In 2006, David Hackenberg, a commercial beekeeper in Pennsylvania, noticed what appeared to be a new disease in his honey bee hives. Bees are susceptible to a range of pests and pathogens with identifiable signatures, such as the parasitic Varroa mites that have devastated US bee colonies since the late 1980s or American foulbrood, a bacterial disease that transforms developing bee larvae nestled in their hexagonal cells into a brown mush. This was different: When Hackenberg opened up his hives, he found his colonies devastated, but without any visible evidence of sick or dying bees or brood. The adult bees had simply deserted the hives, leaving behind what appeared to be a healthy queen bee, her brood, and a handful of young bees. Hackenberg had witnessed an early instance of what would later emerge as a widespread phenomenon of overwintering colony loss in American honey bees, later dubbed Colony Collapse Disorder (CCD). Unexplained colony losses at rates of up to 36 percent have been reported each year since, although the mild winter of 2011-2012 may have contributed to lower losses than in previous years (USDA 2012).

Those familiar with the history of disease research—in either animals or in humans—will know that convincing and widely-accepted explanations for an illness are often disproven by subsequent studies. Researchers’ expectations can sometimes lead them to embrace explanations prior to definitive proof. Such readers will not be surprised that in nearly each year since 2006 a newly published research study appeared to resolve the question of CCD. In 2007, researchers associated CCD with Israeli Acute Paralysis Virus, possibly transmitted by imported packaged bees from Australia (Cox-Foster et al 2007). Researchers implicated the microsporidian Nosema ceranae in 2008 (Higes et al 2008), and a 2009 study proposed that nutritional stress resulting from habitat loss was the culprit (Naug 2009). In 2012 it looked like CCD was instead caused by neonicotinoids, a new class of systemic pesticides used in agriculture, which bees ingest through pollen (Henry et al 2012; Lu et al 2012). Or maybe a new parasite, phorid flies, was causing the hive abandonment seen in CCD (Core et al 2012). Each theory has had its critics, and none has been completely reliable in predicting the collapse of colonies that appear healthy beforehand. However, beekeepers want theories to have predictive power if they are going to invest the time and effort to control any particular pathogen or parasite. Miticides used to treat Varroa infestations, for example, carry their own costs and risks to colonies, so beekeepers have good reasons to use them sparingly if mites alone are not the cause of colony loss.

Understanding and treating the problem of colony loss might appear to be a fairly straightforward matter of identifying the cause of these distinctive symptoms: a problem that, for example, epidemiologists confront on a regular basis in both humans and animals. Samples of beeswax, larvae, and the few bees that remain in hives affected by CCD can be tested for a range of bacteria, fungi, viruses, and environmental toxins. In this view, eventually one variable will prove to be consistently present in diseased colonies and absent in healthy colonies. That single cause can then be eliminated with a tailor-made treatment. Following this logic, the US Department of Agriculture is currently funding surveys focused on identifying a cause of CCD, and Beeologics, a company purchased in 2011 by the biotech giant Monsanto, is marketing novel treatments aimed at neutralizing viruses associated with CCD.

A summary of current research on CCD could easily become a story about the different interests that promote, or dismiss, a range of possible causes of CCD. A narrative of that type would have much to say about the politics of agriculture and bee management practices. Kleinman and Suryanarayanan (2012, 18) have shown how academic priorities and industry interests can lead to a “normatively induced ignorance” in insect toxicology because certain observations and measurements related to pesticide effects are valued over others. But I want to focus on a different problem. This is the difficulty of knowing what a healthy bee colony actually looks like, or establishing the baseline against which a diseased colony might be compared.

The difficulty of precisely characterizing a healthy colony can be perplexing for beekeepers and entomologists who work closely with bees. An experienced beekeeper simply knows when a colony is healthy or sick. (In my research, I am interested in the degree to which these intuitions can be mapped onto measurable factors that entomologists might use in research). Hives with healthy bees smell “like beeswax in the sun” according to one graduate student in entomology. Sick colonies smell different, like rotting dead bees that haven’t been removed from the hive according to usual bee routine because all of the other bees are occupied with being sick as well. Healthy colonies also have a recognizable sound, a contented hum very unlike the disgruntled buzz of a colony that is missing a queen (or isn’t “queenright” in beekeeping parlance) because she has died from disease or injury or has been killed by workers who perceived that she was unwell.

Despite beekeepers’ wealth of tacit knowledge related to colony health, it is nonetheless turning out to be hard to say what, exactly, a healthy bee colony is. One problem is that bees are social insects. Kleinman and Suryanarayanan (2012, 10 and 13) have explained how beekeepers’ own assessments of colony health take into account that a bee colony is a “superorganism” not reducible to the sum of its individual members. If you open up a hive and find a bee with deformed wing virus, that is worrisome, but it doesn’t necessarily mean that the entire colony is affected and will soon succumb to the disease. Honey bees are obsessed with cleanliness—they have reliable hygienic behaviors—and they are good at getting rid of sick bees. These traits are dependable enough that they produce a “social immunity” that may even compensate for honey bees’ relative absence of immune genes (Evans et al 2006). Infections have to reach a certain density in the colony to present a problem for the community as a whole. A sick colony, then, is something other than a colony with some sick bees.

But that isn’t all. Entomologists have conducted surveys of hives in an attempt to catalog all of the microbes present in collapsed hives and identify pathogens present in all cases of CCD. Most colonies carry a significant burden of disease-causing organisms at any given time, with different ones dominating the mix at different times of year (Runckel et al 2011). And in many cases, these are perfectly healthy colonies, which don’t appear compromised in the least. “Being sick” for a colony doesn’t simply mean carrying the organisms that cause sicknesses.

Finally, there is the likelihood that the phenomenon that we call colony collapse disorder may not be caused by a single pathogen, or even a pathogen exclusively, but rather result from multiple stressors acting in concert (Neumann et al 2010). For example, ingesting systemic pesticides may lower the threshold at which bees are able to survive an attack of the intestinal microsporidium Nosema ceranae, making a potentially tolerable infection lethal. That sublethal doses of pesticides can be fatal when combined with other factors is a possibility left out of conventional toxicity assays (Kleinman and Suryanarayanan 2012). Development and climate change lead to meager foraging options, and researchers agree that malnourished bees succumb more rapidly to parasitic mites. Varroa mites, meanwhile, can act as vectors for viruses, rapidly spreading otherwise isolated infections among the bees in a hive (USDA 2005). Overwintering colony losses may be caused by different combinations of factors in different regions of the US, tracking regional differences in microorganisms, climate, pesticide use, and apicultural practice. While the problem of colony loss first identified in 2006 has persisted, cases with the classic symptoms of CCD are rare, suggesting again that colony loss may extend beyond the problem of a single, new syndrome.

At least two of these confounding factors—that harboring pathogens does not automatically mean that an individual bee or colony is sick, and that no single factor may be sufficient to cause colony collapse—have suggestive parallels in human disease ecology. Recent surveys of gut microorganisms in humans have demonstrated that healthy humans routinely carry significant numbers of disease-causing organisms, apparently kept in check by the other microflora present in a well-balanced digestive tract. Doctors have long understood that humans are more susceptible to diseases like tuberculosis when they are also malnourished. Finally, scientists are increasingly concluding that disorders like autism may never resolve into a single discrete disease entity but may represent a range of different disorders, all of which manifest in superficially similar cognitive and behavioral characteristics.

Jake Kosek (2010) reminds those concerned about the health of honey bee populations that to even discuss bee health one must remember that the honey bee is a species biologically shaped and managed by humans, who have bred bees to promote docile temperaments and high honey production, and encouraged them to live in manufactured hives—yet another wrinkle in ascertaining what a healthy bee looks like in “the wild.” Bees themselves are not just bees, but configurations of human agricultural exigencies, crop management practices, beekeeper preferences, and biological constraints.

Kosek’s point is significant, but it is perhaps equally important to recall that CCD occurs in the context of health crises—or at least population declines—in a range of pollinator species. These afflictions range from the rapid population declines that have devastated several US species of bumblebees to white nose disease in bats. Many of these other pollinators also experience the effects of human behavior but have not been reshaped by human artifice the way that honey bees have. It is also key that public and professional uncertainties over the cause of CCD suggest that many of those involved with bees understand the deeply interdependent nature of bee cultivation and human culture, although they might use different terms than the ones employed by social scientists.

Complex and unexplained sicknesses reveal the tenuous nature of “healthy” states in both animals and humans. That some succumb may be an accident of location, life experience, or genetic variation. But the more central issue, and the source of some of the current uncertainty over how to characterize and address pollinator health problems, is how putatively healthy systems, be they hives or human bodies, have become increasingly vulnerable to stress, disease, and disturbance. What counts as healthy, meaning what is measurably healthy, may not be as robust as we might hope.