The BIP Annual Loss Survey -a Conversation with Nathalie Steinhauer

​BIP released preliminary results for the 14th annual survey in June of 2020. This exchange has been edited for length and clarity. 

Nathalie is the scientific coordinator for the Bee Informed Partnership (BIP). From Nathalie to the PAm audience: “If you took the survey, THANK YOU! I would hug you, except, you know, social distances.”

​We have all seen the chart showing the percentage of bees lost over the years. In recent years it has included “Total Annual Loss” in addition to winter loss-a reflection of requests from beekeepers who emphasize the importance now of losses year-round.  Loss rates are estimates of colony turn-over over a season; a mortality rate of colonies and units lost to combinations. It is not a count of the total number of colonies in the country.  
 
The survey began via the Apiary Inspectors of America in 2006 and was taken over by BIP a few years later. Since 2019 Auburn University’s Geoff Williams, who is now the president of BIP, and his Ph.D. student Selina Bruckner, are administering the survey for BIP with assistance from many organizations* and individuals who help get the word out. Winter loss was down 15.5% from last year,  and 6.4% from the historic average. 

Here is the most recent annual loss chart as seen on the Bee Health Collective's web page on Honey Bee Health. Visit the sit at the BeeHealthCollective.org

Grace (PAm): Participation was lower this year than the last few years, can you comment on that? 
 
Nathalie: Indeed. Even before this year, we did notice that participation tends to decline year after year. We are constantly trying to make the experience less painful, but we can't really blame anyone comparing our survey to their tax form.
 
This April, we noticed the lower participation straight away, as we can compare the number of people logging in [to respond to the survey] over time with previous years. We knew this year would be different because of COVID. It is likely it played a role. We also made changes compared to previous years. One major change was to require participants to log in, which for us was a necessary upgrade, but might have been an extra barrier. 
 
In the past, we had ways to identify duplicate entries that were flagged by the system because of similar IP addresses. Because of the new EU rules [on data protection], we couldn't rely on this anymore, [and so we had to make the log in mandatory to avoid duplicate entries]
 
Also, in past surveys, people used to risk logging off if inactive too long on a page and wouldn't be able to finish their entries and had to start over. With the [new] login, they would be able to pick up where they left off, so we thought (hoped?) the improvement would be well received.
 
Still, thanks to the help of our collaborators who relentlessly shared and pushed our survey, we had over 4,000 entries (compared to the last 5 years which averaged at ~ 5400. 

Grace: Can you comment on the overall process of running the survey?
 
Nathalie: We usually start preparing for the survey months in advance. Every year there are a lot of steps that need to happen: update the survey on the online platform (there are always some changes to try and make the survey more intuitive and streamlined while keeping it comparable with past years for comparison sake), test it, create the paper survey to match, print and distribute it, contact all our collaborators, write the announcements that will need to be dispersed, and then when the survey goes live, keep the communications up to encourage a maximum of people to answer and share on their own. 
 
We also have a team on support in case our respondents encounter technical issues. When the online survey closes by the end of April, we still are receiving paper surveys that need to be entered [into the online system].
 
After all the data is in (usually about a month after the survey was officially closed online), comes the validation. We look closely for similar answers to remove potential duplicates, [and] we also look for typos and aberrant answers (for example, answers that would result in more than 100% mortality). We have systems in place so we don't have to decide answer by answer subjectively. We unfortunately also filter out non-US respondents. Maybe in the future, we will include them, but for now, our scope is to look at beekeeping operations in the US.
 
Once the data has been through validation, we calculate our estimates: National estimates, and operation types specific estimates, for each season.
We typically publish a preliminary number at the beginning of May (though we were delayed this year) in our press-release. It is preliminary because, by May, we are typically still receiving paper surveys that need to be entered. Later in the summer, we use the finalized data set to draw our state maps (and also release that publicly through our online platform 
 
At this stage, we usually start to write up the manuscript to be published in a scientific peer-reviewed journal. Selina Bruckner, a graduate student from Auburn is currently working on a summary of the last 3 years [for publication]. And we are also working with a geography Professor from Auburn on a spatial analysis of the last9 years of the winter loss data for stationary, small-scale,  beekeepers! 
​

The “2019/2020 Total Winter All Colony Loss” map is available on the Bee Informed Partnership's data sharing website. This interactive map allows anyone to see how the results break down for each state, including the number of respondents, and number of colonies represented. Visit the website at research.beeinformed.org

 
Grace: It seems winter loss is down pretty significantly from last year, (and down from the year over year average) any thoughts as to why we are seeing that trend? Is it because so many colonies were lost the previous year? 

​Nathalie: Generally speaking, scientists have identified multiple reasons why colonies can be lost. We usually group them in the “4Ps”: parasites, pathogens, pesticides and poor nutrition. Now, why some years come with higher turn-over rates of colonies that other years, that is harder to pinpoint. We know rates fluctuate from year to year. We've actually come to expect a "bad year" to be followed by a "better year" and reciprocally.
 
It is very difficult to speak about one year's winter loss in particular, definitely as more and more data point to the fact that "summer losses" have shown to be anything but trivial.
One thing we can tell from our survey is that here is no indication the loss rates are going into one steady direction: it's not getting better, it's not getting worse, some years are just worse than others.
 
Regarding why this year winter loss was relatively low...... This is where we need to differentiate the data and the hypotheses. The survey's objective is to record the turnover rates of colonies, but we're not really able to get to the causes of those losses with this data. It's not like we have autopsy reports with causes of death for every single colony...We do ask beekeepers their opinion as to what caused their colonies to be lost, but it is rather subjective, and mostly indicative of what the beekeepers perceive are the high risk factors.
 
Concerning this particular year, we have heard anecdotally from various sources that spring 2019 was particularly late and wet, slowing a lot of the development of colonies and queen rearing early in the year, which meant colonies did not grow as strongly as you would have wanted, so that could explain the bad summer losses. Winter 2018-19 was a very bad one on the record, so colonies came out of a bad winter into a poor spring. 
 
When we dig deeper into beekeeper categories, we realized it was mostly large-scale migratory beekeepers that seemed to suffer the most that summer. It is somehow surprising because, though large-scale beekeepers tend to report slightly higher losses in the summer, this year seemed to have been particularly bad.
 
From past survey years, we've come to expect that large-scale beekeepers lose colonies quite evenly in the summer and winter, faring much better than small scale beekeepers in the winter and slightly worse in the summer. This year, however, their summer losses were even higher than their winter losses, which is really not frequent (it only happened one other year on record).
 
Then Winter 2019-2020 came with a relatively low level of losses, as it turned out. It might have been because it was a relatively mild winter, or it could be because last year was so bad it culled all but the strongest colonies...
 
 
One frustrating aspect of any scientific work is that we're always a little late. For example, after this year's survey was over, we heard that spring 2020 was a great one in comparison for queen rearing and colony growth. But we'll have to wait [until] 2021 for the survey to get the numbers on that!



Grace: Why do you think commercial beekeepers tend to lose fewer colonies in the winter [than small scale beekeepers]? Has the practice of storing bees for the winter impacted losses on a national level? (Or do you have any plans to look at winter storage as it relates to losses?)
 
Nathalie: If you look at the losses of commercial beekeepers in the winter and summer, they tend to lose their colonies quite evenly (like I said above), compared to small scale who lose their colonies mostly in the winter.
Again, we know that they do, we don't exactly know why. It could be a number of things.
We know a lot of commercial beekeepers cull their own colonies in the fall in preparation for winter rather than trying to winter high numbers of small colonies (which technically then count in our "summer losses").
 
Winter storage has gained popularity recently and BIP is currently collaborating on a project with WSU led by Dr. Brandon Hopkins that is currently following a cohort of colonies wintering in different conditions.
 
Grace: Are people who lose more colonies more likely to participate in the survey versus those who do not?
 
Nathalie: That, or the opposite, that they wouldn't want to report it. The problem is, how would we know if they don't answer? It is a worry of any non-randomized survey that the respondent pool is biased somehow compared to the target population. We want to represent all managed US honey bee colonies and beekeepers, but we don't have a complete listing to pick a random sample from. So, it is a limitation of our voluntary based survey that we and anyone using our data have to keep in mind.
 
That is also why we put more emphasis on the trends in our data, and comparisons between groups than in the actual number itself.
 
From year to year or region to region, if there is bias, it is unlikely to change drastically from one to the next. That means any difference we see between groups is still meaningful. We have our baseline: ~30% winter mortality over 13 years, and we can see regions and groups that exceed this. We can see groups of beekeepers more at risk than others. In the end, it is all about relative risk: who is most at risk, compared to a baseline.
 
A couple of years ago, the National Agricultural Statistics Service (NASS) decided to start their own Colony Loss survey. They have access to the list of beekeepers who qualify as farmers from the Census of Agriculture, and they are therefore able to select their survey sample randomly from it. It is a chance for us to look at loss estimates from both surveys and see how they compare over time. We will provide a comparison in our next peer-reviewed publication.


Grace: If you could tell beekeepers one thing about the survey what would it be?
 
Nathalie: The survey is meant to document the real world of beekeeping, meaning it is noisy and complex. It is meant to keep a pulse on the health of our honey bee populations, but it is only one tool, in a very large toolbox. Like any tool, it can be useful, but not on its own.
 
From observational to experimental research, basic to applied, surveillance, and monitoring in the field to molecular bio-assays in the lab, every aspect of research builds upon each other, each with their strengths and limitations.
 
Our work is but a fraction of that of a whole community of scientists and extension agents and stakeholders in improving our understanding of honey bee health.
 
We're as thankful to the work of our colleagues as we are of our participants and excited to be able to contribute our part, by asking more questions and maybe getting more answers.
 
Authors note: The Bee Informed Partnership is a supporter of the Bee Health Collective a resource that provides honey bee health information which Project Apis m. oversees. Project Apis m. has supported the Bee Informed Partnership’s Technical Transfer Teams in various ways over the years, you can learn more about that support here. Grace and Nathalie attended graduate school together at the University of Maryland and have been friends ever since. 
 
* “AIA, ABF, EAS, Honey Bee Health Coalition, American Bee Journal, Bee Culture, our PI institutions (UMD, Michigan (MSU), Texas A&M, University of MN, OR State University, and UCD) and many bee clubs) that help us by spreading the word. And we are very thankful for their help!"-Nathalie. 

By: Grace Kunkel
Grace works as a Data Coordinator and assists with communications on the PAm team.  ​