Colony Collapse DisorderA Descriptive Study

A beekeeper walks into an apiary expecting noise — the low, steady hum of tens of thousands of workers. Instead, there is silence. The hives look intact from the outside. When opened, you find capped brood, a queen still present, and honey still stored. But the adult workers are almost entirely gone. There are no piles of dead bees on the ground, no obvious disease, and no clear culprit. Just absence. In the winters of 2006 and 2007, and again the following winter, this scene repeated itself across the United States at a scale that forced the question into public view. Beekeepers were losing a third or more of their managed colonies, and nobody could explain it. Because the dominant feature was the rapid disappearance of adult workers, the syndrome was given a name: Colony Collapse Disorder, or CCD. The name itself tells you something — it emerged before anyone understood the cause. It is operational and symptom-based, derived from what observers could actually see: weak colonies with too few bees to cover their brood frames, young, apparently flightless workers left behind, and delayed invasion by secondary pests like small hive beetles and wax moths, which would normally move in quickly once a colony weakened. Experienced beekeepers recognized the pattern as unusual, but the case definition relied on clinical judgment rather than any single lab test. That’s not a flaw — it’s how outbreak investigations begin. You describe what you see before you can explain it. VanEngelsdorp and colleagues also noted that mass colony losses weren't entirely new. Historical records going back to one thousand eight hundred sixty-nine document at least eighteen episodes of unusually high colony mortality, including a Colorado event in the 1890s called "May Disease," which involved the rapid disappearance of large clusters of bees. What made this different was the scale, the simultaneity across the country, and the stakes — honey bees pollinate treated crops across agricultural systems that feed millions of people. This wasn't a niche beekeeping problem. So the team launched what they called a descriptive epizootiological study — epizootiological meaning epidemiological but for animals. The goal wasn't to test a specific hypothesis. It was to cast a wide net, characterize CCD rigorously, and compare everything they could measure between affected and unaffected colonies. They sampled apiaries in Florida and California, classifying colonies as CCD or control based on field criteria, and then measured sixty-one quantified variables: adult bee physiology, pathogen loads across a broad panel of agents, and pesticide residues. This was the first comprehensive survey of its kind, and the breadth was intentional — when you don't know what you're looking for, you measure everything you can. The clearest signal that emerged was not a single pathogen. It was a heavier, multi-front infectious burden in CCD colonies. The team assayed seven viruses — including Acute Bee Paralysis Virus, Deformed Wing Virus, and Israeli Acute Paralysis Virus — along with two species of Nosema, a gut fungal parasite, trypanosomes, and bacterial markers. What they found was a striking pattern of co-infection. Co-infection with both Nosema species was two point six times more frequent in CCD colonies than in controls. Colonies carrying four or more viruses simultaneously were three point seven times more common in CCD colonies. Breaking that down further, fifty-five percent of CCD colonies had three or more viruses versus twenty-nine percent of controls. At five viruses simultaneously, the numbers were twenty-four percent of CCD colonies against eight percent of controls. The difference at three or more viruses was statistically significant, and the pattern held across the full distribution. Think about what that means. It's not that CCD bees had one particular virus their healthy neighbors lacked. It's that they were carrying multiple pathogens at once, across multiple categories — fungi, viruses, and potentially bacteria — in combinations that healthy colonies were far less likely to show. The mean number of pathogenic organisms per colony was four point three four in CCD colonies versus three point zero in controls. VanEngelsdorp and colleagues raise two interpretations for this, and they don't pick between them. Either CCD colonies were exposed to more pathogens — perhaps through foraging patterns, shared equipment, or movement by commercial beekeepers — or their ability to resist infection had been compromised. The nutritional metrics the team measured didn't differ decisively between groups, so a simple nutrition story doesn't explain it. And crucially, the surviving bees collected from CCD colonies were, by other measures, likely the fittest individuals left — meaning the measured pathogen burden probably underestimates what the colony was carrying during collapse itself. Then there's the pesticide result, which runs in exactly the opposite direction you'd expect. Coumaphos — an organophosphate acaricide that beekeepers apply to control the parasitic mite Varroa destructor — was found at higher levels in control colonies than in CCD-affected ones. It was higher in wax, higher in brood, and higher in adult bees. The p-values ranged from zero point zero zero nine to zero point zero five. Fluvalinate, another approved acaricide, also trended higher in control apiaries. Across the full pesticide screen, fifty different pesticide residues and metabolites turned up in wax samples alone. This is a genuine puzzle. Coumaphos is lipophilic, meaning it binds to fats and accumulates in beeswax over time, and studies cited by VanEngelsdorp and colleagues show elevated wax levels decrease survivorship of developing queens and affect worker bees. So you'd expect higher coumaphos to correlate with worse outcomes. Instead, it's the healthy colonies carrying more of it. The paper doesn't resolve this. It generates hypotheses. One possibility is that CCD colonies came from operations that used fewer in-hive acaricides, so the residue difference simply reflects different management histories. Another possibility is what the authors call a legacy effect — varroa mite loads were comparable between CCD and control colonies at the time of sampling, but heavy mite infestations may have occurred earlier, before miticide treatment, leaving behind immunocompromised colonies even after mite numbers dropped. A third possibility involves resistance: if mites in some populations have developed tolerance to coumaphos, beekeepers applying it may be seeing reduced efficacy while still accumulating residues in their wax — a cycle that could interact with colony health in ways a cross-sectional snapshot can't fully capture. The spatial pattern of how CCD distributed itself through apiaries adds another layer. CCD-affected apiaries contained three point five times the number of dead colonies and three point six times more weak colonies than control apiaries, and these weren't randomly scattered. In a subset of nine apiaries, the team mapped hive positions and compared observed versus expected neighbor pairings. Weak colonies tended to sit next to other weak colonies. Dead colonies clustered near dead colonies. The odds ratio for weak-weak neighbor pairings was two hundred fifty-five. For dead-dead, it was one hundred nine. These are not small effects. Adjacent colonies were sharing a fate. That pattern points in one of two directions: either something was spreading between hives — a contagious agent moving through the apiary — or neighboring hives were being hit by the same environmental exposure simultaneously. Distinguishing those two mechanisms matters enormously for how you'd respond. A contagious agent calls for containment. A shared exposure calls for identifying the source. The spatial data alone can't tell you which it is, but it rules out random chance. Put it all together, and here's what VanEngelsdorp and colleagues can say with confidence: no single variable across sixty-one measured was sufficient on its own to distinguish CCD from control colonies. Not one pathogen, not one pesticide, not one physiological marker. What they found instead was a pattern — heavier co-infectious burden, spatial clustering, and a pesticide signal that runs backward from expectation — consistent with an interaction between multiple pathogens and other stressors. The study accomplished exactly what a first comprehensive descriptive survey should accomplish: it narrowed the field. It told future researchers which hypotheses are worth testing rigorously — the legacy effect of prior varroa parasitism, the dynamics of pesticide tolerance in both bees and mites, and the possibility that immune suppression from one stressor opens the door to proliferation of others. The team explicitly called for longitudinal studies to track temporal sequences rather than single-time snapshots, because what happened to a colony six months before collapse may matter more than what you measure the day you arrive. That ambiguity is not a failure. It’s direction. A study that names no single cause but identifies the shape of the problem — multi-pathogen, spatially clustered, tied to stress interactions — gives the next round of researchers somewhere real to stand. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.