Comments on Flockhart et al. (2017)

Comments on Flockhart et al. 2017 by Chip Taylor, Monarch Watch

Regional climate on the breeding grounds predicts variation in the natal origin of monarch butterflies overwintering in Mexico over 38 years

D. T. TYLER FLOCKHART, LINCOLN P. BROWER, M. ISABEL RAMIREZ, KEITH A. HOBSON, LEONARD I. WASSENAAR, SONIA ALTIZER and D. RYAN NORRIS

Full Article: Global Change Biology (2017), doi: 10.1111/gcb.13589

The paper by Flockhart et al. on the climatic factors that contribute to the natal origins of monarchs that reach the overwintering sites in Mexico is interesting to be sure and it’s a valuable contribution to this topic. That said, it’s not a definitive description of the either climatic factors or origins. Most studies of this type have limitations and, in this case, there are a number of issues that need to be resolved. For example, there are limitations to the precision or confidence given to the assignment of any one individual to a specific geographic origin. Each of these assignments can be considered to be a hypothesis based on current knowledge. The addition of a third isotope to the analysis could change the assignments dramatically and the map as well. In addition, there are other data sets that suggest different causal factors and patterns of origin.

As I said, this paper is a foundation, one that needs to be built upon. That will happen if and when the following issues are resolved.

1) A close examination of the data shows that many of the samples were limited to 40 individuals with less than 30 for a given year in three cases. Only 5 of the of the 20 years were represented by 60 or more individuals. Ok, we don’t need to go into the statistical weeds here, and I’m not good at it anyway, but there is a sample size problem here. Relying on samples of 40 or fewer to represent much of a continent for 15 of the 20 yrs is iffy at best. However, as a first step, small samples sizes are understandable since the analysis of each individual is costly (+/-30$ Canadian).

2) As mentioned, the assignment of individuals to specific regions needs to become more precise. Isotope data has a geographic basis determined by carbon isotope profiles (isoclines) across the country as well as hydrogen isoclines that are determined by the amounts and geographic origins of rainfall as influenced by temperature and topography. The latter tends to vary over time. Further, there are sometimes differences within regions that can be attributed to what an organism has fed on. For example, samples from game animals have been analyzed to determine whether or not they have fed on crops. Along these lines, there are unpublished data that suggest there may be differences in the isotope signatures obtained from monarchs within geographic regions that could result in misassignment. Clearly, to further our understanding of geographic origins, a third or perhaps even a fourth isotope will be needed to assign each individual to a specific location.

Modified from Flockhart, et. al 2017.

3) The data and interpretations presented in this paper differ from those obtained by other means. There are at least three cases where monarch counts differ from the isotope data. Here is a short version of my concerns: A) If the number of monarchs counted each year at Cape May is a measure of the production in the northeast, why aren’t years with high or low numbers at Cape May reflected in the isotope data? B) The eggs per stem data from the Monarch Larval Monitoring Project have been shown to be strongly correlated with the overwintering numbers. It can be said to predict the overwintering numbers to a degree suggesting that the upper Midwest (the 38% area on the isotope map) contributes the majority of the monarchs to the overwintering population – a proportion much larger than 38%. C) The isotope map indicates that 12% of the overwintering monarchs originate from the Northwest; western Dakotas, the Prairie provinces and Montana and Wyoming. How can this be the case since this is a region with low numbers of milkweeds and monarchs? Monarchs only appear to reach this region in good numbers when the spring population is robust together with high temperatures in May and June that promote colonization.

4) The climate analysis does not fit with other data that are known to have a strong impact on monarch production during March and April and later during the summer months. Showing why this is the case will have to wait but this is an issue that also needs to be resolved.

5) And last, but not least, the tagging data has something to say about origins of monarchs that reach the overwintering sites in Mexico. In his post to Dplex-L, Anurag Agrawal assumes that the percentages assigned to each region are equal to production and that area and production and reaching Mexico are linked in a precise way. That’s not the case. The probability of reaching Mexico is a function of origin, but it’s not a linear function. To be a linear function, a certain proportion of monarchs would be expected to reach Mexico in direct proportion to the distance of the starting point to the overwintering sites. For reasons that are not clear, increasing numbers of monarchs need to be tagged at a given latitude for each degree of longitude as one moves eastward. In plain language, more monarchs have to be tagged in the East than in the Midwest for one to reach Mexico. Getting back to the point of geographic production, for 15% of the monarchs reaching Mexico to originate in the Northeast would require production that would be much higher – at least 30% of the total monarchs in a specific migration. This same reasoning and argument applies to the Northwest sector of the isotope map. The 12% and 15% values for these two regions are simply too high.

To clarify my point about differences in recovery rates based on geographic origin, I have included one figure from a preliminary analysis of recoveries obtained following the winter kills at the overwintering sites in the winters of 2001-2002 and 2003-2004. This figure shows the percentage of all monarchs tagged in 2001 distributed over each 5 degrees of latitude and longitude. Note the figure of 25.1% that represents the middle of the corn belt, mostly eastern IA, and all the 5x5s that occur east of 80W (18.3% of all tags applied). The latter represent nearly all of the Northeast and some of the North-Central region. An analysis of the recoveries showed that 33% of the monarchs recovered were tagged in the 5×5 in the center of the corn belt while just 4.6% of the tags were recovered from the area east of 80W. Similar figures were obtained for the tagging and recoveries in 2003-2004 with 32% recovered in the same area of the corn belt and 7.8% from the combined areas east of 80W.

To recap the two years, we are comparing recoveries from areas where 25.1% vs 18.3% of the monarchs were tagged in 2001 and 22.6% vs 18.8% in 2003. The results were similar, with 33% and 32% recovered in the corn belt and 4.6% and 7.8% recovered in the east. This is a striking difference and it strongly suggests that mortality during the migration is much higher for monarchs originating east of 80W. However, before leaving this topic, we have to ask whether the Northeast is underrepresented in the recoveries due to some bias in the tagging. That’s possible. Many of the monarchs in the east are tagged along the coast and the recovery rates for monarchs moving along coast lines are particularly low. Given that potential bias, let’s just compare the inland 5×5 in the east (75-80W, 45-49N) with the corn belt 5×5 for both years. The numbers tagged per recovery were 24 for the corn belt vs 114 for the east in 2001. In other words, taggers had to apply 4.75 times more tags in the eastern 5×5 to have one recovered in Mexico. The difference was not as great in 2003 but those in the east still had to tag 2.2 times more monarchs to have a recovery than their counterparts in the corn belt.

Moving forward, it would be ideal if the isotope data could inform us in greater detail of the natal origins of monarchs that reach Mexico. Such data could be used to more precisely target areas/regions in need of protection and perhaps restoration. This analysis is not quite there yet. At this point, the loss of milkweeds, habitat conversion due to the renewable fuel standard and other factors (see Monarch Watch Blog entries for references), as well as the tagging data, indicate that conservation efforts should focus on what I have referred to as the milkweed/monarch corridor. Based on tagging data, the corridor ranges from 80W to 98W slanting SW through Kentucky, Tennessee and Arkansas and then Texas to the border with Mexico as indicated in the figure below. This figure is similar to one being used by the USFWS to target regions considered to be of greatest concern with respect to monarch conservation.