Varroa resistance in Apis cerana: a review

Varroa is a major world-wide pest to Western honey bees (Apis mellifera), causing huge ongoing losses of colonies every year. Conversely, the Eastern honey bee (Apis cerana) is less vulnerable to the mite having existed alongside it over a long evolutionary period. Research conducted during the 1980s and 1990s, shortly after Varroa had spread across the globe, concluded that the Eastern honey bee was less vulnerable because it displayed higher levels of grooming behaviour, brood removal behaviour and mite infertility than its Western counterpart. However, this review on these Varroa resistance traits in A. cerana indicates that there is surprisingly little evidence for these conclusions. This review explores this evidence and discusses the potential flaws in the studies and the gaps that still remain in our knowledge of Varroa resistance traits in A. cerana.


INTRODUCTION
Varroa is a genus of ectoparasitic mite which parasitises honey bee colonies across the world. Female Varroa mites live on the body of adult honey bees and reproduce in the brood cells alongside the developing honey bee pupae. Varroa garnered attention approximately 70 years ago when the now infamous species Varroa destructor jumped from its original host, the Eastern honey bee (Apis cerana) to the Western honey bee (Apis mellifera) (Oldroyd 1999). This jump was actually a collection of independent species jumps that were made possible because A. mellifera populations were moved into regions in the native range of A. cerana (Roberts et al. 2015;Rosenkranz et al. 2010). Once the parasite switched host, it was accidentally traded worldwide with its Western honey bee hosts, excluding the island of Newfoundland (Shutler et al. 2014), causing widespread colony losses particularly in regions within the Northern hemisphere. To date, Varroa is still a major pest in the Northern hemisphere and financial burden to Western honey bee apiculture (Rosenkranz et al. 2010). Eastern honey bees, on the other hand, suffer fewer negative effects of the parasitisation and generally lack the need for human intervention . Indeed, A. cerana can be described as resistant to the mite, which is defined here as the ability of a Varroa infested colony to survive long term, without control methods administered by humans, within a given environment (Grindrod and Martin 2021).
Over the decades since Varroa spread outside Asia, Varroa resistance has been increasingly observed in western honey bees. Firstly, resistance developed within the Cape and Savannah honey bees in Africa (Allsopp 2006;Nganso et al. 2018) and the Africanised bees of South America (Moretto et al. 1991). Resistance developed fairly quickly in these areas, and this is thought to be because of the lack of chemical control and the more naturalised bee keeping style (Grindrod and Martin 2021). Indeed, this is also true of European honey bees in Cuba, a country which is believed to contain the largest population of resistant European honey bees (Luis et al. 2022). However, resistance has begun to be observed in Europe and the USA (Oddie et al. 2018;Martin et al. 2019;Hawkins and Martin 2021;Grindrod and Martin 2021). Research on these populations suggests that Varroa resistance is the product of a number of resistance traits that regulate the populations of mites within the colony. These traits include brood removal behaviour, which is the removal of dead or diseased or Varroa-infested pupae, grooming behaviour in which bees remove mites from themselves or other individuals, recapping behaviour in which infested cells are opened and resealed and finally mite infertility where mites are incapable of producing viable offspring. Three of these traits brood removal, grooming and mite infertility are the same as those previously reported in A. cerana. The fourth behaviour, recapping, has not been studied in A. cerana as it was more recently discovered by Oddie et al. (2017). Data, from A. cerana, on all four of these traits would therefore be beneficial comparison and be a valuable asset in understanding Varroa resistance in A. mellifera. Recently, an evidence-driven framework was constructed to suggest how these traits may interlink and allow for the development of resistance in A. mellifera (Grindrod and Martin 2021). However, it is not possible to see if the same occurs in A. cerana as there exists relatively little published data, and the data that does exist is from studies with small sample sizes and outdated knowledge including the absence of the identification of Varroa destructor (Anderson and Trueman 2000).
Henceforth, our understanding of the relationship between Varroa and A. cerana and consequently resistance traits is limited and often based on assumptions, which are then used to make further assumptions about A. mellifera. With the continual advancement of Varroa research methodologies, it seems prudent that the relationship between Varroa and A. cerana is re-evaluated. In this review, the major areas of research into Varroa resistance traits, grooming, brood removal and mite infertility, in A. cerana are outlined and discussed to identify gaps and provide suggestions for future research.

GROOMING
Grooming behaviour is often included in the suite of behaviours used in defence against Varroa. It entails adult bees removing mites either from themselves (auto-grooming) or from other adult bees (allo-grooming) using their legs and mandibles (Pritchard 2016). The removal and possible injury of the mites is thought to control the size of the phoretic mite population and thus the overall colony infestation (Moosbeckhofer 1992). However, grooming behaviour is notoriously hard to measure accurately as it relies either on indirect measurements such as mite damage or from potentially subjective direct observations. Despite this, a single study by Peng et al. (1987) (cited almost 300 times in web of science, accessed 17/02/22) appears to have led to the acceptance of A. cerana as the superior groomer over A. mellifera and from this the assumption that grooming is a considerable factor of Varroa resistance. Certainly, at first glance, the results are very enticing as the 99.6% removal of mites by A. cerana vastly overshadows the 0.3% removal by A. mellifera seemingly solving the mystery of why A. cerana are more resistant in one shot. There are, however, a number of reasons to be highly sceptical of the results.

The issue of mite source
Firstly, Peng et al. (1987) used mites from A. mellifera colonies on A. cerana adults. This is likely to have unintentionally exaggerated the results because A. cerana respond much more strongly to mites sourced from A. mellifera colonies compared to their own species (Büchler et al. 1992;Fries et al. 1996;Rath 1991a;Rosenkranz et al. 1993a, b). This heightened response could possibly occur because the mites are a different species (V. destructor) and/ or because the mites have mimicked the original hosts (Apis mellifera) cuticular hydrocarbons (Kather et al. 2015;Le Conte et al. 2015;Martin et al. 2001) which can rapidly be detected as foreign by the new A. cerana hosts (Fries et al. 1996;Rahman et al. 2016;Rath 1991a).

Limitations to direct observation methods
The results may also have been spuriously elevated because, to assess grooming ability, Peng et al. (1987) attempted to directly observe the adult bees undertaking the behaviour. Naturally, this approach is prone to inaccuracy because it is difficult to follow individual mites and to be sure of their fate (Fries et al. 1996). As a result, the authors erroneously considered both the movement of mites from one bee to another and the disappearance of mites to the observer as a successful removal. Indeed, the potential inflation of the results was highlighted nearer the time in a review by Boecking et al. (1993). They also indicated that a constant removal rate as high as 99% would mean that A. cerana colonies would be devoid of mites during periods when drone brood is absent, which is not the case. Nonetheless, despite these shortfalls, the Peng et al. (1987) article is still highly cited with 19 citations in 2021 (web of science, accessed 17 Feb. 22).
Since its publication, only three other studies have sought to repeat or re-evaluate these results (Table I). Büchler et al. (1992) also utilised the direct observation method and used a mix of phoretic and brood mites for A. mellifera colonies. They did improve the methodology by using phoretic A. cerana mites for the A. cerana colonies; however, they chose to source these mites from different A. cerana colonies. This may still affect the results because Varroa mites can mimic the colonies cuticular hydrocarbon profile down to the level of each colony (Kather et al. 2015). Despite this their results appear more realistic, in terms of the earlier criticism by Boecking et al. (1993), with 75% removal rather than 99% for A. cerana. They also found a much greater result for A. mellifera at 48% removal, but both figures need to be interpreted with care as they were based on small sample sizes of 36 and 25 mites, respectively.

Mite damage as a proxy for grooming ability
To avoid the issues with direct observation experiments Fries et al. (1996) used mite damage as a proxy for grooming success. During grooming, mites can endure damage to their idiosoma and legs caused by the bee's mandibles (Rosenkranz et al. 1997;Ruttner and Hänel 1992). Using this as a proxy allowed the experiment to be conducted in a normal, full-size hive compared to the smaller observation hives used in previous studies. Fries et al. (1996) found that, over a 6-h period, 29.6% of introduced mites were damaged by A. cerana and 12.3% by A. mellifera. The result for A. cerana may again have been impacted by the use of A. mellifera mites although the difference between the species is notably smaller than the results of both Peng et al. (1987) and Büchler et al. (1992). The smaller difference could be the result of using an indirect method; however, this is not supported by Rath (1991a). They recorded the number of mites, sourced from A. mellifera, which died and were injured when introduced to adult bees in a cage experiment. Furthermore, Peng et al. (1987) also found a large difference between the two species, 61.7% of introduced mites died on A. cerana in 48 h, whereas only 2.8% died on A. mellifera.
Of those that died, they found that 83% from A. cerana had injuries whilst none of the dead mites from A. mellifera showed any sign of injury.

The uncertainty caused by using a proxy
Measuring grooming indirectly brings its own level of uncertainty to the results because grooming is not the single cause of damage to mites. Mites may also be damaged when infested brood cells are cleaned out (Boecking and Drescher 1991) or by other hive predators such as ants (Bienefeld et al. 1999;Davis et al. 2007) or wax moth (Szabo and Walker 1995). Care also needs to be taken when observing damage to the idiosoma to prevent regular dorsal dimples, a developmental defect, from being confused for grooming induced damage (Davis 2009;Rosenkranz et al. 1997). Also, as with the observation methods, Fries et al. (1996) noted that the artificial introduction of mites into a colony substantially increased the initial mite drop. Additionally, the presence of emerging brood increases the mite fall and mite damage (Hoffman 1995;Lobb and Martin 1997;Martin and Kemp 1997;Rosenkranz et al. 1997) thus adding to the variability of measurements. It is also difficult to conclude whether the damage occurred pre or post mortem; for example, Fries et al. (1996) found that in an A. mellifera colony, 26.4% of naturally fallen dead mites (killed by freezing the combs) had damage but only 9.1% of naturally fallen live mites were damaged, suggesting that either bees injure dead mites or that the injury caused by bees leads to the death of mites.

Summary
It is widely believed that A. cerana perform grooming behaviour to a higher extent than A. mellifera and subsequently that this behaviour plays a large role in the resistance of A. cerana to Varroa. However, this belief is based on only four studies with a disproportionate amount of belief placed on a single, highly cited, study by Peng et al. (1987) that may have falsely elevated results due to flaws in the methodology. Additionally, there are many discrepancies between the methodologies used by the four studies. Grooming is also highly variable both within and between colonies due to the season (Büchler 1993;Mondragón et al. 2005;Moosbeckhofer 1997;Russo et al. 2020), environmental conditions (Currie and Tahmasbi 2008), presence of emerging brood (Hoffman 1995;Lobb and Martin 1997;Martin and Kemp 1997;Rosenkranz et al. 1997) and levels of brood removal behaviour (Kirrane et al. 2018). This variability means that many measurements are required to increase the accuracy of results. Thus, the existing four studies do not provide enough data to accurately suggest the magnitude of the role that grooming plays in Varroa resistance in A. cerana.
Indeed, despite the larger number of studies on grooming in A. mellifera, the results have been highly variable. In some instances, resistant colonies have been found to groom to significantly more (Mendoza et al. 2020) and some studies found a negative correlation between mite damage and infestation rate (Arechavaleta-Velasco and Guzmán-Novoa 2001; Mondragón et al. 2005;Moosbeckhofer 1992;Ruttner and Hänel 1992). Conversely, many others have found the opposite with no significant difference between resistant and susceptible populations suggesting that grooming does not significantly contribute to resistant behaviour (Aumeier 2001;Kovačić et al. 2018;Kruitwagen et al. 2017;Locke and Fries 2011;Nganso et al. 2017;Oddie et al. 2018). Certainly, when the impact of reduced mite fertility is considered, grooming behaviour is not necessary to explain Varroa resistance in A. mellifera (Locke and Fries 2011;Oddie et al. 2017) or in A. cerana in which there is complete infertility of mites in the worker brood (Fries et al. 1994). This is why grooming was not included in the framework proposed by Grindrod and Martin (2021).

BROOD REMOVAL
Brood removal is the archetypal hygienic behaviour in which adult bees uncap and remove dead, diseased or parasitised pupae. Whilst it is used in response to Varroa infestation, it was first described as a response to American foulbrood (Rothenbuhler 1964) and then to chalkbrood (Gilliam et al. 1983). Such brood diseases usually result in the death of the pupae and in turn the release of potent death pheromones such as oleic acid (McAfee et al. 2018). As a consequence of this and the hazards involved in inoculating pupae with diseases, hygienic behaviour has typically been measured using methods that cause the death of the pupae including freezekilled brood (FKB) and pin-killed brood (PKB) methods (Spivak and Downey 1998).

The results and limitations of FKB methodology
A. cerana respond well to FKB; they remove fairly high levels of FKB, 82% in 24 h (Rath and dresecher 1990) and also remove it faster than A. mellifera Shakeel et al. 2020). However, A. cerana colonies can be highly susceptible to the brood diseases sac brood (Abrol 2000;Ai et al. 2012;Hassanyar et al. 2019;Ma 2014;Vung et al. 2020) and American foul brood (Chen et al. 2000). This suggests that high FKB results do not necessarily equate to a high hygienic response to disease which may be because disease transmission can be enhanced by hygienic behaviours particularly due to removing already deceased brood containing highly infective elements.
Additionally, whilst FKB and PKB can offer some insight into the general hygienic capabilities of a colony, they have so far failed to correlate with the results of Varroa-infested brood removal (Boecking and Drescher 1992;Danka et al. 2013;Grindrod and Martin 2021;Leclerq et al. 2018;Martin et al. 2019). Arguably, this is because Varroa rarely kills the developing brood, and thus, the cues used by workers to detect a Varroa infestation are different (Mondet et al. 2021;Spivak 1996).

Artificial mite infestation experiments
Ultimately, due to their observed natural resistance to the mite, A. cerana are generally believed to express a higher level of brood removal behaviour than A. mellifera. However, despite a plethora of anecdotal evidence, this literature search only found three studies that measured the ability of A. cerana to remove cells artificially or naturally infested with Varroa. The first of these papers by Rath and Drescher (1990) found very high removal rates, 97.4% and 91.9%, of A. cerana worker cells artificially infested with live and dead ethanol-washed mites respectively, which indicated the ethanol wash had little, if any, effect. The mites used were again sourced from A. mellifera colonies which, as with grooming, will artificially increase the removal response (Boot et al. 1999;Rosenkranz et al. 1993a, b) due to different chemical profiles. One could argue that the scent of previous hosts would be negated by the ethanol wash of the dead mites; however, even when washed with ethanol, cuticular hydrocarbons, potentially from the original host, are very likely to remain on the mite (da Silva Cunha et al. 2021). Boot et al. (1999) found that A. cerana however removed 84% of mite (sourced from A. mellifera)-infested worker brood cells over 10 days, which whilst lower is still a high result for the removal of Varroa brood in comparison to resistant western honey bees (Grindrod and Martin 2021). Although, a separate experiment by Boot et al. (1999) comparing both species showed that A. mellifera and A. cerana removed a similar percentage of Varroa-infested cells over 4 days, 32% (n = 104) and 29% (n = 131) respectively.
Additionally, the results of Rath and Drescher (1990) may also be somewhat artificially inflated as the results include the cells in which mites had disappeared from as well as fully emptied cells. Whilst workers do seem to be able to remove dead mites and re-seal cells (Rosenkranz et al. 1993a, b) live mites pose more of a challenge to remove and can also exit cells of their own volition whilst the cell is left open. This uncertainty means that the "disappearance" of live and dead mites should ideally be reported as a separate statistic as exemplified by Rosenkranz et al. (1993a, b) and Boot et al. (1999) (Table II). Live mites also "disappear" in A. mellifera colonies (De Guzman et al. 2016). Boot et al. (1999) noted, albeit without numerical evidence, a low removal response in naturally infested colonies, but this may be because of the low infestation rates and thus low levels of stimulus. Conversely, low natural responses may also be the result of the fact that A. cerana mites avoid reproducing in worker brood and thus do not produce the cues necessary to be detected (Mondet et al. 2021). This may also explain why unlike Rath and Drescher (1990) who found an immediate high removal response, Boot et al. (1999) noticed that the removal response of A. mellifera mites was delayed by a couple of days, as time may be needed for reproduction to produce the cues. Accordingly, in A. mellifera, peak removal has been shown to occur roughly 3-5 days post capping (Harris 2007;De Guzman et al. 2016). Although, if a low removal response is due to a lack of reproduction, then it is not easy to explain why Rosenkranz et al. (1993a, b) found a low removal response of 8% in A. cerana when mites were transferred within the same colony (intracolonial) but a high removal response of 50% with mites from a different A. cerana colony (intercolonial). Mites from another A. cerana colony would be likely to avoid reproducing in worker brood to the same degree.

Observations of natural mite infestation
*Cells resealed without mite was not treated as a separate statistic and raw data was unobtainable.

The social apoptosis phenomenon
Indeed, other studies have suggested that brood removal may be stimulated by damage to the pupae rather than scents from the mite. Page et al. (2016) and Lin et al. (2018) discovered that the worker pupae of A. cerana in Thailand and China were more susceptible to wounding and infestation by V. destructor of the Korean haplotype than A. mellifera pupae. The increased susceptibility meant that A. cerana pupae were more likely to be developmentally delayed and die, which would simultaneously prevent successful mite reproduction and provide a signal to worker bees for removal (Lin et al. 2018). As a result, they termed the phenomenon social apoptosis. In support of this, Zhang et al. (2018) discovered a protein in the saliva of mites called Varroa toxic protein, or VTP, which was extremely toxic to A. cerana worker brood but not A. mellifera. However, whilst these results are promising, they seemingly lack support from previous mite reproduction studies in which an enhanced death rate of worker brood was not observed, or at least not recorded (Koeniger et al. 1981(Koeniger et al. , 1983Rath 1991a;Rosenkranz et al. 1993b). Although, in the majority of these previous infertility studies, V. jacobsoni was the infesting mite and may differ to Varroa destructor in terms of the impact of wounding towards its host. It may also be different with other haplotypes of V. destructor (Lin et al. 2018) again highlighting the need for more research.

Summary
The removal of Varroa infested brood is thought to be the cornerstone for resistance in honey bee populations (Grindrod and Martin 2021). Unfortunately, however, there remains a lot missing in our understanding of brood removal behaviour in A. cerana. Firstly, as with grooming, there is very little data concerning the removal of artificially infested cells which combined with the high variability of the behaviour means we do not have a reliable indicator of its relevance to resistance. Additionally, as with grooming behaviour, the methodology used to collect the data varies between studies. Indeed live mites are often sourced from A. mellifera colonies for use on A. cerana, and sometimes, dead mites are used. There are additional data on the removal of FKB; however, as Varroa does not usually kill, the brood FKB ability does not tend to correlate with the ability to remove infested brood. As a final note, there is a distinct lack of clarity concerning the phenomenon entitled social apoptosis, in particular, its prevalence and whether it occurs with both mite species and all the haplotypes.

Infertility of Varroa in Apis cerana worker brood under natural conditions
The definition of mite infertility can include Varroa females producing no eggs at all or Varroa females failing to produce viable, i.e. fully matured and mated, female daughters. The former definition, also known as strict or complete infertility, was used in the studies conducted in the 80 s and 90 s and thus applies to the data reviewed here. Infertility was first reported as a characteristic of the natural infestation of Varroa mites on A. cerana worker brood in Sri Lanka and Java (Koeniger et al. 1981(Koeniger et al. , 1983 and has since been reported in Vietnam (Boot et al. 1997), Papua New Guinea, Java, Irian Jaya (Anderson 1994), India (Rosenkranz et al. 1993b;Twearson et al. 1992). De Jong (1988) did note some rare incidences of reproduction under natural circumstances in worker brood cells in South Korea. Similarly, a recent study by Wang et al. (2020) that investigated 185 A. cerana colonies of six populations in China and Thailand found only one incidence of reproduction in a Varroa mite naturally infesting Apis cerana worker brood in the south of Thailand, and this incidence was not a successful attempt.

The potential causes of Varroa infertility in worker brood
Whilst this infertility is fairly well-documented, the exact cause remains elusive. Research by Grindrod and Martin (2021) on A. mellifera has suggested that a cause of infertility is simply the disruption of reproduction due to brood removal. They suggest that continual high levels of targeted brood removal could cause mites to avoid worker brood in favour of drone brood that is not removed in A. mellifera (Grindrod and Martin 2021) or A. cerana (Harris 2008). In speculation, this may have occurred in A. cerana with the resultant separate evolution of Varroa reproduction and worker pupal development leading to a loss of synchrony in the cycles and thus infertility of the mites. This loss of synchrony could include the loss of specific oogenesis triggers from the pupal host which are normally acquired by the mite when feeding. Although, these triggers could also be lost via selective pressure from mites.
Alternatively, infertility may be a factor of the mites not the pupae. This was suggested by the work of Boot et al. (1999) that showed that mites from an A. cerana colony will not reproduce in the worker brood of another A. cerana or A. mellifera colony if transferred but that mites from an A. mellifera colony will. Rath (1991b) also found it was possible to get Varroa mites from A. mellifera to reproduce on A. cerana worker brood in a lab setting. Boot et al. (1999) propose that the loss of fertility was the result of A. cerana removing reproducing mites more frequently and thus inadvertently selecting for non-reproducing mites. Indeed, a bias toward the removal of reproductive mites is possible because mite reproduction may be required to produce a stimulus that the bees can detect (Mondet et al. 2021).

The fertility of Varroa on A. cerana worker brood during artificial infestation
Importantly, due to the time period of the Boot et al. (1999) study, the mites in question were identified as Varroa jacobsoni. However, at the time in Vietnam, it appears that it was actually Varroa destructor of the Korea and Vietnam haplotypes that were parasitising Apis mellifera and Apis cerana respectively (Fuchs et al. 2000). Therefore, Boot et al. (1999) showed that it is possible for V. destructor of the Korea haplotype, not V. jacobsoni, to reproduce on the worker brood of A. cerana. Since then, this observation has been repeated by Li et al. (2019) who discovered that, in China, Varroa of the Korea haplotype sourced from A. mellifera colonies could reproduce in the worker brood of both honey bee species, albeit at a higher fecundity in A. mellifera. They also found that much like the Vietnam haplotype, the China haplotype from A. cerana colonies was completely sterile in the worker brood of A. mellifera and A. cerana colonies Li et al. (2019).
In contrast, Le conte et al. (2015) found that Varroa destructor from Apis cerana could reproduce in the worker brood of both honey bee species when transferred. However, whilst this study also took place in China, the haplotype of V. destructor infesting A. cerana was the Korean haplotype 3 and not the China haplotype. The Korean haplotype 2 from Apis mellifera also reproduced on both honey bee species. Although, it is worth noting that the sample sizes in this study were very small with only 6 to 10 mites transferred per swap.
Finally, the equal fertility of the Korean haplotype on A. cerana and A. mellifera worker brood in China was also observed by Lin et al. (2018). However, this fertility was facilitated by the removal of adult bees during the experiment. When adult bees were present a high proportion of the infested pupae were removed which Lin et al. (2018) suggest was due to the developmental delays caused by Varroa.

The fertility of Varroa jacobsoni parasitising Apis mellifera
In the absence of competition with V. destructor, in Papua New Guinea (PNG), Roberts et al. (2015) found that it is possible for V. jacobsoni to overtime develop the ability to reproduce in both the drone and worker brood of A. mellifera colonies. Initially, these attempts to reproduce were directed at the drone brood and were largely unsuccessful (Anderson 1994). However, by 2008 V. jacobsoni, was reproducing in high numbers on both drone and worker brood in PNG (Anderson 2008), and later, Roberts et al. (2015) discovered that there had been two independent host shifts of V. jacobsoni onto A. mellifera. Thus, it does appear that the infertility of V. jacobsoni on A. cerana is a product of their relationship that may be reversible if the barriers to reproduction are removed. And in speculation, if these barriers are created in A. mellifera populations, then perhaps the same infertility of mites in worker brood can be established. Although it is important to note that in PNG deformed wing virus is also absent which will alter the relationship between the mite and the honey bee host (Roberts et al. 2020).

Summary
Mite infertility is the most strongly supported trait in A. cerana; however, there is no consensus on its origin or how it is maintained. Mite infertility in worker brood appears to play a large role in resistance and is believed to be the main reason why A. cerana do not succumb to the mite. Given that some studies have found that it is possible to get V. destructor to reproduce on A. cerana, it seems worthwhile to explore this relationship further as it may allude to the cause of the infertility of V. jacobsoni. For example, whether it is the result of brood removal in general or the result of selection by removing reproducing mites and/or the lack of cues in the pupal feed. Additionally, it would be beneficial to understand the differences, if any, between the Varroa species and haplotypes and their ability to reproduce on different honey bee species and castes (Figure 1).

THE GENETICS OF RESISTANCE
In Apis mellifera, there has been a considerable amount of research into the genetic basis of resistance traits; a major goal of this research is to identify molecular markers associated with the traits that could be used to aid the breeding of resistant bees (reviewed in Mondet et al. 2020). Thus far, olfactory sensitivity has generally been highlighted as an integral process for resistance traits . Despite this, there is low agreement in the candidate genes isolated by different studies. This is likely due to a number of factors including the variation in the presentation of behavioural traits, the complexity of the underlying genetics and the different sub-species utilised in these studies Traynor et al. 2020).
As one might expect, the research into the genetic basis of resistance traits is comparatively lacking in Apis cerana although efforts have been increasing as the 'omics methods have become more accessible and affordable (Diao et al. 2018;Ji et al. 2014aJi et al. , b, 2015Ling et al. 2015). Thus far, Ji et al. (2014a, b, 2015) and Ling et al. (2015) have discovered an upregulation in genes related to olfactory sensitivity in A. cerana after mite introduction. This suggests that olfactory sensitivity may play a key role in expression of the resistant traits in both A. mellifera and A. cerana. However, genome comparisons by Diao et al. (2018) discovered that the A. cerana genome contains fewer genes encoding odorant binding proteins and olfactory receptors than the A. mellifera genome. How that translates to the expression of resistance traits olfactory sensitivity is not clear, but it does highlight the need for more research in this area in order to make meaningful comparisons between A. mellifera and A. cerana.

CONCLUSION
Here, the main areas of Varroa-A. cerana research have been presented, and gaps in the research have been highlighted and discussed.
Ultimately, what has become clear is that our assumptions about the ability of A. cerana to perform the resistance traits grooming and brood removal are largely based on only a small number of decade old studies, often using small sample sizes. This is problematic because there is considerable natural variation in the displaying of resistance traits between colonies. Some variation exists naturally within and between populations, but it is amplified by many other factors including the seasons, environmental conditions, mite infestation levels and the methodology used.
Additionally, given its more recent discovery, there are also no published data on the presence of recapping behaviour in A. cerana colonies. This leaves gaps in our understanding of the relationship between A. cerana and Varroa, and it is difficult to relate this to the trajectory of A. mellifera and Varroa's relationship. Thus, there is a need to complete new research to ascertain the level of grooming, infested brood removal, recapping and  Grindrod and Martin (2021); for studies used to calculate grooming in a. and b, see supplementary data. Data for c comes from this study. All grooming averages are based on results using the mite damage proxy mite infertility displayed in A. cerana populations across different regions. Those data could provide important evidence to either support or rebuke the framework of Varroa resistance acquisition presented in Grindrod and Martin (2021). Such data could also be backed by further exploration of the genetic basis of resistance traits in both A. mellifera and A. cerana. Ultimately, achieving a full understanding of the development of resistance in Apis cerana will help inform efforts into developing and maintaining resistance in Apis mellifera populations thus reducing colony losses and dependency on acaricides.

AUTHOR CONTRIBUTION
I. G. conceived this research and collected, analysed and interpreted the data; I. G. wrote the paper with revisions from S. J. M; both authors read and approved the final manuscript.

FUNDING
Funding for this study was provided by Bee Diseases Insurance Ltd.

AVAILABILIT Y OF DATA AND MATERIALS
Summary of grooming data used in Figure 1 is available in the supplementary data file.

CODE AVAILABILITY
Not applicable.

DECLARATIONS
Ethics approval Ethical approval was not needed for this research.

Consent to participate Not applicable.
Consent for publication Not applicable.

Competing interests The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.