• Andy Davis

A breakdown of the new research paper that is a game-changer in monarch conservation

Ladies and Gentlemen, this week I will be reviewing the recent bombshell monarch study that came from the group of monarch researchers at Cornell University, which, as expected, has set off quite a firestorm because of the controversial results they found regarding monarchs and milkweed.

Before I begin, let me be clear – I had nothing to do with this study. I am merely reviewing it here so that everyone can understand it and is aware of its implications.

One more caveat here - given that it was a lengthy paper, with loads of data and results, my summary of it is also lengthy - so go get a cup of coffee and strap in.

To start with, the paper itself is actually not even officially published yet, and already it is generating quite a buzz. The title is “Linking the continental migratory cycle of the monarch butterfly to understand its population decline.” I’ll put a link to it here, which will take you to the journal website, but the paper itself is not freely accessible to the general public. At least the abstract is free. Right now the paper is listed as ‘accepted’ at the scientific journal where it will be eventually published (a journal called Oikos – very well respected). So that means it has been peer-reviewed by multiple experts in this field, and found to be acceptable for publication. Usually, a paper goes through several rounds of revisions before becoming acceptable. In fact, I actually saw a much earlier version of this paper last year, before it went to this journal. The version I saw had some issues, but I can see that this new version is much improved, and I think the authors did well to try to incorporate their work into the existing literature and discussion around monarch conservation. So the paper will most likely be officially published later this year, but until then, the accepted version (a Word document, really) is posted on the journal website.

Speaking of the authors, there were 4 - Hidetoshi Inamine, a Cornell graduate student in the field of the ecology and evolutionary biology, Stephen P. Ellner, and Anurag Agruwal, both professors at Cornell, and James P. Springer, who is the Vice-president of the North American Butterfly Association. I believe that Anurag was the leader of this group, and has been the main point of contact since the paper came out. Stephen Ellner was the statistical guru on this project, I think. Anurag is no stranger to the monarch world – he is a long-time researcher of milkweeds and the critters that use them. From communications with Anurag, this paper represents two years of work on their part (in addition to the thousands of hours spent by citizen scientists).

So the goal of this project was to try to figure out where monarchs are declining (the eastern N. American monarch population), in terms of their multi-stage life cycle. That is, given the clear declines at the Mexican overwintering sites, the authors wanted to know if there were obvious problems during the spring migration northward, or during the summer breeding cycle, or during the fall migration southward, which would explain the declines in Mexico. In other words, where are the links in the chain broken? If this sounds familiar, it’s because it is – a similar paper was published by Leslie Ries and Karen Oberhauser in the 2015 monarch book - "Monarchs in a Changing World." It’s the very last chapter in the book, if you want to go back and re-read it. In that paper, Leslie and Karen had pretty much the same goal, and they even used pretty much the same data sets. AND, the overall outcome of that study was pretty much the same as this new one. So what’s the difference? Well to me it looks like this new one includes a lot more data from those data sets, and their statistical analyses were a bit more rigorous. Plus this new paper has some interesting analyses of herbicide usage over time, which I’ll get into later. On the other hand, having two completely separate researcher groups tackle the same issue independently of one another, and arriving at the same conclusion, provides strength to the conclusions.

The source data for this paper consisted of citizen science data from two fall migration monitoring sites – Peninsula Point, MI, and Cape May, NJ, plus data from the North American Butterfly Association (NABA), which tracks the abundance of adult monarchs. I know for a fact that they obtained permission to use all data sets. Each of these data sets spans about 2 decades, and it looks like their data did not include the 2015 season.

Let me point out a couple of things about the NABA program before I go further. Recall that in a previous blog post I spoke about the NABA data protocols, which is where volunteers go out and count all butterflies they see within a circle with a 15mile radius. I also said that they do this on one weekend during the summer. However, this isn’t entirely true. At some of the sites the surveys are done three times per year, once in the spring, summer and fall, and this is especially true in the more southern regions of the United States, where the butterfly season is very long. This is a very important point which I’ll come back to. Also, this protocol (3 times per year) has only been used in the last 10 years or so, not for the whole 20 year span. In the early years, it was ONLY once in the summer. This also, will become clearer as I go.

So here is a map of the NABA survey locations used in this study, which I obtained with permission from the authors, along with the graphs below. You can see it’s a heck of a lot of sites. Actually, it’s pretty much every site within the NABA database, except for the extreme southeast. I believe the researchers had over 6000 lines of data to work with. It seems the authors did not include any sites from Florida, Georgia or South Carolina, to be sure they could exclude all non-Mexico-bound monarchs from this analysis. I think this is fair, since very few monarchs from that region ever go to Mexico. They also did not look at any sites west of the Rockies, since they wanted to focus on the eastern monarch population.

You’ll see that the sites are grouped into three blocks, which they designated as different regions. They did this to help define the different life stages or regions. Their ‘Midwest’ group is pretty liberal I think, but I don’t think moving these boundary lines here or there would make much difference in their analysis (so no griping about where these lines were drawn!). The different levels of shading for each dot represents overlapping locations. I believe the Ries et al paper in the monarch book used similar region definitions, but they broke the 'Midwest' into two smaller parts, not just one big one.

So recall I said the southern states have count data during the spring summer and fall. The researchers actually capitalized on this in their analyses – they used the overall average counts of all adult monarchs seen during the spring in the south as their index of ‘spring migration size’, which is a very clever way to look at this. I myself have tried to look at the size of the spring migration each year, using Journey North data, but I never thought about looking at the NABA data to do it. So just to be clear - these dots on the map above are not the same as the Journey North sightings. The map above shows where NABA surveys were done over many years.

The other neat thing they did which Ries et al did not do, was to examine the same region (South) in the fall - so they looked at the counts of adult monarchs made at the southern NABA sites during the fall surveys. Then they took the average of all of these counts for each year. Essentially, this gave them an annual index of 'southward migration size', for comparison with the other stages. Unfortunately, these fall counts have only been done for the last decade or so, so it's hard to use them to look for overall declines. In any case, this approach was very clever in my mind too - for years we have been lacking information on the abundance of monarchs during the migration in Texas, which is the choke-point for the entire fall migration. It turns out these data were there all the time!

So to summarize the data, they had information on the overwintering colony size in Mexico for the past 20 years, which they got from WWF, they also had an annual index of spring migration size, then they had counts of adults during the summer in the midwest, there were two datasets of fall migration monitoring data (Cape May and Peninsula Point), then the annual fall counts of adults in the south. Recall that the goal was to see if the numbers of adult monarchs at any one stage are linked with the numbers of adults at the next subsequent stage - like winter to spring, spring to summer, etc.

I won't get into the details of the data analyses - but suffice it to say, they did find some links, and they also found the places where the links broke down. The graph below shows where these links were and were not, by showing the indices of abundance over time at each of the stages. On each graph, if there is a solid trend line, that means the two datasets were statistically related.

I'll go through these a little out of sequence, in a way that makes sense. First, the size of the overwintering colonies each year closely predicted the size of the spring migration (graph A). In fact, this was a very strong link - look at how closely those numbers line up. I think this one little graph (A) is the most powerful graph in this entire study, because it provides clear validation to the NABA surveys. Put it this way - if everyone can agree that the Mexico data are the 'gold-standard' for tracking the population size, then how can a dataset that is perfectly correlated to that be wrong? The other thing this graph (A) says is that we are not losing monarchs during the spring migration.

Graph B shows the spring migration size then predicts (but less strongly) the number of breeding monarchs in the midwest. Then graph D also shows the spring migration predicts the number of breeding monarchs in the northeast. Graphs C and E show that the number of breeding monarchs in the midwest and northeast very loosely predict the number of monarchs counted during fall migration at Peninsula Point, MI and Cape May NJ. Collectively, these graphs tell us that the links between spring and summer are pretty strong, and so are the links between summer and early fall.

Here's where it gets interesting - the links disappear when you go from early in the fall migration (Cape May and Peninsula Point) to late in the fall migration (NABA South). These are graphs F and G above. Then, the number of monarchs in late fall (NABA south) also does not line up with the number in Mexico (H). For good measure, they also included a graph showing there is no significant link between the number of adults in the summer in the midwest, and the size of the overwintering monarchs in Mexico (graph I). Collectively, these graphs tell us that the breaks in the chain are during the fall migration on the way to Mexico.

They included one more graph in their paper, using all of the data sets, which is pasted below. This shows the temporal trends over time in each of the data sets. This graph also tells the same story as the one above. There are significant declines in monarchs at the overwintering sites. There is also a decline in the size of the spring migration - because these two stages are closely linked. In other words, when the overwintering size is large, so is the spring migration, and when it's small, the spring migration is small. So the spring migration is declining because there are fewer monarchs in Mexico nowadays.

All other graphs have no trend line, meaning there was no significant relationship over time, or in other words, no decline was detected. This goes for the summer in the midwest and northeast, the two fall monitoring sites, and even the fall south, although there are fewer years for that one.

There was one other interesting analysis done in the paper, which was to get at the issue of milkweeds and herbicide-tolerant crops. Because the conventional thinking is that monarchs are declining in Mexico due to a loss of milkweeds in agricultural fields, which in turn is from the adoption of herbicide-tolerant crops, the authors tried to specifically test this theory using some data from the United States Department of Agriculture. The authors looked at the annual adoption rates of herbicide-tolerant crops (corn and soybean) from 1996-2015, and used this as a proxy for annual milkweed abundance. I'm going to copy and paste some text here from their paper, so that I don't get the message wrong. In their words, " We expected that increase adoption of herbicide resistant crops leads to increased herbicide use and consequently, decrease in milkweed abundance". They then tested "whether year-to-year changes in adoption correlate with year-to-year changes in monarch population estimates". Also, they used "differenced data, rather than directly correlating monarch populations with herbicide-resistant crop adoption, because any two variables showing a trend over the same time will be correlated. If the relationship is causal, however, annual differences in adoption should correlate with annual differences in monarch populations. No such correlation is observed. As a case in point, the peaks of the summer Midwest and Northeast population indices both occurred in 2006, following a 10-year period in which use of herbicide-tolerant crops rose steadily from 0 to 62.5% of corn and soybean acreage."

So I think I've covered most of the paper by now, or at least the salient parts. Let me try to wrap things up then. From what I can tell, the main conclusions of this project were as follows:

- There is no evidence of a decline in adult monarchs in the breeding season in the midwest or northeast, and the annual monarch numbers do not match up with the annual amount of herbicide applied on agricultural landscapes (remember this is not the same test as a simple correlation).

- there is no evidence of a decline in migrating monarchs at two fall monitoring sites, Cape May, NJ and Peninsula Point, MI.

- The number of adult monarchs counted in the south during spring is a direct reflection of the size of the overwintering colonies (they are mirror images) - so because the overwintering numbers are declining, that causes the the spring numbers to also decline. However, since there was no corresponding decline in the summer, that means the monarchs are rebounding each year from the low starting numbers each spring. On a side note - I'd like to also point out that this same conclusion was reached in a paper I worked on with Elizabeth Howard, using the Journey North data in 2015 - we found that the early spring migration wave appeared to be shrinking over the last two decades, but that the monarchs appear to be rebounding enough each year to fill their traditional breeding range.

- While the Cornell authors did not highlight this, I believe the fact that the spring numbers so closely matched the overwintering numbers is also a sign that the surveys done by the NABA volunteers are accurately measuring monarch abundance.

- This study found that the biggest broken link in the life stages appears to be during the fall migration, on the way to Mexico. In other words, this seems to be the most important life stage for determining how many monarchs we see in Mexico each year. Actually, this same conclusion was reached by Ries and Oberhauser in their chapter in the 2015 monarch book - they also found that the migration was the weak link in the chain. The authors of the Cornell study speculated (only speculation) that there may be a lack of nectar resources during the fall, or adverse climatic conditions, that is hindering the migration.

- Finally (and this is the biggest bombshell), since the number of breeding monarchs show no decline over the last 2 decades in the Midwest and Northeast, the authors concluded that milkweed is not limiting to monarchs. Put another way, monarchs aren't declining in Mexico because of a loss of milkweeds during the breeding season.

OK, I think that's about covered it. Before I go, I'll point out that there is a press release from Cornell that is a handy summary of the paper, and it also has a nifty video-interview with Anurag - http://news.cornell.edu/stories/2016/04/beyond-milkweed-monarchs-face-habitat-nectar-threats


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