We set out to explore the empirical behavior of the ID_risk statistic, and what we found I think is pretty cool! Specifically, it’s interesting to see that IR wolves and FL panthers have by far the highest ID_risk, which makes sense given the high severity of inbreeding depression in these pops 8/n

Paper after paper in species including wolves, dogs, humans, killer whales, etc all seem to tell a similar story. With this in mind, it was no surprise when we found a strong correlation between long ROH and survival/reproduction in ‘akikiki here: www.cell.com/current-biol...
4/n

Maybe my favorite is this paper on Soay sheep where they quantify these effects using SNP and mortality data for ~6000 sheep. For instance, they show that a 1% increase in long ROH reduces first year survival by 12.4%, whereas short ROH have no discernible impact. 3/n doi.org/10.1002/evl3...

I am thrilled to share this paper outlining some ideas I’ve been thinking about for a little while on a simple but powerful approach for predicting risk of inbreeding depression from long runs of homozygosity and non-ROH heterozygosity. 1/n @klohmueller.bsky.social doi.org/10.1016/j.tr...

For ‘akeke‘e – which has recently declined below <100 individuals in the wild due to malaria – we examined the potential effects of ongoing efforts to control mosquito populations on population recovery. We show using a SLiM model that recovery can still occur, but the window is very narrow.

For ‘akikiki – which has just ~40 individuals remaining all under human care – we find that inbreeding has a significant effect on survival and reproduction, which could impact efforts to recover the species.

For the critically endangered ‘akikiki and ‘akeke‘e from Kauai, we generated population-level data to examine the impacts of 99% declines occurring over the last two decades. Indeed, we can directly observe the magnitude of these declines used a LD-based demographic inference approach

This even holds true for the extinct po‘ouli, which we were able to assemble a genome for from banked cells in the Frozen Zoo from the last-known individual of the species. Yet despite high diversity, this individual appears to be relatively inbred, evidenced by numerous long runs of homozygosity

Just 17 out of ~60 honeycreeper species remain, and most are now threatened by the spread of avian malaria in HI. One surprising result from our genomic analysis is that, despite this high degree of conservation threat, honeycreepers in general appear to have quite high levels of genomic diversity.

Very excited to share the first paper from my postdoc at the San Diego Zoo Wildlife Alliance on the population genomics of recovery and extinction Hawaiian honeycreepers:
Paper: doi.org/10.1016/j.cu...
Press release: sandiegozoowildlifealliance.org/PR/hawaiian-...
‪@currentbiology.bsky.social‬

View of the Palisades fire from my apartment in Santa Monica

We also look at the predicted inbreeding load (haploid lethal equivalents) from these models and find that this quantity also depends on the model. Overall, all of the predicted inbreeding loads are well within the range of estimates from mortality data of ~0.7-2.5 9/n

Here, we find that predictions depend on the model: weakly recessive models show a slight increase in load in non-African pops, whereas more strongly recessive models show a slight decrease, perhaps due to purging. 8/n

Next, we sought to use these results to inform simulations revisiting the question of how the out-of-Africa bottleneck may have shaped genetic load in humans. To do this, we selected 3 models encompassing the range of our results 7/n

Finally, when further constraining these models to those with a monotonic decay in h with increasing s, we find just 48 models with good fit. Here, we can see that models with a strong h-s relationship fit the data just as well (if not slightly better) as an additive model 6/n

When constraining to models with good fit, just 303 models remain, also having a range of possible h and s values for each bin. However, it appears that moderately del muts have a lower bound of h of 0.15, and strongly del muts have a lower bound of 0.05 5/n

However, these models with a single h probably aren’t very realistic, so we next explored the fit of 4096 models with varying h for each bin of the discrete DFE. Many of these models have very poor fit (right panel) and the resulting DFEs vary widely (middle panel) 4/n

We began by fitting models with a single h for all mutations to the nonsyn SFS in humans. Lots of models can fit the data, with the exception of highly recessive models with h<0.15. Also, as h becomes more recessive, s becomes more strongly deleterious, as expected. 3/n

Out today is our genomic and simulation analysis of pacific fin whale populations! Amazing work led by Sergio Nigenda-Morales and Meixi Lin www.nature.com/articles/s41...