Following the previous season's staggering 60% honeybee colony losses sustained by commercial beekeepers in the United-States, a group of researchers at the USDA recently conducted a study on the impact of viral infections on honeybee mortality. Among other findings, the study concluded that:

• Colonies that suffered severe losses in 2025 had elevated loads of the Deformed wing virus (DWV) of strain A, the Black queen cell virus (BQCV), and the Acute bee paralysis virus (ABPV)
• Young and healthy adult bees that received diluted viral doses showed elevated mortality after 36 hours
• Varroa mites from all collected samples showed resistance to amitraz, a well-known miticide used to control varroa

We believe that the study's results offer interesting insights on how honeybee colonies respond to viruses. However, the observed resistance to amitraz has already been reported in other studies (R4P Network, 2016; Bahreini et al., 2025). Just as antimicrobial resistance has become a major public health concern (World Health Organization, 2023), the widespread use of chemical pesticides can drive the evolution of resistance in pest arthropods—a phenomenon shown time and time again in farmland and forest ecosystems (Hawkins, 2018).

Further, while viruses are a likely end-stage cause of colony mortality, as the USDA researchers state, the rising trend of honeybee colony losses is a multifactorial problem that cannot be solely attributed to viral pathogens. Other factors such as climate change and habitat loss, among others, can also exacerbate the susceptibility of honeybee colonies to parasite or viral infections, thus increasing mortality. In this context, we believe that the beekeeping community and stakeholders mustn't lose focus on other pressing issues associated with honeybee mortality. 

As a company deeply rooted in the commercial beekeeping sector, and amid growing calls for practical solutions to prevent honey bee mortality, Nectar commends the USDA’s efforts to investigate these urgent issues. We believe that constructive engagement from the wider scientific and technological community can help enrich the conversation and accelerate progress. Hence, after thoroughly reading the USDA-ARS study and reflecting on our own findings regarding the impact of beekeeping practices on colony survival, we wanted to provide a constructive review of the results, discuss their broader implications, and share suggestions for advancing research on honeybee health.

Viruses Contribute to Honeybee Colony Declines

What the Study Tells Us

In the first part of the study, USDA researchers detail how they found a high prevalence of RNA viruses in colony samples provided by six large commercial beekeeping operations. Specifically, out of 113 colonies (72 weak and 41 strong), 71.68% had the ABPV, 75.22% had the BQCV, and 77.88% had the DWV of strain A. While these numbers show an important and widespread load across samples, there were no differences in pathogen load among weak and strong colonies. In addition, researchers did not find evidence of differences in pathogen load among beekeeping operations, suggesting that all operations experienced similar degrees of pathogen infection.

As expected, bees showing symptoms of behavioural impairment (n=38) had notably higher viral loads of both DWV-A and DWV-B compared to asymptomatic bees (n=28). In addition, none of the asymptomatic bees were infected by the DWV-B virus, suggesting a potential causal link between the presence of these viruses and honeybee mortality. A subsequent laboratory essay showed that 44% of healthy and young adult bees receiving a highly diluted viral dose (0.00000001%) from a morbid bee ended up dying within 36 hours. However, it is important to remember that these results were obtained experimentally using direct injection, which isn’t how bees get infected. Yet, it highlights how virulent these isolated pathogens can be.

Ultimately, the research brings us to its core message: that all varroa samples (n=39) collected from 18 colonies (13 alive, 5 dead) across five operations showed a genetic marker for amitraz resistance. Researchers claim that this resistance trait was far less common in a published U.S. survey from 2023 (Rinkevich et al., 2023).

What the study fails to address

When we read the research paper, the first thing we noticed was that the sampled colonies from six beekeeping operations were inherently biased, as all faced significant losses in 2025 (over 60%). One of the most important steps before carrying out any statistical analysis of data is to collect random samples that are representative of the true population. This means that collected samples should cover the full width of the phenomenon under study — in this case, colony mortality. 

While this does not necessarily invalidate the USDA-ARS study, it is an important caveat that should be highlighted when interpreting their results. For instance, we shared in a recent blogpost that some of the operations that Nectar works with suffered around 30% losses on average — that is, half the losses reported by the operations surveyed in this study. Further, as the researchers in the USDA-ARS study stated themselves, there is substantial variation in reported losses from operations included in the PAm survey. So to uncover the bigger picture, it is critical that studies analyze samples that are representative of the broader beekeeping industry landscape, including beekeeping operations that did not sustain losses as high as the ones reported in PAm’s study.

In our opinion, the most important limitation of the study is that 1) its small sample size limits replicability and 2) that the analyses are currently not reproducible. Replicability and reproducibility are major principles underpinning the scientific method, and we believe that to move forward in beekeeping research—especially where the results of studies often lead directly to field-level actions among beekeepers—studies need to provide the means to replicate and reproduce the results.

We identified the following concerns that may hinder replicability in the USDA-ARS study:

• Its data comes from a small sample of operations that suffered heavy losses at a given point in time (i.e., snapshot). 
• The results do not account for seasonal effects nor include a reference-season for comparison. Without a reference, we cannot exclude the possibility that the high viral loads of DWV-A at the time of collecting samples were simply due to chance. 
• Moreover, low sample sizes can overestimate statistical effects, increasing the chances of detecting false positives. 

Ultimately, the fact that a small data sample was collected from six operations that suffered heavy losses in a single season raises concerns of potential biases that are currently not addressed in the study's conclusions. Given these concerns, we suggest that the results from this study should be published in a peer-review journal once replicability is ensured, by collecting samples over more than a single season and across a larger spectrum of operations.

In terms of reproducibility, the authors shared the software and packages used to perform their analyses. While this is an important first step, it is unfortunately insufficient for other researchers to fully reproduce the results. To strengthen reproducibility and transparency, the following steps could be taken (see the detailed guidelines by the Center for Open Science):

• The raw data and metadata is (if possible) deposited on a remote repository freely available to the public. Otherwise, a mock dataset allows users to reproduce the results
• The software, code, and package versions used to produce the results are reported and deposited on a remote repository freely available to the public (e.g., GitHub)
• The computational environment to produce the results is versioned and shared, allowing for the results to be reproduced outside of the computer used to generate them

We understand that this work is pre-peer review, but we encourage alignment with Open Science principles. Many scientific disciplines are moving towards an Open Science approach in order to maintain and improve trust in the scientific community. Reproducible practices strengthen the credibility and utility of research—particularly for interdisciplinary collaboration and policy influence. This is especially important given the recent reproducibility crisis that has plagued many disciplines including medicine, economics, and psychology, to name a few (Ioannidis, 2005; Open Science Collaboration, 2015; Baker, 2016). Fortunately, solutions are at our disposal to make beekeeping science more transparent and reproducible (Roche et al., 2021; Gomes et al., 2022; Braga et al., 2023).

Viruses: Cause or Symptom of a Larger Problem?

The results of the USDA-ARS study may prompt an urgent search for solutions to the issue of virus-induced bee mortality. For instance, stakeholders may feel compelled to explore alternatives to amitraz by funding research for the development of new synthetic pesticides. Of course, the use of pesticides within an integrated pest management approach will continue to play a role in addressing the problem of varroa parasitism. Yet, we believe that these observations raise a fundamental question: are viruses really the cause of higher colony losses… or the symptoms of a larger phenomenon?

Over the past decades, honeybee mortality has been steadily increasing, and numerous studies have pointed out that resource availability, landscape configuration, and air quality, all contribute substantially to honeybee health and survival (Vanderplanck et al., 2019; Zhang et al., 2023). At Nectar, given our software’s integration with multiple large-scale migratory operations, we have recently been able to quantify the effects of beekeeping practices on honeybee colony mortality. By leveraging our capacity to provide quantitative measures of beekeeping practices, we provide evidence that colonies moved for pollination at a high rate have shorter lifespans. While this does not unravel a causal effect on mortality, we can hypothesize that moving hives a lot, and placing them at high densities in homogeneous landscapes (e.g., monoculture farms), may facilitate the dispersion of varroa parasites, stress colonies, and ultimately exacerbate infection and mortality. 

As different studies point out, most of the viral transmission is done by varroa (i.e., transmission between bees is inefficient), and the growing epidemic of the honeybee viruses is essentially mediated by the trade and movement of colonies (Wilfert et al., 2016; Martinez-Lopez, 2022). In addition, a recent study using Nectar’s database adds to the growing body of evidence that certain crops, such as cranberries and maize used in commercial pollination, have detrimental effects on honeybee survival (Vadnais et al., 2025). Hence, if the industry wants to properly understand the reasons behind the increasing high losses that it is facing, studies such as the one published by the USDA-ARS should also aim to explore and quantify the role of beekeeping practices on honeybee health.

Environmental effects on honeybee health, such as the continued rise in forest fires accompanied by the deterioration of air quality, are additional concerns that the scientists should continue to address. Poor air quality is linked to reduced foraging efficiency and increased mortality in honeybees (Thimmegowda et al., 2020; Phillips et al., 2021; Coallier et al., 2025). Yet, providing access to high quality environments with sufficient vegetation can alleviate these adverse effects (Vanderplanck et al., 2019; Coallier et al., 2025), outlining how simple solutions can help in preventing/reducing colony mortality. If we think about varroa, studies show that climate change promotes range shifts in their distribution across landscapes, facilitating their capacity to parasitise honeybee colonies (Giliba et al., 2020; Smoliński et al., 2021). Moving forwards, it becomes clear that preventing varroa infection requires integrated solutions that go beyond the use of chemical pesticides.

Where should we go from here?

Moving forward, we encourage all beekeeping researchers to leverage state-of-the-art approaches to data analysis. Such approaches would combine advanced statistics with predictive (AI) and causal modeling to unravel how the big picture unfolds. In general, we believe that the beekeeping industry cannot afford to overlook the broader picture when assessing the factors contributing to honeybee colony mortality. Technology is at hand to provide answers, and we must actively consider all contributing factors rather than merely assuming their influence.

To support this, we propose the following:

1. Silos between the industry’s stakeholders need to be broken and synapses must be created to combine expertise from the research in academia, beekeepers, non-for-profits and private organizations, such as Nectar, offering large-scale data collection and tracking capabilities. 
2. When designing collaborative efforts to solve the multiple and existential challenges that this industry is facing, the expertise from fields outside of the traditional bee industry research ecosystem is essential, including computer science, epidemiology, ecology, geography, as well as social sciences. 
3. Organizations with experience in tackling issues such as climate change and evolving agricultural landscapes should be included in beekeeping research. To solve these complex issues, we need to think outside of the box.

At Nectar, we do our best to embody those objectives by combining multiple in-house disciplines – data science, artificial intelligence, honeybee biology, and software engineering – to build the expertise required to deliver actionable solutions that help commercial beekeepers meet the moment. When we realize that our internal knowledge has limitations, we seek out collaborations with research partners that will push us out of our comfort zone in order to learn. 

We recognize that increasingly high losses are the new norm—only by adopting a holistic approach aimed at identifying the causal agents of these losses can we address this crisis. As we are starting to witness early signs of success in this battle, hope is on the horizon.