Natural selection for varroa resistant honeybee colonies

Gepubliceerd op
28 januari 2014

Following the observation of colonies of honeybees surviving in the wild (North America, Tom Seeley; France, Yves LeConte) or obtained through ‘planned naturalizing’ (Gotland, Sweden, Ingemar Fries), we started a comparable experiment in 2007. In 2007 a group of offspring colonies from the Gotland population was taken and brought to the Island of Tiengemeten. A newly set up group of a ‘Dutch mixture’ of colonies was placed in 2008 in a remote area (remote on a Dutch scale), the Amsterdam Water Dunes (AWD), at the North Sea cost, close to Vogelenzang. In both groups we ceased varroa control. However, to the 2008 Dutch mixture group we added a group of control colonies, which were managed similarly, but including varroa control twice a year.

In short the approach comes down to following: Colonies that grow and develop well in spring, and which build and develop a drone comb are used for the new generation. The other are discarded. From the colony the queen is removed by making a small artificial swarm including the old queen. The colony will then build emergency quueen cells, which will be mature after 14 days. On that day we do split the colonies into 4-5 equal size (about 3 frames with bees) baby colonies, each receiving one just emerged queen. The mating of the queens will take place in the two remote area’s Tiengemeten and the AWD, where the own drones (from the developed comb mentioned above) are supposed to be the majority. At splitting the colonies also the varroa population is split in more or less equal sub-populations per baby colony. After mating of queens the young colonies have to build up a population very fast to be able to survive the next winter (only time from about mid-July through end September).

In the first years we saw a steep increase in the varroa infestation, and many colonies collapsed, but after a few years the population of bee colonies started growing again. This was similar to the observations made before by Fries for the Gotland population. Simultaneously the varroa infestation decreased. Figure 1 shows the number of colonies that at the end of summer were strong enough to survive the next winter. Noteworthy that, after a bottleneck in 2008 the TG group continued to increase in size, except in 2012. In 2013 there was recovery. A bottleneck is a strong (sudden) reduction of the population, caused by a change in an environmental factor (in this case stopping varroa control), by which only the most robust or adapted individuals which can cope with the change, survive (survival of the fittest). In the AWD group there seemed to be a first bottleneck in 2010, however the severest bottleneck was to come in 2011: only 4 of the 56 wintered colonies survived the winter. Happily we were nevertheless able to winter 10 offspring colonies. The group of control colonies does lose only very few colonies each year, therefore we limit every year the control group to 20 colonies. Otherwise this population would grow with a factor four each year.

Figure 1. The number of colonies being strong enough to winter during the years. Varroa control had stopped in 2007 (TG) and 2008 (AWD). The control group is not represented because it is limited to 20 colonies each year.
Figure 1. The number of colonies being strong enough to winter during the years. Varroa control had stopped in 2007 (TG) and 2008 (AWD). The control group is not represented because it is limited to 20 colonies each year.

Mite infestation

Mite infestation is established twice a year, once in summer at the start of the growth of the new baby colonies, and once in winter (Dec/Jan), by taking a sample of ~200 bees and washing out the mites and counting both the bees and the mites. Both in July and in December we have a good estimation of mite infestation because in both cases the colonies are brood less, and therefore all mites reside on the worker bees. After taking a bee sample in the control group the colonies were treated against varroa by spraying (summer) or trickling (winter) with oxalic acid. The result of the mite infestation is given in Figure 2. The sample of winter 2013 has been taken, but the numbers of mites still have to be counted.

Figure 2. The number of mites per 100 worker bees in the colonies of Tiengemeten, AWD and the control group from Lelystad, in summer and winter, since July 2008.
Figure 2. The number of mites per 100 worker bees in the colonies of Tiengemeten, AWD and the control group from Lelystad, in summer and winter, since July 2008.

The summer observation is always from a just starting baby colony (~3000 bees). The sample of the following winter is from the same but grown colonies (between 6000 – 15000 bees). Often the infestation in July is higher than in December, indicating that the bee population has grown faster than the mite population, which dilutes the mite infestation. Although varroa is not controlled, the infestation of the TG group stays generally between 5 and 10% (except summer 2011; why?). The infestation of the AWD was initially higher but is now also approaching 5%. However, in 2011 we had a huge increase in infestation from July to December in the AWD group, simultaneously in the TG group the infestation did decrease (by something the bees did (?)). Possibly there is a role for differences in varroa resistance, but the environmental conditions on different sites may also be involved. It is at least probable that the high losses of AWD colonies in winter 2011 (Figure 1) were related to the high mite infestation in December 2011. In 2013 the infestation of the baby colonies was again between 5 and 10%. The mite infestation of the control group shows that using twice a year an oxalic acid treatment against varroa is adequate to keep the infestation below 5 % throughout.   

Reproduction success of Varroa destructor (student projects of Michiel Glorius 2011, Thijs Gerritsen 2012 and Anne van Woerkom, 2013)

From the described developments it appears that something ‘has happened’ with the population success of the mites in the experimental groups. Something hampers unrestricted growth of the mite populations. One could think of many mechanisms leading to the observed effect; maybe the mites are recognized by the bees and removed, maybe bees with mites on their bodies fly out and abduct the mites. But there can be also not so obvious and possibly invisible mechanisms by which mites in ‘resistant’ colonies would become less successful. For instance the larvae are less suitable for the mites as a food source (through some chemical, or their composition), by which the mites would grow slower. Or through other traits of the bees the mites do produce fewer offspring than normally. In 2011 Michiel Glorius found that the mites in ‘selected’ colonies produced fewer offspring per mother mite, and in more occasions the mother failed to produce a son first. The latter would lead to daughters remaining unfertilized, indeed there was a greater share of the adult female mites not reproducing.  

In a similar way Thijs Gerritsen investigated in the late summer of 2012 the reproduction success of mites in colonies with the queens of 2011. Just alike in the experiment of Michiel (above) combs with larvae just before capping were introduced in a colony with a high mite infestation on the worker bees (‘mite shower’), after capping these were returned to their colonies of origin. Just before emergence the combs were frozen, and after defrosting the cells were opened and the varroa families observed.

In the TG colonies more often than in the control and AWD colonies the male mite was missing, and the number of female offspring was reduced. Noteworthy the reproduction success was much lower in these experiments than in the earlier by Michiel: this will have been caused by the season, since it is known that reproduction success of mites is reduced in early and late seasons. The latter experiment was done in September, the former in July.

Reproduction success in drone brood

A discriminating difference between varroa in the western honeybee (A. mellifera) and the Asian honeybee (A. cerana) is reproduction in worker brood: in A. ceranae this does not happen. In that case the mite can only reproduce in drone brood. This difference is the reason why varroa can reproduce so quickly in western honeybee colonies: worker brood is always amply available, drone brood only a limited period and in limited amounts. However, mite reproduction success per mother is higher in drone brood, and more daughters reach maturity.

Reflecting on the situation in A. cerana it is interesting to speculate whether resistance for varroa would also become evident in drone cells or only in worker cells. At least in our ‘selection system’, the selection pressure is on worker brood, since every year only one drone comb is produced per colony, and since in the small baby colonies no drones are produced.

Therefore in spring 2013 Anne van Woerkom investigated the reproductive success in drone brood. The colonies were the same as those used by Thijs in 2012. Colonies were stimulated to build and develop a drone comb. We just tested for the natural population of mites present in the colonies. The natural infestation of the drone cells varied from 7 to 25% (about 25 cells per colony were checked, per treatment 4 colonies were used). No differences were found between the three groups (Tiengemeten, AWD, control); not in the numbers of offspring per cell and per mother mite, not in the fraction of fertile daughters (= grown up in presence of a male), not in the fraction of cells with no male present, nor in the frequency of not reproducing female mites. In all cases there was a lot of variation between colonies within the groups. The number of offspring (daughters + son) varied from 2.5 to 5.2. The maximum number of mature daughters that can develop per cycle in a drone cell is 5, in a worker cell 3.

Where do we stand now?

After six years of cruel harmony between honeybee colonies and populations of the mite Varroa destructor it looks as if the bees have developed some kind of resistance: the mite infestation rates do not increase as fast as they did in the beginning, it is possible to keep the population of bee colonies at strength without controlling the varroa mite (although the Varroa controlled population of colonies can grow by a factor 4 per year!).

In the next three years of the EU ‘Honey’ program we will continue selecting with these colonies. However, just continuation is not enough: we need to learn why and how these bees survive with varroa, and how or why varroa thrives worse. There seems to be an effect on reproduction success: in the next years we shall try to establish the reproductive success in worker and drone brood in the same colonies simultaneously.

In the population of ‘resistant’ bees from Gotland the mites had a lower rate of entering the cells to reproduce. We do not know yet for our populations. Besides the reproductive success the longevity of mites could also play an important role: if the chances are greater to be removed from the colony, from the cells by hygienic behaviour, or from the bees by grooming, this might also challenge the population growth of the mite.

In 2014 we will try to estimate differences in grooming of mites. And of course in the end we aim to find out which genes are involved in the selected adaptations.