Reduced thermal variability in cities and its impact on honey bee thermal tolerance

Introduction

Urbanization is one of the most profound drivers of current environmental change (Shochat et al., 2006; Grimm et al., 2008). Currently, more than half of the human population lives in cities, and this figure is expected to rise in the upcoming decades, particularly in developing countries of Africa, Asia and Latin America, resulting in an expansion of urbanized areas worldwide (United Nations, 2014). Thus, it is important to investigate species’ responses to such rapid changes in order to understand the present and future impacts of urbanization. The number of studies targeting the ecological impact of urban areas on biodiversity has increased in the past few years, most of them examining the effect of urbanization on species assemblages with a focus on the role of the urbanized landscape as a filter for the community (Chown & Duffy, 2015; Diamond et al., 2015; Harrison & Winfree, 2015; Ouyang et al., 2018). However, many species can occur in both rural and urban environments, and the consequences of the urban environment on the responses of these populations (e.g., behavioral and physiological) are poorly understood (Diamond et al., 2017; Banaszak-Cibicka et al., 2018).

One of the most distinct consequences of urbanization, and a clear example of anthropogenic microclimate modification, involves a decline in thermal variability and an increase in average ambient temperature in cities when compared to surrounding rural areas (Oke, 1982; Pickett et al., 2001; Gaston, Davies & Edmondson, 2010). This urban heat island effect results primarily from the removal of vegetation and a higher absorption of solar radiation due to impervious surfaces such as buildings, parking lots and roads, and also from direct heat generation related to human activity as well as reduced wind speeds because buildings restrict air circulation (Rizwan, Dennis & Liu, 2008). The duration and magnitude of the temperature differential between urban and rural areas depend on city size, population density and the spatial heterogeneity of the urban landscape, mainly related to the percentage of human-made surfaces (Pickett et al., 2001), and vegetation type and cover (Imhoff et al., 2010). Although urban heat islands have been noticed since 1833 (Howard, 1833), studies of their effects on animal thermal responses are remarkably limited (Brans et al., 2017; Diamond et al., 2017; Hamblin, Youngsteadt & Frank, 2018).

Insects are highly dependent on ambient temperature to ensure their biological functions (Angilletta, 2009). Their capacity for thermal regulation is limited, and their persistence in the environment and resistance to unfavorable temperatures depend in large degree to their thermal tolerance (Chown & Nicolson, 2004). Thermal tolerance is a trait delineated by the coldest and hottest temperatures at which organisms can maintain muscle control (critical thermal minimum, ‘CT_min’ and critical thermal maximum, ‘CT_max’, respectively). It can be modified through short-term processes such as phenotypic plasticity (e.g., acclimation) and long-term processes such as evolutionary adaptation (Chown & Nicolson, 2004; Angilletta, 2009). The thermal tolerance breadths of organisms (i.e., the difference between CT_max and CT_min) often match the temperature extremes they are likely to encounter. The evidence supporting this pattern comes from studying species across latitudinal and altitudinal geographical gradients, with thermal tolerance limits being narrower for organisms living in lower latitudes and elevations due to their reduced seasonal and diurnal temperature fluctuations compared to environments with more variable climatic conditions (higher latitudes and elevations) (Janzen, 1967; Chown & Terblanche, 2006; Liefting, Hoffmann & Ellers, 2009; Overgaard et al., 2011; Sunday, Bates & Dulvy, 2011; Araújo et al., 2013).

Honey bees are heterothermic insects that are present across urban and rural environments throughout the world (Cane, 2005) and are being affected by climate change (Le Conte & Navajas, 2008). Individuals are born as strict stenothermic larvae completely dependent on the adult bees for heat (Heinrich, 1993). Newly born adult females perform hive duties in the brood nest where temperatures are maintained near 34 °C (Jones et al., 2005). Honey bee females develop the capacity to generate endothermic heat from muscle contractions and begin foraging after a few days of hatching as adults (Free & Spencer-Booth, 1960; Heinrich, 1993). Foragers are capable of regulating their body temperature across a range of ambient temperatures (Heinrich, 1980). Honey bee foragers lose neuromuscular activity (i.e., enter into chill-coma) when thorax temperatures are below their CT_min (∼10 °C) (Esch, 1988; Goller & Esch, 1990a; Goller & Esch, 1990b) and above their CT_max (∼50 °C) (Kovac et al., 2014). Thermal breadth can however be altered by plasticity (e.g., acclimation) (Free & Spencer-Booth, 1960; Free & Spencer-Booth, 1962) and evolutionary change (Chuda-Mickiewicz et al., 2009; Kovac et al., 2014) on different time scales. Below their CT_min, foragers cannot generate heat for colony thermoregulation and flight initiation, and have reduced survival if exposed to these low temperatures for prolonged periods (Free & Spencer-Booth, 1960; Esch, 1988; Goller & Esch, 1990a; Heinrich, 1993). Foragers have lower chill-coma temperature than nurses, drones or queens because they stay away from the brood nest and acclimate to the cooler ambient temperatures in the internal peripheral areas of the hive, particularly at night (Free & Spencer-Booth, 1960; Goller & Esch, 1990a; Goller & Esch, 1991).

We investigated whether reduced thermal variability and an increase in ambient temperature caused by urbanization can influence the limits of thermal tolerance in the honey bee (Apis mellifera L.), one of the world’s most important crop pollinators (Klein et al., 2007). We hypothesized that ambient temperature in our urban study site would be less variable and on average higher compared to rural settings. As a consequence honey bees from urban areas would exhibit narrower climatic tolerance breadths, with higher tolerance to extreme warm temperatures and lower tolerance to extreme cold temperatures than bees from rural areas. We also investigated whether the difference in thermal sensitivity between urban and rural bees resulted from an acclimation response. Since larger body size individuals are known to be able to tolerate more extreme temperatures (Pereboom & Beismeijer, 2003; Oyen, Giri & Dillon, 2016; Peters et al., 2016), honey bee body size was measured.

Materials & Methods

Site selection

Our study was conducted in the metropolitan area of Pachuca, Hidalgo, Mexico, which covers approximately 100 km². It is located between 19°50 and 20°10N, and between 98°41 and 98°57W, at 2,400–2,800 m above sea level. It has a population of 512,196 inhabitants, and has one of the fastest growing rates in the country (INEGI, 2010). It has a semi-arid climate, with a mean annual temperature of 15 °C and a mean annual precipitation of 367.6 mm (Gómez-Aíza & Zuria, 2010). We randomly selected 8 urban and 8 rural sites separated by at least 1 km (Fig. 1) (Waddington et al., 1994; Baum et al., 2011; Leonard et al., 2018). We measured the percentage cover of impervious surfaces (houses, buildings, roads, parking places) in urban and rural sites in buffers of 500 m radius, centered at each sampling site, using a 2015 WorldView-2 (Digital Globe) high-resolution satellite image and ArcGis (ver. 10.2, ESRI). For the purpose of this study, we defined urban sites as those having more than 60% impervious cover (72.4 ± 2.1, mean ± se), while rural sites had less than 20% impervious surface (11.4 ± 2.5, mean ± se) (Fig. 1).

Results

Ambient temperature evaluation

We could only recover dataloggers or ambient temperature data form 10 of the 16 sites (5 urban and 5 rural). Ambient temperature range size was significantly narrower in urban compared to rural sites (t = −2.45, d.f. = 8, p = 0.040) (Fig. 2A). The narrower ambient temperature range in the urban landscape did not result from differences in mean daily maximum temperatures among urban and rural sites (t = −0.54, d.f. = 8, p = 0.610) (Fig. 2B), but from differences in higher mean daily minimum temperatures in the urban areas compared to the rural sites (t = 3.48, d.f. = 8, p = 0.008) (Fig. 2C). Average daily temperatures were significantly higher in urban compared to rural sites (t = 3.99, d.f. = 8, p = 0.004) (Fig. 2D).

Maximum and minimum temperature limits and acclimation effects on thermal tolerance

Honey bees from urban sites had significantly narrower tolerance breadths than bees from rural sites (t = −2.53, d.f. = 14, p = 0.024) (Fig. 2E). CT_max did not differ significantly between urban and rural bees (t = −0.78, d.f. = 14, p = 0.448) (Fig. 2F), but CT_min (log₁₀ transformed) (Fig. 2G) was significantly higher in urban compared to rural bees (t = 2.44, d.f. = 14, p = 0.029).

CCRT (log₁₀ transformed) was affected by acclimation temperature (F = 30.63, d.f. = 3,45, p = 0.0001). It took bees exposed to 20 °C significantly less time to recover from chill-coma than bees exposed to 24 °C, and bees exposed to 20 and 24 °C had significantly lower CCRTs than bees exposed to 34 °C (Fig. 3). There was no significant relationship between honey bee body mass and CT_min (r = 0.029, n = 16, p = 0.915), CT_max (r = 0.023, n = 16, p = 0.934), and thermal tolerance breadth (r = 0.005, n = 16, p = 0.986). Body weight did not differ significantly between urban and rural bees (t = 0.153, d.f. = 8, p = 0.881).

Discussion

The relationship between the temperature extremes to which individuals are exposed and their thermal tolerances has been mainly documented in studies conducted across large-scale temperature gradients, with thermal tolerance breadth being narrower in environments with reduced temperature fluctuations (e.g., lower latitudes and elevations) compared to environments with more variable climatic conditions such as higher latitudes and elevations (Angilletta, 2009; Araújo et al., 2013). Our overall results show that differences in thermal variability along smaller spatial scales such as urban-rural gradients can influence species’ thermal tolerance breadths.

The decrease in ambient temperature variability and increase in average ambient temperature in urban sites compared to rural sites found for the metropolitan area of Pachuca coincide with results documented in other cities (Oke, 1982; Gaston, Davies & Edmondson, 2010; Imhoff et al., 2010). Similar maximum daytime temperatures in urban and rural habitats have been documented for cities with arid and semi-arid climates due to scarce vegetation and the presence of bare soil exposed to solar radiation in the rural landscapes (Shepherd, 2006; Bounoua et al., 2009; Imhoff et al., 2010), as is the case for the open scrub forest in the rural areas of the metropolitan area of Pachuca (Carbó-Ramírez & Zuria, 2011; Sánchez-Echeverría, Castellanos & Mendoza-Cuenca, 2016). Differences in ambient temperature range across urban and rural sites in our study site paralleled those of honey bee thermal tolerance breadth, and resemble a pattern found across large geographic areas (Chown & Terblanche, 2006; Araújo et al., 2013; Sunday et al., 2014). Further, studies along geographical gradients have shown in terrestrial insects that narrower thermal tolerance breadths in more stable climates (lower latitudes and altitudes) result from changes in CT_min, rather than in CT_max (Addo-Bediako, Chown & Gaston, 2000; Overgaard et al., 2011; Hoffmann, Chown & Clusella-Trullas, 2013; Bozinovic et al., 2014), a macrogeographic pattern that parallels the microgeographic pattern observed among honey bees in the metropolitan area of Pachuca.

The empirical studies that have reported that thermal performance and tolerance of terrestrial animals differ between urban and rural sites have documented higher CT_max in urban compared to rural individuals (Angilletta et al., 2007; Diamond et al., 2017; Warren, Bayba & Krupp, 2018), which differs from our results with honey bees. The discrepancy between our results and those reported in these studies is likely due in part to differences in daily maximum temperatures among urban and rural habitats. The three aforementioned studies were conducted in cities surrounded by high biomass forest vegetation, where urban habitats tend to have higher maximum temperatures than rural habitats (Imhoff et al., 2010), which can lead to urban individuals having higher thermal tolerances than rural individuals (Angilletta et al., 2007). In contrast, urban and rural sites in arid and semiarid cities with low biomass vegetation tend to have similar maximum ambient temperatures (Imhoff et al., 2010), which could lead to similar higher thermal tolerances in urban and rural individuals, as was found in our study for honey bees. In addition, the maximum ambient temperatures registered in our urban and rural sites were considerably lower than the CT_max reported for honey bees (Atmowidjojo et al., 1997; Kovac et al., 2014); this study), while the maximum ambient temperatures registered in the urban sites in the previous three studies did overlap with the CT_max reported for ants, which was their study group (Angilletta et al., 2007). More studies investigating the thermal responses of different taxa across cities with different climates will provide more insight into how organisms respond to urban climates (Hamblin et al., 2017; Youngsteadt et al., 2017).

Few studies have reported critical thermal limits of bees, particularly lower critical limits (Free & Spencer-Booth, 1960; Esch, 1988; Goller & Esch, 1990a; Goller & Esch, 1990b; Chuda-Mickiewicz et al., 2009; Oyen, Giri & Dillon, 2016; Peters et al., 2016; Oyen & Dillon, 2018). CT_min is related to body size in some bee species (Heinrich, 1993; Oyen, Giri & Dillon, 2016) however, we did not find a relationship between honey bee body mass and critical thermal limits. Differences in nutritional state can influence lower critical thermal limits in bees (Free & Spencer-Booth, 1960), but see (Oyen & Dillon, 2018). However, honey bee foragers’ nutritional condition (amount of carbohydrates, lipids and proteins) does not vary with urbanization in the metropolitan area of Pachuca (Hernández-Medina, 2018) and we did not find that honey bees from urban and rural sites differ in body size (an indicator of environmental quality in bees; Banaszak-Cibicka et al., 2018). Moreover, some studies have suggested that habitat quality in the city might be better for bees than rural landscapes, since the former may provide with richer and year-round food resources, higher and more stable temperatures, and less environmental pollution (Banaszak-Cibicka et al., 2018). Thus, it is unlikely that differences in nutritional state among urban and rural honey bees can explain differences in honey bee thermal tolerances in our study.

Honey bee foragers have 1–2 °C lower chill-coma temperatures than nurse bees as a result of their acclimation to the lower temperatures in the cooler internal peripheral areas of the nest, particularly at night (Free & Spencer-Booth, 1960; Heinrich, 1993). Winter bees also have 1–2 °C lower chill-coma temperatures than summer bees (Free & Spencer-Booth, 1960). Although urban and rural sites were relatively close to each other (3–10 km), we did find a ∼1 °C difference between the CT_min of urban and rural bees. Thus, it is likely that the lower night ambient temperature found in our rural sites may account for the lower CT_min of rural foragers. This is partially supported by the acclimation effect found in honey bees exposed to laboratory experimental temperatures that differed by 4 °C, which was the difference between the lowest mean temperatures registered in rural and urban sites.

The lower CT_min of rural foragers could allow them to generate endothermic heat from muscle contractions more readily at lower temperatures, be more active, and provide heat for colony thermoregulation. In contrast, it is possible that urban bees may not need to invest as much energy and will consume less honey as a result of experiencing higher minimum and similar maximum ambient temperatures (Free & Spencer-Booth, 1958; Free & Spencer-Booth, 1960; Heinrich, 1993; Vollet-Neto, Menezes & Imperatrix-Fonseca, 2011). Since bees provided with artificial nest heating and lower thermal variability produce more brood, honey, and pollen as a result of reducing the energetic costs in colony thermoregulation (Free & Spencer-Booth, 1958; Wineman, Lensky & Mahrer, 2003; Erdogan, Dodologlu & Emsen, 2009; Vollet-Neto, Menezes & Imperatrix-Fonseca, 2011), future studies should be conducted to determine if the microclimates that result from urban heat islands can enhance honey bee colony performance.

Conclusions

Our results show that differences in thermal variability along small spatial scales such as urban-rural gradients can influence species’ thermal tolerance breadths. We also show that thermal tolerance breadth in urban honey bees can be modified in response to acclimation. Further research addressing the effects of increasing temperatures and decreasing climatic variability caused by urban heat islands on physiological, ecological, and evolutionary processes will permit improved predictions of species responses in urban and future climates.

Supplemental Information