Evolution studied broadly. AI in service of the biology.

This is the part of the lab with the longest continuous track record. We build machine-readable karyotype databases, run phylogenetic comparative methods on them, and ship the analytical tools as R packages. The current headline: birds, long treated as the textbook case of chromosomal stasis, sit above the global median once microchromosome dynamics are resolved. Stasis tracks life history, not taxonomic group.

Beetles alone display more sex chromosome systems than any other animal order. We use that diversity to study what drives sex chromosome birth, how fast Y chromosomes degenerate once they stop recombining, and what sexually antagonistic alleles do to genome architecture in the process. Methods work draws on dosage-compensation data, comparative cytogenetics, and population genetic theory.

Comparative work across Coleoptera and Carnivora found that effective population size (via range size) predicts karyotype evolution rates better than any measure of selective pressure. That's uncomfortable for classic adaptive accounts of chromosome evolution and increasingly useful for thinking about how evolution actually proceeds in shrinking populations of conservation concern.

We re-analyzed over 1,600 line-cross datasets spanning plants and animals using an information-theoretic framework (SAGA) we built for this kind of work. Epistasis was detected in the majority of crosses. That settles a decades-old Fisher-vs-Wright debate empirically, and it reshapes how we think about the response to selection.

Two reference-quality genomes in two years: Chrysina gloriosa (G3, 2024) and the endangered long-armed scarab Cheirotonus formosanus (Ecology and Evolution, 2026), the latter yielding the first putative Y-linked scaffold in the genus. A separate project found that domesticated galliform lineages are more compatible as hybrid parents than their wild relatives, a genomic signature of relaxed selection on reproductive isolation.

A general-purpose pipeline for building trait datasets from the primary literature. Starts from a keyword search across PubMed, OpenAlex, bioRxiv, and Crossref. Retrieves full-text PDFs through a 12-source cascade. Extracts structured trait data with mandatory double-entry verification (two independent extractions must agree). Writes to CSV only when validated. Generalizes from version one, which was built only to collect karyotype data, to any trait at any phylogenetic scale.

Tealc is a persistent, tool-using AI agent running locally on Claude plus LangGraph. In daily use now for administrative work: email, calendar, literature search across PubMed, bioRxiv, and OpenAlex, and persistent notes that survive restarts and new machines. Phase 2 — in active development — makes it always-on: a background scheduler assembles a structured briefing every morning at 7:45am covering new citations, overnight preprints in chromosome evolution, flagged emails, and a 90-day grant countdown. Work done before the PI sits down. Phase 3 adds R execution against ape, phytools, and geiger, so Tealc can run phylogenetic comparative analyses directly, read the output, and interpret results. Phase 4 is a manuscript and grant machine: structured critique, journal targeting, and point-by-point response to reviewers. The long arc — Phase 7 — is a lead investigator that carries an evolutionary-biology project from hypothesis to manuscript draft, with optional human review at the checkpoints. Each phase is instrumented to surface where current models break on real scientific reasoning. That is the publishable result.

If AI tools can read our work, they can use it to help other researchers. So we publish everything in formats agents can consume: llms.txt and a longer llms-full.txt as a single-file snapshot of the whole site; JSON exports of every database; structured JSON-LD on every page; and an open data/ directory. It is a small amount of extra effort per page and it costs nothing to readers who do not care.

A course-based undergraduate research experience where each student picks a clade, extracts data with agents, runs comparative analyses, and writes up their results. The spring 2026 cohort is in progress and several of their projects are tracking toward publication.

A concentration we built with Texas A&M Biology to give students a formal credential alongside their degree. Required courses cover AI fundamentals, computational biology, data literacy, and critical evaluation of model outputs. Designed so it ages well as the tools change.

Grad 101, BIOL 682 readings, the R seminar archive, the AI-tools guide, the prompting guide, phylogenetics 101, experimental design notes, and a growing reading group archive. If we use it to train someone, it's on the site.

63,682 karyotypes, 55 eukaryotic clades. Dysploidy rates vary by a factor of 844. Birds, long treated as the textbook example of chromosomal stasis, exceed the global median once microchromosome dynamics are accounted for. Chromosomal stasis tracks life history, not taxonomic group.

I opened the lab in 2017. My background is in theoretical population genetics and cytogenetics, and most of my field time is spent chasing scarab beetles. I also chair the TAMU EEB interdisciplinary doctoral program.

Nobody in this lab is working on a warmed-over PI-assigned project. Students (including first-year undergraduates) are put on questions where their contribution is real and authorable. The alumni list is long now, and it's the thing I'm most proud of.

If the research program here lines up with what you want to do, read the expectations page and then email me. Strong computational grounding helps but is not required. Willingness to read primary literature, break things, and recover is required. TAMU EEB runs on its own application calendar; TAMU Biology on the department calendar.

Enroll in the CURE for a course-based way to join, or email directly with a short pitch for a project you'd like to work on. We accept undergraduates every semester. Lab member expectations (10 hrs/week, lab-meeting participation, taking ownership of projects) apply from day one.

Every dataset the lab produces is machine-readable and licensed for reuse. If you want to collaborate on a project that uses one of these agents (especially TraitTrawler on a new clade or trait) we are actively looking for partners. Email is the fastest path.