Introduction

Acarapis woodi, a tracheal mite that infests honey bees, was first reported in the European honey bee, Apis mellifera, in the UK (Rennie 1921). The mite spread to the rest of Europe and all over the world in the twentieth century (Otis 1990; Wilson et al. 1997; Sammataro et al. 2000). Infested hives of European honey bees suffer damage from the mites, including low honey production and low pollination capacity, and a heavy load of A. woodi results in colony loss during winter (Eischen 1987; Eischen et al. 1989; Otis and Scott-Dupree 1992; McMullan and Brown 2009). The annual economic damage caused by the mite, including the cost of surveillance, was estimated as 1 million US$ in the USA and 0.8 million US$ in Canada (Otis 1990).

When A. woodi invades a new region, the mites rapidly increase their population and spread their distribution. For example, rapid spread of the tracheal mite was observed in the USA, with 17 states being infested during a 13-month period (Delfinado-Baker 1985). A few decades after the first epidemic, the number of mite-infested colonies decreased and the mite prevalence, represented by the proportion of infested bees in a colony, had decreased in several countries. For example, in the UK, mite prevalence was about 50–60% in 1925 but decreased to 20–30% in 1946 (Morison et al. 1956). Moreover, it dropped to less than 10% in the 1960s (Bailey 1985). A similar decline was reported in Spain (Orantes and García-Fernández 1997), Greece (Bacandritsos and Saitanis 2004), the USA (Moore et al. 2017), and Norway (Stachurska-Hagen et al. 2018).

Only one report on A. woodi infestation of the European honey bee in Japan has been published (Kojima et al. 2011b). By polymerase chain reaction (PCR) analysis, the authors identified a DNA fragment of tracheal mite in European honey bee. On the other hand, Japanese honey bee, Apis cerana japonica, has been infested by tracheal mites since 2010 (Kojima et al. 2011a; Maeda 2015, 2016; Maeda et al. 2015; Maeda and Sakamoto 2016). In 2014, most native honey bee colonies infested by tracheal mites were in central and eastern Honshu (Maeda 2016; Maeda and Sakamoto 2016) and a few infested colonies were found in western Honshu, but no tracheal mites had been detected on Japan’s other two main islands, Shikoku and Kyushu (Maeda and Sakamoto 2016). The possibility of a rapid spread of tracheal mites to western Japan, as occurred in other countries such as the USA (Delfinado-Baker 1985), is concerning. The Japanese honey bee is not distributed in Hokkaido or the Ryukyu Islands, including Okinawa (Yoshiyama and Kimura 2018), so we excluded these islands from our survey in the current study.

In Japan, many hobby beekeepers capture wild Japanese honey bees in spring and keep them in hand-made box hives (Yoshida 1997a, 1997b). If hobby beekeepers were able to easily notice signs of mite infestation, it would assist researchers in identifying the spread of the mite’s distribution. However, because the mites are too small to see with the naked eye and they remain inside the bees, they can be directly detected only by observation under a binocular microscope after dissection, or by PCR analysis (Sammataro et al. 2013). In the European honey bee, two signs are used to diagnose tracheal mite infestation: bees crawling on the ground near their hives and bees with disjointed hindwings, called K-wing (Sammataro et al. 2000, 2013). The presence of K-wing is positively correlated with mite prevalence in Japanese honey bees (Maeda 2016), and crawling bees are also found on the ground around dwindling Japanese honey bee colonies (Maeda 2016). Therefore, these two signs, K-wing and crawling bees (both illustrated in Maeda et al. 2015), may be useful for preliminary diagnosis of A. woodi infestation by hobby beekeepers.

In the current study, we investigated the expansion of the distribution of tracheal mites in Japan. To test whether mite infestation of Japanese honey bees and European honey bees has declined, we tracked the share of infested hives (percentage of infested colonies among all tested colonies) and mite prevalence (percentage of infested honey bees in a colony) from 2013 to 2018. In addition, we conducted a nationwide survey in winter 2018–2019 in which we collected honey bees and asked hobby beekeepers whether they had observed K-wing or crawling bees in their colonies to evaluate whether their observations could be used to diagnose infestation by tracheal mites.

Materials and methods

Bee sampling

To investigate the status of tracheal mite infestation in Japan before 2013, we examined honey bees collected from 1992 to 2013, preserved in 99.5% ethanol and maintained under refrigeration (–20 °C) at the Institute of Livestock and Grassland Science, National Agriculture and Food Research Organization. Thirty-two colonies of the European honey bee and nine colonies of the Japanese honey bee were provided for the mite detection procedure. More than 20 bees were used for each colony.

We also examined worker bees of both A. cerana japonica and A. mellifera collected after 2013 from all over Japan with the cooperation of hobby beekeepers and apiarists. Most Japanese honey bees were collected from beehives of hobby beekeepers. The European honey bees were collected by both apiarists and hobby beekeepers. More than 20 bees were collected in front of each beehive with forceps or by using an insect net. If no bees emerged from the beehives because of low temperature or for other reasons, we collected them directly from inside the hives. As a preliminary step, we compared mite infestation between bees inside the hive and foraging bees and found that the difference in infestation rates in each beehive was much smaller than the variation among beehives.

To identify whether A. woodi had invaded remote islands of the Japanese archipelago, from 2016 to 2018 we also collected Japanese honey bees from eight small remote islands: Oki, Iki, Gotō, Koshiki, Yakushima, Tanegashima, Amami, and Tsushima. In the current study, we consider these islands are too far from the main islands for the bees to migrate there on their own. We collected more than 20 workers per colony. In addition, we asked hobby beekeepers, apiarists, and residents whether any beehives had been intentionally moved, and, if so, where the hives had been moved from. We also asked the beekeepers whether they observed K-wing and crawling bees in 2018.

To evaluate the efficacy of observation of K-wing and crawling bees by hobby beekeepers for the evaluation of tracheal mite infestation, we conducted a nationwide survey from October 2018 to February 2019. We requested 280 hobby beekeepers to collect 20 workers of the Japanese honey bees from one of their colonies, and we also asked them whether they had observed K-wing or crawling bees in those colonies. Crawling and K-wing can be caused by other factors (Bailey 1999; Shutler et al. 2014). Therefore, observation by hobby beekeepers may include results caused by other factors. However, the purpose of this study is to test whether such observations by the hobby beekeepers could be used to diagnose infestation by tracheal mites, even if their observations are not perfect.

Governmental reports

We referred to governmental reports ‘Animal disease: the outbreak situations in Japan and OIE statuses,’ published each year since 1937 by the Ministry of Agriculture, Forestry and Fisheries (2019a) with detailed data for each prefecture available for 1998 onward (Ministry of Agriculture, Forestry and Fisheries 2019b), for complementary data. According to the Act on Domestic Animal Infectious Diseases Control, acarapiosis—the disease caused by the tracheal mite (OIE 2008)—must be reported in each prefecture. Veterinarians confirm the presence of the tracheal mites by microscopic examination or DNA analysis, and they use several signs, such as colony diminishment and crawling bees around beehives, to diagnose acarapiosis. Bee species are not specified in these reports, so we requested information from the Livestock Health and Sanitation Office of each prefecture (July 2019) to confirm the bee species of all samples.

Microscopic examination of bees for tracheal mites

All bees collected after 2013 were preserved in 70% ethanol in a refrigerator (4 °C). We removed the head and thoracic collar with fine forceps, as described by Lorenzen and Gary (1986), and observed the tracheal tubes under a stereomicroscope at 20–80 × magnification. If we observed mites in the tracheal tube or heavy melanization of the tracheal tube, we considered the bee to be infested by tracheal mites. Our preliminary observations indicated that the melanized tracheas always contained tracheal mites or their debris and that no other factors caused the melanization. Therefore, melanization of the trachea would be a reliable sign for the presence of tracheal mites. However, because melanization occurs at the later stages of infestation, we observed all tracheas in detail in order to identify those in the earlier stages of infestation.

Statistical analysis

The percentage of mite-infested colonies was calculated for the following traditionally recognized regions of Japan: Tohoku, Kanto, Chubu, Kinki, Chugoku, Shikoku, and Kyushu (Fig. 1). Although the Kyushu region includes the Ryukyu Islands, we excluded the Ryukyus from our survey because the Japanese honey bee does not occur there. We also excluded Hokkaido for the same reason. The Tohoku, Kanto, Chubu, Kinki, and Chugoku regions are located on Honshu Island, and Shikoku and Kyushu are separate large islands.

Fig. 1
figure 1

Range expansion of tracheal mites in Japanese honey bees in Japan from 2010 to 2018. Japan has 47 prefectures that are grouped into eight regions: Hokkaido, Tohoku, Kanto, Chubu, Kinki, Chugoku, Shikoku, and Kyushu (from north to south). We excluded Hokkaido and the Ryukyu Islands from our survey because the Japanese honey bee is not distributed there. We used the governmental records and analyzed stocked samples as complementary data to our survey data. The first detection of the tracheal mite in Nagano (2010), Shiga (2011), Nara (2016), Shimane (2016), Tottori (2016), and Ishikawa (2017) prefectures were from the governmental reports. NA: not assessed

Full size image

Mite prevalence was represented by the percentage of infested bees in each colony. In this study, prevalence was calculated as the number of mite-infested individuals relative to the total number of tested individuals from each colony. The average mite prevalence in mite-infested colonies was calculated for every region.

Multivariate logistic regression models were prepared to test whether observations of K-wing or crawling bees by hobby beekeepers were related to colony infestation by tracheal mites. The model assumed a binomial distribution in which the response variable was the presence or absence of infestation and the explanatory variables were K-wing, crawling bees, and their interaction [glm: infestation ~ K-wing + crawling bee + K-wing * crawling bee, family = binomial (link = logit)]. Akaike’s information criterion (AIC) was used to select the best model. All statistical analyses were carried out using the open software package R v.3.6.0 (R Core Team 2019).

Results

Mite infestation of Japanese honey bees

We dissected 224 preserved Japanese honey bees from nine colonies collected from 1992 to 2009 and found no tracheal mites. From 2013 to 2018, we investigated 3494 Japanese honey bee colonies, including those included in the nationwide survey in 2018. Results from 470 colonies collected in 2013 and 2014 were previously published as a distribution map of A. woodi in Japan (Maeda 2016; Maeda and Sakamoto 2016). We used these data along with data of the samples collected before 2012 and data from the governmental reports, to draw an accurate, updated distribution map. Our analysis confirmed that, as reported by Maeda (2016) and Maeda and Sakamoto (2016), tracheal mites had already been found in 10 prefectures by 2013 (Fig. 1). The A. woodi distribution expanded each successive year. Tracheal mites were first detected in the Kyushu region in 2015, and by then A. woodi had spread throughout Kyushu Island. By the end of 2018, A. woodi had been found in most parts of Japan, except for Wakayama and Kochi prefectures.

Governmental reports

The governmental reports showed that 252 colonies of honey bees in Japan were infested by tracheal mites from 2010 to 2018 (Ministry of Agriculture 2019b). We confirmed that 247 colonies were Japanese honey bees, two colonies were European honey bees, and the species of the other three colonies were not known by the Livestock Hygiene Service Center of each prefecture. One European honey bee colony was found in Ibaraki Prefecture, and the other was in Hiroshima Prefecture. In the case of the infestation in Ibaraki, none of the other eight colonies in the same apiary were infested. For the three colonies of unknown species, two were in Tochigi Prefecture (2016 and 2018) and one was in Aichi Prefecture (2015).

Invasion of small remote islands

Mite-infested Japanese honey bee colonies were found in the Gotō Islands (one of 10 colonies) and Koshiki Islands (two of four colonies; Supplementary Table 1). We found no tracheal mite on the remote islands of Oki, Tsushima, Iki, Tanegashima, Yakushima, or Amami (Table 1). Interviews revealed that the Japanese honey bee had been intentionally introduced to most of these islands, except for Tsushima and Yakushima (Table 1). The colonies on six of the eight islands were introduced from a nearby prefecture on one of the main Japanese islands or from a nearby small island. The colonies on Amami and Oki were introduced from further away (Chiba and Hiroshima prefectures, respectively).

Table 1 Colony introduction, based on interviews
Full size table

Changes in the percentage of mite-infested colonies and mite prevalence in infested colonies

In all tested colonies, the percentage of mite-infested colonies remained fairly constant (38–49%) from 2013 to 2018 (Fig. 2). In the Kyushu region, the percentage of mite-infested colonies showed an increasing trend (Fig. 2, Supplementary Table 2), whereas no obvious trends were observed in any other region.

Fig. 2
figure 2

Changes in the percentage of mite-infested Japanese honey bee colonies from 2013 to 2018. Percentages of all tested colonies infested (Total), and those in each region. The sample sizes (number of tested colonies in each year) are indicated at the top of each graph

Full size image
Table 2 Coefficients and odds ratios of the explanatory variables (K-wings and crawling bees)
Full size table

The average mite prevalence in the infested colonies was highest in 2013 (62%) and remained around 50% (49–54%) from 2014 onward (Fig. 3). There was no recognizable trend in mite prevalence in any of the regions (Fig. 3, Supplementary Table 3).

Fig. 3
figure 3

Changes in average mite prevalence in infested Japanese honey bee colonies from 2013 to 2018 in all of Japan (Total) and each region. The sample sizes (number of mite-infested colonies in each year) are indicated at the top of the graph

Full size image

Observations by hobby beekeepers and mite infestation

In the nationwide survey of observations of K-wing and crawling bees by hobby beekeepers, we obtained data for 206 colonies from 43 prefectures. Among colonies where both K-wing and crawling bees were observed, 52.1% of the bees were infested by tracheal mites (Fig. 4), whereas in colonies where neither K-wing nor crawling bees were observed, the mite prevalence was 14.8%. In colonies where a single sign, either K-wing or crawling bees, was observed, mite prevalence was 33.6 and 40.2%, respectively.

Fig. 4
figure 4

Percentage of individual mite-infested Japanese honey bees in colonies exhibiting certain behaviors, K-wing and/or crawling, as evaluated by hobby beekeepers. The colonies without such behaviors are shown as ‘normal’ in the figure. Sample sizes are indicated at the top of the graph: number of colonies (number of individual bees)

Full size image

The logistic regression analysis showed that models including the explanatory variables K-wing and crawling bees were significant (Table 2), but their interaction was excluded from the best model according to the AIC (Supplementary Table 4). In the logistic regression model including K-wing and crawling bees as explanatory variables, the estimated odds ratio was 1.8211 for K-wing and 2.7262 for crawling bees (Table 2).

Mite infestation of European honey bees

We dissected 5794 European honey bees collected from 1997 to 2019 from 289 colonies. We confirmed that only one colony was infested by the tracheal mite, and the infestation occurred in 2016 (T. Sasaki, pers. comm., 2016). There were no signs of acarapiosis, such as population decline, so this colony was not listed as having acarapiosis in the governmental report for that year (J. Nakamura, pers. comm., 2016). We used these data along with data from the governmental reports, to draw an accurate, updated distribution map of European honey bee (Fig. 5).

Fig. 5
figure 5

Distribution of tracheal mites in European honey bees in Japan. Only three colonies were found to be infested by tracheal mites, one each in Ibaraki, Tokyo, and Hiroshima Prefectures. NA, no available data. Although Kojima et al. (2011b) reported Acarapis woodi DNA in a European honey bee, they did not report where the colony was collected. Thus, their data are not represented on this map

Full size image

Discussion

Our survey revealed that A. woodi rapidly spread across the Japanese islands of Honshu, Shikoku, and Kyushu between 2010 and 2018. According to Nakamura (1988), there were no tracheal mites in Japan before 1988, and we did not find any tracheal mites in bees collected before 2009. However, we investigated only nine colonies of Japanese honey bees collected before 2009, so we could not verify that there were no tracheal mites in Japan earlier than 2010. But considering the quick spread of the mites after 2010, it is plausible that the first invasion of A. woodi might not have been long before 2010.

No tracheal mites were observed in 11 tested colonies in Kochi Prefecture and 30 tested colonies in Wakayama Prefecture. More colonies should be investigated in these prefectures to confirm that A. woodi has not invaded. In Wakayama, traditional beekeepers have long used hollowed-out logs, called gouras, as beehives. In the traditional beekeeping style, people set out a goura and wait for honey bees to arrive and build their hive. Because A. cerana japonica is abundant in Wakayama, people did not need to introduce bee colonies from other prefectures. In addition, a goura is too heavy to transport. Thus, these conditions of abundant honey bees and the traditional beekeeping style might have kept Wakayama Prefecture free of tracheal mites. On the other hand, many A. mellifera colonies are kept in Wakayama during winter because of its warm climate, which is good for overwintering European honey bees. If European honey bees had been infested by tracheal mites and had been able to serve as a vector of the mites to Japanese honey bees, then the Japanese honey bees in Wakayama might have become infested by the mite in early years, for example, around 2010.

Our findings indicate that in every year since 2013, more than 40% of all Japanese honey bee colonies have been infested by the tracheal mite, and mite prevalence in all infested colonies has been constant at around 50%. In other countries, infestation of the European honey bee by tracheal mites has decreased in a few decades (Bailey 1985; Orantes and García-Fernández 1997; Bacandritsos and Saitanis 2004; Moore et al. 2017). However, we did not observe declining trends in either the percentage of infested colonies or in mite prevalence. We hypothesized that infestation might be declining in some local areas, especially those where the tracheal mites were first found, but we did not find any local declines. These results suggest that Japanese honey bees are still under threat of infestation by tracheal mites. The decline in infestation in the European honey bee was observed in long-term surveys, for example, 30 years in the USA (Moore et al. 2017) and 55 years in the UK (Bailey 1985). Thus, we need to continue nationwide surveillance of Japanese honey bees to observe potential infestation declines in the future.

For long-term surveillance, the cooperation of beekeepers is essential. Apis cerana japonica is a wild species, although some colonies are kept by hobby beekeepers. Most hobby beekeepers capture swarming bee colonies in spring and keep them in hand-made box hives. It is easier for researchers to collect Japanese honey bees from these managed colonies than to find and sample wild Japanese honey bee colonies.

In the current study, we showed that observation by hobby beekeepers can be effectively used to identify mite-infested bee colonies. For example, the possibility of mite infestation in a colony in which the beekeeper has noticed K-wing is about 1.8 × (odds ratio) higher than the possibility in a colony where no K-wing was observed. In colonies where crawling bees are observed, the likelihood of infestation is about 2.7 × higher than in colonies where this behavior is not observed. Preliminary diagnosis by hobbyists can therefore be effectively used for identifying mite-infested Japanese honey bee colonies. However, we should note that even if beekeepers do not observe K-wing or crawling bees, about 15% of the bees in the hive may be infested by tracheal mites. Crawling bees are apparent only during winter or early spring, and only when mite prevalence is very high (Sammataro 2013), and mite-infested bees do not always exhibit K-wing (Maeda 2015). Moreover, observations of crawling bees and K-wing are not sufficient for diagnosing the bee colonies as having acarapiosis. Other factors such as viral infections are known to cause similar signs in European honey bees (Sammataro 2013; Shutler et al. 2014). In Japan, final diagnosis must be done by a veterinarian.

All of the small remote islands investigated in the current study are at least 20 km away from Japan’s main islands. A honey bee does not fly more than 20 km when foraging (Couvillon et al. 2015) or swarming (Seeley and Morse 1978). Therefore, natural migration of the mites across more than 20 km of water is not likely to occur. However, people intentionally move Japanese honey bee colonies long distances because beekeeping of this species is fashionable among retired people.

Our survey indicated that Japanese honey bees were intentionally introduced to most of the small remote islands investigated. We must note, however, that we interviewed a small number of beekeepers, so we might have underestimated the introduction of Japanese honey bees. We found mite-infested colonies on Koshiki Island. Beehives were introduced to the Koshiki Islands from 2015 to 2018. During this period, Japanese honey bee colonies in the nearest prefecture, Kagoshima, were already infested by mites, so the introduced bees might have been infested as well. We found one mite-infested colony in the Gotō Islands. Before 1990, there were wild Japanese honey bees in the Gotō Islands, but in 2004, no one found Japanese honey bees there (Y. Nakamura and T. Miyazaki, pers. comm.). To enjoy beekeeping, more than 10 colonies had been brought to the Gotō Islands in 2008, but because no tracheal mites were detected in Japan in 2008, those introduced colonies would not have been infested. Considering these situations, we speculate that additional colonies of the Japanese honey bee might have been introduced to the Gotō Islands after 2016 from Kyushu or another region of Japan.

On the six other remote islands, we did not find any tracheal mites. Tsushima and Yakushima have no introduced honey bee colonies, and thus it is plausible that there are no tracheal mites there. The colony introduced to Tanegashima was from Yakushima, where no tracheal mites were found. In the case of Oki and Iki, the colonies had been intentionally introduced before A. woodi had invaded the source region of the introduced colonies. Therefore, no tracheal mites were detected on these islands. On Amami, a colony was introduced from Chiba Prefecture in 2016, where there was a heavy infestation of tracheal mites at that time. Fortunately, however, it appears that the introduced colony was not infested by the mites, because we confirmed in April 2019 that a colony derived from the introduced one had no tracheal mites.

When Japanese honey bee colonies are infested with tracheal mites, the overwinter colony mortality is drastically increased (Maeda and Sakamoto 2016). For example, the average mite prevalence was 50% in the current study, leading to a predicted overwinter colony mortality of about 80% (Maeda and Sakamoto 2016). Currently, about 40% of Japanese honey bee colonies are infested by tracheal mites, raising concerns that the population of Japanese honey bees may be decreasing. To control tracheal mites on European honey bees, some chemicals have been registered and used abroad, for example, in the USA and Canada (Sammataro et al. 2013). On the contrary, no veterinary medicine is available to control the tracheal mites in Japan. In addition, it is not possible to treat wild colonies of Japanese honey bees with any chemicals to control tracheal mites. Therefore, the most effective way to control the rapid spread of tracheal mites may be to stop the intentional movement of beehives.

In contrast to the epidemic situation in Japanese honey bees, European honey bees in Japan have rarely been found to be infested by A. woodi. The reason why the tracheal mite is rarely found from the European honey bee is still under investigation. Danka and Villa (2000) showed that two-thirds of A. mellifera queens sold commercially in 2000 had mite resistance. In addition, some miticides used for controlling varroa mites might reduce the number of tracheal mites (Eischen et al. 1986). These genetic and chemical factors might explain the decreasing mite prevalence in the European honey bee. Sakamoto et al. (2017) reported that Japanese honey bees are more susceptible to infestation by tracheal mites than European honey bees and focused on grooming behavior as the resistance mechanism. To examine the decline of tracheal mite infestation from the viewpoint of the resistance traits such as grooming behavior, such behaviors need to be compared before and after the decline. No one observed grooming behavior against the tracheal mite in the European honey bee before the honey bee became resistant to the mites. However, Sakamoto et al. (2019) have investigated grooming behavior and susceptibility of Japanese honey bees to tracheal mites. Long-term research on the relationship between A. woodi and A. cerana japonica will help us to understand the evolution of parasite-resistance behavior.