Thread by @TysonWepprich

This new @PNASNews paper estimates 100+ year monarch butterfly and milkweed abundance, which is cool but…1/
pnas.org/content/116/8/…

I'm sending this letter to @PNASNews in response because of a flaw that changes the story for monarchs. Will explain briefly in thread 2/
biorxiv.org/content/10.110…

The authors have been forthright with me after I expressed my concerns. Their sharing of data and code made this reanalysis possible, for which they deserve loads of credit. 3/

Why do I care? It's a controversial topic bc more Roundup -> less milkweed over recent decades is well established. This means less food for monarch caterpillars. But what about before herbicide-resistant crops? 4/

The paper argues that both milkweed and monarchs started to decline around 1950 and that bigger farm size is primary cause for their declines (instead of Roundup). 5/

Big, if true! I don’t doubt that farm management or consolidation before Roundup influenced milkweed and monarch populations. But … 6/

What is their monarch abundance trend based on when no monitoring existed until recently? ~1000 museum specimens over 115 years across the eastern USA. 7/

I research insect abundance and it's really hard! Insect data is noisy and plagued by sampling issues and I have a hard time estimating 20-yr abundance trends from 24000 systematic surveys 8/

Often museum data is the only data, and it has many neat uses for ecology/evolution research. Check out this special issue edited by @EmilyMeineke! 9/ royalsocietypublishing.org/toc/rstb/374/1…

Is museum data reliable for abundance? In this case no. I don't know when monarch decline started, but the reported trend is wrong (and arguing that Roundup can't be blamed because of this trend is misleading). 10/

Why is the claim that monarch decline started mid-century wrong (based on museum records)? 1. Biased sampling unaccounted for in the paper. 2. When corrected, the trend doesn't match recent systematic monitoring. 11/

In paper, relative abundance in each year = monarch records / all Lepidoptera records in same region. Standardizing by collection effort makes sense, but… 12/

Moth collection and butterfly collection are very different: nighttime light traps vs daytime netting. Why not standardize by butterfly records? 13/

Here, their trend reproduced on left vs the trend when standardizing by butterflies on right: the mid-century decline disappears and monarch records are increasing! 14/

Note: monarch abundance is not actually increasing. In this case, museum records are not representative of the real world. 15/

Why is dividing by butterflies vs all Lepidoptera showing such a different trend? It's such a small choice in the analysis (which was an honest mistake) 16/

Moth sampling increases as a proportion of museum collections at the same time as their monarch trend starts declining 17/

I would guess that any butterfly species would have the same hump-shaped trend when standardized by butterfly + moth records, due to this change in sampling. 18/

Why the sudden change in museum collections? I'm guessing due to light traps getting popular in the 1950s. That history is here: 19/
simonleather.wordpress.com/2015/08/06/ent…

Milkweed has 30X more records so maybe that trend is robust. And agricultural management pre-gmo is a key part of the monarch/milkweed story that is understudied. 20/

My conclusion about exactly when Monarch butterflies started declining: unknowable based on these museum records. But they aren't doing so hot over the time when we have good data. 21/ journals.plos.org/plosone/articl…

Overall lessons I've learned: estimating insect abundance is hard. Check data for sampling biases, especially when not from systematic monitoring. Share code to ensure accuracy. 22/

Link to code 23/23
gist.github.com/tysonwepprich/…