top of page
  • Andy Davis

New study of roadkill monarchs shows some highways in Texas are "slaughterhouses"

Hello everyone,

Last week, a new study was published in a very important scientific journal, Biological Conservation, and I guess you could say this new study falls under the category of "things-humans-are-doing-to-damage-the-migration." Today, I'll give you a summary of the paper, based on my read and interpretation of their data. I'm afraid that none of this is good news...

The paper was titled, "Spatial risk assessment of eastern monarch butterfly road mortality during autumn migration within the southern corridor", and a link to it is here. However, this paper is not freely-available to download. Again I say, hey, monarchs scientists, don't publish your work in closed journals! Incidentally, the paper was authored by the following individuals: Tuula Kantola, James L. Tracy, Kristen A. Baum, Michael A. Quinn and Robert N. Coulson.

The paper is a summary of 2-year project where the goal was to determine how many monarchs are killed on roadways in the lower part of the migration flyway, that is, Texas, Oklahoma and northern Mexico. This was indeed a lofty and ambitious goal, and very, very important for monarch conservation. The authors appeared to be well-aware of the rising concern over the loss of migratory monarchs each fall, or at least, they indicated this in the introduction of the paper. They stated that this was a big part of the rationale for doing this project, and they (correctly) stated that there are not that many other projects like this that specifically focused on fall-migrating monarchs. I've blogged about most of the relevant past research, and you can find those blogs here. The upshot of all of the previous work is that a hell of a lot of monarchs are killed on roads each fall while trying to get to Mexico.

The authors (or those who collected the data) conducted a series of surveys along specific roadsides in Texas, and a little bit of Oklahoma. To the right is a map from the paper that shows the locations of the surveys, and the boundary of the "study region". They called this study region the "central funnel", which refers to the fact that as the monarchs fly southward throughout the eastern flyway, they get funneled into a smaller and smaller area when they reach this region. This is an important point to remember, because it means, unfortunately, that they become concentrated along roadways in this region.

The other thing to note about this map is that even though their survey points were confined to Texas and Oklahoma, they extrapolated their results from those surveys to this much larger area. This was possible because they used a very sophisticated software program (called Maxent) that allows for this. I'll come back to that.

Their surveys at each of these locations were conducted 4 times during the fall of 2016, and again in 2017. They were done between 10 Oct and 4 Nov in 2016, and between 3-27 Oct in 2017. They stated that these dates were chosen so that the surveys could be done AFTER the peak of the migration had passed, so that would (in theory) allow the researchers to collect monarch specimens that had "accumulated" after the peak migration wave had passed through. I totally get why they did this. This project was likely done by a Masters student, and time was probably limited, so they likely had to maximize the survey effort. But (big but here), this strategy may, or may not, have worked. I'll come back to this.

So for the surveys, the researchers essentially walked along the side of the road, for exactly 100 meters at each location, and collected all dead monarchs they found. This seems pretty straight-forward. They did this on one side of the road, and then extrapolated to the other side. They also collected other information like the number of cars on the road, the landscape characteristics of the area, the local climate, and the number of homes, things like that. This was so that they could try to pin down what types of areas are most prone to roadkills. In other words, where are the roadkill "hotspots"? This was of importance for obvious reasons.

Here is what they found:

Over both years, they surveyed a total of 16.1 km of roads (a total of 161,100 meter-long transects). In 2016 they surveyed 8.8 km and in 2017, they surveyed 7.3 km of roads. They collected a total of 581 dead monarchs during both years, which averages out to 3.4 dead monarchs per 100m of road. This is equivalent to 340 monarchs per km. But, this is an average of both years. They actually found most of those monarchs (546) during the first year, and only a small number (35) in the second year. And, in the first year, most of those monarchs were found at a roadkill "hotspot" - 499 (91% of the total) were collected along Interstate 10, between Sonora and Sheffield. They calculated that about 66 monarchs were killed per 100 meters along this highway during 2016. That translates to 660 per km. For prospective, imagine taking a 10 minute walk, and finding a dead monarch every 5 paces... If you picked them all up, after your walk you'd have a grocery bag full of dead monarchs.

On the one hand, their data showed that this highway, Interstate 10, was a virtual monarch slaughterhouse during one year of their study. But weirdly, it wasn't during the second year. But, the authors did point out that this same location has been reported to be a hotspot by citizen scientists in the past, although those were anecdotal reports (i.e. not surveys).

Where was this hotspot anyway? I'll paste a map below that I pulled from the internet, which is what I figured based on their description.

Next, is a zoomed-in screenshot of one section of this interstate, that shows the landscape of this region...

And finally, here is a photo that I also grabbed from the web of a random section of I-10, which shows what it looks like from a driver's (or monarch's) perspective...

Do you see what I see in these shots? That's right. Nothing. This is a very lonely, uninteresting road. There is absolutely nothing along the sides of this road, or at least, nothing that would indicate why the monarchs were getting creamed along this stretch of road. There isn't even any flowers!

Recall that the authors did a statistical assessment of the characteristics of roads and surrounding landscape that point to the highest roadkills, using that computer program. This actually showed that the locations with the largest roadkills are places like this above - roads with few human developments and arid landscape. And also, they found that the highest roadkills were in locations with low traffic density! In other words, the monarchs are getting creamed on these lonely stretches of road pictured above. The authors did point out that these conclusions may have been a statistical artifact of the fact that the monarch flyway may simply be going through these regions, and not necessarily through the human developments. So in the end, it looks like the authors were not really satisfied that their surveys could accurately predict where the roadkill hotspots are. Or at least, it didn't seem like they could pinpoint anything in particular that would be like a warning sign, or something that we could use to help mitigate this issue.

I mentioned that most of the monarchs were collected in the first year and they hardly found any in the second season. They interpreted this to mean that monarch roadkills are inherently variable - sometimes the conditions are just right, and sometimes they aren't. I believe this. For example, we know that during strong headwinds, monarchs tend to fly low to the ground (where they could be killed) and they fly way high during light winds. Also, if the bulk of the migration happened to be flying through a region, say, during rush hour, then you'd probably have some carnage.

While I thought that the data from this project were sound, and I do applaud the authors for undertaking this study, I did have a couple of gripes. For one, the fact that they found so few roadkills during the 2017 season gives me some pause. Either this means that there really were very few monarchs killed on roads that year (even at the same hotspots along I-10), or that the researchers missed the migration somehow. The fact that they surveyed for roadkill AFTER the main peak of the migration had passed is a little iffy. This assumes that any monarchs that did get killed would still be present on the side of the road when the researchers did come around to count. This may not be the case - I recall reading a study on this, where researchers determined that most butterfly carcasses disappear after a couple of days, from scavenging, ants, etc. So while their surveying was probably done as best as they could do it, it's possible that they missed a lot of carcasses. In other words, their estimates of mortality are probably way low.

Speaking of way low, let me speak a little on their extrapolations. They took those roadkill numbers (per 100m), and then tallied up the length of roads in this region. Based on their numbers, they stated the following: "Extrapolating roadkill rates across the Central Funnel, we estimated averages of 3.0 ± 0.7 and 1.1 ± 0.3 million (mean ± SD) road-killed monarchs for 2016 and 2017, respectively." This is apparently an extrapolation to the "central funnel" region, which by my estimate, is an area that covers approximately 20-25% of the entire fall flyway of the eastern population, depending on which regions you classify in the flyway. I'm not sure why the authors did not extrapolate their numbers further. Perhaps they figured that their data were applicable only to this region of the country (i.e. areas with this landscape). However, I have no problems doing so here, as I explain below.

Over the years, because of my interest in all things migration-related, I've been keeping track of the new research on this subject as it comes out, and I've recently been putting together a summary of it, for just this very reason. I'm next going to show you a figure that I'm working on for an upcoming project, where I've combined all available estimates of road mortality that have been published, and even one that was an unpublished Masters thesis. I found that one online at the university repository. Keep in mind, this figure in itself is not yet "published science", but it is based on all available science to date. So it is not wrong, or open to interpretation. Take a few minutes to absorb this...

As you can see, I've tried to report the known estimates of road mortality in numbers that are on the same scale (i.e. dead monarchs per km of road). For the Kantola paper, I used their overall estimate of 3.4 monarchs per 100 meters and scaled up. And also, I extrapolated all of the numbers given in these studies into something on the same timescale (per month). This is so that one can see how each place studied compares. Doing so, it is obvious that the scope of the problem is much larger as one moves down the flyway, likely because of the concentration issue. Note also, that I included a handy number on the left, which indicates the total amount of roads in each of the pink states (I counted). This is not even counting the Mexican roads.

So have you done the math yet? If you use the first estimate from McKenna et al in 2001 - 3 monarchs per km, and then extrapolate that to the entire pink region, you get 10.2 million. But that's probably not accurate, since this low estimate from the Midwest doesn't account for the concentration factor in the central funnel. If you use the 7.7 figure from the Oklahoma masters thesis, then you get 26.2 million. Then if you use the estimate of 34 monarchs per km from Kantola et al, then you get 115.6 million! Then, there is the anecdotal estimate from Mexico... maybe it would be best if we didn't even go there...

There are probably a hundred ways to slice this, but in all cases, you get some ridiculously large numbers of roadkills. I guess it's safe to say that the number of monarchs killed on roads during the fall migration each year is EASILY in the tens of millions. Compare this to the size of the overwintering colonies for the last few years - 50-70 million - and then you can see why this is a huge issue.

Lastly, based on the science we have to date, including this new paper, it is clear that road mortality is a major conservation issue for monarchs. But, the real problem is that we don't know what to do about it. In this new paper, the authors pointed out a couple of strategies that have been employed elsewhere and for other species. One strategy is to place nets along roads that appear to have the highest risk. Another is to reduce the speed limit on such roads. Either of these could be things to try, although perhaps the real limiting step would be getting local governments on board. And maybe the local people who drive on these hotspot roads need to be on board too. Clearly, there's a lot a work ahead.

OK, so I think this about covers the new study, plus my own take on it. I wish I could offer something witty or uplifting to end this...Thanks for ready, anyhow.



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

bottom of page