SeppeKuehnLab
banner
seppekuehnlab.bsky.social
SeppeKuehnLab
@seppekuehnlab.bsky.social
Physics and biology of microbial communities. Structure, function, evolution. Center for the Physics of Evolving Systems. Dept. of Ecology and Evolution. UChicago. @NITMB, @CLS. kuehnlab.org
25. Is it general? Are these regimes particular to our soils? No – we see hallmarks of the very same regimes in papers back to the 50s, and repeating the experiment on other soils gives similar results!
July 17, 2025 at 6:36 AM
23. In Regime III, the model claims that rare taxa are driving metabolism. Why? No activity in the CHL conditions, but fast uptake in the no-drug conditions. Remarkably, sequencing data show many rare Firmicute ASVs exploding out of the background in regime III.
July 17, 2025 at 6:36 AM
22. The model predicts nutrient limitation governs growth in regime II. In that case we should be able to remove the limitation by amendments. When Kiseok added carbon (but not N, S, P) the nitrate utilization rates jump up & are predicted by the model! Mechanisms were predicted…
July 17, 2025 at 6:36 AM
20. What we found is that growth is dominated by ASVs from a few phyla. Remarkably, the patterns of which pH perturbations each phylum grows under reflects the boundaries of the functional regimes we find via the model/nitrate utilization dynamics
July 17, 2025 at 6:36 AM
17 When you plot the two parameters - initial biomass, limiting nutrients – you get this remarkable pattern which captures the functional regimes of nitrate respiration in response to pH perturbations across soils. There are just a few ways the collective can process nitrate under variation in pH.
July 17, 2025 at 6:36 AM
July 17, 2025 at 6:36 AM
15. The model included a nutrient (not nitrate) limiting growth. The model has just two parameters – the quantity of initial functional biomass and the quantity of nutrient limiting the growth of that biomass.
July 17, 2025 at 6:36 AM
9. Kiseok brought soils into the lab, made slurries, and ran 1,500 microcosm experiments—perturbing pH to 13 levels and adding nitrate. Nitrate dynamics were measured in triplicate, with and without CHL, a translation inhibitor that arrests microbial growth.
July 17, 2025 at 6:36 AM
8. We wanted to study the response of nitrate utilization to short and long-term changes in pH. So @kiseokmicro.bsky.social found a USDA site in Washington State with a large pH gradient (4-8) and consistent land-use history. He sampled across the gradient (thanks David Huggins/USDA!)
July 17, 2025 at 6:36 AM
6. Building on work by @karnagowda.bsky.social , we focused on nitrate utilization under anaerobic conditions—where microbes use nitrate as an electron acceptor to make ATP, converting it to N₂ (denitrification) or ammonia (DNRA), key processes in agriculture and wastewater treatment.
July 17, 2025 at 6:36 AM
New paper in @nature.com! With @kiseokmicro.bsky.social , Siqi Liu, Kyle Crocker, Jojo Wang, Mikhail Tikhonov & Madhav Mani — a massive dataset and simple model reveal a few conserved regimes that capture how soil microbiome metabolism responds to perturbations. www.nature.com/articles/s41...
July 17, 2025 at 6:36 AM
So Kiseok Lee figured out how to measure BPA in soils. We contaminated soils with BPA in the lab and measured whether our synthetic communities could degrade in the presence of native microbiomes – and YES!
March 31, 2025 at 1:30 PM
A recent review from Michael Silverstein, Daniel Segre, Jennifer Bhatnagar at BU proposed that if a niche is weakly occupied in a natural environment (BPA is hard to degrade) then the chance of a “probiotic” engrafting and remediating might be higher. doi.org/10.1111/gcb....
March 31, 2025 at 1:30 PM
For that we turned to landscapes. Predicting AUC from community composition – it works quite well! But predictions get worse at higher concentrations (toxicity) - why? www.cell.com/cell-systems...
March 31, 2025 at 1:28 PM
We found low-rank structure in a matrix where rows are communities, columns BPA concentrations, and entries AUCs. Due to this low rank, we could use the ‘netflix’ algorithm to ‘impute’ held out data…enabling us to predict performance of communities across BPA concentrations quite well!
March 31, 2025 at 1:28 PM
Mahmoud and Kiseok then assembled ~70 random communities from this library and measured the dynamics of BPA degradation in time across 5 different initial concentrations of BPA – to see how communities handled the increasing toxicity that comes with increasing BPA.
March 31, 2025 at 1:27 PM
Mahmoud and Jonathan took a field trip – turns out BPA production is centered in Indiana, and Chicago is rife with contaminated sites. A few quick samples from these places, some enrichments…and they had found a set of 16 strains that either could degrade or were implicated!
March 31, 2025 at 1:26 PM
We then did a serial dilution experiment that showed what our theory proposed. On short timescales (rapid fluctuations), guilds are cohesive, but on longer timescales, this cohesion breaks down.
March 26, 2025 at 12:09 PM
Instead, we found that when environmental fluctuations are fast, guilds respond cohesively - with positively correlated abundance dynamics. But, when fluctuations are slow, then intra-guild competition takes over and strains within the same guild become negatively correlated.
March 26, 2025 at 12:08 PM
We defined a simple consumer-resource model that had this type of structure - two guilds. The overlap matrix defines how similar each pair of strains is in their resource preference - here guilds are clear as well.
March 26, 2025 at 12:08 PM
We then took these methods to data from time series in the ocean we found that integrated coherence provided strong signals for phylogenetically related taxa. This is a suggestion that these methods have some power!
March 24, 2025 at 8:33 PM
What Xiaowen showed was that integrated measures of cross-power spectra between species dynamics provided very strong measures of resource-preference overlap. These metrics also allowed measuring guild-structure in complex communities.
March 24, 2025 at 8:32 PM
Xiaowen setup consumer-resource models to reflect guild-like resource preference structure – groups of strains with similar resource preferences being driven by time-varying resources. This leads to a resource-overlap matrix which quantifies how much two strains resource preferences intersect.
March 24, 2025 at 8:31 PM
We've arrived in the Netherlands for a sabbatical and gotten settled in Utrecht. Already had exciting interactions with people at VU, UUtrecht, AMOLF, and Delft. Excited for many European adventures in the coming months! Gratuitous picture of the Dom tower in Utrecht tonight....
January 17, 2025 at 6:44 PM
Work by former student Chandana Gopalakrishnappa (now MIT post doc) on algae bacteria interactions. Cover of Cell Systems with Commentary from Calatrava, Kilonzo, Hom. www.cell.com/cell-systems...
September 18, 2024 at 6:35 PM