The final paper of my UoS postdoc is out now in PNAS! How to detect developmental plasticity in the fossil record:
Featuring
@tomezard.bsky.social
@jamesmulqueeney.bsky.social
@katsamenis.bsky.social
@clivetrue.bsky.social
@thefosterlab.bsky.social
among others
www.pnas.org/doi/10.1073/...
Some great work by @aniekebrombacher.bsky.social! Super cool to be part of this collaborative project!
New paper led by @aniekebrombacher.bsky.social using x-ray CT and laser ablation to detect plastic environmental responses in fossil individuals www.pnas.org/doi/10.1073/pnas.2421549122 featuring @jamesmulqueeney.bsky.social @clivetrue.bsky.social @thefosterlab.bsky.social
No worries. I have been working with these methods for a while so good to try and demonstrate their potential and help others utilise them. Best of luck with things!
Overall, we show the landmark-free methods like Deterministic Atlas Analysis can yield correlated results with traditional manual landmarking & semilandmarking methods. We use the results to propose how to improve these methods in the future!
Finally, we compare the downstream estimates of morphological disparity, evolutionary rates and phylogenetic signal. Here, we find significant correlations between methods for almost all metrics.
We then compare heatmaps of how shape is measured using each method. To do this we compare the shape of different crania with estimated mean shapes of each analysis (landmark top, DAA below). Here is shown for the specimen, Cacajao calvus.
We then compare the correlation between the two methods using Euclidean distances, the Mantel test and the Procrustean randomisation test. We find the methods correlate and that using Poisson meshes significantly improves the correlation between methods.
We then visually compare the results of the manual landmarking analysis (754 landmarks and semilandmarks; top left) with the landmark-free approach using 45, 270 and 1,782 control points (top right to bottom left). We notice differences mainly for Cetacea and Primates.
Once overcoming modality issues, we apply the method to 322 placental mammal crania as seen in Goswami et al. 2022 (science.org/doi/full/10....). Here, we apply and test a range of kernel widths, which changes the number of control points used to compare shape (replacing landmarks).
We focus on comparing an implementation of Large Deformation Diffeomorphic Metric Mapping (LDDMM) called a Deterministic Atlas Analysis (DAA) with manual landmarking techniques. We first notice that data modality affects the results, which we overcome by making Poisson meshes.
Thrilled to share my second PhD paper with @tomezard.bsky.social & @evoswami.bsky.social focused on the application of landmark-free geometric morphometrics! 🎉 Excited to contribute to advancing shape analysis techniques in biology. Check it out here: doi.org/10.1186/s128...
In our March Newsletter we hear from UK-IODP scientists Cedric John, Lewis Grant & @jamesmulqueeney.bsky.social, who've each developed novel applications of machine learning to facilitate rapid, accurate classification of IODP samples
issuu.com/uk-iodp/docs...
@ecord.bsky.social @anzic.bsky.social
a 3d rendering of the skull and cranial muscles of the Eurasian oystercatcher (Haematopus ostralegus)
turns out you can do some pretty cool stuff with the power of diceCT, SPROUT (www.biorxiv.org/content/10.1...), and the SmARTR pipeline (www.sciencedirect.com/science/arti...)
Reminder of the January 8th deadline for this reef palaeoecology PhD project at @sotonoceanearth.bsky.social with @tomezard.bsky.social + @chrisgoatley.bsky.social :)
www.findaphd.com/phds/project...
Highly recommend this very interesting new paper. I'll be reading it again--and then probably again & again. It's insightful, idea rich, & thought provoking even in places where I disagree w/ interpretations! 🧪 #EvoSky
"Three modes of evolution? Remarks on rates of evolution and time scaling"
Figure showing diverse datasets analysed with SPROUT, including a skink skeleton, aardvark skull, human heart, concrete block, and foraminifera.
New paper! For all of you working with 3d scans (e.g. micro-CT, MRI), check out SPROUT, a rapid open-source tool for generating segmented and parcellated data, meaning your scans are separated into the individual elements without any manual labelling or training. www.biorxiv.org/content/10.1...
Our new paper on #AI for evolutionary morphology is out! This massive team effort covers the history of AI for studying morphology, reviews new tools, provides many case studies & a prospectus for using AI to progress diverse topics in evolutionary morphology. academic.oup.com/iob/article/...
Laptops: check. Coffee: check. Giant box of homemade focaccia: check. And with all hands on deck, we are now ready to finish off our paper describing SPROUT, a new semi-automated (emphasis on the automated, for real this time) method for parcellating CT images into 3D segments. Stay tuned!
