Noxious stimulation induces self-protective behavior in bumblebees

Summary It has been widely stated that insects do not show self-protective behavior toward noxiously-stimulated body parts, but this claim has never been empirically tested. Here, we tested whether an insect species displays a type of self-protective behavior: self-grooming a noxiously-stimulated site. We touched bumblebees (Bombus terrestris) on an antenna with a noxiously heated (65°C) probe and found that, in the first 2 min after this stimulus, bees groomed their touched antenna more than their untouched antenna, and more than bees that were touched with an unheated probe or not touched at all did. Our results present evidence that bumblebees display self-protective behavior. We discuss the potential neural mechanisms of this behavior and the implications for whether insects feel pain.


INTRODUCTION
Nociception is the detection and processing of noxious stimuli 1 and can be identified from recording neural activity or behavior associated with the activation of nociceptive circuits. 2,3Insects have both nociceptors and nociceptive neurons that detect mechanical, thermal, and chemical noxious stimuli, 4,5 and they respond behaviorally by moving away from and avoiding noxious stimuli. 6,7elf-protective behavior-behavior with the aim to protect a body part from further noxious stimulation-is seen in response to noxious stimulation in many species, including humans.Examples include tending to, guarding, self-grooming, or rubbing a noxiously-stimulated body part.In humans, this can be seen, for example, when you grab and rub your bumped toe to reduce the pain caused by the nociceptive processing.In insects, there are no quantitative studies of self-protective behavior (such as self-grooming) directed toward a noxiously-stimulated site. 8In fact, anecdotal reports claim that insects do not protect their injury sites, and that insects continue to walk, feed, and mate normally after injury. 9,10][13] In other animals, self-protective behavior is widely reported.Rats (Rattus norvegicus) rub their face after being injected with a noxious substance 14 and some bird species groom limbs that have been injected with a noxious substance (e.g., Pyrrhura molinae 15 ).There are similar findings of fish (Oncorhynchus mykiss) rubbing an area that was treated with a noxious injection into the gravel and the sides of their tank. 16Some invertebrates have also been observed performing self-protective behavior, in the form of self-grooming a noxiously-stimulated site.For example, Asian shore crabs (Hemigrapsus sanguines) will rub a claw that has been injected with formalin. 17Similarly, shore crabs (Carcinus maenas), 18 prawns (Palaemon elegans), 19 cuttlefish (Sepia pharoaensis), 20 and octopuses (Octopus bocki) 21 will groom or scratch a body part where acetic acid has been applied.The latter will also respond with self-grooming an area on their arm after it was squeezed with serrated forceps for up to 20 s. 22 As noted previously, there is no evidence of insects self-grooming noxiously-stimulated body parts.However, insects are known to selfgroom in non-noxious contexts, for example during general cleaning, 23 and when removing dust particles (e.g., in the German cockroach Blattella germanica 24 ), pollen grains (e.g., in bees 25 ), and parasites such as mites (e.g., in honeybees, Apis mellifera 26 ).Further, after noxious stimulation, insects may also generally groom themselves more all over, or change their grooming pattern.For example, after having their antenna amputated, red mason bees (Osmia bicornis) groom their head and body, although, in the study where this was found, no site-specific measurements were found/taken, nor was there a non-noxious control to compare to. 27ome anecdotal reports claim that insects do not protect their injury sites, 9,10 but there are also some reports suggesting the contrary.Such reports, however, have not yet been supported by quantitative or statistical analyses. 8,28For example, when pinched on the abdominal ll OPEN ACCESS proleg, moth larvae (Manduca sexta) reportedly turn their heads to the wound, and repeatedly touch the area with their mouthparts, but this behavior was not measured or compared to a control. 29The same was the case in a study on cockroaches (Periplaneta americana) which were reported to groom their wounds following an abdominal puncture. 30Since both reports of the absence and the existence of self-protective behavior in insects are not supported by quantitative measurements or analyses, 28 a robust, experimental assessment of self-protective behavior in response to noxious stimuli in insects is required.][11][12][13] In this study, we tested whether Bombus terrestris bumblebees display a type of self-protective behavior: selectively grooming a noxiouslystimulated body part.For each bumblebee, we either briefly touched one antenna with a noxious stimulus (a 65 C heat probe), or a nonnoxious tactile stimulus (an unheated probe), or we did not touch either antenna (control).We recorded self-grooming behavior toward both antennae for 25 min.If bees specifically groom a site of noxious stimulation, we would predict more grooming on the noxiously-stimulated antenna than the other antenna.We would not expect this difference in bees touched with an unheated probe, nor by bees that were not touched.

Self-grooming in the 25-min observation period
We first tested whether, within the whole 25-min observation period, there was a difference between grooming durations on the touched and untouched antennae, and, if so, whether this difference was larger when the probe was noxiously heated.For the 25-min observation period, bees groomed their touched antennae significantly more (touched: 18.11 G 26.79 s; untouched: 2.22 G 3.57 s; t 5792 = 5.922; p < 0.001; N = 40), regardless of whether the stimulation was noxious or non-noxious tactile (no significant effect: t 5792 = 0.056, p = 0.955; N = 40; no significant interaction: t 5792 = À0.224,p = 0.822; N = 40; Figure 1).Therefore, over the 25 min, grooming was directed toward the touched antenna, but not the noxiously-stimulated antenna specifically.We also observed a significant interaction effect of sex on the total grooming duration over the 25 min, with workers (females) grooming their touched antenna (and not their untouched antenna) for significantly longer than males

DISCUSSION
Our results provide the first quantitative evidence of self-protective behavior in insects.In the first 2 min after noxious stimulation on an antenna, bees groomed this noxiously-touched antenna more than their untouched antenna and more than control (untouched) bees groomed either of their untouched antennae.The same results were not found in bees that were touched with a non-noxious, tactile stimulus.Further, noxiously-stimulated bees groomed their noxiously-touched antenna for longer than the tactilely-stimulated bees groomed their tactilelytouched antenna.8][19] A reason for this timing might be that the nociceptive processing ceased after around 2 min; this would likely change with a higher intensity of the noxious stimulus than we used here.An association between grooming and the cessation or onset of nociceptive processing has been previously noted in mice, in response to nociceptive formalin injection.2][33] By analogy, we suggest that, in our study, the first 2 min corresponded to an acute phase of grooming in response to the noxious heat stimulation.Based on this evidence, future research should investigate the neural processing of noxious heat stimulation in insects, and how the temporal characteristics of the self-grooming might relate.
If grooming directed toward a noxiously-stimulated antenna happens in the first 2 min after stimulation, one might expect to also find a significant increase in grooming within the first minute.Here, we did observe an increase in grooming on the noxiously treated antenna in the first minute, but this increase was not statistically significant after correcting for multiple comparisons (Figure 1).This could reasonably be explained by our use of the Holm-Bonferroni correction, which has a high risk of false negatives. 34n the first 25 min after stimulation, the bees' sex had a significant effect on how long they groomed their touched antenna, regardless of whether the stimulus was noxious or not, with workers (females) grooming their touched antenna for longer on average than males did.There are currently no studies investigating sex differences in self-grooming behavior in bees, but male bees do not groom pollen off their bodies, suggesting that they might be less equipped for self-grooming in response to something touching their body. 35n interesting future line of research would involve investigating the neural underpinnings of our findings.The neural processing of the noxious heat might be similar to that seen in honeybees, where nociceptive signals in the antennae are detected by thermo-sensory neurons and carried to the antennal lobe. 36As for the neural circuits of self-grooming, these have, of course, only been studied in the context of general, non-noxious self-grooming.For example, research on Drosophila melanogaster has identified neurons in the antenna that project to the ventral brain and antennal descending neurons that, if stimulated, cause antennal grooming. 37Nociceptive self-grooming in bees might use similar neural mechanisms, but more research is needed.
What might our results mean for the topic of insect pain?Firstly, we need to clarify whether and how self-protective behavior might relate to pain.0][41] One reason for this association is that self-protective behavior seems to reduce the feeling of pain in humans 42,43 and is not merely a reflexive behavior.For example, self-touch has been found to reduce the painful perception of heat, even when this ''heat pain'' is caused by an illusion that leads participants to perceive pain without there being any nociceptive stimulus. 44This shows that self-touch reduces pain specifically, rather than nociceptive processing.
There are, however, some historic studies in frogs and dogs with severed spines where noxious stimulation of extremities induces leg movements that are roughly directed toward the site of stimulation, suggesting that nociceptive reflexes might underlie some sort of selfprotective behavior. 45,46However, the animal pain frameworks clarify that self-protective behavior should be directed toward the injury site 40,47 and, in these studies, the leg movements are not directed specifically to the site of noxious stimulation.This might mean that general self-grooming in response to injury might be able to occur via nociceptive reflex loops in the spinal cord, but directing the response specifically to the site of injury may require some sort of brain processing.It should also be noted that these studies lack solid experimental measures, such as quantified behavior, mention of sample size, formal analysis, or a control experiment using non-noxious stimuli or healthy animals, so the results cannot be directly compared to our study.Moreover, it is clear that the behavior we observed requires the brain, since noxious stimulation of the antenna feeds directly into the antennal lobe of the bee brain. 36n conclusion, the self-protective behavior displayed by the bees in our study both requires the brain and is akin to a behavior that is associated with pain in humans and other animals.What does this mean for the likelihood that bees can feel pain?Our study shares with others (including those on vertebrates) the challenge that it is currently impossible to be certain about whether a behavior includes the affective component of pain.Therefore, to assess whether an animal can feel pain, it is valuable to collect evidence from multiple different lines of neural, behavioral, and psychological investigations to shift probabilities for or against. 39,40,48Self-protective behavior is included as one of eight criteria for the evidence of pain in other animals. 40Before our study, Adult Hymenoptera already fulfilled four of these eight indicators of pain, namely they have nociceptors 49 and sensory integrative brain regions, 50 display motivational trade-offs, 51 and learn from aversive experiences. 52Our study thus provides evidence for a 5th criterion, self-protective behaviour, so Hymenoptera might now be considered in this framework to show ''strong evidence for pain.''Therefore, our study is one of a set of studies that have found evidence of traits indicative of pain in Hymenoptera.These studies, when taken together, give reason to increase our confidence that bees may feel pain.[11][12][13]

Limitations of the study
We suspect that the self-grooming we observed with this set-up may only be a fraction of the bees' natural response, when not under stress or in a novel environment, since stress and novel contexts have been found to reduce the expression of behaviors after noxious stimulation in insects (honeybees 53 ), similarly to other taxa (humans 54 ; rodents, 55 fish, 56 birds, 57 and snails 58 ).The experiment contained multiple novel and/or potentially stressful experiences and environments for the bees.For example, the stimulation itself involved them climbing onto metal forceps, being lifted out of the nest box, and immobilized during the stimulation-all potential stressors.Further, bees were isolated from the nest and other colony members during testing, and their normal route back to the nest was blocked.In future experiments, observing bees in the nest post-stimulation may lead to the identification of their more naturalistic behavior in response to noxious stimulation.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Figure 1 .
Figure 1.Mean duration of grooming for the untouched and touched antenna per each minute after noxious or tactile stimulation *p < 0.001; Wilcoxon test.Error bars represent the standard error of the mean.

Figure 2 .
Figure 2. Boxplot of duration of grooming on each antenna for each stimulation type group Boxplot boundaries indicate the 25th and 75th percentiles; the whiskers indicate the minimum and maximum values within 1.5 times the interquartile range.Crosses indicate values outside this range (boxplot outliers); triangles indicate the mean; lines indicate the median.*p < 0.001; Wilcoxon test.
TABLE d RESOURCE AVAILABILITY B Lead contact B Materials availability B Data and code availability d EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS d METHOD DETAILS B Treatments B Behavioral analysis d QUANTIFICATION AND STATISTICAL ANALYSIS iScience Article