For many animals, abnormal temperatures during development, either elevated or depressed, are known to have profound impacts on adult phenotype. This sometimes takes the form of phenotypic plasticity, such as the reptile temperature-dependent sex determination systems (Singh et al., 2020) or the spring/summer morphs of certain butterfly species (Gilbert, 2001). Other times, deviation from an optimum developmental temperature is negative, causing adult morphological deformation (Groh et al., 2004; Gerard et al., 2018), behavioural impairment (Van Damme et al., 1992; Tautz et al., 2003), deficient neurological development (Groh et al., 2004) and physiological disruption (Watkins, 2000; Singh et al., 2020).

Across exothermic vertebrates, non-adaptive deviations in adult phenotypes due to abnormal developmental temperatures have been well investigated when temperature stress occurs during the entirety of development (Burger, 1991; Van Damme et al., 1992; Watkins, 2000; Micheli-Campbell et al., 2011; Sfakianakis et al., 2011; Singh et al., 2020). Zebra fish embryos reared in warmer temperatures swim better as adults than their cooler compatriots, but show no discernible morphological differences (Sfakianakis et al., 2011). In contrast, hatchlings of the turtle Elusor macrurus from cooler incubation temperatures are the better swimmers (Micheli-Campbell et al., 2011; Singh et al., 2020). Even in insects, such as damselflies, exposure to elevated temperature throughout development has been shown to affect both wing shape and flight performance (Arambourou et al., 2017). The solitary bee Osmia bicornis suffers increased adult mortality, as well as decreases in prepupal weight, under hotter developmental temperatures (Radmacher and Strohm, 2011).

Exposure to elevated ambient temperatures has also been shown to affect bee colonies, albeit in a species-dependant manner. Honeybees that experience heat stress during development do not show any morphological indicators (Jones et al., 2005) (though see Groh et al., 2004 for reports of morphological disfigurement) but display learning and memory impairment (Tautz et al., 2003). Furthermore, adult honeybees reared under naturally occurring but undesirable temperatures had fewer mushroom body microglomeruli in the region dedicated to olfactory learning (Groh et al., 2004), suggesting that their capacity to learn olfactory stimuli would be impaired. Bumblebees subjected to heat stress over the duration of development have modified wing shapes as adults (Gerard et al., 2018), although it remains unknown if these or any other potential morphological variations have any behavioural consequences.

While we now have a good general understanding of how exposure to abnormal temperatures during the entirety of development affect a range of organisms, we still know little about the effects of abnormal temperature exposure during only a part of development. Understanding such effects, however, is becoming increasingly important if we are to understand how animals will be impacted by global warming. Global warming is increasing the frequency of heatwaves (Perkins-Kirkpatrick and Lewis, 2020) – prolonged periods of excessive heat (Perkins and Alexander, 2013; Perkins-Kirkpatrick and Lewis, 2020) – a weather phenomenon that may only impact an animal over a portion of their development. Animals like bumblebees that grow their colonies over spring and summer, when extreme heat events are most common, are particularly vulnerable to excessive temperature events. While the influence of elevated temperatures over a fraction of development is relatively unknown, investigations into the impacts of heatwaves specifically indicate they may have a profound effect on all organisms (Stillman, 2019).

Short-term heat-shock in insects often causes reproductive disruptions (Sales et al., 2018; Chen et al., 2019; Martinet et al., 2020), particularly impacting male fertility by altering spermatozoa (Sales et al., 2018; Martinet et al., 2020). In O. bicornis, exposure to elevated temperatures during pupation only disrupted male mate attraction behaviour (Conrad et al., 2017). In eusocial insects, capable of intranidal temperature regulation, the effect of partial developmental heatwaves has only been investigated in Africanized honeybees (Medina et al., 2018; Poot-Báez et al., 2020). A naturally occurring heatwave prior to pupation caused smaller adults (Poot-Báez et al., 2020) whereas a simulated heatwave during pupation affected wing shape, altered the strength of fluctuating asymmetry and lowered the age at onset of foraging (Medina et al., 2018).

The stochasticity of heatwaves means that eusocial insect larvae will be at different developmental stages at the onset of a heatwave and will therefore experience differential exposure to elevated temperatures. Having uneven exposure to elevated temperatures may still affect colony performance. Here, we begin to address the functional consequences of elevated temperatures during varied periods of mid and late development in the bumblebee Bombus terrestris. Bumblebees are an important model for this work because they tightly regulate both their colony (Schultze-Motel, 1991; Jones and Oldroyd, 2006; Gardner et al., 2007) and brood temperatures (Vogt, 1986; Weidenmüller et al., 2002). The proper development and function of sensory systems in bumblebees is essential for their pollination activities as they rely heavily on visual, olfactory, gustatory and mechanosensory information to navigate between flower patches and their colony (Dyer and Chittka, 2004; Orbán and Plowright, 2014; Wilmsen et al., 2017; Sprayberry, 2018) and to optimise their foraging (Spaethe et al., 2001; Chittka et al., 2003; Kulahci et al., 2008). Furthermore, bumblebees are already experiencing range shrinkage and population declines in both Europe (Kerr et al., 2015) and North America (Cameron et al., 2011; Kerr et al., 2015) as a result of a warming climate. The capacity for this genus to adapt to hotter temperatures may be physiologically constrained; Bombus vosnesenkii workers show no geographical variability regarding their critical thermal maximum (Pimsler et al., 2020), indicating a lack of local adaptation. We predicted that elevated temperatures over a portion of development, akin to a partial developmental heatwave, would impair behavioural responses to relevant sensory stimuli and cause morphological changes in adult worker bumblebees. Our results indicate that, despite having no discernible impact on morphology, partially elevated temperatures impair bumblebee responses to important sensory information, likely leading to decreased colony fitness.

There was no significant interaction between temperature and stimulus type (LRT, X₆ = 4.04, p = 0.62) nor was there a significant effect of fresh mass (LRT, X₁ = 0.02, p = 0.89). There was no significant effect of days taken to eclose or days to onset of behavioural testing (LRT, X₁ = 3.27, p = 0.07) There was a significant effect of both temperature (type II ANOVA, X_59,1 = 5.22, p < 0.03) and stimulus type on the likelihood of bees responding in accordance with the expected criteria (type II ANOVA, X_59,6 = 0.03, p < 0.001).

Post-hoc pairwise comparisons revealed that bees reared at 26°C made a greater number of expected responses compared with bees reared at 33°C (Figure 2, Z = 2.29, p = 0.02). Irrespective of temperature treatment, bees were more likely to respond in accordance with the expected criteria to UV light, honey water odour, quinine, an air puff and brown sugar water than they were to white sugar water and white light (Figure 2, Z > 3.73, p < 0.01).

There was no significant difference in the mean size of any of the measured morphological features (antennae length, eye area, forewing length, forewing width, proboscis length, body mass) between temperature treatments (Figure 3, T-test, t_{22–60,17–58}, < 1.85, P > 0.17).

We explored the functional consequences of elevated rearing temperatures during a period of development – such as those likely to be induced by heatwaves arising from global warming – on bumblebee behaviour by testing their initial responses to various novel and familiar sensory stimuli (Table 3). Bees that spent a portion of their development time at 33°C made 8% more unexpected responses than those reared at 26°C, a difference that would represent a significant negative effect on behaviour. Though these bees were naïve and tested in laboratory conditions, it is not implausible to posit that this could lead to many incorrect choices when extrapolated across the lifetime of a colony. Although maladaptive responses to relevant sensory stimuli could potentially be rectified by experience, previous work showing that honeybees raised at elevated temperatures have impaired memory (Tautz et al., 2003; Jones et al., 2005), suggests that this process may also be affected. The effect of elevated temperatures on the response to sensory stimuli was independent of the time the bees spent at that temperature. This suggests that elevated developmental temperatures can disrupt adult behaviour, independent of their duration – a shorter period of exposure to elevated temperatures is as detrimental as a longer period. This effect of elevated temperature on behavioural responses to sensory stimuli did not appear to be related to morphology; we found no differences between body masses or the size of important sensory structures such as antennae, eyes, wings or proboscides between the two treatment conditions (Figure 2).

Morphological changes, such as organ shape (Gerard et al., 2018), body size (Gerard et al., 2018) or levels of fluctuating asymmetry (Klingenberg, 2015) are often employed as indicators of environmental stress. We found that behavioural impairment occurred without a morphological correlate, indicating that the approach of using morphological differences to indicate environmental stress alone is not necessarily sufficient and other factors, such as behaviour, should be considered during such investigations. The elevated temperatures presented during the mid- and late stages of development in this study did not induce any morphological changes, unlike in Gerard et al. (2018), where a constantly elevated temperature throughout development caused changes to wing morphology. Since we exposed bumblebees in later stages of development to heatwave conditions, it is likely that body size and relative organ sizes were already developmentally fixed (Tian and Hines, 2018). A useful extension of this study would be to examine the effect of elevated temperatures during early development and assess the extent of behavioural and morphological alteration.

The impairments in behaviour observed here, as a result of a heatwave-like event, likely have a neurological, rather than morphological basis. It is unclear if the impairment lies in sensory structures, ascending or descending neurons or in areas of the brain. Tautz et al. (2003) and Jones et al. (2005) concluded that decreased temperature during development caused subtle neurological changes that impaired short-term memory in honeybees. Further evidence for this can be found when examining mushroom bodies of honeybees reared outside of their optimum temperature (Groh et al., 2004). The number of microglomeruli (distinct synaptic complexes) fell as rearing temperature deviated away from the naturally held optimum. This effect was most pronounced within the olfactory input region, while the visual region of the mushroom body was less affected (Groh et al., 2004). Our data are consistent with these findings – a subtle neurological impairment could explain the poorer responses observed in bumblebees reared partly at 33°C. Further, short-term memory in honeybees (Tautz et al., 2003; Jones et al., 2005) was examined via the proboscis extension reflex, a common method for establishing learning and memory in bees (Giurfa and Sandoz, 2012). One of the responses of the bumblebees in this study was the extension of their proboscides (Table 2), indicating that the short-term memory issues reported in honeybees reared at abnormal temperatures might be mechanistically or neurologically similar to the unexpected behavioural responses we observe.

Although we generally observed an effect of elevated developmental temperature on the behavioural responses of bees, there was variation in the extent of the impairment depending on the sensory stimulus. For example, the proportion of expected responses to brown sugar water was similar in both temperature groups (85% at 33°C, 75% at 26°C), while the response differed more markedly in response to white sugar water (70% at 33°C, 50% at 26°C). It is unclear why there was a difference in the response to these two stimuli but it possibly relates to the fact that the brown sugar water had a “honey-like” odour while the white sugar water did not. There was also a difference in how the two temperature groups responded to the two light stimuli that were presented. The bees in this study were reared in the dark and had thus never seen light before, which may explain the higher proportion of unexpected responses to white light. Though, this does not explain the higher proportion of expected responses to UV light, which was equally novel to them. As the stimuli were always presented in the same order, it is also possible that the differences in the proportion of expected responses are related to the order of stimulus presentation. Nonetheless, this does not explain differences between the proportion of responses observed in the two temperature groups to each stimulus as both groups experienced the same order of presentation. Further experimentation and analysis would be required to ascertain why bees reared under elevated temperatures were better able to respond to some stimuli than others.

It is unclear at what temperature B. terrestris colonies begin to experience heat stress. The value appears to be somewhere between 26°C (Weidenmüller et al., 2002) and 32°C (Vogt, 1986). Whether the bees or the colonies were indeed under heat stress in this study is somewhat moot concerning what our results indicate – the elevated developmental temperatures were sufficient to induce an effect on adult behaviour. Heat stressed brood are an obvious and likely explanation behind our results, however, even if colonies were not under heat stress, the indication that sub-stressful elevated temperatures can cause behavioural impairment should be troubling to anyone paying attention to current trends in global climate.

Our data also do not indicate whether the elevated temperature experienced by the colony was the direct cause of the behavioural impairment we observed. Bumblebees can regulate the air and the brood temperatures independently (Vogt, 1986). We measured only the ambient temperature inside the colonies and did not directly measure the brood temperature; therefore it is possible that the brood were kept above or below the temperature set by the incubator, despite the air temperature within the colonies being consistently 33°C or 26°C. The temperature of the brood would need to be directly measured to assess the capacity of the bees to combat the overwhelming air temperature. If the brood are reared at temperatures different from ambient, it would indicate that there are indirect temperature effects on developing bees, likely a result of modified action by nurses. It is known that increased thermoregulatory behaviour lessens the attention that nurses provide to brood (Vogt, 1986). The observed behavioural impairments may therefore be caused directly via the temperature of the brood or indirectly via temperature causing disruption across the whole colony. Either proximate mechanism indicates that elevated temperatures over portions of development create behavioural impairments in adult bumblebees.

Our finding that bees experiencing periods of elevated temperature during development make sub-optimal responses to relevant sensory stimuli is more evidence of the dangers of climate change to heterothermic animals, which are susceptible despite their capacity to independently regulate their temperature. Bees in our experiment experienced elevated temperatures during their mid- and late developmental stages, mimicking heatwave periods, the risk of which are increasing across the northern hemisphere (Qiu and Yan, 2020) and increasing in frequency and magnitude across the globe (Perkins-Kirkpatrick and Lewis, 2020). Therefore, our data provide an indicator of how current and near-future climate change may affect bumblebees. Responding poorly to stimuli, relative to bees reared in more optimal temperatures, could come at a greater energetic cost and result in early worker death, which is a source of stress for a colony (Bryden et al., 2013). Furthermore, a reduced capacity to make appropriate responses to sensory stimuli will affect the speed at which correct decisions are made, impacting foraging efficiency and therefore colony fitness (Spaethe et al., 2001; Chittka et al., 2003; Kulahci et al., 2008).

Our findings add to an existing body of work that indicates that global warming will have a significant effect on pollinators (Shrestha et al., 2018), though few experiments have explicitly examined the effects of elevated rearing temperature on adult behaviour. The impairments presented here show that elevated rearing temperatures affect innate behavioural responses to sensory stimuli. Our experiment tested basic behavioural responses but the results indicate that elevated rearing temperatures are likely to affect more complex behaviours, as well as learning and memory (Tautz et al., 2003; Jones et al., 2005). Further investigations are necessary to understand if these behavioural impairments are mitigated by age, the effects of increased learning and to uncover the mechanisms by which impairments occur.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

MG, ZJ, CR, and EB designed the study. CP, ZJ, VW, ZM, MG, and CR collected the data. AM acquired data and provided technical micro-CT expertise. CP performed the analysis and drafted the manuscript with EB. All authors critically revised the manuscript and gave approval for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.