• Andy Davis

New statistical study of overwintering colony size shows the importance of "greenness" dur

Hello folks,

As the title of this post shows, there was a brand new study published this week, in a very prestigious scientific journal (Proceedings of the National Academy of Sciences), and I'll be telling you all about it today.

I have to say up front, I was personally very excited to read this paper, since it relates to the topic of migration, and especially, on how important it is to the monarch population (something I've been saying for years!). Here is a link to the abstract, but unfortunately the paper itself is not free to download - https://www.pnas.org/content/early/2019/03/14/1805114116. Hopefully, I will be able to give you the gist of the study, and its findings here.

The paper is titled, "Multiscale seasonal factors drive the size of winter monarch colonies", and the study was a collaboration between the following researchers - Sarah P. Saunders, Leslie Ries, Naresh Neupane, M. Isabel Ramírez, Eligio García-Serrano, Eduardo Rendón-Salinas, and Elise F. Zipkin. It looks like Sarah Saunders led this project, while she was a postdoc in Elise Zipkin's lab at Michigan State. The Zipkin lab has led some other very exciting (also statistical) work on monarch spring migration, which I've blogged about in the past.

Let me first set the stage for this study. If you've been keeping up with the science around the monarchs lately, you probably know of the long-running debate that has been going on among the monarch scientists (myself included) that is focused on why the overwintering colonies in Mexico are shrinking. There has been a lot of research into this issue, a lot of online arguing, and it is arguably (see what I did there?) one of the biggest scientific mysteries out there today! In a nutshell, there is some research that seems to point to the culprit being a loss of milkweeds during the breeding season, but then other research seems to show this isn't the problem after all. It also doesn't help that there seems to be some researchers who are actively working to suppress findings that don't show what they want (see the study on this here)! The crux of the issue is that there is a very clear negative trend in the 20+ year dataset on the overwintering colony size, but this same negative trend is not present in all of the long-term datasets on monarch abundance from the summer. That has led some to speculate that the real reason the colonies are shrinking is that something is happening to migrant monarchs en route to Mexico.

OK, now enter this new study. Here, the authors specifically wanted to test the idea that the migration is the problem for monarchs. They did this using a very robust statistical approach (which I won't get into much here). They collated and amassed a very large amount of citizen science and other data on monarchs across a 12 year span (2004-2015), including the Mexico colony size data, which the folks at WWF Mexico contributed (thanks to Eduardo Rendon). Essentially, the researchers tried to include all possible variables that they thought could be influencing the numbers of monarchs that reach the Mexican overwintering sites. You may be asking, why only use 12 years of data? It's because some of the data they used only spanned this time frame. Below is my summary of the variables, or datasets they used in their tests.

- They included data from the North American Butterfly Association, which counts adult monarchs in the summer; this was to estimate how large the summer population was each year.

- They then included annual numbers of migratory roosts each year, as reported to Journey North. This was to estimate the overall size of the migratory cohort each year.

- Next, they used yearly estimates of OE prevalence (although there were issues with this that I'll explain later), to see if the disease impacts how many monarchs reach Mexico.

- They included annual temperature during the fall migration (I think this was to see if cold temps lead to monarch mortality?)

- They also included data on the degree of forest cover around the winter colonies - this was smart, since most of Lincoln Brower's research points to how important this is, as a "blanket" for the winter monarchs. Including this variable in the test essentially asked if more forested areas support larger colonies.

Now, here is the really important variable - they were able to derive an estimate of "greenness" of the migration flyway each year, using large-scale, landscape data they obtained from the USGS (https://www.usgs.gov/). This agency of the government houses a massive amount of imagery (taken from satellites in space), and from these images, you can tell how green the landscape at any location is, for any time of the year. The researchers obtained greenness data on two specific regions of the fall flyway, which are indicated on the map below. I assume their estimates were some kind of average index of hundreds of landscape images, because these regions are huge.

This annual greenness index basically was a proxy for nectar resources, because in years when there is more precipitation, the landscape is greener, and more lush. Conversely, in dry years, you have fewer flowering plants, etc. I thought this was a very clever way to get at this issue. While folks like me have been saying for years that the real problem monarchs are facing is a lack of nectar during fall migration, we actually didn't have any way of tracking this! And apparently, this index has been used in many prior studies of other insects, and it tends to correlate well with nectar availability.

I won't get into the stats here, because, frankly, they were over my head! I trust the researchers here knew what they were doing, and from what I know from their previous work, this team seems to be the experts in the room when it comes to this.

Let's cut to what they found.

The results of the statistical analyses showed that there were three variables that most predicted the size of overwintering colonies each year - the counts of breeding adults from the NABA data, the greenness of the flyway, and the density of forest around the colonies. The other variables (number of roosts, fall temperature, OE infection rate) did not explain as much as those first three. It was surprising to me to see that the number of roosts did not predict the overwintering colony sizes, but it looks like the roost numbers were redundant when the counts of summer adults were included in the analyses. In other words, they both were correlated, and so only one ended up being significant.

Here is a simpler way to explain their statistical finding - of the 12 years they examined, the years with the largest winter colonies were those that start with a large summer population (obviously!), but that also have optimal migration resources in the fall (greenness). Once the monarchs arrive at the winter region, they appear to form the largest colonies in places that have the thickest forest. That much seems obvious too.

The authors appeared to conduct some followup tests, where they analyzed how well the summer counts predicted the winter colonies, in the absence of the greenness index. They found that the statistical correlation was not as good if you don't consider the fall migration greenness.

The bottom line - large wintering colonies are a product of the summer population size, fall migration resources and forest density around the colonies. If the summer population is high, but the conditions during the fall migration are poor, then you get a small wintering colony. Conversely, you can't get a large wintering colony if the summer population is low, even if the fall conditions are optimal.

The take-home message of these results are this - the fall migration is a critical time for the monarch. They need to have ample resources during this period to sustain their journey, in the form of nectar.

Now, let me offer some musings on these results for a bit.

It seems clear that the migration greenness thing is critical in order to have large overwintering colonies. But, what causes greenness in the first place? Precipitation! Or the lack thereof. We need to remember that this study considered the entire flyway when they determined greenness - recall that in years when the entire flyway was more green, there were large wintering colonies. The entire flyway only becomes greener during years when there is ample precipitation. So if you think about it, this finding suggests that what monarchs really need during fall migration is simply for mother nature to be good to them, and that it has little to do with us puny humans, or our gardens.

The other interesting thing about this migration greenness index, is that they showed that it itself has not really been declining over the 12 years considered in this study, or at least, not enough to explain the severe declines at the wintering sites. In some years the autumn greenness is high, and in others it's low, but overall, there is not much of a decline in this variable. See the graph below, which I cut and pasted from their study.

Recall that the wintering colony size is predicted by the summer population size, autumn greenness and forest cover at the colonies. So if the NABA data do not show a long-term decline in summer monarch abundance, and this greenness also doesn't show a decline (or not that much, anyway), that means neither of these two variables fully explain (statistically speaking) the decline in winter colony size. Technically, they can predict when we will see large colony years, or small ones, but from a statistical standpoint, they do not fully explain why the colonies are declining over the long-term. So, one could argue that there must be something else that is going on here, and that was not accounted for in this statistical project. Off the top of my head, I can think of a few possibilities - such as an increase in "alternative" winter destinations over time, increasing road mortality (I've blogged about this a lot), greater numbers of monarchs simply dropping out of the migration along the way, or maybe an increase some other disease that bugs can get, and that wasn't considered here. These are just a few thoughts. I could EASILY think of more. Suffice it to say that there is still more to be learned here.

Finally, about the disease issue - let me also say one thing about how they tested for the effects of OE prevalence in this study. Recall that the authors included an annual estimate of OE for the entire population in their statistical models, to try to test if higher OE in any given year leads to reductions in winter colony size. This is indeed a valid question, but (big but here), their estimate of OE was from samples of monarchs at the overwintering colonies! In other words, they were testing if the size of the overwintering colonies could be predicted by the prevalence of OE at the colonies. The monarchs who arrive at the colonies with OE are those that were hardy enough to survive the trip. What they really should have done is considered the population-level prevalence of OE at the beginning of the fall migration (end of summer). Their stats showed that OE was not a significant predictor of winter colony size, but I'm sure this is why.

OK, that's my gripes and musings on this study. Don't get me wrong, overall, I thought this paper was a big step forward in advancing our collective knowledge about the year-round life cycle of monarch populations. I hope this work spurs more like it.

And kudos to the authors for a great study!


Direct link to this blog entry:




The science of monarch butterflies

A blog about monarchs, written by a monarch scientist, for people who love monarchs