2013).
In addition, the time spent by bees collecting floral resources and the behavioral repertoire of defense and agonistic interactions likely has important consequences on the fitness of both native and nonnative pollinators. There are records of intraspecific differences in the behavioral repertoires of introduced bee species such as A. mellifera (Wilms & Wiechers 1997; Schlumpberger & Badano 2005) and Bombus terrestris L. (Hymenoptera: Apidae) (Spaethe & Weidenmüller 2002), as well as those from families of native bees including Halictidae, Megachilidae, Colletidae (Batra 1978), Anthophoridae (Stone 1995), and the family Apidae, tribe Euglossini (Dressler 1982).
Although studies of agonistic interactions between non-native and native bees have been done in tropical environments (Jha & Vandermeer 2009; Downing & Liu 2012), few studies deal with the spatial convergence of native and non-native bees in floral resources in arid and semiarid areas, where bee diversity is relatively high (Minckley et al. 2000; Golubov et al. 2010). The genus Opuntia has diversified extensively in Mexico in arid and semi-arid habitats. Bees from several native bee genera including Lithurgus, Diadasia, Melissodes, Bombus, Agapostemon, and Megachile visit Opuntia flowers (Osborn et al. 1988; McFarland et al. 1989; Mandujano et al. 1996; McIntosh 2005; Reyes-Agüero et al. 2006; Mandujano et al. 2013).
The interaction between native bees and A. mellifera has been studied mainly in agricultural areas (Pinkus-Rendon et al. 2005; Peters & Carroll 2012) and it is unknown if A. mellifera, introduced in the central region of Mexico in approximately 1760 (Labougle & Zozaya 1986), affects the behavior of native bees when interacting over flowers in other habitats. To understand the patterns of competitive interactions in arid environments, we observed native and non-native bees in the Tehuacán desert in México to determine whether individuals of the oligolectic bee Lithurgus littoralis Cockerell (Hymenoptera: Megachilidae) displayed agonistic behavior to A. mellifera and to conspecifics in flowers of Opuntia huajuapensis Bravo (Cactaceae).
We hypothesized that the convergence of A. mellifera and native bees at the same time and on the same flower would lead to an increase in the frequency and duration of agonistic behavioral displays. This would likely affect the behavior of native and A. mellifera bees during each conspecific or heterospecific encounter. Thus, our aim was to record the behavioral responses of L. littoralis and A. mellifera bees and their interactions in flowers of the cactus O. huajuapensis to address the following research objectives: (1) to describe the behavioral repertoires, (2) to compare them between species, and (3) to compare them in native male and female bees when they converge on O. huajuapensis flowers.
Materials and Methods
STUDY SITE
The study was conducted in the northern extreme of the Tehuacán desert, in the north central portion of the states of Veracruz and Puebla, in the locality known as “Frijol Colorado” (09.6078944°, -97.3821500°). This area is within the arid Cuenca Oriental Basin in the Mexican Trans-volcanic belt (INEGI 1998) at an altitude of 2,300 m, an average annual rainfall of 500 mm, and a mean annual temperature of 12 °C to 18 °C. The climate type is 1 of the wetter types in semiarid zones in Mexico (García 1988). A lava flow outcrop characterizes the geology of this site dominated by the arborescent monocotyledons Nolina parviflora (Kunth) Hemsl. (Asparagaceae) and Yucca periculosa Baker (Asparagaceae) (Dávila et al. 2002).
Temperature (°C), relative humidity (%), and wind speed (m per s) were measured at the beginning and end of each filming period at the floral level with a portable digital weather tracker (Kestrel 4000; Nielsen-Kellerman, McKellar, ACT, Australia). The micro-environmental conditions of flowers during the study were similar among recording times. The mean temperature was 24 ± 2.3 °C (± 1SE; coefficient of variation [CV] = 9.6%). The minimum and maximum temperatures were 13 and 28 °C, respectively, and the humidity was 35 ± 7.6% (min. 16%, max. 53%; CV = 3%) and wind speed was 2 ± 1.1 m per s (± 1SE; min. 0 m per s, max. 17 m per s; CV = 56%).
STUDY SPECIES
The cactus Opuntia huajuapensis occurs in the states of Puebla, Veracruz, and Oaxaca in Mexico. It has 5 to 6 cm-long yellow flowers with greenish yellow segments in the perianth (Bravo-Hollis 1978). This species blooms from May through Jun, when their flowers represent the only resource available for various pollinators, and with the native bee L. littoralis and the non-native A. mellifera as the most conspicuous species. The mean flower area in the study site was 15 ± 6.5 cm2 at a height of 52 ± 17.6 cm (SE). The average time that flowers remained open was 1.4 ± 0.64 d as recorded in 60 flowers within the study site.
In the study area, feral colonies of A. mellifera visit the flowers of O. huajuapensis, and the native oligolectic bee L. littoralis was observed only during the flowering season of O. huajuapensis. Lithurgus littoralis can be recognized by its dark color and sexual dimorphism with females larger than males (13–16 and 10–13 mm, respectively).
Lithurgus littoralis was identified by Hugo Eduardo Fierros López, Centro Universitario de Ciencias Biológicas y Agropecuarias (CUCBA), Departamento de Botánica y Zoología, Universidad de Guadalajara, Zapopan, Jalisco, México and voucher specimens were deposited in the Colección Entomológica of José Luis Navarrete-Heredia.
BEHAVIORAL REPERTOIRE
Consecutively, for 5 d, at the beginning of the O. huajuapensis flowering season in late May 2009, we filmed 150 flowers from different plants with 2 camcorders (Sony Handycam® Camcorder, model number: DCR-DVD610; Sony Corporation, Tokyo, Japan) using a 10× zoom. The cameras were placed 1 m away from the target flowers to avoid interfering with bee behavior. The videos were recorded between 11 AM and 12 PM. Recording sessions lasted 3 min and were finished each d after each operator recorded 75 flowers. Bees were present in all cases and selected flowers had full open petals and stigma, anthers with pollen, and lacked evidences of florivory. The weather conditions in all cases were favorable for insect activity.
Each video file was analyzed frame by frame with an InterVideo WinDVR Recorder 4.5 program to annotate the behavioral repertoire of the bees (Campos-Jiménez et al. 2014). For statistical analyses, behavioral patterns were classified into 10 categories: (1) “floral visit” indicates contact of the legs with some structure of the flower; (2) “feeding” indicates collection of nectar or pollen; (3) “exploratory flights”, which include sustained, fast zigzag, horizontal, S-form, U-form and near the flower flights without making contact with the flower; 4) “bee walk” on either stamens or petals of the flower; 5) “body cleaning” either head or abdomen; 6) “agonistic behavior” including fighting, biting, body contact, lunge or approach between bees in or close to the filmed flower; 7) “no activity” implying bee immobility while it was posing in the flower; (8) “rapid movement of abdomen” is selfexplanatory; (9) “proboscis extension” while in the flower without doing another behavior in the flower; and (10) “leaving the flower” was moving away from the flower for at least 1 s. We also recorded the species and number of bees that performed the different behaviors, and the duration of each behavior.
INTERACTIONS BETWEEN SPECIES
The intra and interspecific interactions among bees were recorded from the videos as the number of events in which 2 species of bees visited a flower at the same time and the frequency of intra and interspecific agonistic behaviors. We also recorded the amount of time bees stayed on the flower after receiving aggression from either intra or interspecific individuals independently of the exhibited behaviors after aggression. The numbers of bees recorded per behavioral group were independent during counting because only the first behavioral event was registered.
STATISTICAL ANALYSES
The behavioral repertoire was analyzed using a Clench model, which is a mathematical function that describes the accumulation curves (Dias et al. 2009). We fitted a nested ANOVA using general linear models (GLM) to relate the number of bees, time spent per visit, and feeding duration as dependent variables (y) to bee species a 3-level factor accounting for A. mellifera females and L. littoralis males and females) and flower identity. The model was y = bee species + bee species [filmed flower] + error, with bee species nested within the filmed flower. We considered each filmed flower as an independent observation. The dependent variables (xi) were square root transformed to meet the assumptions of normality and homogeneity of variances (Zar 1996). Comparisons among means were done using Tukey honest significant difference tests (P < 0.05). To evaluate the time spent by bees in flowers, the following Pearson correlations were carried out: (1) time spent in flowers by female vs. male bees and by native species vs. A. mellifera and (2) feeding duration of female of A. mellifera vs. the duration of agonistic behavior by male bees. The agonistic interaction between bee species was compared using percentages. All analyses were performed in JMP 6.0 (SAS 2005).
Table 1.
Behavioral repertoire of Apis mellifera (A) and of males (L♂) and females (L♀) of Lithurgus littoralis exhibited on flowers of Opuntia huajuapensis. The 1st column indicates the 10 behavioral groups into which the 25 behavioral patterns were merged.

Results
BEHAVIORAL REPERTOIRE
We registered 25 patterns of behavior and grouped them for analysis (Table 1). There were 15 patterns of behavior in the A. mellifera repertoire (Clench model: a = 0.63; b = 0.04; asymptote: 12.5; r2 = 0.95; Fig. 1A). The repertoire of L. littoralis regardless of sex included 25 behaviors (Clench model: a = 4.69; b = 0.22; asymptote: 22.5; r2 = 0.96; Fig. 1B). There were 16 behaviors in the female L. littoralis repertoire (Clench model: a = 0.57; b = 0.05; asymptote: 13; r2 = 0.86; Fig. 1C). Finally, there were 23 behaviors in the repertoire of L. littoralis males (Clench model: a = 4.67; b = 0.25; asymptote: 21; r2 = 0.97; Fig. 1D).
Fig. 1.
Behavior accumulation curves of bees in 150 flowers of Opuntia huajuapensis. A: Apis mellifera (1) and Lithurgus littoralis (2). B: L. littoralis females (3) and L. littoralis males (4). Dotted lines indicate the 95% confidence intervals.

The number of bees for each behavioral group varied between species and sex. Female bees of L. littoralis visited more flowers but A. mellifera bees displayed a higher percentage of feeding behavior compared with female native bees. The exploratory flight was the most common behavior of male bees coupled with walking in the flowers and body cleaning. Nearly 14.6% of male native bees displayed agonistic behaviors in contrast to less than 1% of female native and nonnative bees (Table 2). When the bees were feeding, they tilted their heads toward the base of stamens to collect nectar or assumed horizontal postures to collect pollen or to rest. A diagonally tilted position with the head pointing away from stamens and nectaries seemed to be a defensive posture from which, according to our field observations, bees were more prone to initiate attacks.
NUMBER OF BEES ON FLOWERS
On the 150 flowers, 684 bees were recorded and the number of bees varied by species and sex (A. mellifera: n = 123; L. littoralis males: n = 495; L. littoralis females: n = 66). The 1-way nested ANOVA indicated that there were significant differences among groups (species and sex). The means comparison tests (P < 0.05) indicated that the mean number of L. littoralis males per flower (3.2 ± 0.2) was significantly higher than that of A. mellifera (0.8 ± 0.09) and L. littoralis females (0.4 ± 0.05) and that there were no significant differences between L. littoralis females and A. mellifera (F = 3.2; df = 2, 149; r2 = 0.81, P < 0.001). A few Diadasia sp. (Hymenoptera: Apidae) were observed but the numbers were not sufficient for statistical analyses and no A. mellifera males were recorded at any time during the experiment.
DURATION OF BEHAVIORS
Apis mellifera spent more time feeding than L. littoralis, with the females of L. littoralis feeding only half the time spent by A. mellifera. Lithurgus littoralis males were inactive on the flowers a greater percentage of time compared with females (Table 3). Furthermore, male bees invested less time in direct resource defense spending only 1.53% of their total time on agonistic behavior (Table 3).
The time spent in flowers by A. mellifera was significantly higher than that recorded for native females and males (F = 3.8; df = 2, 149; r2 = 0.79; P = 0.01; Fig. 2A). In this regard, there were no differences between sexes in L. littoralis (P > 0.05; Fig. 2A). The duration of feeding behavior was significantly different between species and sexes (F = 17; df = 2, 149; P <0.001). Means comparison tests (P < 0.05) indicated that the feeding duration was statistically similar between A. mellifera and L. littoralis females, but significantly higher when compared with L. littoralis males (Fig. 2B).
CORRELATION OF BEHAVIORS
There was no correlation between the time spent by female and male native bees and the time spent by A. mellifera (N = 138, r = 0.03, P > 0.05). There also was no correlation between the time spent by male and female L. littoralis (N = 138; r = 0.04, P > 0.05). Finally, the time that L. littoralis females fed on the flower was not significantly correlated with the time used by L. littoralis males for agonistic behavior (N = 138; r = 0.06, P > 0.05).
INTERACTIONS BETWEEN SPECIES
In 78 of the 150 filmed flowers of O. huajuapensis, there were 107 instances with more than 1 bee on a flower at the same time, 9% involving individuals of A. mellifera. The presence of an A. mellifera and a L. littoralis male in the same flower occurred in 24% of the occasions but A. mellifera and L. littoralis females never shared the same flower. The coincidence of L. littoralis males and females in the same flower accounted for 43%, and pairs of L. littoralis males accounted for 23% of all the cases.
Agonistic behaviors on O. huajuapensis flowers occurred in all 107 instances where pairs of bees were recorded. Native males directed 55% of their agonistic behavior toward native females, 24% toward other native males, and 28% toward A. mellifera. Out of 26 agonistic events by L. littoralis males against A. mellifera, only 5 resulted in A. mellifera leaving the flower. When a pair of native males coincided, both flew away immediately from the flower and in only 1 instance did 1 of them, although it could have been another native male (we were unable to identify the individual), occupy the flower again. When a native male and a native female coincided, the male would bite the thorax of the female, but the female would remain feeding in the flower.
Discussion
We evaluated the feeding and agonistic behavior of A. mellifera and L. littoralis males and females on O. huajuapensis flowers, the first study of its kind on the relationship between these species. We found that agonistic behavior between L. littoralis males was more frequent when they were in the same flower with L. littoralis females, but there was no displacement. The displacement of A. mellifera was rare, but it occurred when 2 native males were on or near the same flower.
Table 2.
Bee numbers and corresponding percentages of behaviors displayed on Opuntia huajuapensis flowers by Apis mellifera and by Lithurgus littoralis males and females according to the behavioral group. Because we recorded only the first behavioral event, the numbers of bees per behavioral group were considered to be independent events.

BEHAVIORAL REPERTOIRES
The Clench model allowed us to show that the behavioral repertoire of L. littoralis males was far richer (as shown by a higher asymptote) than that of the similar repertoire of L. littoralis and A. mellifera females, with most of their activities consisting of floral visits, collection of floral resources and exploratory flights, similar to other megachilid female bees (McKinney & Park 2012). The behavioral repertoire of L. littoralis males was more diverse. It included exploratory flights, which would allow them to assess the presence of other bees on flower resources. These males also displayed a larger proportion of agonistic behaviors compared with L. littoralis and A. mellifera females, including up to 7 different behaviors exhibited when coping with other bees on the same flower.
In most species of Megachilidae, males and females have different foraging strategies. Males take no part in cell provisioning or brood care and feed only to maintain their ability to copulate (Johnson & Hubbell 1974; Alcock et al. 1977; Eickwort 1977). This would explain the greater time spent by L. littoralis males in short flights and agonistic behavior, possibly as a strategy to compete with other males for access to females. In contrast, females need the energy to fly, to invest resources in eggs (Stone 1995), and to provision them with resources far beyond their own metabolic requirements. Therefore, females invest more time feeding than males, and their behavioral repertoire should allow them to optimize search patterns to obtain floral resources.
Table 3.
Time (s) and corresponding percentage of the behaviors displayed on Opuntia huajuapensis flowers by Apis mellifera and by Lithurgus littoralis males and females based on data extracted from 150 three-min videos. Because we recorded only the first behavioral event, the numbers of bees per behavioral group were considered as independent events.

BEHAVIORAL RESPONSES
The behaviors performed by A. mellifera and L. littoralis in the flowers of O. huajuapensis varied among individuals. Our data show that 70% of A. mellifera fed only on nectar and that 16% performed exploratory flights that allowed them to distinguish flower characteristics and to evaluate resources (Townsend-Mehler et al. 2011). The flight over flowers performed by L. littoralis females would allow them to assess floral traits and resource availability and to avoid native males as reported for Anthophora plumipes Pallas (Hymenoptera: Apidae) by Stone (1995). This contrasts with the short exploratory flights exhibited by males, possibly a strategy to find females and to drive away other bees from their territory.
There was no evidence of agonistic behavior against conspecific or heterospecific bees by A. mellifera when visiting O. huajuapensis flowers, a behavioral pattern reported in other environments (Schaffer et al. 1983; Huryn 1997; Campos-Jiménez et al. 2014). Nevertheless, the feeding time spent by A. mellifera could lead to resource depletion and interference competition against native bees (Polatto & Chaud-Netto 2013). Such an extensive use of resources may give A. mellifera a competitive edge against other species (Schaffer et al. 1979; Paini 2004).
Fig. 2.
Time spent (A) and mean feeding duration (B) in flowers of Opuntia huajuapensis by Apis mellifera females and Lithurgus littoralis females and males. No A. mellifera males were recorded at any time during the experiment. Vertical bars indicate 95% confidence intervals.

The lack of intraspecific agonistic behaviors of A. mellifera observed in our study could be a consequence of colony-speci?c identity by the mixture of chemical compounds that are carried on the surfaces of all workers and allow discrimination between self and non-self (Breed et al. 1995).
Males of L. littoralis were as aggressive as other oligolectic bees (e.g., Ptilothrix fructifera Holmberg [Hymenoptera: Apidae]; Oliveira & Schlindwein 2010) and megachilid bee species (Seidelmann 1999). We assume that the agonistic behavior between native males allows them to monopolize floral resources (Nagamitsu & Inoue 1997; Rivera-Marchand et al. 2008) and to compete for females (Alcock et al. 1978; Oliveira & Schlindwein 2010). The large proportion of male attacks toward females was consistent with previous observations on solitary bees (Stone 1995) in which agonistic behavior allowed males to drive away receptive females from the flower to copulate with them at ground level or to drive off gravid females in favor of receptive females.
The presence of immobile native male bees on cactus flowers seems to be a guarding behavior, as reported for the oligolectic bee Diadasia rinconis Cockerell (Hymenoptera: Apidae; Ordway 1987). This behavior likely functions as a means of interference competition, preventing the male from investing energy in direct confrontations with other males. This would involve a trade-off, as the male would reduce its risk and energy expenditure against opponents by reducing feeding time, which is short according to our records. Agonistic bees often benefit by having continuous access to the resource in contrast to passive bee species (Roubik et al. 1986), but this did not seem to be the case in the L. littoralis males because they spent little time on nectar consumption.
INTERACTIONS BETWEEN SPECIES
Apis mellifera individuals do not respond to the aggression of L. littoralis males, as reported when A. mellifera and Trigona corvina Cockerell (Hymenoptera: Apidae) interact (Roubik 1981). Our findings differ from those in which native species avoid the flowers when A. mellifera displays constant aggression towards them (Pinkus-Rendon et al. 2005). About 15% of native males attacked other bees during short intervals to drive them away from cactus flowers. In contrast, female native bees did not visit flowers if A. mellifera was present, possibly because the flower resources have been depleted, making their visit unprofitable. It is important to determine whether A. mellifera reduces resource availability such that it affects the feeding behavior and survival of L. littoralis, as reported for other native bees (Horskins & Turner 1999; Thomson 2004), or if it only reduces the reproductive success of the plant without compromising the fitness of native bees (do Carmo et al. 2004).
The behavioral repertoire of A. mellifera is less diverse than that of L. littoralis males. However, exploratory flights and feeding of A. mellifera and L. littoralis females are the dominant behaviors, likely due to their role in the collection of floral resources. We found that a large proportion of male bees displayed brief episodes of agonistic behavior and invested more time on defensive behavior against other bees, possibly as a strategy to increase access to receptive females on the flowers of O. huajuapensis
Acknowledgments
We thank R. Méndez-Alonzo (CICESE) for suggestions and comments, Hugo E. Fierros López at Universidad de Guadalajara for bee identification, and 2 anonymous reviewers for their constructive comments and suggestions. We also thank the IIP-UV for the use of its facilities, and the students of Universidad Veracruzana: P. Del Moral-Cervantes, J. González-Gálvez, I. Rivera, and C. Suárez-Ramírez, for their help in the field. Research was funded in part by CONACyT postgraduate scholarships # 223551 to ASA-A, # 377111 to GHM and projects UVER-PTC-223 and PFA C-703/2013 (UV) provided to AJM. Mention of trade names does not imply recommendation or endorsement.