The effect of urbanisation and seasonality on wild bee abundance, body size and foraging efforts

Anthropogenic changes highly impact the world's biodiversity. An important human‐driven change to natural environments is increasing urbanisation, which is responsible for decreasing suitable habitats for many wild species, including bees. In this study, we investigate if three levels of urbanisation (low, medium and high) affect body size, foraging efforts and abundance of the sweat bee Agapostemon virescens. Overall, A. virescens was more abundant in medium‐urbanised sites. Second‐generation females (summer bees) were more abundant than overwintered (spring bees) at all levels of urbanisation. According to body size, female bees were larger in highly urbanised sites and male bees were larger in medium‐urbanised sites. According to foraging efforts, we observed an increase in wing wear during spring and a decrease during summer. It was also found a female‐biased sex ratio under high urbanisation and a male‐biased in low urbanisation sites. Our results suggest that highly urbanised sites can still provide sufficient nesting and foraging resources for A. virescens. In addition to our findings of higher bee abundance in low and medium urbanised sites, we suggest that maintaining different levels of urbanisation and heterogenous landscapes within a populous city might have a more positive impact on wild be sustainability.


INTRODUCTION
Human populations have become increasingly concentrated within cities (Grimm et al., 2008;Seto et al., 2011). Urban populations currently exceed 50% of the global population and are projected to reach about 68% by the year 2050 (United Nations, 2018). On a global scale, urban areas are expanding twice as fast as their populations (Angel et al., 2011). This positions urban development as an important anthropogenic alteration of the landscape, representing one of the main causes of habitat loss and extirpation (Grimm et al., 2008;Seto et al., 2011). Because urbanised areas are continuously growing (through sprawl and densification), their presence is generally associated with reductions in biodiversity (Szab o et al., 2022). Therefore, assessing wild and managed species' response to this growing humanmediated threat is imperative to facilitate appropriate conservation actions to better protect vulnerable species.
Urbanisation is especially detrimental to wildlife as it is associated with the degradation, fragmentation and conversion of habitat into artificial impervious surfaces (reviewed in Ayers & Rehan, 2021).
These changes likely negatively affect the foraging, density and survival of many urban-dwelling animals (Brant et al., 2022;Harrison et al., 2019;Hernandez et al., 2009;Marzluff, 2001;McIntyre, 2000;Shochat et al., 2004). Conversely, it has been proposed that cities may serve as dynamic and heterogeneous landscapes capable of connecting humans and providing ecosystem services (e.g., pollination) at multiple scales (Grimm et al., 2008;Pickett et al., 2004). According to this idea, urbanisation might be beneficial for many taxa, including small, highly mobile animals such as bees which may take advantage of the increased heterogeneity and variety of habitats in cities (McKinney, 2002;Savard et al., 2000). In fact, large and diversified city parks have been shown to harbour a greater diversity of pollinators compared to agricultural and natural habitats (Banaszak-Cibicka et al., 2018;Prendergast et al., 2022). For example, a study in parks in the highly populous San Francisco, USA, showed higher support for bumblebee (Bombus sp.) abundance compared to wild parks surrounding the city (McFrederick & LeBuhn, 2006). Additional examples are the diverse bee assemblage harboured by urban community gardens in New York City and the increased wild bee diversity related to higher impervious surfaces in the populated Lyon, France (Fortel et al., 2014;Matteson et al., 2008). Many types of urban habitats, especially residential gardens and parks, may be considered resourcerich for bees, offering them ample foraging and nesting opportunities (Banaszak-Cibicka et al., 2018;Blackmore & Goulson, 2014;Matteson et al., 2008). Thus, the response to urbanisation in bees seems to be idiosyncratic, and highly dependent on the species and on the type of urban habitat.
Urbanisation also acts as an environmental filter, affecting bees according to their functional traits. For example, increasing impervious surfaces (asphalt and concrete) may disproportionately affect bees that nest below ground as it removes the bare ground cover they require for nesting (Geslin et al., 2016;Threlfall et al., 2015). This is the case for the ground-nesting genus Homalictus, which occurs more frequently in green spaces surrounded by lower impervious surface cover within a city (Threlfall et al., 2015). Likewise, small and mediumbodied bees are reported to be more dominant in highly developed city centres, with reduced green areas and less available food (Banaszak-Cibicka & _ Zmihorski, 2012). This is expected given that low amounts of food available contribute to lower maternal brood provisions and translates into smaller offspring body sizes (Bosch & Vicens, 2002;Lawson et al., 2016). Additionally, smaller bees would carry lower pollen and nectar loads back to their nests, impairing brood provision (Chole et al., 2019). The sex ratio is also affected by increasing urbanisation. The theory of Conditional Sex Allocation predicts that under harsher conditions, offspring would be male-biased given that it is the sex with higher relative fitness (Frank & Swingland, 1988). In fact, the European beewolf Philanthus triangulum displayed a male-biased investment ratio under scenarios of low food availability (Strohm & Linsenmair, 1997). Similar results were found in wild bees dwelling in an urban-to-rural gradient in Michigan, USA, where the investment sex ratio was male-biased as urbanisation increased (Fitch et al., 2019). However, it is also important to stress that the opposite can occur, and patchily distributed resources with higher isolation in cities might benefit large-bodied bees and a femalebiased sex ratio. This is a consequence of adults capable of accessing distant yet rewarding patches as a consequence of their greater flight distance capabilities (Ferrari & Polidori, 2022;Greenleaf et al., 2007;Theodorou et al., 2020).
Increasing urbanisation might also affect foraging activity in bees, given that impervious surfaces intensify foraging efforts by increasing collision probabilities with foliage, in addition to being highly associated with mortality (Cartar, 1992;Foster & Cartar, 2011). Another functional trait highly influenced by urbanisation in bees is seasonality, with lateremerging bees thriving more than those that emerge earlier in cities (Wenzel et al., 2020). This is partly explained by the mismatch in plantpollinator phenology that might occur in cities due to climate change (Goulson et al., 2015). Although a report has shown that bees can parallel phenological changes with the plants they pollinate (Bartomeus, Park, et al., 2013). Another hypothesis is that later-emerging bees also benefit from the greater amounts of available provision that naturally increase through time. For instance, the sweat bees Lasioglosum calceatum and Halictus rubicundus presented a correlation between body size and season length, where larger body size was found when the season was longest (Davison & Field, 2017). Therefore, species able to shift phenology during a prolonged foraging season and those capable of producing multiple generations a year would be less affected (Goulson et al., 2015;Wenzel et al., 2020;reviewed in Ayers & Rehan, 2021).
Sweat bees of the genus Agapostemon (Halictidae) are soil nesting (Roberts, 1973). The genus comprises species exhibiting a variety of social behaviours, including solitary and aggregation nesters (Roberts, 1973). The bicoloured striped sweat bee, Agapostemon virescens Fabricius, is a medium-bodied bee that nests in aggregations below ground and is widespread across North America (Abrams & Eickwort, 1980). This generalist species provide pollination services to many crops and wild plant species (Gardiner et al., 2010;Gibbs et al., 2017;Sivakoff et al., 2018). Previous studies in sweat bees, including Halictus ligatus and A. virescens, have revealed variation in body size and seasonality according to urbanisation and land use change to their natural habitat (Brant & Camilo, 2021;Nooten & Rehan, 2022) and have reported high abundance in urban farms and vacant lots within cities (Sivakoff et al., 2018). Here, we investigate the effects of urbanisation on the body size, foraging efforts and seasonality of the ground-nester A. virescens dwelling in urban landscapes. Our objectives were to (i) investigate how body size and foraging efforts are affected by urbanisation intensity and seasonality (spring and summer) and (ii) examine how abundance is affected by urbanisation and seasonality. We hypothesize that in highly urbanised sites (higher impervious surface), bees will: (i) present smaller body size and higher wing wear (a proxy for foraging effort), (ii) present lower total abundance and (ii) present a male-biased investment ratio. We also hypothesize a higher abundance of bees during summer (later-emerging bees).

Study species life history and seasonality
A. virescens is a communal species that form aggregation nests shared among many females. Several females occupy a single burrow, although each individual autonomously builds cells and provisions their brood. A. virescens produce one brood per year, although two generations co-occur throughout the year. Overwintered females (hereafter called spring females) usually first emerge from hibernation in mid-May. These females continue foraging for the following 2 weeks after emergence and begin digging nests. This first generation of females gradually disappears after July. The daughters of overwintered females (hereafter called summer females) usually start emerging in early August and are active until October, the time when most bees start hibernating to emerge in the following year (Abrams & Eickwort, 1980). Males are present from late July to September but typically do not overwinter and are hardly seen during spring months (Abrams & Eickwort, 1980). For our study, we defined spring females as those collected from May to July 21st and summer bees as collected from July 22nd to October. These dates were selected based on the phenology plot in Figure S1, which shows a gap in bee abundance, likely associated with this species' seasonal behaviour of parenting gradually disappearing during summer until the emergence of daughters. We also associated these dates with the first appearance of males, highly correlated to the first emergency of daughters (Abrams & Eickwort, 1980).

Study area and sampling method
This study was carried out in Toronto, Canada ( to calculate the percentage of land cover of our sampling points (Land Information Ontario, 2019). We defined low-urbanised sites (n = 10) as having less than 25% of impervious surface, medium-urbanised sites (n = 10) as having between 25% and 75% of impervious surface, and high-urbanised sites (n = 9) as having more than 75% of impervious surface within the buffer. A. virescens individuals were sampled twice per week in each site using sweep nets, pan traps and blue vane traps throughout this species' activity (May-October). These bees do not show morphological castes or discrete size variation classes, although a difference can be found between male and female body sizes (Abrams & Eickwort, 1980). Therefore, for all the analyses, we considered males and females separately. Information on bees collected from each sampling method can be found in Table S1.

Body trait measurements
For every bee, we determined the sex and measured body size using a stereo microscope (Nikon SMZ1270) according to two body traits: head width and intertegular distance (ITD). Head width corresponds to the distance between the outer margins of the compound eyes (Rust, 1991), and ITD is defined as the distance measured between each tegula (Cane, 1987). To determine whether the urbanisation intensity was affecting bees' foraging efficiency, we measured wing wear, here determined as the missing portions seen in the apical and posterior borders of both forewings (Mueller & Wolf-Mueller, 1993).
We classified the levels of wing wear on a scale from zero to five according to a protocol proposed by Mueller and Wolf-Mueller (1993), where zero describes wings with margins completely intact, and five are wings with margins showing total obliteration.

Statistical analyses
We tested for the normality of our data using the Shapiro-Wilk test of normality, then plotted a quantile-quantile (Q-Q) plot to assess if residuals were normally distributed. We then tested possible correlation among the body trait variables by performing a Pearson's correlation test in R v 4.2.1 (R core team 2022). Head width and ITD showed a high correlation (r = 0.89, t = 72.52, df = 1341, p-value <0.001, Figure S2). Although head width has been successfully used as a proxy for body size, there might be differences due to allometry (Cane, 1987).

Urbanisation and seasonality effects on body size
Body size, as measured by ITD, in A. virescens females ranged from 1.29 to 2.64 mm (mean ± SE; 2.09 mm ± 0.01) and from 1.28 to F I G U R E 2 Total number of collected individuals of Agapostemon virescens according to sex and urbanisation intensity. Low urbanised sites were male-biased (62%, χ 2 = 25.42, p < 0.01), and high urbanised sites were female-biased (41% male, χ 2 = 57.67, p < 0.01).
F I G U R E 3 Total number of collected individuals of Agapostemon virescens according to seasonality and urbanisation intensity. Bars represent mean standard errors. Letters indicate statistically significant differences in spring and summer abundance in low urbanised sites (Tukey's, t = 1.87, p = 0.04).

Urbanisation and seasonality effects on foraging efforts
Wing wear, a proxy for foraging effort, was higher during spring (mean X = 0.91 ± 0.07, Figure S5) compared to summer (mean X = 0.84 ± 0.06; t-test; t = À1.68, p = 0.05) and higher in males (1.014 ± 0.08) compared to females (0.539 ± 0.06), but not statistically significant (t-test; t = 0.20, p = 0.42). Females had wing wear increasing during spring (t-test; t = 6.20, df = 9, p < 0.01) and decreasing during summer (t-test; t = 10.43, df = 9, p < 0.01). More specifically, spring females had greater wing wear accumulation at medium urbanisation sites (ANOVA with Tukey's; F = 3.83, df = 2, p = 0.02, Figure 6), whereas summer females had greater wing wear in high urbanised F I G U R E 4 Body size (measured as intertegular distance, ITD) across urbanisation intensity habitats for females and males Agapostemon virescens. Bars correspond to standard error, and small letters represent values significantly different according to the posthoc Tukey test between low and high-intensity levels for females ( p = 0.01).
F I G U R E 5 Body size (measured as intertegular distance, ITD) across urbanisation intensity habitats for spring and summer females of Agapostemon virescens. Bars correspond to standard error. Letters indicate significant differences according to the Tukey test between low-medium intensity ( p = 0.01) and low-high intensity (p < 0.01) for summer females.

DISCUSSION
This study investigated how the sweat bee Agapostemon virescens responds to an urbanisation gradient according to abundance and body traits. Following our predictions, we found higher bee abundance in medium-urbanised sites and during summer but found the contrary for body size patterns, where larger-bodied bees were found in highly urbanised sites. We found a female-biased ratio at high urbanisation, whereas low urbanisation sites were male-biased. We also found that wing wear, a proxy for foraging effort, increased in spring and decreased in summer females. Overall, these data suggest that wild bees are affected by their local environment and that the scale of the impact is season and sex-dependent.
We found a higher abundance of A. virescens at low and medium levels of urbanisation. This is perhaps unsurprising given that the increasing development of cities reduces and fragments habitat, which is highly detrimental to bee species richness and abundance (Winfree et al., 2009;reviewed in Ayers & Rehan, 2021). Additionally, higher abundance and species richness have been found in public parks compared to residential neighbourhoods, suggesting that green spaces can provide critical and abundant resources for wild bees (Threlfall et al., 2015). We also found an overall higher abundance of A. virescens during summer. This could be a result of the bee's life cycle, where spring bees are usually the overwintered inseminated females, and summer bees are the subsequent daughter generation and are, therefore, naturally more numerous (Abrams & Eickwort, 1980). More specifically, a significantly higher abundance of summer bees versus spring bees was found in low-urbanised sites (more green spaces). A similar pattern was found for other groundnesting bees in the genus Halictus, where, in an urban landscape, abundance increased according to the amount of flowers (Bennett & Lovell, 2019). Thus, low and medium levels of urbanisation seem to be creating favourable environments for the maintenance of numerous individuals of ground-nesting bees such as A. virescens, presumably by providing abundant floral resources and reducing intensive human activity (Threlfall et al., 2015).
We found larger-bodied bees in highly urbanised areas. These sites had a contrasting result of comprising less abundance yet holding larger-bodied bees. One explanation is assigned to the habitat fragmentation present in highly urbanised environments (suitable habitat patchily distributed) holding only a limited number of individuals (Swenson & Franklin, 2000). Albeit larger bees would be benefited from their ability to fly the longer distances needed to reach patches with greater foraging and nesting resources (Ferrari & Polidori, 2022;Theodorou et al., 2020). Thus, urban areas can act as environmental filters for small-bodied bees whilst large-bodied bees prosper (Ayers & Rehan, 2021). For example, the buff-tailed bumblebee Bombus terrestris is larger in body size in highly urbanised cities presenting a fragmentation background (e.g., those with a high density of roads and human infrastructure; Theodorou et al., 2020). Another advantage of larger-bodied bees compared to smaller bees is their ability to transport more pollen and nectar simultaneously, which would be beneficial in urban and managed environments (Goulson et al., 2002;Willmer & Finlayson, 2014). This has been reported for the groundnesting bee Andrena nasonii, in which larger females transported almost 40% more pollen in highly managed areas (Renauld et al., 2016). Also, in contrast to simplified landscapes, cities can offer a mosaic of different habitat types that results in high heterogeneity, F I G U R E 6 Wing wear across urbanisation intensity for spring and summer females of Agapostemon virescens. Letters indicate significant differences according to the Tukey test between medium-high intensity ( p = 0.02) for spring females. Summer females did not show significant differences according to wing wear and urbanisation intensity ( p = 0.06). Bars correspond to standard error.
which may enable bees to access greater opportunities for food and nesting resources (Bennett & Lovell, 2019;as reviewed in Prendergast et al., 2022). Interestingly, in 2016 Toronto became the first 'Bee City' in Canada, meaning that the City Council started adopting a series of steps to raise awareness of bees' importance (City of Toronto, 2018).
Among those steps are habitat enhancement for pollinators and the creation and connection of green spaces within the city. Urban areas are of high concern for bee conservation, and policymakers must propose strategies to provide higher habitat quality and guarantee enough reachable food resource opportunities.
The sex ratio was female-biased in high urbanisation and malebiased in low-urbanisation areas. Depending on which direction the sex ratio goes toward (male or female-biased), it can influence the success of dispersal and colonisation of new habitats (Cote et al., 2007).
Females are considered better effective pollinators than males, partly because male bees often disperse longer distances in search for mates, whereas females tend to disperse shorter distances from their nest (L opez-Uribe et al., 2015;Ne'eman et al., 2006;Wolf et al., 2012). Also, we found larger-bodied bees in highly urbanised sites, and larger females tend to shift their sex ratio toward daughters (Seidelmann et al., 2010). Thus, the strategy of producing more females is likely related to improving foraging efficiency in fragmented areas with reduced foraging opportunities.
We found foraging effort to be limited by levels of urbanisation in female bees. Higher wing wear was found in medium and high urbani-

SUPPORTING INFORMATION
Additional supporting information can be found online in the Supporting Information section at the end of this article.