In January, 2023, beekeepers in the state of Florida began reporting unexpected losses of their managed honey bee colonies, and simultaneously an unusually large number of weak colonies. The reports came from two separate groups of beekeepers. The first report was from commercial queen producers in South Florida who sent videos of adult bees exhibiting morbidity. These queen breeders know each other but keep bees separately. They experienced unexpected whole colony losses, and an almost complete loss of their forager bees (older bees comprising up to half of the colony). The second wave of reporting occurred several weeks later from a larger operator in northern Florida who reported similar symptoms as the queen breeders during the first wave. We created a network of affected operators using targeted sampling “snowball” methodology1 by encouraging initial reporters to self-recruit amongst their colleagues to assess losses across the network of commercial operators. We quickly determined through interviews that reported losses and morbidity were unusually high and were causing severe economic harm to affected operators.
Information from interviews was used a priori to direct sampling protocols for an immediate sampling expedition. Since beekeepers reported sudden losses of their adult bee populations sampling focused on the collection of adult bees. Because of worries that a survivorship bias may occur, efforts were made to locate colonies which still had symptomatic bees, and to capture those bees alongside asymptomatic nestmates.
Upon arrival, we found operators across the state with thousands of unexpectedly weak colonies in their commercial apiaries (Figure 1). These operators had inspected and graded their colonies for the upcoming almond pollination in California. These colonies must have a minimum of 10 frames of bees and 6 frames of brood (developing bees) to meet grade, with many operators sending colonies with upwards of 20 frames of bees. When beekeepers returned to their colonies in January, they found thousands of those colonies had suddenly dwindled to two-to-three frames of brood over the course of three weeks (Figure 1).
Beekeepers self-reported mosquito spraying as a possible culprit for the losses. There was an increase in mosquito spraying after Hurricane Ian, and every county in which bees were located had sprayed a combination of larvicide and adulticide sprays through a combination of ground and aerial dispersal. Nevertheless, the most persistent observation in surviving colonies was the presence of Varroa destructor, a parasitic mite that causes direct damage to bees by feeding on their fat bodies and hemolymph2 and indirect damage by vectoring viruses when they feed on adult bees and developing brood3. Despite repeated treatments, Varroa was detected in colonies across the study (71.6% of colonies with at least one detection). Alcohol washes were performed on each colony in the study and varied greatly around the mean (M = 2.57%, SD = 3.88%). There was no significant difference in mite levels between colonies of different strengths (H(2) = 0.0298, p = 0.985). Figure 2 shows the percentage of mites detected on adult bees. Beekeepers reported applying miticides multiple times in October and November, with follow-up treatments reported by most operators into January. However, clinical signs of mite infestation were observable in 44.6% of the colonies sampled (N = 74). Additionally, many healthy appearing asymptomatic colonies, not showing any signs of disease or typical mite damage, harbored infestations of Varroa higher than economic thresholds.
The alcohol wash was not deemed a universally reliable method to report what Varroa infestations might have recently been since beekeepers had been repeatedly treating. For example, two operations had reported treating up to two weeks prior to our sampling. Those colonies reported very low levels of Varroa even though clinical signs of high mite levels were visible in the sampled colonies. Other methods to detect Varroa were implemented. Worker bees were inspected at emergence for Varroa. Some colonies had a high level of infestation in their worker brood (0% - 90%, M = 15.78%, SD = 23.43%).
Pathogens in sampled colonies
Colonies were screened for pathogens using standard methods4. Screenings measured levels of key viruses, the microsporidian Nosema, and trypanosome parasite that are known to affect honey bee health and have been associated with historical bee losses. All colonies had at least one pathogen detected, with the majority of colonies having multiple detections (95.8%). The most prevalent pathogen detected across all colonies was Nosema ceranae, followed by Deformed wing virus (DWVA), and then Black queen cell virus (BQCV). Other pathogens that were detected were detected at low frequencies (Table 1). Israel acute paralysis virus (IAPV), a pathogen previously associated with colony collapse5, was not detected in any of the sampled colonies. Another paralysis virus, Kashmir bee virus (KBV) was also not detected, while Chronic bee paralysis virus (CBPV) was rarely detected. However, Acute bee paralysis virus (ABPV), a paralysis virus associated with Varroa destructor was present in 20.8% of sampled colonies.
Table 1 Prevalence of 10 pathogen targets in colonies of different strengths from Florida
Prevalence of pathogens in sampled colonies
|
Colony Strength
|
DWV
|
DWVB
|
ABPV
|
BQCV
|
CBPV
|
IAPV
|
KBV
|
SBV
|
N.cerana
|
Tryp
|
Strong
|
1
|
0.385
|
0.231
|
0.692
|
0
|
0
|
0
|
0
|
1
|
0.615
|
Medium
|
0.864
|
0.409
|
0.182
|
0.864
|
0
|
0
|
0
|
0.045
|
1
|
0.5
|
Weak
|
0.87
|
0.46
|
0.216
|
0.76
|
0.027
|
0
|
0
|
0.054
|
0.972
|
0.405
|
All Colonies
|
0.889
|
0.431
|
0.208
|
0.778
|
0.014
|
0
|
0
|
0.042
|
0.986
|
0.472
|
There was no significant difference in viral levels between colonies of different strengths (Table 1), nor was there a strong correlation between colony size and viral levels. Several pathogens were associated with each other. DWVA was associated with BQCV, DWVB, ABPV and N. ceranae. These relationships were positive, moderate in strength and significant (Table 2). ABPV had a similar association with N. ceranae, as did N. ceranae with trypanosomes.
Table 2 Associations of target pathogens (using Spearman coefficient, df =70, insignificant associations are printed in normal typeset, bold p < 0.05, bold* p < 0.001).
|
DWVA
|
DWVB
|
BQCV
|
ABPV
|
N. cerana
|
Tryp
|
DWVA
|
|
0.42*
|
0.26
|
0.36*
|
0.37*
|
0.22
|
DWVB
|
|
|
0.041
|
0.15
|
0.082
|
0.11
|
BQCV
|
|
|
|
0.009
|
0.17
|
-0.026
|
ABPV
|
|
|
|
|
0.32
|
0.093
|
N. cerana
|
|
|
|
|
|
0.36
|
There was a significant, positive correlation between the percentage of mites found in emerging workers, or the alcohol wash, and key pathogens that affect honey bee health (Table 3).
Table 3 Associations of target pathogens and infestation levels of adult bees or emerging brood (using Spearman coefficient, df =70, insignificant associations are printed in normal typeset, bold p < 0.05, bold+ p < 0.001).
|
DWVA
|
DWVB
|
BQCV
|
ABPV
|
N. cerana
|
Tryp
|
Alcohol wash
|
0.46*
|
0.32
|
0.19
|
0.27
|
0.18
|
0.33
|
Percent of emerging workers with mites
|
0.66*
|
0.61*
|
0.38
|
0.24
|
0.35
|
0.28
|
Economic harm
Twelve beekeeping operations were interviewed and sampled in this study. Of those operations, nine were commercial operators who regularly provided pollination services. Two remaining operations commercially produced queens (2), while one was a non-migratory honey producer within the state of Florida.
Each operation responded to an economic harm survey, which was designed to uncover the immediate loss of income incurred by these events and the cost to recover. Beekeepers were asked to only include immediate loss of income, and not to include extended losses beyond March, 2023. When including costs associated with extreme morbidity operations were asked to report the cost to fix or replace these dwindled colonies.
Outright loss of colonies from November, 2022, to February, 2023, was estimated through interviews with operations. Colonies lossed due to Hurricane Ian were excluded from these estimates. Losses between November, 2022 and the beginning of February, 2023 ranged from 20% to 88.9% (M = 48.1%, SD = 22.7%). This represented a total 20,567 colonies lost across 12 operations in a span of approximately three months. An estimated 21,671 colonies remained, with 6,727 (31.5%) of these colonies being economically unviable units. These dwindled colonies could not provide any service within their operations as they did not meet minimum requirements for pollination contracts, queen production, or honey production.
Beekeeper responses varied greatly with respect to how they would manage these dwindled colonies. Responses were reflective of the circumstances and resources unique to each operation. Two of the smaller operators believed keeping their dwindled colonies alive posed a greater risk to their healthy colonies than the cost to euthanize them. They responded by euthanizing their dwindled colonies. The larger operators allowed the dwindled colonies to persist in their holding yards. Beekeepers will repair or replace these dwindled colonies, potentially introducing new pathogens into their operation by the large scale purchase of bees.
At the time of interviews, beekeepers were immediately focused on readying large colonies for transport to almond pollination in California. Beekeepers responded that they would see how many of the dwindled colonies would continue to die. When asked how they would fix or replace the dwindled colonies, answers varied. Three operators said they will or were considering purchasing replacement colonies to replace their dwindled colonies. The remaining beekeepers would fix the colonies using resources within their own operations. Two operators responded they had neither the financial resources to replace nor the bee resources within their operations to fix their dwindled colonies. Dwindled colonies represented 44% to 100% of the remaining colonies of these operations. The total cost to replace or fix the 6,818 dwindled colonies was estimated to be $908,630, and was based upon interviews with each operation.
Loss of income from commercial pollination
Severe losses and morbidity resulted in the immediate loss of pollination income for 9 of the 12 operations in this study. Operations sent 34.1% to 100% fewer colonies to almonds than anticipated (M = 70.4%, SD = 22.9%). This represented an estimated loss of $3,451,420 anticipated income from almond pollination across the nine operations (figure 3).
Loss of income from queen and honey production
The three other operators produced honey and or queens instead of engaging in commercial pollination services. We surveyed for their immediate loss of income by covering anticipated production during early spring, and arrived at an estimated loss of $825,700. We combined the estimates from all three operations purposefully to maintain anonymity of the beekeeping operations.
Loss of employment
Five of the 12 operations will have layoffs (42%), resulting in eight jobs lost.
The economic harm incurred by these beekeepers was severe. Several operators lost 90% or more of their anticipated income when colonies either died suddenly or dwindled to economically unviable clusters. Estimated gross immediate loss from pollination services from the nine migratory operators in the survey was $3,541,420 dollars. Immediate loss of this income resulted in the loss of eight jobs. Colonies which dwindled to economically unviable clusters will be replaced or fixed through normal beekeeping practices by operators. Operators estimated that replacement or fixing of these colonies would cost $875,960, with an average cost that varied greatly between operations, reflecting beekeeping practices and resources available within each operation. For example, one operator euthanized their dwindled colonies, fearing they could be reservoirs for potential disease transmission to healthy colonies. As a result, their cost to fix those colonies was relatively low ($20 per colony). Other operators reported they would fix their dwindled colonies using resources within their operation. Several other operators said they would purchase replacement colonies simply because they did not have enough viable colonies within their own operation to pull resources from. At least two operators will not purchase new colonies because they lost all operating income.
Reporting survey results
Private, personal information was provided during interviews with stakeholders. Raw data is not being provided to respect the anonymity of operations which shared sensitive, personal, information during these interviews. The size, location and production of an operation can provide enough information to single out individuals. As a result, we made two groups when presenting our data to maintain anonymity. The nine commercial pollinators were grouped together, and then data from their interviews were supplied together. To protect the anonymity of the queen breeders and honey producers in this study, their survey results were grouped together. When reporting personal information, we used the term “multiple operators” or “other operator” and should be inferred as: at least one.