The parasitic mite Varroa-jacobsoni is one of the most serious pests of Western honey bees, Apis mellifera. The mites parasitize adult bees, but reproduction only occurs while parasitizing on honey bee brood. Invasion into a drone or a worker cell is therefore a crucial step in the life of Varroa mites. In this thesis, individual mites, the population of mites and characteristics of honey bee brood cells have been studied in relation to invasion behaviour. In addition, a simple model has been developed to study which selective forces may have shaped the strategies used by Varroa mites with respect to the allocation of invasion and subsequent reproduction over drone and worker brood cells of the honey bee.Invasion behaviour of individual Varroa mites (chapter 1).Preceding invasion, Varroa mites are carried close to a suitable brood cell by a honey bee. The mite moves directly from the bee into the selected brood cell, crawls between the larva and the cell wall, and moves on to the bottom of the cell. At the moment of leaving the bee, the mite cannot touch the larva. It still has to cover the distance from the cell rim to the larva, which measures 4-7 mm in cells that are attractive to the mites. Thus, information to decide whether to stay on the bee or to invade a brood cell is acquired at a distance from the larva, possibly by a volatile chemical or by differences in temperature. Since invasion only occurs when a bee brings a mite close to a suitable brood cell, the chance of being carried close enough may well limit the number of mites that invade. If so, population growth of the mites is limited in turn, because the mites reproduce exclusively inside brood cells.Invasion of brood cells by a population of Varroa mites.Invasion into worker brood cells (chapter 2 & 3)Within a day after emergence from a brood cell, i.e. the moment when Varroa mites begin their residence period on adult bees, some mites invade a new brood cell. The percentage of mites on adult bees that invade per day depends on the number of cells suitable for invasion and on the number of bees in the colony, regardless of the time that the mites have stayed on adult bees. The more cells and the fewer bees, the higher is the percentage of mites that invade per day, as expected when invasion is limited by the chance of being carried close enough to a suitable brood cell. This can be understood as follows. Since only one or a few bees can be near a cell simultaneously, the chance of being carried close enough for invasion increases when the number of brood cells increases. In addition, in a smaller bee colony, but with the same number of brood cells, the mites are spread over a smaller number of bees. The number of bees that come close enough to a suitable cell stays the same, and therefore the mite's chance of invasion is increased.The percentage of mites that invade per day decreases when young open brood, still too young to be suitable for invasion by mites, is present. This decrease in invasion rate may arise because the mites prefer to be carried by young adult bees, which are likely to stay in the brood nest area. Within the brood nest area these young bees are divided over areas with brood cells that are suitable and areas that are unsuitable for invasion by mites. Hence, an increase in the amount of unsuitable open brood may keep part of the preferred young bees away from the suitable brood cells, and may thus decrease the invasion rate.If invasion is limited by the chance of being carried close enough to a brood cell, the spatial distribution of the mites inside the colony may affect invasion. In the areas where invasion occurs, the mite density on the bees will decrease. The mites will redistribute spatially by movement of the bees that carry them and by moving from bee to bee, but depending on the rate of this phoretic process invasion will be more or less limited. However, the rate of invasion was similar in bee colonies in which either 600 brood cells were available for mite invasion during one day or three times 200 brood cells were available during three days, whereas the colonies were comparable in all other respects. Thus, on a time scale of days the process of redistribution of mites inside the colony seems to be fast enough to prevent a significant effect of the period of brood cell availability on the rate of invasion. This is important for application of biotechnical control methods in which brood combs are introduced into the colony to trap mites. The 'trapping combs' are removed from the colony and the mites inside the cells are killed. Our results have shown that the number of cells used for trapping the mites is crucial, whereas the period during which the cells are available to the mites is of minor importance.Invasion into drone brood cells (chapter 4)Invasion by a population of mites into drone brood cells is similar to invasion into worker cells, except that invasion into &one cells is a much faster process. When invasion is compared between colonies with either exclusively worker cells or exclusively drone cells, Varroa mites invade a drone cell about 12 times more frequently than a worker cell. Hence, when both types of brood cells are available a biased distribution of 12 times more mites in drone cells than in worker cells is expected based on the differential frequency of invasion. This expected bias is larger than the bias actually found in colonies with both types of brood cells, which measures on average 8 times more mites per drone cell than per worker cell. The lower actual bias when compared to the expected one may be understood as follows. In normal honey bee colonies invasion into drone and worker cells is probably more or less segregated in time. Since the frequency of invasion is much higher per drone cell than per worker cell, the number of mites on bees will decrease much faster during periods when drone cells are abundantly present. Fewer mites will invade drone cells than expected when a constant number of mites on bees is assumed. Hence, the actual distribution over drone and worker cells may be less biased than expected from the differential frequency of invasion per cell. In addition, the biased distribution is sufficiently explained by the differential frequency of invasion per cell alone. There is no reason to believe that mites respond to the presence of nearby drone brood cells by refraining from invasion into worker brood cells, thus causing the biased distribution over drone and worker cells. Since the rate of invasion into drone brood cells is high, a trapping method using drone combs may be very effective in controlling the Varroa mite. When no other brood is present, 462 drone cells were estimated to be sufficient to trap 95% of the mites in a colony of 1 kg of bees.Effect of the period spent on adult bees on reproduction of the mites (chapter 5)No correlation has been found between the length of the period that Varroa mites stay on adult bees (1-20 days) prior to invasion and the total number of offspring per mite, the number of viable daughters per mite, the fraction of mites without offspring, and the fraction of mites that produces only male offspring. Thus, reproduction of the mites is apparently independent of the period that the mites reside on adult bees prior to invasion into brood cells.Mortality of mites during the period spent on adult bees (chapter 5)Mortality of Varroa mites, as measured by counting mites fallen on the bottom of the hive, occurs primarily right after emergence from the brood cell. When brood containing mites emerges during one day, 18% of the mites that have been present on the emerging brood is found on the bottom of the bee hive at the end of that day. Part of these mites may already have died inside the capped brood cells, and have fallen down after cleaning of cells by the bees. At the second and third day following emergence, respectively 4% and 2% of the mites on adult bees is found on the bottom, whereas from the fourth day on (up to 23 days) only 0.6% of the mites on adult bees is found on the bottom per day. Since the number of mites on the bottom of the hive will be strongly associated with the number of freshly emerged mites, counting the number of dead mites on the bottom may be a useful tool to estimate infestation levels in honey bee colonies.Attractiveness of brood cells to Varroa mitesThe attractive period of worker and drone brood cells (chapter 6)Worker brood cells are attractive to Varroa mites from 15-20 hours preceding cell capping, whereas drone cells are attractive from 40-50 hours preceding cell capping. Since the attractive period of drone cells is 2-3 times longer than that of worker cells, drone cells are consequently expected to be invaded 2-3 times more frequently. Actually, a drone cell is invaded 12 times more frequently than a worker cell. Hence, more factors must be involved in causing this difference in frequency of invasion. When the frequency of invasion is proportional to the surface of a brood cell, more mites are expected per drone cell due to its 1.7 times larger surface than a worker cell. Taken together, this would result in a 3.4-5.1 times higher frequency of invasion, which is clearly much lower than the 12 times actually found. Therefore, the higher frequency of invasion into drone cells may be attributed for an important part to differences in the information mites use to select a cell for invasion, either quantitatively or qualitatively.Effect of larva-cell rim distance on attractiveness of brood cells (chapter 7)Varroa mites are not randomly distributed over different types of cell which contain similar larvae. Per cell, more mites invade into shorter and narrower cells than control cells, whereas fewer mites invade into longer and wider cells. The period during which cells are attractive to mites varies among the different cell types, and whether in a certain type of cell more or fewer mites are found in comparison to control cells, is correlated with the length of the attractive period of that type of cell. The type of cell also affects the distance from larva to cell rim in the period preceding cell capping. When this distance is larger in comparison to control cells with larvae of the same age, the attractive period of the brood cells is shorter and vice versa. Since in all cell types the distance from larva to cell rim continuously decreases preceding cell capping, this negative correlation suggests that there is a critical larva-rim distance under which brood cells are attractive to the mites. Then, the length of the attractive period of brood cells depends on the moment this critical distance is reached. The distribution of mites over different cell types in turn results from differences in the attractive period. In normal drone and worker brood cells the critical larva-rim distance for invasion is 7-8 mm.Effect of methyl palmitate on attractiveness of brood cells (chapter 8)Since Varroa mites decide at some distance from the larva whether to stay on a bee or invade into a cell, they may well use a volatile chemical to select a brood cell. A few aliphatic esters, predominantly methyl palmitate, have been claimed to be this volatile signal for the mites for two reasons. The mites respond to the esters in an olfactometer (Le Conte et al., 1989), and the levels of the esters in worker and drone larvae may explain that drone cells are attractive during a longer period and are invaded more frequently than worker cells (Trouiller et al., 1992). However, invasion itself is unaffected by application of methyl palmitate to brood cells. In addition, analysis of headspace volatiles above attractive brood cells showed hundreds of components in the volatile blend, but in only 2 of 17 analyses a trace of methyl palmitate was found. Hence, there is no reason to believe that methyl palmitate is used as a signal for invasion by the mites.Reproductive strategy of Varroa mites (chapter 9)Since reproductive success of Varroa mites is higher in drone cells than in worker cells, the question arises why the mites do not restrict invasion to drone cells. Therefore, a simple model using population growth as a fitness measure has been developed to study under which circumstances specialization on drone brood would be a better strategy to adopt than reproduction in both types of cell. For European A.mellifera, the model suggests that if mites have to wait less than 7 days on average before they can invade a drone cell, specialization on drone brood would be a better strategy. This is close to the estimated waiting time of 6 days. Hence, small differences in reproductive success in drone and worker cells, and in the rate of mortality may determine whether specialization on drone brood will be promoted or not. In European A. mellifera colonies, Varroa mites invade both drone and worker cells, but specialization on drone brood cells seems to occur to some extent because a drone cell is more frequently invaded than a worker cell. In the parasite-host association of V. jacobsoni with African or Africanized A. mellifera or with A. cerana, the mites also invade both drone and worker cells, but the mites specialize on drone brood with respect to reproduction since a large percentage of the mites in worker brood do not reproduce. Only in the parasite-host association of Euvarroa sinhai, a mite closely resembling V . jacobsoni, and A.florea specialization is complete because these mites only invade drone brood.Does current knowledge of invasion behaviour help in controlling the Varroa mite?Our research on invasion behaviour did not result in a method in which Varroa mites are controlled using attractant or repellent chemicals. We still have to identify the signal the mite uses to invade a brood cell, although we know that mites perceive this signal at a distance from the larva and that the larva-cell rim distance affects the response of the mites to it. However, our results on invasion behaviour are useful to understand the possibilities and limitations for improvement of biotechnical control methods. We now know how many drone or worker cells are needed in a 'trapping comb' to catch a certain percentage of the mites in a colony. In theory, control methods that make use of 'trapping combs' are simple. In practice however, the methods may become complicated because application is integrated with other activities by the beekeeper like building of new colonies and swarm prevention. In addition, application of biotechnical control methods is usually labour intensive. Our results can be applied to design the simplest method that is sufficiently effective. This will remain an important application in future. Since much research is nowadays directed to breed honeybees that are less susceptable to Varroa mites (Woyke, 1989b; Büchler, 1994; Moritz, 1994), the effectiveness of control methods needed for control may decrease, which allows simplification of control methods. By combining simple 'trapping comb' methods and breeding of Varroa -less susceptable honey bees, there is a clear perspective for beekeeping without the use of acaricides to kill Varroa mites.

The parasitic mite Varroa-jacobsoni is one of the most serious pests of Western honey bees, Apis mellifera. The mites parasitize adult bees, but reproduction only occurs while parasitizing on honey bee brood. Invasion into a drone or a worker cell is therefore a crucial step in the life of Varroa mites. In this thesis, individual mites, the population of mites and characteristics of honey bee brood cells have been studied in relation to invasion behaviour. In addition, a simple model has been developed to study which selective forces may have shaped the strategies used by Varroa mites with respect to the allocation of invasion and subsequent reproduction over drone and worker brood cells of the honey bee.Invasion behaviour of individual Varroa mites (chapter 1).Preceding invasion, Varroa mites are carried close to a suitable brood cell by a honey bee. The mite moves directly from the bee into the selected brood cell, crawls between the larva and the cell wall, and moves on to the bottom of the cell. At the moment of leaving the bee, the mite cannot touch the larva. It still has to cover the distance from the cell rim to the larva, which measures 4-7 mm in cells that are attractive to the mites. Thus, information to decide whether to stay on the bee or to invade a brood cell is acquired at a distance from the larva, possibly by a volatile chemical or by differences in temperature. Since invasion only occurs when a bee brings a mite close to a suitable brood cell, the chance of being carried close enough may well limit the number of mites that invade. If so, population growth of the mites is limited in turn, because the mites reproduce exclusively inside brood cells.Invasion of brood cells by a population of Varroa mites.Invasion into worker brood cells (chapter 2 & 3)Within a day after emergence from a brood cell, i.e. the moment when Varroa mites begin their residence period on adult bees, some mites invade a new brood cell. The percentage of mites on adult bees that invade per day depends on the number of cells suitable for invasion and on the number of bees in the colony, regardless of the time that the mites have stayed on adult bees. The more cells and the fewer bees, the higher is the percentage of mites that invade per day, as expected when invasion is limited by the chance of being carried close enough to a suitable brood cell. This can be understood as follows. Since only one or a few bees can be near a cell simultaneously, the chance of being carried close enough for invasion increases when the number of brood cells increases. In addition, in a smaller bee colony, but with the same number of brood cells, the mites are spread over a smaller number of bees. The number of bees that come close enough to a suitable cell stays the same, and therefore the mite's chance of invasion is increased.The percentage of mites that invade per day decreases when young open brood, still too young to be suitable for invasion by mites, is present. This decrease in invasion rate may arise because the mites prefer to be carried by young adult bees, which are likely to stay in the brood nest area. Within the brood nest area these young bees are divided over areas with brood cells that are suitable and areas that are unsuitable for invasion by mites. Hence, an increase in the amount of unsuitable open brood may keep part of the preferred young bees away from the suitable brood cells, and may thus decrease the invasion rate.If invasion is limited by the chance of being carried close enough to a brood cell, the spatial distribution of the mites inside the colony may affect invasion. In the areas where invasion occurs, the mite density on the bees will decrease. The mites will redistribute spatially by movement of the bees that carry them and by moving from bee to bee, but depending on the rate of this phoretic process invasion will be more or less limited. However, the rate of invasion was similar in bee colonies in which either 600 brood cells were available for mite invasion during one day or three times 200 brood cells were available during three days, whereas the colonies were comparable in all other respects. Thus, on a time scale of days the process of redistribution of mites inside the colony seems to be fast enough to prevent a significant effect of the period of brood cell availability on the rate of invasion. This is important for application of biotechnical control methods in which brood combs are introduced into the colony to trap mites. The 'trapping combs' are removed from the colony and the mites inside the cells are killed. Our results have shown that the number of cells used for trapping the mites is crucial, whereas the period during which the cells are available to the mites is of minor importance.Invasion into drone brood cells (chapter 4)Invasion by a population of mites into drone brood cells is similar to invasion into worker cells, except that invasion into &one cells is a much faster process. When invasion is compared between colonies with either exclusively worker cells or exclusively drone cells, Varroa mites invade a drone cell about 12 times more frequently than a worker cell. Hence, when both types of brood cells are available a biased distribution of 12 times more mites in drone cells than in worker cells is expected based on the differential frequency of invasion. This expected bias is larger than the bias actually found in colonies with both types of brood cells, which measures on average 8 times more mites per drone cell than per worker cell. The lower actual bias when compared to the expected one may be understood as follows. In normal honey bee colonies invasion into drone and worker cells is probably more or less segregated in time. Since the frequency of invasion is much higher per drone cell than per worker cell, the number of mites on bees will decrease much faster during periods when drone cells are abundantly present. Fewer mites will invade drone cells than expected when a constant number of mites on bees is assumed. Hence, the actual distribution over drone and worker cells may be less biased than expected from the differential frequency of invasion per cell. In addition, the biased distribution is sufficiently explained by the differential frequency of invasion per cell alone. There is no reason to believe that mites respond to the presence of nearby drone brood cells by refraining from invasion into worker brood cells, thus causing the biased distribution over drone and worker cells. Since the rate of invasion into drone brood cells is high, a trapping method using drone combs may be very effective in controlling the Varroa mite. When no other brood is present, 462 drone cells were estimated to be sufficient to trap 95% of the mites in a colony of 1 kg of bees.Effect of the period spent on adult bees on reproduction of the mites (chapter 5)No correlation has been found between the length of the period that Varroa mites stay on adult bees (1-20 days) prior to invasion and the total number of offspring per mite, the number of viable daughters per mite, the fraction of mites without offspring, and the fraction of mites that produces only male offspring. Thus, reproduction of the mites is apparently independent of the period that the mites reside on adult bees prior to invasion into brood cells.Mortality of mites during the period spent on adult bees (chapter 5)Mortality of Varroa mites, as measured by counting mites fallen on the bottom of the hive, occurs primarily right after emergence from the brood cell. When brood containing mites emerges during one day, 18% of the mites that have been present on the emerging brood is found on the bottom of the bee hive at the end of that day. Part of these mites may already have died inside the capped brood cells, and have fallen down after cleaning of cells by the bees. At the second and third day following emergence, respectively 4% and 2% of the mites on adult bees is found on the bottom, whereas from the fourth day on (up to 23 days) only 0.6% of the mites on adult bees is found on the bottom per day. Since the number of mites on the bottom of the hive will be strongly associated with the number of freshly emerged mites, counting the number of dead mites on the bottom may be a useful tool to estimate infestation levels in honey bee colonies.Attractiveness of brood cells to Varroa mitesThe attractive period of worker and drone brood cells (chapter 6)Worker brood cells are attractive to Varroa mites from 15-20 hours preceding cell capping, whereas drone cells are attractive from 40-50 hours preceding cell capping. Since the attractive period of drone cells is 2-3 times longer than that of worker cells, drone cells are consequently expected to be invaded 2-3 times more frequently. Actually, a drone cell is invaded 12 times more frequently than a worker cell. Hence, more factors must be involved in causing this difference in frequency of invasion. When the frequency of invasion is proportional to the surface of a brood cell, more mites are expected per drone cell due to its 1.7 times larger surface than a worker cell. Taken together, this would result in a 3.4-5.1 times higher frequency of invasion, which is clearly much lower than the 12 times actually found. Therefore, the higher frequency of invasion into drone cells may be attributed for an important part to differences in the information mites use to select a cell for invasion, either quantitatively or qualitatively.Effect of larva-cell rim distance on attractiveness of brood cells (chapter 7)Varroa mites are not randomly distributed over different types of cell which contain similar larvae. Per cell, more mites invade into shorter and narrower cells than control cells, whereas fewer mites invade into longer and wider cells. The period during which cells are attractive to mites varies among the different cell types, and whether in a certain type of cell more or fewer mites are found in comparison to control cells, is correlated with the length of the attractive period of that type of cell. The type of cell also affects the distance from larva to cell rim in the period preceding cell capping. When this distance is larger in comparison to control cells with larvae of the same age, the attractive period of the brood cells is shorter and vice versa. Since in all cell types the distance from larva to cell rim continuously decreases preceding cell capping, this negative correlation suggests that there is a critical larva-rim distance under which brood cells are attractive to the mites. Then, the length of the attractive period of brood cells depends on the moment this critical distance is reached. The distribution of mites over different cell types in turn results from differences in the attractive period. In normal drone and worker brood cells the critical larva-rim distance for invasion is 7-8 mm.Effect of methyl palmitate on attractiveness of brood cells (chapter 8)Since Varroa mites decide at some distance from the larva whether to stay on a bee or invade into a cell, they may well use a volatile chemical to select a brood cell. A few aliphatic esters, predominantly methyl palmitate, have been claimed to be this volatile signal for the mites for two reasons. The mites respond to the esters in an olfactometer (Le Conte et al., 1989), and the levels of the esters in worker and drone larvae may explain that drone cells are attractive during a longer period and are invaded more frequently than worker cells (Trouiller et al., 1992). However, invasion itself is unaffected by application of methyl palmitate to brood cells. In addition, analysis of headspace volatiles above attractive brood cells showed hundreds of components in the volatile blend, but in only 2 of 17 analyses a trace of methyl palmitate was found. Hence, there is no reason to believe that methyl palmitate is used as a signal for invasion by the mites.Reproductive strategy of Varroa mites (chapter 9)Since reproductive success of Varroa mites is higher in drone cells than in worker cells, the question arises why the mites do not restrict invasion to drone cells. Therefore, a simple model using population growth as a fitness measure has been developed to study under which circumstances specialization on drone brood would be a better strategy to adopt than reproduction in both types of cell. For European A.mellifera, the model suggests that if mites have to wait less than 7 days on average before they can invade a drone cell, specialization on drone brood would be a better strategy. This is close to the estimated waiting time of 6 days. Hence, small differences in reproductive success in drone and worker cells, and in the rate of mortality may determine whether specialization on drone brood will be promoted or not. In European A. mellifera colonies, Varroa mites invade both drone and worker cells, but specialization on drone brood cells seems to occur to some extent because a drone cell is more frequently invaded than a worker cell. In the parasite-host association of V. jacobsoni with African or Africanized A. mellifera or with A. cerana, the mites also invade both drone and worker cells, but the mites specialize on drone brood with respect to reproduction since a large percentage of the mites in worker brood do not reproduce. Only in the parasite-host association of Euvarroa sinhai, a mite closely resembling V . jacobsoni, and A.florea specialization is complete because these mites only invade drone brood.Does current knowledge of invasion behaviour help in controlling the Varroa mite?Our research on invasion behaviour did not result in a method in which Varroa mites are controlled using attractant or repellent chemicals. We still have to identify the signal the mite uses to invade a brood cell, although we know that mites perceive this signal at a distance from the larva and that the larva-cell rim distance affects the response of the mites to it. However, our results on invasion behaviour are useful to understand the possibilities and limitations for improvement of biotechnical control methods. We now know how many drone or worker cells are needed in a 'trapping comb' to catch a certain percentage of the mites in a colony. In theory, control methods that make use of 'trapping combs' are simple. In practice however, the methods may become complicated because application is integrated with other activities by the beekeeper like building of new colonies and swarm prevention. In addition, application of biotechnical control methods is usually labour intensive. Our results can be applied to design the simplest method that is sufficiently effective. This will remain an important application in future. Since much research is nowadays directed to breed honeybees that are less susceptable to Varroa mites (Woyke, 1989b; Büchler, 1994; Moritz, 1994), the effectiveness of control methods needed for control may decrease, which allows simplification of control methods. By combining simple 'trapping comb' methods and breeding of Varroa -less susceptable honey bees, there is a clear perspective for beekeeping without the use of acaricides to kill Varroa mites.