Mortality of Varroa Destructor in Honey Bee (Apis Mellifera) Colonies During Winter

Apidologie 32 (2001) 223–229 223 © INRA/DIB-AGIB/EDP Sciences, 2001

Original article

Mortality of Varroa destructor in honey bee (Apis mellifera) colonies during winter

Ingemar FRIES*, Silvia PEREZ-ESCALA

Department of Entomology, Swedish University of Agricultural Sciences, Box 7044, 75007 Uppsala, Sweden

(Received 29 August 2000; revised 30 January 2001; accepted 23 February 2001)

Abstract – The change in infestation levels of the mite Varroa destructor Anderson and Trueman on adult bees during periods with little or no brood rearing (late October/early November to early Febru- ary) was investigated in 10 colonies for two consecutive years in a Swedish climate (N57°06’E18°16’). The results do not support the hypothesis that mites become concentrated on the remaining bees as bees die off from the winter cluster. When the number of all mites recovered from dead bees or from debris was used to calculate mites per dead bee, the level of infestation per bee was not significantly different between samples of live bee and dead bees. For modelling purposes, we presently find no reason to differentiate the mortality rates of bees and mites during periods when there is no or limited amounts of brood in the colonies, although the connection between bee mortality and mite mortality may not be as direct as previously assumed.

Varroa destructor / population dynamics / winter mortality

1. INTRODUCTION growing information on mite biology, there also is good data on the intrinsic growth rate The biology of the mite Varroa destruc- of the mite population during periods of tor Anderson and Trueman, 2000 (formerly brood rearing (Calatayud and Verdu, 1995; V. jacobsoni Oudemans) in colonies of the Kraus and Page, 1995; Marcangeli et al., western honey bee (Apis mellifera) has been 1995). However, mite mortality is poorly described in some detail (Boot et al., 1994; understood during periods of mite popula- Donzé and Guerin, 1994; Martin, 1994; tion growth, and even more so during periods Martin, 1995; Donzé et al., 1996). Parallel to with no brood rearing and mite population

* Correspondence and reprints E-mail: [email protected] 224 I. Fries, S. Perez-Escala decline. For purposes of modelling, mite We studied winter mite mortality in a population dynamics, data on mite mortality cold climate. In particular, we investigated is critical. whether mites become concentrated on the remaining bees as bees die off during winter. Based on assumptions and available data, there have been three attempts to develop mathematical models of the population dynamics of V. destructor mites (Calis et al., 2. MATERIALS AND METHODS 1999; Fries et al., 1994; Martin, 1998). It has been assumed in two of these models To investigate the changes in infestation that mites and bees have a similar death rate level of mites on adult bees during periods during winter (Calis et al., 1999; Fries et al., with little or no brood rearing, we equipped 1994), yielding a daily mite death rate of 10 colonies with net screen bottoms for col- 0.004 per day during periods of no brood lection of mites below the screen (30 mm rearing. The third model used a winter mite between net and collection tray), and for mortality of only 50% of other models collection of bees dying inside the colony (0.002) (Martin, 1998). The latter death rate above the screen (20 mm between net and was based on the average of two citations lower frame bars), for two consecutive years. in the literature in which the number of mites The colonies were also equipped with an that died with bees outside the colonies was not considered (Korpela et al., 1993; external cage that allowed collection of bees Moosbeckhofer, 1991). The death rate of leaving the hives to die outside the colonies. 0.004 was based on the assumption that If mites dislodged from bees collected out- approximately half of the population of bees side the hive, such mites were not collected. may die in a climate similar to Scandinavia Before the experiment started each year, all (Avitabile, 1978) and that mites are ran- colonies were examined and found free from domly distributed in the winter cluster sealed brood. At the onset of the experi- (Ritter et al., 1989) and do not change host ment, a sample of approximately 1 dl of as the bees die (Müller, 1987). Recent data, bees (~ 250) were taken from each colony however, indicate that mites may change and examined for mites. Dead bees to be their host during winter and that they actu- examined for mites and mites in the debris ally may leave dying hosts before the bees were collected approximately every two drop from the cluster (Bowen-Walker et al., weeks until the end of the experiment, when 1997). Bowen-Walker et al. (1997) studied a sample of 1 dl of live bees again was taken the distribution of mites in one colony dur- from each colony and examined for mites. ing a period when they assumed there was no brood rearing under British climate con- The experimental site was located on ditions and found an increase in infestation Gotland (N57°06’; E18°16’). The first year, level of the adult bees from 12.9% in Octo- the experiment started on October 20 and ber–December to 36.8% in January and was terminated on February 2 (mean daily February. Bowen-Walker et al. (1997) inter- temperature ± standard deviation; 1.94 ± preted this result as a differential mortality 3.31 °C). The second year the experiment rate of mites and bees. Later, Bowen-Walker started on November 16 and was terminated and Gunn (1998) used the data from the on February 5 (mean daily temperature ± same colony and concluded: “As overwin- standard deviation; 2.04 ± 2.45 °C). The tering bees start to die of old age and dis- experiments were terminated in early Febru- ease, more and more mites, potentially ary because honey bees may initiate some carrying disease agents, will become con- brood rearing at this time under similar cli- centrated on the remaining bees”. mate conditions (Avitabile, 1978). Varroa mite mortality during winter 225

3. RESULTS mortality, any change in infestation rate of live bees due to mites leaving dying hosts The numbers of mites recovered per live will be difficult to detect. Nevertheless, if bee, in the debris, on dead bees, and on dead mites do become concentrated on remain- bees and debris combined are tabulated in ing bees as bees die off, we would expect Table I. a linear relationship between the variables in There was no significant change in the Figure 1. However, the correlation coeffi- infestation levels in the colonies between cient is low (r = –0.12) and not significant the initiation and termination of the experi- (P = 0.62) indicating that no linear rela- ments in either of the two years (for each tionship between the variables exists. In year; P > 0.05, paired t-test, 9df). The same other words, an increase in bee mortality was true when the data from both years were does not seem to be linked to an increase in combined (P > 0.05, paired t-test, 19df). The live bee infestation levels. infestation level of the dead bees was sig- From the presented data (Tab. I, Fig. 1), nificantly lower than the level on the live we find no support for the conclusion pre- bee samples during both years (P < 0.001, sented by Bowen-Walker and Gunn (1998) paired t-test, 9df). However when the num- that mites become concentrated on the ber of mites recovered in the debris was remaining bees as bees die from the winter combined with the mites found on dead bees cluster. What is obvious, however, is the to give the recovered number of mites per large degree of variation between colonies dead bee, the difference was not significant with an apparent increase in infestation rate for either of the years (P > 0.05, paired in some colonies and an apparent decrease of t-test, 9df), or for the two years combined similar magnitude in others (Tab. I). Sam- (P > 0.05, paired t-test, 19df). To avoid dam- pling errors may explain some of the aging the colonies, they were not checked observed variation, in particular during the for brood at the end of the experiment in first year with a low infestation level. In early February. We assume that brood rear- the second year the variation between ing was very limited in early February under colonies prevailed in spite of a higher infes- the prevailing weather conditions. Thus, it is tation level, indicating that much variation plausible that minimal, if any, mite repro- may be due to individual differences duction occurred during the experiment con- between colonies. This variation remains sidering the climatic conditions and the fact unexplained. that all colonies did not have sealed brood at the onset of the experiment each year. In the present experiment, we can not In Figure 1 a scatter diagram of the determine if the mites found on dead bees change in infestation level in live bees at fell from the cluster and remounted dead the start and at the end of the experiment bees. Nor can we determine if mites in the (end level minus start level) vs. the number debris fell from the cluster or from already of dead bees is plotted. The correlation coef- dead bees. When all mites recovered were ficient is low (r = –0.12) and non-signifi- considered together, the number of mites cant (P = 0.62) indicating that there is no per dead bee was not significantly different linear relationship between the two vari- from number of mites per live bee over all ables. colonies. There was a non-significant ten- dency that the number of mites recovered per dead bee was smaller than the number of 4. DISCUSSION mites recovered per live bee. We collected all dead bees both inside and outside the A problem with the presented data is hive entrance, but if mites fell from bees on the level of bee mortality. With low bee the outside, they were not retrieved (31% 226 I. Fries, S. Perez-Escala s recovered before after bees on dead bees in debris per dead bee Number of mites recovered per live bee before and after the experiments, number dead bee, mite 1 0.013 0.018 1 19912 0.0143 0.0066454 0.008409 5 0.037 839 0.013 0.018 1 6 0.049 0.030 0 809 7 0.07610969 0.052 0.010 7 4 8 0.044 5 0.012 2 9 0.04527 0.004 1 0.029 0.008 10 0.012487 15 862 3 1 195 48 0.085 2541 0.055 157 2 0.349 4 36 16 2 0.067 3 0.024 20 3 0.408 0.486101 0.006 1 4 66 0.228 222 20 29 0.015 5 2235 1 6 178 0.024 0.012 7 0.133 5 0.054 0.058 628 0.079 0.090294 0.200 19 1 0.011 100 9 0.206 0.151 0.131 3 904 819 26 285 36 0.151516 0.105 974093 33 0.025 18 0.241 103 0.154 0.044 0.152 1 0.005 1 1 0.111 1 9 35 68 92 0.167 0.009 0.086 143 0.225 76 19 37 0.057 2 290 0.136 0.132 49 68 0.051 Year 1 Year 2 Table I. in colony debris, and total number of recovered dead mites per bee. Colony Mites per live bee, Mites per live bee, Number of dead Number of mites Number of mites Total recovered mites Varroa mite mortality during winter 227

mortality may not be as direct as previously assumed (Calis et al., 1999; Fries et al., 1994). Further studies of mite mortality, both inside and outside of bee colonies, during periods with and without brood are war- ranted to further refine the predictive value of existing models of the mite population dynamics.

Figure 1. Scatter diagram of infestation increase Résumé – Mortalité de Varroa destructor of live bees between the start and the end of the dans les colonies d’abeilles domestiques experiment (end infestation rate minus start infes- (Apis mellifera) au cours de l’hiver. tation rate) vs. number of dead bees recovered. Bowen-Walker et al. (1997) ont étudié la The correlation coefficient (r = –0.12) is not sig- répartition de l’acarien Varroa destructor nificant (P = 0.62). Anderson et Trueman dans une colonie d’abeilles au cours d’une période supposée sans production de couvain dans les condi- of the dead bees were collected outside the tions climatiques britanniques. Ils ont trouvé hives). Thus, the number of mites per dead une augmentation de l’infestation des bee may have be slightly underestimated, ouvrières passant de 12,5 % en octobre– further weakening the concentration hypoth- décembre à 36,8 % en janvier–février. Ils esis. Factors which may add to the differ- ont interprété ce résultat comme étant dû à ences between this study and the results of une différence du taux de mortalité des Bowen-Walker et al. (1997) could be dif- abeilles et des acariens. Plus tard, sur la base ferent opportunities for brood rearing during des données de cette même colonie, Bowen- the studied period (climate) as well as dif- Walker et Gunn (1998) ont conclu : « Quand, ferent opportunities for mites to re-mount durant l’hivernage, les abeilles se mettent à adult bees after falling from the cluster (due mourir de vieillesse ou de maladie, de plus to differences in bottom board construction). en plus d’acariens, vecteurs potentiels The variation found between the level of d’agents pathogènes, se concentrent sur les infestation among colonies emphasizes the abeilles restantes ». importance of sample size when comparing Nous avons étudié la mortalité hivernale the mortality rate of mites and bees. The dans une région à climat froid (N57°06’ ; conclusion that mites become concentrated E18°16’) et recherché particulièrement si on the remaining bees as bees die off is les acariens se concentraient sur les abeilles based on data from one colony only survivantes au fur et à mesure que la mor- (Bowen-Walker et al., 1997). The data pre- talité des abeilles augmentait au cours de sented here, based on 20 colonies under l’hiver. Durant deux années consécutives conditions when brood rearing is unlikely nous avons comparé sur 10 colonies dépour- or very limited, and in which mites that fell vues de couvain le taux d’infestation des through the bottom screen probably did not abeilles en début d’expérience (de fin re-enter the bee cluster, does not support octobre ou mi-novembre) et en fin d’expé- this conclusion. For modelling purposes, we rience (début février). Les abeilles mortes presently find no reason to differentiate the et les débris de la colonie ont été récoltés mortality rates of bees and mites during peri- environ chque semaine durant toute l’expé- ods when there is no or limited amounts of rience pour y rechercher les acariens morts. brood in the colonies, although the connec- Le tableau I donne le nombre d’acariens par tion between bee mortality and mite abeille vivante, le nombre d’acariens trouvés 228 I. Fries, S. Perez-Escala dans les débris et sur les abeilles mortes et et al. (1997) untersuchten die Verteilung leur somme, qui donne le nombre d’aca- von Milben (Varroa destructor Anderson riens par abeille morte. On n’a trouvé de und Trueman) in einem Bienenvolk inner- modification significative (P > 0,05, test-t halb eines Zeitraums, für den sie annahmen, par paires, 9df) pour aucune des deux dass unter britischen Klimabedingungen années, ni pour les deux années réunies keine Bruterzeugung vorhanden war. Hier- (P > 0,05, test-t par paires, 19df). Pour cha- bei fanden sie eine Zunahme des Befalls der cune des deux années le taux d’infestation Bienenarbeiterinnen von 12,5 % im Oktober des abeilles mortes était significativement und Dezember auf 36,8 % in Januar und plus faible que celui des abeilles vivantes Februar. Sie führten dieses Ergebnis auf eine (P > 0,001, test-t par paires, 9df). Si l’on unterschiedliche Sterberate von Bienen und regroupait les acariens trouvés dans les débris Milben zurück. Später schlossen Bowen- et ceux sur les abeilles mortes, pour obtenir Walker und Gunn (1998) auf Grundlage die- le nombre d’acariens trouvés par abeille ser Daten, dass “Wenn überwinternde Bie- morte, la différence n’était significative pour nen durch Alterung oder Krankheit sterben, aucune des deux années (P > 0,05, test-t par mehr und mehr Milben als potenzielle Trä- paires, 9df), ni pour les deux années prises ger von Krankheitserregern auf den übri- ensemble (P > 0,05, test-t par paires, 19df). gen Arbeiterinnen konzentriert werden”. Les différences trouvées entre colonies sou- Wir untersuchten die Wintersterblichkeit in lignent l’importance de la taille de l’échan- einem kalten Klimabereich (N57°06’; tillon lorsque sont comparés les taux de mor- E18°16’) und untersuchten besonders ob die talité des acariens et des abeilles. La Milben sich mit fortschreitender Bienen- conclusion, selon laquelle les acariens se sterblichkeit auf den überlebenden Bienen concentrent sur les abeilles survivantes, n’a konzentrieren. In zwei aufeinander folgen- été tirée qu’à partir de l’étude d’une seule den Jahren untersuchten wir in 10 brutfreien colonie (Bowen-Walker et al., 1997). Les Völkern im späten Oktober/November die données présentées ici, obtenues à partir de Befallsrate der lebenden Bienen und vergli- 20 colonies dans des conditions où la pro- chen sie mit der Befallsrate im frühen duction de couvain était improbable ou très Februar. Tote Bienen und das Gemüll wur- réduite, ne confirment pas cette conclusion. den etwa wöchentlich bis zum Versuchsende Actuellement il n’y a donc aucune raison, gesammelt und auf tote Milben untersucht. lorsqu’on construit des modèles de popula- Die Anzahl von pro lebender Bienen gefun- tion, de considérer séparément la mortalité dener Milben, von Milben im Gemüll, auf des abeilles et des acariens au cours des toten Bienen und die Summe von Milben périodes où la production de couvain est im Gemüll und auf toten Bienen sind in nulle ou réduite. Pourtant la relation entre Tabelle I aufgeführt. In keinem der beiden la mortalité des abeilles et la mortalité des Jahre gab es eine signifikante Änderung acariens n’est probablement pas aussi directe (P > 0,05, gepaarter t-Test, 9df), auch nicht qu’on le supposait jusqu’à présent (Calis wenn beide Jahre zusammengefasst wurden et al., 1999 ; Fries et al., 1994). (P > 0,05, gepaarter t-Test, 19df). In beiden Jahren war die Befallsrate der toten Bienen Varroa destructor / dynamique popula- signifikant niedriger als die der lebenden tions / mortalité hivernale Arbeiterinnen (P < 0,001, gepaarter t-Test, 9df). Wenn die Milben im Gemüll mit den Milben auf den toten Bienen zusammenge- Zusammenfassung – Sterblichkeit von fasst wurden, um die Anzahl gefundener Varroamilben (Varroa destructor) in Milben pro toter Biene zu ermitteln, ist die- Honigbienenvölkern (Apis mellifera) ser Unterschied allerdings weder in einem während des Winters. Bowen-Walker der beiden Jahre signifikant (P > 0,05, Varroa mite mortality during winter 229 gepaarter t-Test, 9df) noch in beiden Jahren Oud. on overwintering honeybee (Apis mellifera) zusammengenommen (P > 0,05, gepaarter workers and changes in the level of parasitism, Parasitology 114, 151–157. t-Test, 19df). Calatayud F., Verdu M.J. (1995) Number of adult Die zwischen den Völkern gefundenen female mites Varroa jacobsoni Oud. on hive debris Unterschiede belegen die Wichtigkeit einer from honey bee colonies artificially infested to ausreichenden Probengröβe bei der Unter- monitor mite population increase (Mesostigmata: suchung der Sterbensraten von Milben und Varroidae), Exp. Appl. Acarol. 19, 181–188. Bienen. Die Schlussfolgerung, dass die Mil- Calis J.N.M., Fries I., Ryrie S.C. (1999) Population modelling of Varroa jacobsoni, Apidologie 30, ben sich auf den übrigbleibenden Bienen 111–124. konzentrieren wenn die Bienen sterben Donzé G., Guerin P.M. (1994) Behavioral attributes wurde auf Grund der Untersuchung nur and parental care of Varroa mites parasitizing hon- eines Volkes gezogen (Bowen-Walker et al., eybee brood, Behav. Ecol. Sociobiol. 34, 305–319. 1997). Die hier dargestellten Daten, erstellt Donzé G., Herrmann M., Bachofen B., Guerin P.M. auf Grundlage von 20 Völkern unter Bedin- (1996) Effect of mating frequency and brood cell gungen, in denen Bruterzeugung unwahr- infestation rate on the reproductive success of the honeybee parasite Varroa jacobsoni, Ecol. Ento- scheinlich oder sehr begrenzt ist, unterstüt- mol. 21, 17–26. zen diese Schlussfolgerung nicht. Zur Zeit Fries I., Camazine S., Sneyd J. (1994) Population gibt es daher bei der Erstellung von Popu- dynamics of Varroa jacobsoni: a model and a lationsmodellen keinen Grund, die Bienen – review, Bee World 75, 5–28. und Milbensterblichkeit zu Zeiten geringer Korpela S., Aarhus A., Fries I., Hansen H. (1993) Var- oder keiner Bruttätigkeit getrennt zu betrach- roa jacobsoni Oud. in cold climates: population ten. Dennoch ist die Beziehung zwischen growth, winter mortality and influence on survival of honey bee colonies, J. Apic. Res. 31, 157–164. Bienensterblichkeit und Milbensterblich- Kraus B., Page R.E. (1995) Population growth of Var- keit vermutlich nicht so direkt wie bislang roa jacobsoni Oud. in Mediterranean climates of angenommen wurde (Calis et al., 1999; Fries California, Apidologie 26, 149–157. et al., 1994). Marcangeli J., Eguaras M., Fernandez N. (1995) Pop- ulation growth of Varroa jacobsoni (Gamasida: Varroa destructor / Populationsdynamik / Varroidae) in colonies of Apis mellifera (Hymenoptera: Apidae) in temperate climates using Wintersterblichkeit Camazine’s model, Apiacta 30, 13–19. Martin J. (1998) A population model for the ectopara- sitic mite Varroa jacobsoni in honey bee (Apis mel- REFERENCES lifera) colonies, Ecol. Model. 109, 267–281. Martin S. (1994) Ontogenesis of the mite Varroa jacob- Anderson D.L., Trueman J.W.H. (2000) Varroa jacobsoni Oud. in worker brood of the honey bee Apis soni (Acari: Varroidae) is more than one species, mellifera L. under natural conditions, Exp. Appl. Exp. Appl. Acarol. 24, 165–189. Acarol. 18, 87–100. Avitabile A. (1978) Brood rearing in honeybee colonies Martin S. (1995) Ontogenesis of the mite Varroa jacob- from late autumn to early spring, J. Apic. Res. 17, soni Oud. in drone brood of the honey bee Apis 69–73. mellifera L. under natural conditions, Exp. Appl. Boot W.J., Sisselaar D.J.A., Calis J.N.M., Beetsma J. Acarol. 19, 199–210. (1994) Factors affecting invasion of Varroa jacobsoni (Acari: Varroidae) into honey bee, Apis mel- Moosbeckhofer R. (1991) Varroaverluste während der lifera (Hymenoptera: Apidae), brood cells, Bull. Überwinterung, Bienenvater 112, 300–303. Entomol. Res. 84, 3–10. Müller M. (1987) Befallsentwicklung der Milbe Var- Bowen-Walker P.L., Gunn A. (1998) Inter-host trans- roa jacobsoni in den Wintermonaten, Allg. Dtsch. fer and survival of Varroa jacobsoni under simu- Imkerztg. 21, 15–22. lated and natural winter conditions, J. Apic. Res. Ritter W., Kerkhof U., Pätzold S. (1989) The distri- 37, 199–204. bution of Varroa jacobsoni Oud. in the winter clus- Bowen-Walker P.L., Martin S.J., Gunn A. (1997) Pref- ter of Apis mellifera carnica, Apidologie 20, erential distribution of the mite Varroa jacobsoni 516–517.