Urbanisation and wing asymmetry in the western honey bee (Apis mellifera, Linnaeus 1758) at multiple scales

Introduction

The unique interplay between humans and the environment in cities generates novel anthropogenic stressors such as artificial lighting, air pollution, and elevated temperatures that can profoundly affect organisms at multiple levels of biological organization (Johnson & Munshi-South, 2017; Williams, Hahs & Vesk, 2015; Alberti, Marzluff & Hunt, 2017). Such anthropogenic stressors can induce physiological responses such as cell death (Lodovici & Bigagli, 2011), and elevate stress hormone concentrations (Ouyang et al., 2015; Regoli et al., 2006). Furthermore, they may alter behaviours, including daytime activity (Dominoni & Partecke, 2015) and foraging efficiency (Gate, McNeill & Ashmore, 1995), and significantly influence the composition of biological communities, resulting in increased or decreased numbers of certain functional groups including predators and scavenging individuals (Davies, Bennie & Gaston, 2012). Given the large number of anthropogenic stressors present in cities and the magnitude with which they affect species, it is imperative we find sensitive indicators (i.e. biomarkers) that can detect detrimental effects before populations are irreversibly impacted.

Biomarkers may act as functional measures of exposure to anthropogenic stressors and environmental conditions more generally, as such, they can indicate deteriorating environmental quality and, ultimately, serve as precursors to additional physiological, behavioural, and population-level impacts (Hook, Gallagher & Batley, 2014, Hagger et al., 2006). Traditional biomarkers including abundance, reproductive output, and survival are typically time consuming and cost-prohibitive to quantify. As such, it is unsurprising urban ecologists have spent a great deal of the last two decades testing the utility of a suite of potential biomarkers including physiological and phenotypic proxies of fitness such as the quality or concentration of hormones (e.g. testosterone, Crago et al. (2011)) and organism size and mass (e.g. morphology and body size of European blackbirds, Terdus merula (Linnaeus, 1758, Passeriformes, Turdidae); Evans et al. (2009)).

One biomarker that shows increasing promise is fluctuating asymmetry. This type of asymmetry occurs when developmental errors resulting from exogenous environmental conditions experienced during embryogenesis produce asymmetry in morphological traits normally distributed around a mean of zero (Palmer & Strobeck, 2003). Fluctuating asymmetry occurs in bilaterally symmetrical body parts such as wings (Hoffmann, Collins & Woods, 2002; Hoffmann et al., 2005), head capsules (Leamy et al., 2015), jaws (Wiig & Bachmann, 2014), and scales (Brown et al., 2017). It is distinct from other types of asymmetry including directional asymmetry, where the population mean of the left right differences do not equal zero, and antisymmetry, where asymmetry is the norm but asymmetry towards the left and right sides are equally common, that is normally distributed around a mean of zero (Klingenberg, 2015). Fluctuating asymmetry is often considered a meaningful index of environmentally induced developmental stability (De Anna, Bonisoli-Alquati & Mousseau, 2013) and is associated with several fundamental life history components including growth, fecundity, and longevity (Møller, 1997). Additionally, fluctuating asymmetry occurs across several taxonomic boundaries including plants, vertebrates, and invertebrates (Klingenberg, 2015). Following larval exposure to certain environmental conditions under standardized conditions (i.e. laboratory), several insect taxa may exhibit increased or no change in wing size and/or wing shape fluctuating asymmetry. In the bumblebee Bombus impatiens (Cresson, 1863, Hymenoptera, Apidae), for example, exposure to CO₂ increases wing shape asymmetry (Klingenberg et al., 2001), whilst low temperatures increase wing shape but not wing size asymmetry in the noctuid moth, Helicoverpa punctigera (Wallengren, 1860, Lepidoptera, Noctuidae), (Hoffmann, Collins & Woods, 2002). Furthermore, pollen deprivation has no impact on wing size asymmetry in neither male or female honey bees (Apis mellifera) (Szentgyörgyi et al., 2017).

Attempts to quantify fluctuating asymmetry in the field and associate variation with environmental conditions including gradients of contamination or urbanization have yielded mixed results (Nunes, Araújo & Marchini, 2015; Weller & Ganzhorn, 2004; Lutterschmidt, Martin & Schaefer, 2016; Banaszak-Cibicka et al., 2018). Indeed, red mason bees (Osmia bicornis, Linnaeus 1758, Hymenoptera, Megachilidae) collected along a zinc, cadmium, and lead contamination gradient in Poland show no difference in wing size or shape asymmetry (Szentgyörgyi et al., 2017). Similarly, euglossine bees (Eulaema nigrita, Lepoletier 1841, Hymenoptera, Apidae) collected from agricultural landscapes where pesticide use is prevalent did not record higher fluctuating asymmetry than individuals collected from a tropical savannah (Pinto et al., 2015). In carabid beetles, individuals display a species dependent effect of urbanization on elytra size asymmetry with ubiquitous species like Nebria brevicollis (Fabricuis, 1792, Coleoptera, Carabidae), and Leistus rufomarginatus (Duftschmid, 1812, Coleoptera, Carabidae) exhibiting no effect, yet other species including Platynus obscurus (Herbst, 1784, Coleoptera, Carabidae) and Leistus ferrugineus (Linnaeus 1758, Coleoptera, Carabidae) showing increased asymmetry with decreasing distance towards the city centre (Weller & Ganzhorn, 2004). Quantifying changes (if any) in the fluctuating asymmetry of species encountering urban areas and exposed to anthropogenic stressors of varying intensities is a critical step in determining the utility of morphometric techniques and the candidate species most appropriate for assessing asymmetry variation.

We assessed fluctuating asymmetries in wing size and wing shape, among adult western honey bees (A. mellifera) collected from colonies located along an urbanization gradient in the Greater Sydney Region. The western honey bee is an excellent model organism to study the effects of anthropogenic stressors and urbanization more generally due to its ubiquity across different landscape types including cities. Furthermore, wing asymmetry is common in western honey bees and frequently used to investigate the consequences of ploidy and hybridization (Smith, Crespi & Bookstein, 1997). We predicted wing shape and size fluctuating asymmetry would increase as anthropogenic stressors including anthropogenic land cover, road proximity, and housing density increased. We also sampled anthropogenic stressors at multiple spatial scales, allowing us to test how the spatial scale at which data are recorded influence associations between anthropogenic stressors and fluctuating asymmetry. We predicted the effects of anthropogenic stressors on fluctuating shape and size asymmetry would be consistent at both fine and coarse spatial scales. Multi-scale approaches like this should be a key consideration in urban ecology given they can substantially affect the interpretation of biological patterns. For example, urbanisation at fine (i.e. 200 × 200 m plots) but not coarse (i.e. 3 × 3 km) spatial scales explains variation in the fitness (i.e. survival) of Lasiommata megera (Linnaeus, 1767, Lepidoptera, Nymphalidae) caterpillars (Kaiser, Merckx & Van Dyck, 2016).

Methods

Sample collection and measurement

We sampled 22 permanent disease-free colonies managed by amateur bee keepers across several local government areas within the Greater Sydney Region (Fig. 1). All colonies were located within the Sydney Basin Bioregion, an area characterized by temperate climatic conditions and sandy soils derived from Hawkesbury Sandstone (NPWS, 2003). Mean annual temperature in this region is 16 °C and 1,000 mm average annual rainfall. In each case, bee keepers collected between 10 and 15 dead bees found at the colony entrance (i.e. landing board) each morning over a week-long collection period, from the 8th to the 14th of February 2016. The disposal of dead nest mates from within the colony by honey bee ‘undertakers,’ is a common behavioural response that facilitates colony hygiene (Visscher, 1983). These dead bees are typically older and have likely transitioned through all stages of the honey bee temporal caste system including cell cleaner, nurse, and forager (Langstroth, 1857). By using these bees, we attempted to reduce temporal dependent differences in exposure to environmental stressors during development. Asymmetry and, especially, fluctuating asymmetry appears during development and is believed to reflect developmental instability (Palmer, 1996). After collection by bee keepers, bees were stored in ethanol until analyses.

Before morphometric analyses, the forewings of 10 randomly selected individuals per colony were detached at the base of each wing, placed on a microscope slide and secured with a coverslip. Bees with damaged wings (i.e. wings where all 19 wing vein junctions were not present) were excluded. Asymmetry has been verified in small sample sizes including <10 individuals (Palmer, 1996). Each slide was photographed at 12.5 × magnification using a compound light microscope and microscope Camera Leica MC170 HD attachment. Using tPSDig software (Rohlf, 2001), we digitized photographs and collected a set of 19 landmarks positioned at wing vein junctions (Fig. 2). To reduce measurement error, this process was repeated, independently, two times and the average taken. Right wings were also mirrored to be comparable to left wings. In all cases, measurements were made by the same person.

Results

Wing shape asymmetry

For the three km buffer, colony-road proximity, anthropogenic land area, vegetation area, and number of people significantly predicted wing shape asymmetry (F_(5,23) = 3.17, p = 0.04, R² = 0.43; Table 1). However, only nearest straight-line distance from colony to road added significantly to the prediction, p = 0.01 (Table 1). Hives closer to roads contained bees with higher wing shape asymmetry values (Fig. 3).

Table 1:

Parameter effect estimates.

Buffer	Explanatory term	Beta coefficient	t-value	p-value
Three km	Nearest straight line distance from hive to road	−0.60	−2.79	0.01
	Anthropogenic land area	−0.43	−1.52	0.15
	Number of people	0.05	0.15	0.88
	Vegetation area	−0.39	−1.36	0.19
500 m	Nearest straight line distance from hive to road	−0.57	−2.69	0.02
	Total road length	0.21	1.06	0.30
	Anthropogenic land area	−0.57	−2.69	0.02
	Number of dwellings	−0.29	−1.25	0.23
	Vegetation area	−0.57	−2.15	0.047

DOI: 10.7717/peerj.5940/table-1

Note:

Parameter effect estimates (with associated t-statistic and p-values) for variables collected from three km and 500 m buffers surrounding each hive and included in multiple linear regressions analysing variation in Procrustes distance.

For the 500 m buffer, colony-road proximity, anthropogenic land area, vegetation area, total road length, and number of dwellings significantly predicted wing shape asymmetry (F_(5,21) = 4.34, p = 0.01, R² = 0.58). Colony-road proximity, vegetation area, and anthropogenic land area all added statistically significantly to the prediction (Table 1). Hives closer to roads contained bees with higher wing shape asymmetry values (Fig. 3). Hives surrounded by less vegetation contained bees with higher wing shape asymmetry values (Fig. 4A), and colonies surrounded by more anthropogenic land area contained bees with lower wing shape asymmetry values (Fig. 4B).

Discussion

Colony location significantly affected both wing size and wing shape asymmetry. Furthermore, the effect of specific landscape features on wing shape asymmetry differed depending on the spatial scale considered. This finding is novel and highlights the important spatial scale considerations needed to assess the applicability of biomarkers including fluctuating asymmetry in the future. In our study, within the three km buffer, the only factor adding significantly to the model was colony-to-road proximity whilst, within the 500 m buffer, colony-to-road proximity and two additional factors (anthropogenic land area and vegetation area) added significantly to the model. Ultimately, our findings indicate colonies located in areas surrounded by more urban features (e.g. proximal roads, less vegetation) contain bees with higher wing shape asymmetries. Whilst these findings differ from those conducted on non-Apis bees (E. nigrita) (Pinto et al., 2015) our results are consistent with those of previous work demonstrating increased wing asymmetry in western honey bees inhabiting urban compared to less urban forested areas (Nunes, Araújo & Marchini, 2015).

Wing shape differences have previously been reported between western honey bees collected from several forest and contaminated landscape populations, including urban areas (Nunes, Araújo & Marchini, 2015). These studies grouped sites categorically as either urban or rural (Nunes, Araújo & Marchini, 2015; Silva et al., 2009). To appreciate the more nuanced effects site specific differences may have on wing asymmetry, we collected continuous measures of several landscape traits typically favoured or lost through urbanization. We suspect using continuous rather than categorical predictor variables may be more appropriate especially when the focal species is likely to respond to fine scale changes in habitat including vegetation and impervious surface. Future studies would benefit from including variables collected over fine spatial scales such as the diversity of vegetation patches directly surrounding colonies, and quality of food resources.

We found associations between road proximity and wing shape asymmetry. Similar results have been reported in the western fence lizard, Sceloporus occidentalis (Baird & Girard, 1852, Squamata, Phrynosomatidae), with individuals inhabiting sites containing off-highway vehicular roads reporting higher bilateral scale pattern asymmetry (Tull & Brussard, 2007). The specific mechanism(s) associated with associations between how proximal colonies were to roads and wing shape asymmetry in honey bees remains unclear. From a physiological perspective, we suspect the air pollutants occurring near roads and associated with road use (e.g. oxides of nitrogen and carbon) may be important. Carbon dioxide, a common traffic-related air pollutant, for example, significantly affects development in the moth Manduca sexta (Linnaeus, 1763, Lepidoptera, Sphingidae), increasing egg development time parabolically (Kerr, Phelan & Woods, 2013). Furthermore, CO₂ exposure increases wing shape asymmetry in the bumblebee B. impatiens (Klingenberg et al., 2001).

Why wing asymmetry sizes and wing shape asymmetries differed between colony locations remains enigmatic, however, we suspect features of the hive and/or management practices used by bee keepers may be important. Practices increasing pesticide exposure risk including frequent spraying and/or multi-pesticide use are associated with increases in asymmetry in several bee species including the western honey bee (Abaga et al., 2011) and Scaptotrigona aff. Depilis (Moure, 1942, Hymenoptera, Apidae) (De Souza Rosa et al., 2016).

Given associations between wing shape asymmetry and vegetation and anthropogenic land area were detected at 500 m but not three km we suspect wing shape asymmetry may be resistant to changes in landscape at coarser spatial scales. This could also reflect decreased food quality and/or lower diversity and abundance of flowering plants directly surrounding each colony (i.e. within a 500 m radius). The fact wing size asymmetry was unaffected by landscape traits at either spatial scale, is unsurprising. This measure is generally constant between populations exposed to several environmental stressors including CO₂ and cold temperatures (Hoffmann, Collins & Woods, 2002).

Conclusion

Overall the present results indicate wing shape asymmetry in western honey bees may be related to how proximal colonies are to roads and depending on the spatial scale considered (e.g. 500 m vs. three km), other habitat variables including vegetation area and anthropogenic land area. While additional environmental conditions may also have contributed to the fluctuating asymmetry observed (e.g. temperature, pesticides, or parasitism), our findings provide support for an effect of roads. Given that roads generate light, noise, and air pollution and increase edge effects by dividing habitat, combinations of these factors should be incorporated into future observational studies. It is difficult to avoid confounding factors in such studies, however, combinations of field and manipulative laboratory studies (e.g. greenhouse reared colonies are exposed to known concentrations of traffic related air pollutants) increase the confidence with which we can establish a cause-effect relationship. One additional avenue for future work is the potential fitness consequence of the asymmetry observed. To our knowledge, this is the first study to assess the effects of spatial scale on associations between measures of wing asymmetry and landscape traits associated with urbanization. Understanding how spatial scale and the factors favoured or lost through urbanisation affect biomarkers like wing asymmetry is an important step required to quantify the utility of morphometrics for assessing urbanization impact.

Supplemental Information