The varroa mite (VM), Varroa jacobsoni (Oudemans), is an economically important external parasite of honey bees (Apis mellifera L). VM develop and reproduce in the sealed brood cells of honey bees where they feed on the hemolymph of bee pupae. Depending on the level of parasitism, VM usually kill entire colonies in from 12 months to 4 years following initial infestation.

Varroa-tolerant honey bees have been identified in Europe, South America and Mexico via a number of studies (see Buchler, 1994; Medina, 1998; Moretto et al., 1991,1995; and Rutter, 1991 ). Observed varroa-tolerance appears to depend upon a variety of heritable and environmental factors acting independently and in concert. These factors include sterility of female mites, shortened developmental time of worker pupae, worker grooming of siblings, workers chewing free roaming mites, hygienic behavior, and death of mites trapped in unopened drone cells. Several viral diseases may be transmitted by VM confounding the impact of the parasite infestation (Allen and Ball, 1996).

In an earlier three-year study conducted in a single untreated apiary of 36 colonies in Central Arizona (Erickson et al., unpub.), four colonies survived with very low levels of infestation. The remaining 32 were killed by VM and robbed out. The survival of these four colonies prompted the current twoyear study to determine whether a varroa-tolerant honey bee population could be developed and maintained using conventional beekeeping practices. Concurrent infestations of the honey bee tracheal mite (HBTM), Acarapis woodi (Rennie), were present in the test colonies. Hence, we had the opportunity to determine whether there were interactions between the two parasite populations. Additionally, in our second year of study, we were able to identify four beekeepers who have been attempting to select for varroa-tolerance, much like ourselves. We solicited samples from these beekeepers and present corroborating data from these collateral studies. Finally, we present corroborating data obtained from studies of feral colonies in Arizona (Loper, 1997).

Colonies identified as potentially varroa-tolerant or queens from such colonies were obtained from several cooperating beekeepers across southern and central Arizona. Selection of these colonies/queens was based on survivorship in untreated apiaries or on noteworthy low VM infestation levels in apiaries wherein other colonies were highly infested. Daughter queens reared from those in the earlier study conducted by Erickson et al. (unpub) were included. All colonies/queens were established in standard Langstroth hives beginning in November of 1995 and the queens marked. None of these colonies had been treated with Apistan@ within the preceding 12 months or more. VM populations were allowed to develop unchecked by any form of intervention throughout the study period. All colonies were managed using the normal commercial management practices of the beekeeper (L. Hines).

The study site consisted of three isolated apiaries located in southeastern Arizona (Figure 1). Survivor colonies were placed in the central apiary (T). A second apiary (S) was located 3.9 miles (6.3 km) to the northeast, while the third (B) was 1.4 miles (2.3 km) south of T. Apiaries B and S were separated by 5.3 miles (8.6 km). There were no other managed colonies in the area. Feral colonies, if present in the study area, must also be considered survivors. (Note, VM have been present in Arizona since 1992)

Mite populations were determined every two months throughout the study (see below). Beginning in the spring of 1996 colonies with the lowest VM and HBTM infestations were identified. Daughter queens were reared from these colonies, open-mated in apiary T and used to requeen the colonies in apiaries S and B. Thereafter, queens reared from selected program colonies were open-mated in one of the three apiaries or at a third isolated mating area where all of the colonies had been requeened with "program" queens. "Program" colonies with VM infestations exceeding 15% or HBTM infestations exceeding 10% were removed from the study. Colonies were requeened with "program" queens whenever the queen appeared to be failing. Supersedure queens were marked and allowed to remain. Colonies eliminated from the "program" or lost over winter were replaced in the spring and throughout the summer with newly identified survivor colonies or with established colonies headed by "program" queens.

Samples of ~ 125 - 200 bees were gathered from the face or top bars of brood combs using a modified battery powered vacuum cleaner (Erickson and Edwards, 1990). Each sample of bees was immediately preserved in 70% ETOH. In the laboratory each sample was individually spray-washed three times in tap water to remove all mites from the sample, the mites and bees were then counted and the data converted to mites/100 bees.

Thirty bees were removed at random from each sample in ETOH to determine the level of HBTM infestation for each colony/treatment/date. HBTM infestations were estimated using the procedures of Delfinado-Baker (1984). Prothoracic collars were removed from each of the 30 bees, clarified for 24-36 h in five percent potassium hydroxide at 39' C, and examined at 100 x magnification with a stereo- microscope. Infestation is reported as the percent of the 30 bee sample with one or more mites.

The coefficient of correlation for interaction between VM and HBTM populations was determined.

In the fall of 1995 a number of queens were obtained from a Georgia bee breeder and used to requeen colonies maintained at Tucson, Arizona in standard Langstroth hives. These queens came from the same Georgia breeder described below, but were not from his experimental apiary. Nine of these colonies were selected at random and sampled monthly from March, 1996 through March, 1997.

In 1997, four beekeepers/bee breeders, who have been using varying approaches in similar attempts to select and breed for VM resistance/tolerance, submitted samples for analysis on a one time only basis. These beekeepers/bee breeders operate in Georgia, New Mexico, central Arizona (Phoenix) and southern Arizona (Amado). All samples were collected during a 45 day period between August 27 and October I I and analyzed to determine VM plus tracheal mite infestation levels.

The Georgia bee breeder has maintained one untreated experimental apiary of -45 colonies in isolation for 10 years. Each spring he replaces dead colonies with divides from the survivors. Populations are equalized and the queen-less divides are allowed to rear their own queens which mate in the area. Twenty-four of these colonies were randomly selected and sampled. The New Mexico beekeeper has never treated for varroa. He is a commercial beekeeper who allowed -80 percent of his 800 colonies to die. However, he now has 800 to 850 strong colonies derived by splitting survivors. The thirty colonies sampled were randomly selected from nine apiaries. The North Phoenix area beekeeper, whose colonies have been established largely from swarms picked up in the area, is a hobbyist who has been keeping bees for four years. This beekeeper has never used Apistan@, but does apply a mixture of cooking and essential oils every two months. Eleven colonies were sampled. The Amado commercial beekeeper has never used Apistan. Here VM and HBTM susceptible colonies are allowed to die and are replaced with divides from surviving colonies. This beekeeper has used other non-chemical parasite management strategies including the use of brood combs with cell diameter smaller than the industry norm. Colonies in two apiaries were sampled. None of these beekeepers routinely monitor their mite infestation levels by means other than visual inspection.

Also reported herein are three years of parasitic mite data obtained from a long-term study of what was initially an extensive population of feral (unmanaged) colonies in southern Arizona, north of Tucson (Loper, 1997). Surviving colonies in this population were sampled between October and December, 1995-1997.

Seven hundred thirty-three samples (representing 118,512 bees washed and counted to determine VM infestation levels and 21,990 bees dissected to determine HBTM infestation levels) were examined. VM and HBTM infestations for apiaries 13, S, and T over the two-year study period are presented in Figures 2 & 3 and Table 1. The two-year means and ranges for VM infestation levels were 6.8 (0-50.3), 6.0 (0- 35.7), and 7.7 (0-38.4) mites/] 00 bees, respectively, for apiaries B, S, and T. Means and ranges for HBTM infestations were 6.4% (0-50%), 4.3% (0-36.7%), and 4.1% (0-36.7%), respectively. (Note, colonies with VM and HBTM infestation levels above our established limits of acceptability were sampled and included in the data cited above before they were eliminated from the study.) There was no significant correlation/interaction between VM and HBTM populations (p>0.05).

Mean parasitic mite infestation levels for the Tucson colonies headed by Georgia queens are presented in Figure 4 and summarized in Table 1. Mean VM and HBTM infestation levels in the managed colonies sampled late in 1997 are also summarized in Table 1. Overall, mean VM infestations ranged from 4.1-8.8 mites/100 bees, while HBTM levels ranged from 0.6-9.5 percent.

Annual VM and HBTM infestation levels in feral colonies are summarized in Table 2. Mean VM infestations ranged from 0.2-4.4-5.7 mites/100 bees, while HBTM levels ranged from percent. In the summer of 1993 there were 191 colonies at this study site. Their number dwindled to as few as 12 in the spring of 1996. This population seems to have stabilized and may be making a very slow recovery.

This study was not undertaken to develop a varroa-tolerant strain of honey bees. Rather, it was intended to determine if varroa-tolerant honey bee colonies could be readily identified in the southwestern United States and whether beekeepers can develop varroa-tolerant honey bees using conventional beekeeping practices. We found it relatively easy to find varroatolerant colonies. Our results demonstrate that it is indeed possible to select and produce strains of honey bees with improved varroa tolerance out of existing domestic stock. Corroborating data from our collateral studies clearly show that some beekeepers/bee breeders are currently doing just that. In our varroa-tolerant population the overall mean level of VM infestation was between 6 and 7 mites/100 adult bees; results from the collateral studies were similar. Moreover, our results corroborate those of similar studies (see Buchler, 1994; Medina, 1998; Moretto et al., 1991,1995; and Ruttner, 1991) wherein reported mean VM infestation levels are well below 10 mites/100 bees.

Gatien and Currie (unpub) have found that, in Canada, spring VM levels above I percent impact colony honey production, spring levels between 2 and 5 mites/100 bees may significantly reduce honey production, while levels above 20 mites/100 bees eliminate the production of surplus honey in Canada (Gatien and Currie unpub). Our mean spring populations ranged between 2-5 mites/100 bees with some individual colonies having no detected mites in the spring and early summer.

Although we have found it relatively easy to identify colonies possessing moderate to high levels of varroa-tolerance, we have not been able to identify specific factors that contribute significantly to the observed tolerance to either VM or HBTM. Our empirical observations compel us to conclude, as others have, that both environmental and genetic elements contribute. While identification of such factors would be highly useful, this knowledge is not essential for beekeepers to develop varroa-tolerant populations.

The colonies in apiaries B, S, and T were totally dependent upon native plants for nectar and pollen. Abnormally low precipitation during 1996-97 led to a significant reduction in the availability of these resources. Little surplus honey and pollen was harvested and some colonies had to be fed to ensure their survival. Thus, the colonies experienced a significant level of stress throughout the study period. From time to time, this stress contributed to reduced brood rearing. Even so, it was most interesting to note that from time to time VM could not be detected in some "program" colonies.

Some may wonder about the amount of influence that the presence of Africanized honey bees in southern Arizona may be having on our apparent ability to develop a more VM tolerant line of bees. For comparison, we can point out that in another study area (many miles away from apiaries B, S, and T) where highly Africanized bees have been maintained in isolation by the Research Center for over two years, we have seen very little indication that these nearly pure Africanized Honey Bees are any less susceptible than the colonies in our selective breeding program. In fact, considerable numbers of the AHB colonies have succumbed with VM damage appearing to be a major cause. Moreover, in yards B, S, and T the beekeeper (L. Hines) makes a concerted effort to eliminate from the proaram any colonies that show significant AHB characteristics. From these two facts, the assumption can be made that Africanization probably has had relatively little bearing on whatever success we have had.

Our HBTM data provide further support for the studies of (Danka and Villa, 1996; Loper et al., 1992) who demonstrated that it is possible to develop HBTM-tolerant honey bees.

Scientists as well as beekeepers have wondered whether there might be some interaction between VM and HBTM populations infesting the same colony. That none was found is particularly significant, especially given the large size of our data base. We know of no other published studies that address the question of interactions between VM and HBTM populations.

What we in this study and others in similar efforts appear to have done is accelerate the process of natural selection that would normally take place without beekeeper intervention (e.g. mite control measures such as the use of Apistan). Ultimately, one must wonder whether the perpetuation of susceptible colonies via chemical control is in the long run counter productive. Either way, most beekeepers will want to see more longterm research before they feel comfortable in significantly reducing reliance on chemical control for VM. To this end, we and others engaged in similar efforts must monitor the stability of our varroa-tolerant populations over time and identify ways of enhancing factors contributing to varroa-tolerance.

In the interim we feel that it would make good sense for beekeepers to start a nationally coordinated selection and breeding program wherein only colonies with elevated VM levels are treated and drone flight from treated colonies is restricted or the colonies eliminated. This would reduce beekeeper reliance on chemical control measures and retard the development of VM resistance to miticides. Area wide participation and cooperation among beekeepers would be a prerequisite for such a program. As professor Ruttner (1991) so aptly pointed out, that which remains is for the beekeeping community to work together to make varroa-tolerant honey bee populations a reality.

The authors wish to thank all of the beekeepers who so willingly participated in this study and Dr. Gerald Loper who provided the data on the feral colonies. We also thank G. Mills, M. Chambers, and J. Crowe.

LITERATURE CITED

Allen, M.; Ball, B. 1996. The incidence and world distribution of

Erickson, E. H., and Edwards, J. F. 1990. Manual for field collection and documentation of feral honey bee colonies