February 2026 science summary

Greetings,

A new paper I'm a co-author on just came out in Conservation Letters. It's about what counts as a "rapid evidence assessment" and how to do one well. Here's a 275 word blog about it: https://sciencejon.blogspot.com/2026/01/new-paper-rapid-evidence-assessments.html , the full paper (~3400 words) is here: https://conbio.onlinelibrary.wiley.com/doi/10.1111/con4.70005 , and I've summarized it below.

For any other map nerds out there; Esri has a new web map to make it easy to see 40 years of USGS land cover data at https://links.esri.com/LCExplorer . There's a blog about it at https://www.esri.com/about/newsroom/arcnews/40-years-of-usgs-land-cover-data-in-arcgis-living-atlas

If you know someone who wants to sign up to receive these summaries, they can do so at http://subscribe.sciencejon.com (no need to email me).

FRESHWATER PROTECTION:
Until very recently, we didn't have a decent estimate of which rivers were protected across the United States! Comte et al. 2026 describes how the first such database (the National Protected Rivers Assessment) was compiled and launched in Feb 2025. Since freshwater-specific protection is rare it relies mostly on assumptions about how likely different kinds of terrestrial protection are to effectively conserve the 5 key ecological attributes (KEAs) of rivers (flow, water quality, connectivity, habitat, and fish/wildlife/plants/etc. - see Fig 1). They find 19% of river length (12% of CONUS river length) has "viable" protection, although only 0.9% is comprehensively protected. Note their bar for "viable" protection (good enough) is fairly low - a score of 1.25/5 on their index. The index combines length and KEAs, so a 1.25 could mean 25% of the river provides protection for each of the 5 KEAs (overlapping or distinct), or all the river has protection for 1 KEA, and 25% has protection for another, but the other 3 KEAs are unaddressed. Hypothetically a river w/ 100% protection on 4 KEAs would score as "comprehensive" protection but could lack any protection from water withdrawals that would make the river run dry. See Fig 4 for rivers in the best shape that are most important for drinking water. This is a super useful resource. Explore the data at https://map.myriver.americanrivers.org/


CARBON AND WATER FOOTPRINT OF AI:
Xiao et al. has granular state projections of demand for AI through 2030, plus the carbon and water footprint of AI in each state. They recommend trying to steer data center growth to four states (TX, MT, NE, SD) given relatively low water scarcity and abundance of renewable energy (and potential to expand wind and solar). With the middle case assumptions by 2030 AI's water footprint would only be ~0.2% of current US crop water footprint, and the energy consumption would be ~2% of current total electric power generation. AI can have important local impacts and is growing fast, but national impacts are still projected to be relatively small.


IMPACT EVALUATION:
Neugarten et al. 2025 is a good overview to how to evaluate if conservation worked or not. They define impact evaluation and key terms like counterfactuals (what would have happened w/o conservation), confounders (variables that make understanding impact harder), and cover different types of evaluation (randomized experimental, quasi-experimental, and qualitative methods). They also discuss why looking at trends alone can be misleading (wildlife population might be dropping, but would have dropped more w/o action). They conclude recommending impact evaluation for projects that are high stakes, expensive, over big areas, untested, and/or assume additionality.


RAPID EVIDENCE ASSESSMENTS:
Webb et al. 2026 (I'm a co-author) proposes a consensus definition of what should count as a "rapid evidence assessment" (or REA) in conservation. It can be hard to find the sweet spot when assessing evidence. Too quick and dirty and you can get wrong answers, but too rigorous and the results 1) may come too late to be useful and 2) take a lot of resources that could be spent on multiple smaller studies. The paper has our final definition, recommended steps for a REA, and Table 1 has a nice little guide to picking what level of rigor may be the best fit in different circumstances.


REFERENCES:
Comte, L., Olden, J. D., Littlefield, C., Dickson, B. G., Zablocki, J., & Moryc, D. (2026). National assessment of river protection in the United States. Nature Sustainability. https://doi.org/10.1038/s41893-025-01693-8

Neugarten, R., Rodewald, A., Eklund, J., & O’Garra, T. (2025). An introduction to impact evaluation for conservation. Conservation Science and Practice, 7(11), 1–9. https://doi.org/10.1111/csp2.70169

Xiao, T., Nerini, F. F., Matthews, H. D., Tavoni, M., & You, F. (2025). Environmental impact and net-zero pathways for sustainable artificial intelligence servers in the USA. Nature Sustainability, 8(12), 1541–1553. https://doi.org/10.1038/s41893-025-01681-y

Webb, J. A., Schofield, K. A., Cook, C. N., Fisher, J. R. B., Cheng, S. H., Christie, A., Cooke, S. J., Dubois, N. S., Frampton, G., Macura, B., Nichols, S. J., Richards, R., Aicher, R. J., Mason, S., Anderson, E., Betley, E., Borsuk, M., Busch, J., Carlson, S., … Ridley, C. E. (2026). A Standardized Definition of Rapid Evidence Assessment for Environmental Applications. Conservation Letters, 19(1), 1–7. https://doi.org/10.1111/con4.70005


Sincerely,
 
Jon
 
p.s. This is a four pound turnip. I bought it at the farmer's market out of curiosity, and split it across two recipes and it was tasty!