Further reflections on bee monitoring: The debate over lethal sampling

Zach Portman
11 min readNov 2, 2023


It’s now been over two years since I published my original paper on the state of bee monitoring and over one year since I posted previous reflections. I’m now back with the next installment on what I see as one of the biggest unresolved issues in the field of bee ecology and monitoring: the issue of lethal sampling.

When I first put my thoughts to paper for this post, I had a number of disparate bee-monitoring-related topics I wanted to touch on, but it quickly became apparent that the issue of lethal sampling is one of the most important and pressing issues facing the field.

In the past year or so, there have been numerous papers published that discuss the issue of lethal sampling in bee studies and advocate for less-lethal methods and ways to better utilize existing data and specimens. This issue is particularly relevant for bee monitoring, since many of the most outspoken proponents of bee monitoring advocate for large-scale lethal sampling. For example, the classic proposal by LeBuhn et al. (2013) estimated their monitoring scheme for the US would kill 624,000 bees in the first five years. As I’ve discussed before, the approach of killing large numbers of bees for the sake of monitoring and conserving them is ineffective and rather paradoxical and I do not support it.

The debate over lethal sampling is also important because of the rise of passive lethal traps, such as bowl traps and especially vane traps, which allow bee biologists to collect huge numbers of bees with relatively little effort. And it seems like there are a growing number of projects and studies that are deploying these traps, often in an irresponsible manner.

Here, I want to discuss this issue head on, and in particular highlight multiple recent papers that have called attention to this important debate. I also critique a paper that seems to serve as blanket justification for many studies that lethally sample bees, and I discuss the rise of what I call “extra-lethal” bee studies. Finally, I have some thoughts and hopes about the future of the field.

(meme by Ian Lane, modified from comic by Jake likes onions)

Recent papers on the debate over lethal sampling

One interesting recent trend in the field is the number of papers calling for the use of non-lethal, less-lethal, or more sustainable methods when studying bees.

Montero-Castaño et al. (2022) published a review article covering best-practices for bee research and the broader issue of lethal collection of bees. They documented how the total number of studies employing lethal capture is increasing, many of them are catching large numbers of bees using both passive traps and netting, and the majority of studies are not engaging in best practices of data sharing and specimen deposition. They also offer up some solutions and recommendations for researchers to follow, including a helpful flowchart in Figure 2.

Miller et al. (2022), in an article in American Entomologist, highlighted the issue of how increased conservation concern about bumblebees may be leading to more lethal sampling. They document how the rate of lethal sampling in North America has accelerated and recommend scientists use non-lethal methods when possible and urge more rigorous standards when it comes to specimen collection and deposition. A reply to this paper was published by Camillo (2022), though from what I can see, the reply seems to agree with the major points of the original paper and also advocates for less-lethal sampling strategies such as not collecting bumblebee queens.

Dorian and Crone (2022) published a thought-provoking preprint that advocates for population ecology studies that use field identifications. They also raise a number of good points, such as how there is an imbalance in the field of bee ecology due to the glut of community-level data but a lack of knowledge of the life history of species that would allow us to properly interpret that data.

Note that as a taxonomist (who has often critiqued bad identification methods), I fully support the use of field identifications in the context of known bee populations. For example, field identifications at nest aggregations are straightforward for most species, and are possible even for difficult-to-identify species as long as one knows which other similar bee species are present.

Most recently, Barrett et al. (2023) highlighted the work of many of the other scientists discussed here and they discuss the issue of lethal sampling in the broader context of insect welfare.

Overall, there are a few common themes across these papers: the number of lethal studies on bees is growing, the total number of bees killed for research is increasing rapidly, and scientists could be doing a much better job adopting best-practices to minimize lethal capture and to make already-collected specimens available for further study. It’s also interesting how there seems to be a generational gap, where the push to use less-lethal methods and adopt better research practices seems to be largely driven by younger scientists, whereas these issues (at least in my experience) don’t seem to even be on the radar of many established and older scientists.

Does lethal sampling affect bee populations? A critique of Gezon et al. (2015)

It’s a truism that we affect systems when we measure them. When we are attempting to measure bee populations by lethally sampling them, we are directly impacting the very thing we are seeking to measure. Many scientists are aware of this and try to account for it, but others simply dismiss it or ignore the issue altogether.

When the issue of lethal sampling is debated, one paper is inevitably brought up, and often used as a blanket justification for the view that lethal sampling does not harm bees or affect bee populations: “The effect of repeated, lethal sampling on wild bee abundance and diversity” by Gezon et al. (2015).

However, there are many reasons why the results from that paper would not necessarily apply to most cases where bees are lethally sampled. For example:

  • The study was conducted in relatively unfragmented and undisturbed habitat in the Rocky Mountains in Colorado. As a result, the sampled areas could likely be easily recolonized from adjacent areas.
  • The analyses did not look at the response of any individual bee species, and instead examined community metrics (i.e. overall patterns of abundance and diversity). As a result, any ill-effects of lethal sampling on particular bee species could be masked by species that are less impacted by lethal sampling.
  • The sampling regime tested is much lighter than most similar bee studies, with each sampling event consisting of 45 minutes of netting and ~9 hours of pan trapping, and a total of only 5 sampling events in 2009, 7 in 2010, and 7 in 2011.
  • The study has a fundamental confounding issue that makes me seriously question the validity of the results. Specifically the study compared 6 sites sampled from 2009–2012 with 17 sites sampled only in 2012. Importantly, as noted in the paper, 2012 had “extreme drought conditions” that “resulted in extremely low floral resources during that summer.” This is important because it has previously been found that catch rates in pan traps are higher in drought years (for example, see Mayer & Kulhmann 2010 and Baum & Wallen 2011), presumably because there are fewer flowers around to compete with the traps. Figure 1 from Gezon et al. is shown below, showing the catch rate of bees per year. Note how 2012 (highlighted in red) is clearly an outlier year with much higher capture rates in pan traps.
Figure 1 from Gezon et al. (2015), with a red box added around the year 2012, which was both an extreme drought year and the only year that compared sites that had been previously sampled with unsampled sites.
  • Lastly, one overlooked aspect of the Gezon et al. (2015) study is that when they say “lethal sampling for bees using pan traps and netting”, they are defining pan traps and netting in a different sense than almost every other bee study because they made a deliberate effort to catch and release bees that could be identified visually (which was most bumblebees). This is in contrast to most other bee studies where bumblebees are collected like any other bees. In fact, I would classify the Gezon et al. study as “less-lethal” since they sought to reduce lethal capture, yet this aspect is almost universally ignored by studies citing it to claim lethal sampling of bees does not affect their populations.

All that being said, I do think that the authors of the Gezon et al. paper do a pretty good job pointing out the limitations of their findings and conclusions in the discussion, as they note the uniqueness of the habitat, high turnover of species in the Rocky Mountains compared to other areas, and they mention how they did not examine the response of any individual bee species.

However, in the broader bee literature, these limitations seem to be universally ignored when the paper is used as a blanket justification for all kind of intensive lethal sampling of bees. The paper is invoked to justify a range of bee-collecting efforts, from small studies to large-scale lethal bee monitoring. The paper is also used to justify very different sampling methodologies (such as trap nests and vane traps), to the extent where I suspect many researchers citing the paper have not actually read it.

The rise of extra-lethal bee studies

In contrast to the work highlighting non- or less-lethal sampling discussed earlier, there are a number of recent studies that I would term “extra-lethal”. These are studies that leave out vane traps continuously (without taking them down) for extended periods of time.

Some studies simply make no attempt to justify their extra-lethal sampling methods. Others justify their trapping methodology by citing the Gezon et al. (2015) paper, even though that paper explicitly states: “our results apply to sampling bees every two weeks and cannot be used to assess whether more frequent sampling, for example weekly or even daily, may negatively affect bee populations or communities.”

For example, a study by Galpern et al. (2017) sampled bees continuously for a 48-day period using vane traps. Galpern et al. justified this by stating: “Our sampling activities in this region are unlikely to have impacts on population health, given the low spatial density of traps (>3 km spacing), and evidence that repeated sampling of a similar magnitude has no long-term effect on pollinator populations (Gezon, Wyman, Ascher, Inouye, & Irwin, 2015).” However, they report collecting 866 bumblebee queens, indicating a relatively high level of impact.

Another study by Turley et al. (2022) set out vane traps continuously from April to October over six years as part of a bee monitoring program. Rather unsurprisingly given the extreme levels of collecting, they found declines in most bee species that were collected in high enough numbers to analyze. Looking closer at the raw data, there are dramatic declines in many bumblebee species, and it appears that some species, such as Bombus perplexus and Bombus vagans, have undergone precipitous drops over the course of the study. The data from the common bumblebees is plotted below:

The patterns of collected bees over time for the six most common bumblebee species in the study.

The pattern seen across all combined bumblebees is especially alarming. The study does not break down the bumblebees by caste, so it is unknown how many queens were collected, but I suspect queens were collected in large numbers given how early the vane traps were set out.

The number of collected specimens for all bumblebees combined.

In that paper, which collected the bees as a broader bee monitoring project, the authors largely dismiss oversampling as a potential reason for the declines they found in many different bee species.

However, a previous study by Gibbs et al. (2017), issued explicit warnings about the dangers of oversampling with blue vane traps (BVTs):

A recent study suggests that passive sampling for bees using colored bowls does not impact bee communities, owing to consistent communities being found over multiple years and at similar sites never before sampled (Gezon et al. 2015). Our results suggest that this may not be true of BVTs, as the extremely high abundance of some bee species captured may be problematic.
Furthermore, there is strong evidence that BVTs may negatively affect those bee species that they preferentially capture. The effectiveness of BVTs means they have value for sampling bee communities in other contexts. However, they should be used with care (e.g., limiting the number and duration of use) or excluded from sampling protocols in favor of other less destructive methods, like bee bowls (Gezon et al. 2015), in areas where bee populations may be sensitive to local extirpation or when at-risk species, such as Bombus affinis Cresson (Hatfield et al. 2016), are known to be present.

Given the results and warning by Gibbs et al., declines due to oversampling should be the primary hypothesis to explain declines when bees are sampled with vane traps that have been left out continuously. This level of sampling — with vane traps set out continuously for months — is extreme and frankly irresponsible.

Conclusion and the future

One thing that struck me, as I was perusing through the papers that had cited Gezon et al. (2015), was the number of papers that used that study as a way to explicitly dismiss concerns about lethal sampling without any consideration given to how their methods and study system might differ. Basic logic dictates that sampling bees in different circumstances will not have the same results as the Gezon et al. study. For example, sampling in rare, isolated, and small patches of habitat will obviously have a much greater risk of wiping out local populations of bee species.

As scientists who study bees, I don’t think any of us feel particularly good about the fact that much of our research requires lethal sampling. I’ve certainly grappled with that myself, especially when I see projects that kill absurd numbers of bees for little to no scientific benefit (and the many studies that are never finished). The field of bee ecology hosts a wide diversity of studies, ranging from studies that use less-lethal methods and engage in best-practices for depositing specimens, to studies that kill huge numbers of bees and do not engage in best practices. Unfortunately, I think that many scientists are all too eager to dismiss any concern about lethal sampling, since the alternative — that we might actually be harming the very organisms we care about — is a scary idea to face.

However, despite everything I am mildly optimistic about the future, especially given the attention given to this issue by the scientists who I highlighted in the earlier section on the debate over lethal sampling. In the clash between those advocating for less-lethal sampling and those performing extra-lethal sampling, I think it remains to be seen which side will predominate. Though the entire debate may become moot in the future as the increasing number of bees of conservation concern (such as the endangered rusty patched bumblebee) render extra-lethal sampling methods increasingly indefensible.

In the end, I hope that scientists will be more thoughtful and deliberate about how and why they collect bees. I encourage people to read the recently published papers on the topic (Montero-Castaño et al. 2022, Miller et al. 2022, Dorian and Crone 2022, Barrett et al. 2023), as they have clearly thought deeply about these issues and the implications for the broader fields of bee ecology and biology. And I hope that scientists will stop using a single study as blanket justification for all manner of lethal sampling.



Zach Portman

I am scientist who studies bees. My research covers the identification, biology, evolution, and conservation of native bees.