Reflections on the state of bee monitoring
Looking back two years after the publication of “The state of bee monitoring in the United States”
A bit over two years ago (May 2020), along with my coauthors Bethanne Bruninga-Socolar and Dan Cariveau, I published what has arguably been my most influential paper, entitled “The State of Bee Monitoring in the United States: A Call to Refocus Away From Bowl Traps and Towards More Effective Methods.”
Lately I’ve been thinking a lot about bee monitoring again, especially since there continue to be calls to implement large-scale bee monitoring programs in the US and elsewhere. In some ways the field is moving in good directions, with many scientists pursing new and innovative approaches but at the same time, other researchers seem intent on doubling-down on the same old ineffective techniques.
Here, I’ve put down on paper some of my reflections on our impact, address some of the misunderstandings of our work, highlight some good research that has come out since, discuss the taxonomic bottleneck, and speculate on the future of bee monitoring.
Why we wrote it and the intended audience
This paper was written during an inflection point in my career as I was transitioning to a general bee taxonomist after spending graduate school focused primarily on the taxonomy of the genus Perdita. I was encountering more and more projects that tried to monitor bees by simply placing out lots of bowl traps (also commonly called pan traps).
As I learned about more of these bee monitoring projects, a clear pattern emerged. Most of them bowl trapped a lot of bees but then got stuck in various stages of incompleteness (usually in the specimen identification phase), with the project abandoned and the results never published.
A big driver of this pattern is that there are many official resources out there that all say the way to monitor bees is to put out bowl traps. Some guides that appear to have been particularly influential on that front include “The Very Handy Manual,” official guides from the US Fish and Wildlife Service, and official guides from the UN Food and Agriculture Organization.
However, when you delve into these resources, you find that while they are extremely detailed about how to put out traps and catch a ton of specimens, they get very handwavy and vague when it comes to identifying specimens and analyzing the data. This leads many researchers into a trap, where they catch a lot of bees using bowl traps, but then hit a wall when it comes to identifying the species, analyzing the data, and storing the specimens.
I was finding the failed projects were generally being done by a combination of well-meaning conservation professionals and scientists. As a result, I wanted to write something that would reach those people. Something that when they search for guidance on how to monitor bees, they don’t just find a chorus of resources promoting bowl traps as the go-to method to monitor bees. From that perspective, I think it was a success, because it caused a lot of people to question the conventional wisdom of how to monitor bees.
The biggest surprises
For me, the biggest surprise was the lack of any kind of negative response to our paper. I was expecting some of the proponents of large-scale monitoring using bowl traps to stand up and say “No! You’re wrong!” However, I have yet to see any pushback from any, in either public or private. I think the main reason for this is that all of our main points were correct — bowl traps are bad at measuring bee abundance and populations, bowl traps have biases and their catch is influenced too much by the environment, and there is no good solution to solve the bottlenecks in taxonomy and storage.
The other surprise was how highly cited the paper has been, much more so than any of my other papers. I think a big reason is that we were essentially just stating what many people in the bee world have been saying for a long time but there haven’t been any good papers to cite about the flaws of bowl traps. My sense is that most of the citing papers cite the paper as a way to say “we acknowledge bowl traps are biased but we used them anyways.”
One particularly surprising aspect has been the number of bee monitoring papers that conspicuously avoid citing it or acknowledging the primary points in our paper. They simply ignore what I’ve come to think of as the four inconvenient truths of bee monitoring:
- The presence or absence of coflowering plants influences the species and abundance of bees collected in passive traps in ways that are not well understood.
- Especially for social species, abundance is not a good proxy for population size. There are many cases where this inconvenient fact can be ignored, but monitoring is not one of them.
- You cannot or should not monitor rare or declining species by killing large numbers of them.
- The taxonomic bottleneck is real and has no easy solutions. You cannot monitor bees if you cannot identify them.
The biggest misunderstandings
One of the biggest misunderstandings of our work is that some people seemed to think we were advocating for simply swapping out netting for bowl trapping in large-scale bee monitoring. This is incorrect, and as we point out in the paper, all of the commonly-used methods (bowls, vanes, and netting) suffer from the same root problem of being bad at measuring abundance. The focus on bowl traps was due to the number of resources promoting their use in bee monitoring. Had those resources been promoting netting or vane traps as a way to perform large-scale monitoring of bees, we would have focused primarily on those methods instead.
One thing that continues to puzzle me is that many scientists do not acknowledge the basic concept that all the common methods for capturing bees — bowls, vanes, and netting — do not provide unbiased measures of abundance. Because bowl and vane traps are attractive to bees, their catch rates will be influenced by the preferences of the bees and the flowers in the surrounding environment in ways that are not well-understood and that are not consistent across species. Netting has the same issue, since netting is performed at flowers, which again, attract bees. That being said, netting is often preferable to passive trapping because it at least allows researchers to target specific slices of the bee fauna as opposed to passive traps, which can capture an absurd amount of bees and other bycatch.
The other frustrating aspect of the conversation is that many papers advocate for bowl traps or other passive methods because they are the “most effective” method, where “most effective” is defined as the method that catches the highest abundance or raw diversity of bees. It’s frustrating to see that point made repeatedly when the metric we should be using to evaluate monitoring methods is their ability to detect change over time. In fact, catching more bees is often counterproductive for monitoring efforts, because it creates a glut of specimens that contribute to the identification and storage bottlenecks, not to mention that we want to limit the catch of sensitive or declining species. If I could magically change one aspect of this conversation, it would be for people to stop defining the effectiveness of methods simply by their ability to catch the highest abundance and diversity, especially when a disproportionate source of that diversity is often a ton of Lasioglossum species and other common halictids.
Interesting new research
One of the positive aspects of this whole experience has been the amount of interesting and useful new research has subsequently examined different aspects of bee monitoring. I like to think that our work played at least some role in inspiring this further research and it’s been nice to see many of our main points supported by more robust research.
Here, I want to take a moment to highlight a few of the papers that I enjoyed reading and I think are especially interesting and relevant to the bee monitoring conservation.
One paper I particularly like is a study led by researchers at North Carolina State University that used mark-recapture methods to test the efficacy of bowl trapping and netting. They found that bowl traps did not collect bees in proportion to their actual abundance and they found that netting actually did a much better job quantifying abundance. It would be nice to see this for more species, but it’s encouraging to see research on the effectiveness of commonly-used methods and further support that bowl traps are particularly poorly-suited for monitoring bees.
Another paper demonstrating some key biases of bowl traps was led by researchers at MPG Ranch in Montana. They showed that the surrounding floral community significantly affected the richness and abundance of bees caught in bowl traps. Interestingly, these patterns were not consistent across bee genera or seasons, with the effect of co-blooming plants most pronounced in spring.
Another study I found interesting was led by a researcher at Deakin University in Victoria, Australia who examined different ways to monitor pollination in berry crops in polytunnels. They found that depending on the methods used, the results looked very different. For example, they found honeybees to be one of the most important pollinators, though extremely few of them ended up in pan traps. I like this study because it’s a good example of monitoring that focuses on a specific type of data — in this case flower visitation and pollination — and tested different approaches. This is in contrast to many studies that collect bees but end up with little knowledge of what the bees were actually doing.
The taxonomic bottleneck
I continue to consider the taxonomic bottleneck to be the single biggest barrier to bee monitoring in the US. As a taxonomist, I’m probably a bit biased, but I also like to think my experience performing taxonomic research and identifying tens of thousands of bees a year allows for some unique insights. It is painfully apparent to me that there is an acute shortage of taxonomic specialists to accurately identify bees as well as update the woefully out-of-date taxonomic and identification resources. What little taxonomic research is being done is being performed by grad students, postdocs, or Canadians. The number of US-based bee taxonomists conducting taxonomic research on US bees is at an all-time low.
For the people who are identifying bees (both taxonomists and non-taxonomists), they are relying on taxonomic research that is incredibly out of date. For example, many bee groups in the US were last revised by T.B. Mitchell, in his Bees of the Eastern US published in 1960 and 1962. While he did excellent work, how much research from the 1960’s do you trust uncritically? We know those older works have big gaps and other issues such as numerous cryptic and undescribed species.
In addition to the lack of expert taxonomists and taxonomic research, a huge proportion of identifications are being performed by non-experts using error-ridden, non-peer-reviewed resources and keys, such as the keys on discoverlife. This results in various combinations of serious misidentifications or the most difficult to identify bees never get identified.
Lately it’s been encouraging that a growing proportion of researchers recognize that the taxonomic bottleneck represents one of the biggest and most intractable barriers to effectively monitoring bees. At the same time, however, recognition of the problem does not seem like it has led to solutions. It’s frustrating how many researchers will not even take the most basic steps to support taxonomists, such as citing the taxonomic papers used to identify their bees (though often when researchers fail to cite taxonomic works from the primary literature, it’s because they didn’t use them and their identifications should be considered unreliable, to put it mildly).
Looking ahead
Overall I think our paper was a success and achieved our primary objective of putting a dissenting scientific opinion in the literature. I’m sure there will continue to be numerous half-baked projects that simply put out bowl traps with no question and insufficient preparation and planning, but based on what I’ve seen, the sheer number of projects like that seems to be abating somewhat. Though, on the other hand, the number of half-baked projects that simply put out vane traps seems like it may be increasing. As I’ve discussed elsewhere with my colleague Vince Tepedino, the projects based on putting out many passive traps have numerous flaws, including the fact that they kill many bees without providing much information on the causes of bee declines or how to help bees thrive.
Overall I believe the future of bee monitoring will not be large scale passive trapping. The taxonomic and storage bottlenecks we laid out in our paper are not going to be resolved in the near future. So the question becomes: what slice of the bee fauna can we actually monitor? In a more recent paper we proposed bumblebees as the best target for large-scale bee monitoring in the US. And while that proposal is debatable, what is not debatable is the fact that we cannot monitor all bee species.
It continues to baffle me that there are researchers who seem to believe that we can somehow monitor all or most bee species. I suspect a big reason for this is that many researchers getting into bee monitoring have not read up on or engaged with the extensive library of research on monitoring in general. This is especially apparent when researchers suggest doing things that many general monitoring papers explicitly warn against. In particular, the practice of collecting data and then searching for a question to answer is a practice that is heavily criticized in the monitoring literature (see Nichols and Williams 2006) yet it is frequently proposed as a way to monitor bees. If you are looking to delve into the monitoring literature, I recommend the book “Effective Ecological Monitoring” by Lindenmayer and Likens as a good place to start.
I foresee the biggest challenge for bee monitoring will be these diverging philosophies about how to monitor bees. Currently it looks like bee monitoring is diverging into three main strategies: first are the researchers who want to collect as many bees as possible and then mine the data for potential declines, second are researchers pursing more limited approaches on one or a few species that are hypothesis-driven and based on natural history, and third are those who are focusing almost entirely on bumblebees.
I obviously favor the latter two approaches because they are achievable and can produce data that can inform bee conservation. For a couple interesting examples see this research on bumblebee reproduction and the rare bee Andrena parnassiae. I think that trying to shift the field away from the “collect as many bees as possible” approach will continue to be one of the biggest challenges facing bee monitoring efforts. This is because our current academic and funding system rewards the type of science that can be achieved in short funding cycles and quickly produce some kind of data, regardless of whether that data is actually meaningful or actionable. Overall, however, I’m encouraged by a lot of the high-quality and natural-history-based research that seems to be flourishing in many corners of the bee world.