Can we make computers see genomes?

Five years ago, I started learning about this AI thing and how to use it. I got especially excited about the idea that we can basically reformat any kind of data to something that we already have a pretty good neural network trained on. At that time, I was interested in methods that could be used to do DNA-based identification in the field. For example, by extracting DNA and sequencing just a few sequences from a whole genome, so we could quickly identify what something is (like a bug, or a leaf). And that’s how a small side project ended up in something that can identify all the species with publicly available whole-genome DNA data.

VarKoder workflow, adapted from de Medeiros et al (2025)

I encoded a really small amounts of genome sequences from a sample as an image (on the order of 1% of random sequences from the whole genome). I then applied some tricks to train image classification models to infer taxonomy from those images. It works much better than widely used DNA-based species identification methods, from domains of life to subspecies (including things that are mixtures of organisms such as lichens). We can also tell where a sample of environmental DNA came from using this method. Finally, it also works well with degraded DNA, so our collections could be great for creating training sets for the vast amounts of diversity that currently cannot be identified based on DNA. The AI model has been trained on all of the taxa that had publicly available data online as of 2 years ago when I first submitted the paper, so it is ready to use in practice. And it can run on laptops using data from portable DNA sequencers, so no need for a fancy setup.

In addition to practical use, this paper shows us that there is something that varies between genomes that can be captured by this random DNA sequencing: we are not comparing the same genes across species, and we are not comparing whole genomes. The differences may be very subtle, and we don’t know for sure what they are. It is probably related to the repeated parts of the genomes. In any case, most of our genomics research relies on comparing the same genes across species, so there is a hint that we should look outside this box. 

The method and the data are described in the following publications:

Asprino, R.C., Cai, L., Yan, Y., …, de Medeiros B.A.S. A curated benchmark dataset for molecular identification based on genome skimming. Sci Data 12, 906 (2025). https://doi.org/10.1038/s41597-025-05230-2
de Medeiros, B.A.S., Cai, L., Flynn, P.J. et al. A composite universal DNA signature for the tree of life. Nat Ecol Evol (2025). https://doi.org/10.1038/s41559-025-02752-1

And the software is available here:
https://github.com/brunoasm/varKoder

High-throughput mothing

"Lucy Guarnieri (PhD candidate at the Ohio State University) came to the Field Museum for a 2-week visit to image her collection of over 5,000 moths. Lucy studies how different types of insects use urban habitats and how urban green space can be managed to provide the greatest benefits for humans and wildlife. Her moth collection is the result of several years of collecting throughout the Columbus Metro Parks system. With the help of the Elizabeth Postema, Tony Eisel and Francis Dela Pena, she used the GIGAMacro imaging system, and a series of fine-tuned computer vision models, to get images and body measurements for all 5k+ moths. Using these data, she hopes to compare functional and morphological diversity across different urban habitats.

View fullsize

View fullsize

Member's Nights 2025

Last week we hosted thousands of Field Museum members for a tour behind the scenes. Our lab showed the diversity of colors in insects, our digitization efforts, and our research on beetle color evolution and weevils!

Abigail Magland is now PhD Candidate!

Abigail has successfully defended her thesis proposal on the evolution of palm flower beetles communities last week!

Lots of exciting work ahead combining systematics and ecology to understand how these intricate interactions evolve!

Congratulations, Abigail!

Insect flower visitors lab at ESA 2023

This is the first year that multiple lab members attended the ESA meeting in National Harbor. To be fair, it is the first year that the lab counts with multiple members!

Insect flower visitors lab members at ESA

Bruno de Medeiros presented his ongoing research on applying large language models (LLMs) in taxonomy.

Diego Souza presented on his progess in using adaptive sampling in nanopore sequencers to generate phylogenetic datasets.

Diego calmly reading a book in perfect lighting after getting done with talks in the first day (photo by Elizabeth Postema)

ElizabethPostema and Abigail Magland joined the lab more recently, presenting on their previous research: an ingenious poster on UV reflectance in swallowtail butterflies and a a presentation on the evolution of eyes in beetle flower visitors.

Elizabeth presenting her poster with visible and UV photos.

Abigail presenting on evolution of flower beetles

First Chicago summer wrap-up

Annual cicada noise is beyond its peak, and almost all grasshoppers around are adults showing their beautifully colored wings. Summer is coming to an end!

This was also the first summer of the de Medeiros Insect Flower Visitors lab at the Field Museum, marked by a lot of activity.

We participated in the first Bioblitz at the Rice Garden, organized together with the Learning Center and the Action Center of the Field Museum.

Diego Souza and Allisa Doucet preparing insects collected during the Rice Garden Bioblitz

We visited local natural areas, such as the James Woodworth Prairie and the Warren Woods Field Station.

Elizabeth Postema, Bruno de Medeiros and the James Woodworth Prairie director Alan Molumby collecting insects

We searched for monarchs to update the Field Museum learning resources.

Bruno showing a monarch caterpillar in the Rice Garden

And we interacted with the museum public in person and through the media.

Bruno showing “Surprising Pollinators” at the Grainger Science Hub

As we recede into our offices (and a bit of international fieldwork) for the next few months, we look forward to planning more exciting activities for next summer!

In the news: weevil brood pollinators

Julien Haran, Gael Kergoat and I just published this review on weevil brood pollinators:

Haran J, Kergoat GJ, de Medeiros BAS. 2023 Most diverse, most neglected: weevils (Coleoptera: Curculionoidea) are ubiquitous specialized brood-site pollinators of tropical flora. Peer Community Journal, 3 https://doi.org/10.24072/pcjournal.279

In the article we review what we know about brood pollination in weevils and trying to understand why something so obvious for anyone that has looked closely has been missed both by many pollination ecologists and many weevil experts.

Much to my surprise, it caught a lot of media attention, so hopefully many more people will consider weevils when looking at flowers in the future! Here are a few of the news articles mentioning our review:

https://phys.org/visualstories/2023-05-weevils-long-nosed-beetles-unsung-heroes.amp

https://www.popsci.com/environment/weevil-beetle-pollination/

https://www.earth.com/news/weevils-the-unknown-and-underappreciated-pollination-heroes/

Ok, this last photo is NOT one of the pollinators, but I hope some day we will change what you can find when searching the internet for palm weevils!

More generally, it is nice to see more attention given to the role of insects as pollinators and other ecosystem functions. For example, The Nature Conservancy put together this nice page about pollinators in the Midwest: https://www.nature.org/en-us/about-us/where-we-work/priority-landscapes/great-lakes/stories-in-the-great-lakes/peculiar-pollinators/

In the News: what is a species, anyway?

Bruno spoke to CNN about species concepts and limits, and Anchylorhynchus ended up with a photo there!

See the news piece here: https://www.cnn.com/2023/05/12/world/what-is-a-species-explainer-scn/index.html

Visit to O'Brien weevil collection at ASU

This week, Bruno and Aline visited one of the most important weevil collection in the USA: the Charles O’Brien collection at Arizona State University.

Its importance comes not only from its impressive size, but also the amazing state of curation, benefitting from the years of careful work and passion by Charles O’Brien. Incidentally, he and the current of the collection - Nico Franz - have sone a lot of taxonomic work in flower-associated weevils, rendering it especially relevant for these groups.

Aline has found a lot of material for the ongoing phylogeny of Eugnomini, which we will soon take to the Field Museum. Also a great opportunity to see friends - some of them for the first time without a Zoom screen in the middle - and to take a break of Chicago weather.

Overall a great trip, and many important results will come from it!

Bruno and Aline can’t get enough weevils at the O’Brien collection

Pollinator beetle workshop at Universidad Ikiam

The past week I have been in Ecuador visiting my colleague Maria Cristina Peñuela at Universidad Ikiam in Tena, Ecuador. Cristina's students have been working with palm pollinators for their theses and therefore collected a lot of beetles, with weevils being a large part of them. As anywhere in the Amazon, those species turn out to be mostly undescribed, and in some cases it is even hard to fit them in the known genera of flower weevils.

Sampling beetle pollinators in Jatun Sacha reserve

Therefore, we partnered up to do a hands-on workshop on pollinator beetles, including discussions on their ecology and evolution and an overview of the taxonomy of main groups. At the end of the week, we had built a curated reference collection of pollinator beetles from the Napo province, which will support future work on ecology and systematics.

Lab component of the workshop, in which we mounted and identified palm flower beetles

This was not only my first time in Ecuador, but the first time in years teaching in person and visiting the Amazon. It was a great relief from the pandemic experience and really joy! It turns out it was also the beginning of the first semester with in-person classes at Ikiam too, so students were also highly motivated. I hope this was just the first step in a long-term collaboration and also the first of many initiatives to facilitate work on tropical beetle pollinators.

End of the workshop, with Maria Cristina Peñuela, students and the three boxes of insects forming a reference collection for further studies

How to sample Cecropia flower weevils

Cecropia is a genus of very common Neotropical plants known as guarumo (spanish) or embaúba (portuguese). They are very well known by their defensive mutualism with ants and generally thought to be wind-pollinated. But the reality is that pollination has never been studied in most cases.

It turns out that, hiding in plain sight, there are weevils intimately associated with flowers of Cecropia. This is the genus Udeus. Aline Lira is a PhD student at the Universidad Federal Rural de Pernambuco in Brazil, who I am co-advising, and she is currently working on the systematics, natural history and evolution of this genus.

Udeus eugnomoides in male flowers of Cecropia peltata in Panama.

The number of described species in the genus is certainly very small compared to the real existing diversity, and so far pretty much every new species of Cecropia that we sampled has resulted in a different species of Udeus.

It is appalling how rare these weevils are in collections, since in nature their host plants are extremely abundant and the weevils themselves are quite abundant on their flowers. Maybe because of their small size and colors blending well with the flower background? To encourage people to collect more samples, Aline has put together a guide available in Portuguese here: http://www.ppgea.ufrpe.br/sites/ppgea.ufrpe.br/files/entomonews_008.pdf

Basically, you can find adult weevils always in male inflorescences and, in some species of Cecropia, also in the female inflorescences. The larvae can be found inside branches of old male inflorescences. Because these are shed by the plant, this is the easiest way to sample them: find a decaying male inflorescence in the leaf litter, put in a container and wait a week for the adults to emerge.

If you are reading this text and find any Udeus in your backyard, please reach out!

Pandemic Weevil Description

This month, Jennifer Girón, myself and collaborators have described a new species of weevil from Panamá. This was my first time describing a species of Entiminae, but not the first attempt. My first undergrad project was to describe a weird entimine the my advisor Sergio Vanin had found in the collection. When he contacted Analía Lanteri to request more samples in case she had some, we discovered that she was already working on a description, which ended up published here:

This time around, I was the one contacted, and the reason was the covid19 pandemic. Jennifer is the real expert on Entiminae here, but it was logistically impossible for her collaborators in Panama who found the beetle feeding on crops to send samples in 2020. Since I had started a postdoc at STRI by then, she asked for my help to get the samples from Chiriquí and describe the species together.

This turned out to be much more challeging than usual: I arrived at STRI right before Sars-Cov-2. By then I needed a special request to go to the lab and to visit collections. I had no dissection equipment because mine was (and still is) stranded at my previous job at Harvard University. With a lot of help from Don Windsor and Annette Aiello at STRI, I was able to use their equipment, do the dissections and imaging. Jennifer and I discussed characters over e-mail, and we finally realized something very cool: our new species is just one of many at higher elevations in Central America! Most of these are currently identified as Epicaerus inaequalis in collections, but for now we only gave a name to the species found near Volcan Baru in Chiriqui, since it might turn out to be an important pest.

You can find more in the paper:

Atencio R, Barba A, …, de Medeiros BAS, Girón J. 2022. A new species of Epicaerus Pascoe, 1881 (Coleoptera: Curculionidae: Entiminae: Geonemini) associated with potato cultivars in Tierras Altas de Chiriquí, Panama. Zootaxa 5115: 103–121. http://doi.org/10.11646/zootaxa.5115.1.7

Using OpenStreetMap for fieldwork hiking

One year later than planned, I am finally spending some time In Barro Colorado Island (i. e. BCI) to conduct to do the research I planned to undertake at STRI. I just finished placing bags to differentially exclude pollinators and in a few days we should start understanding something about the evolution of weevil brood pollinators in generalist palms!

But this is not what I’d like to write about today. Instead, I’d like to share a tool that I have been using for a while and I think other people who spend a lot of time in trails could make use of it. This is OpenStreetMap, and I’m a fan.

Have you ever wished that you could use your GPS to navigate through trails as well as it does through streets? How to add trails that are not available in online maps?

Bagged flowers of the palm Oenocarpus mapora

This is a problem I encountered, for example, here in Barro Colorado. Even though the trails are well mapped, when I set my GPS to go to a specific point it attempts a straight line. This is bad for planning routes and for me to have a realistic idea of how much I have to walk, so ideally I would want it to route through trails. Below I explain how I put the trail information in a map in my GPS, and actually made it available to everyone in the process.

What is OpenStreetMap?

OpenStreetMap is akin to a Wikipedia of maps: anyone, anywhere in the world can edit, and through time it gets more and more accurate. This content is distributed freely under a Creative Commons license. This means that there is a whole community using OpenStreetMap data for all sorts of products, including maps that can be loaded in Garmin GPS units, the brand that I use.

How to load a map from OpenStreetMap to Garmin?

Since there is a large community of knowledgeable people working on optimizing maps for Garmin devices, all you need to do is to follow their advice. The ideal map depends on your intended usage (like driving, sailing or hiking) and region of the world. Check out this page for details.

For the record, I am currently using a map of Central America optimized for hiking and mountain biking: https://openmtbmap.org. They update weekly, so all I have to do is to edit OpenStreetMap to make sure it includes the trails that I need. Then a week later I can download a nice map that works in my GPS device. I did this last week, and now my GPS shows the exact distance through trails and instructions on how to get to the palm trees in which I placed bags:

How to add trails to OpenStreetMap?

The very first page of OpenStreetMap there is a button named EDIT. There you can use satellite images or GPS tracks that have been uploaded to the website to trace trails manually. In my case, I did not need to do that since the Smithsonian Institution already provides maps for BCI trails. Fortunately, they are distributed under a Creative Commons license as well, so I could use that data as base to create trails in OpenStreetMap. Just like Wikipedia, my edits are tracked with an ID and there one can see where the information came from: https://www.openstreetmap.org/changeset/113428019

Since I already started with a file containing edited tracks and names for all trails, I did not manually edit them in the website. Instead, I used JOSM, a program to edit OpenStreetMap that runs in my computer. Consulting the documentation of JOSM, in about an hour I figured out how to transform the raw data from STRI into the kind of data that shows that these are connected hiking trails. The program did a few automated checks and fixes to make sure everything looks right before uploading, and done.

Then I only had to wait a few days until someone used those new additions to OpenStreetMap to create a map suitable for my Garmin device, and follow their installation instructions.

This all seems hard, is there another way to help?

Yes! If while working in the field you have your GPS on all the time, like I do, it will generate traces (or tracks). Those are files that show where you have been, for example, throughout one day. You can download those traces from you gps (for example, using Garmin Explore or Garmin Basecamp) and upload them to OpenStreetMap. Other users can then use the traces to manually refine maps. For example, here I show a trace from one day I was in the field last week, and we can see that the BCI trail map does not match it exactly. If we had a number of these traces from many users (to account for GPS errors), someone could improve this map manually.

OpenStreetMap edit window showing mapped BCI trails (yellow) and GPS tracks from one day of my fieldwork (pink)

Weevil Workers Meeting 2021

I really disliked most of my experiences with online meetings since March 2019. With one exception: smaller conferences without concurrent sessions and held in regular zoom meeting format, with a chance for live interaction between everyone.

This is the case of the Weevil Workers Meeting, bringing together a lot of people interested in weevils and in learning from each other’s talks. This is the kind of conference that I think is enriched by the online format, with the possibility for direct interaction between amazing people that would be very hard to have all at the same time in the same physical space. It is an amazing community and a very fun meeting.

More info at: https://www.curculionoidea.org/weevil-workers-meetings

Gigante Field course

How to do a Field course while the covid19 pandemic is still raging?

Some people at STRI have taken the challenge! Daniel Medina and Rosannette Quesada have all my praise for taking the lead in making this 2-week-long course to happen, even if online, after a a hiatus of one year in the traditional Gigante Course.

I volunteered to lead a student project, and ended up deciding to use iNaturalist as the tool to connect students living across Panama. We created a project to record observations of insect flower visitors, learnt to upload observations and to export data for a project to do a simple test of insect preference for flower colors.

In addition to the teaching experience, it was one more opportunity to test ideas on how to use the data in the platform!

Screenshot of the project home page in iNaturalist, showing student observations.

Picudos en La Prensa

The story on STRI website has just been picked up by La Prensa, one of the largest newspapers in Panamá!

So many of the important plants in Neotropical forests are either pollinated or harmed by weevils (or both!), it is great to spread the word out a little bit.

The article can be accessed here: https://www.prensa.com/impresa/vivir/polinizadores-de-palmeras-gorgojos-no-tan-malvados/

Screenshot of La Prensa website

Story on STRI website

This week, a story on my work about pollinator weevils written by Beth King was published at the STRI webpage, you can check it out here (in English or Spanish): https://stri.si.edu/story/palm-pollinators

Ana Endara also produced these beautiful videos on the two sides of the palm/weevil interaction.

The fact that adults are pollinators:

And the weird morphology and behavior of their larvae:

I really enjoyed both videos, I hope you like them as well!

Teaching about weevils while 3,000 miles away

Here we are in the second year in a row in which a trip to Brazil is unlikely. Too bad, but at least the zoom world allowed be to teach a lecture today on weevils to the graduate student at Federal University of Paraná. Having araucarias, Butia palms and all their weevil flower visitors around would have been even better, but in any case it was great interacting with the students and teaching in Portuguese for the first time in many years!

Screenshot of the weevil lecture at UFPR

Pollinator Weevils at Coleoptera Meeting 2021

A group of students and postdocs from Brazil organized in March a Virtual Coleoptera Meeting (Encontro de Coleoptera). It featured a great group of speakers from multiple countries and speaking 3 languages. All presentations have subtitles in English (or in Portuguese, if presented in another language).

Here’s my talk encouraging everyone in Brazil to look for weevils in their backyard palm flowers:

There were also great surprises when it comes to weevil pollinators. Priscila Sanz-Veiga talked about a weevil species that breeds on fruits of Ouratea spectabilis (Ochnaceae), and apparently adults are flower visitors. It does not look like they are harmful to fruits at all, could they be unrecognized brood pollinators?

Alexandre Medeiros presented on the natural history of species of Oxycorinus weevils (family Belidae) pollinating parasitic plants in the family Balanophoraceae. This is a lineage of weevils for which very little is known, and it seems that in this case there is more than one species involved and their larvae may be feeding on different tissues (underground tubercle or floral stem):

A few other talks to highlight were Roberta Valente showing how she has been greatly advancing the study of palm flower weevils and other beetles in the Amazon, starting basically from scratch:

And Jennifer Girón presented on the ambitious goals of Colombian coleopterists to build a checklist of Coleoptera in one of the most diverse countries in the world!

These are just a few examples, many more great talks are available at the conference YouTube Channel: https://www.youtube.com/channel/UCDRwt-x6B_j9C9OCE4NjNNw/videos

Interview about Amazonia 4.0 project

This week I talked to the “Problemas Brasileiros” magazine about my role in the design of the Amazon Creative Lab of Genomics under the Amazonia 4.0 Initiative. I hope we will have more (and exciting) news on that soon: the idea is to build a portablelab for capacity building on DNA sequencing using automation and real time sequencers, targeted to non-biologists living in Amazonia. Together with a database for these sequences under development in a collaboration with researchers from the University of São Paulo, we hope this can foster the generation of new markets, new products and new research that will ultimately help us to increase the value of standing forests and improve livelihoods.

Here is the link to the interview (in Portuguese): https://revistapb.com.br/meio-ambiente/industria-amazonia-4-0/