Happy New Year!
As always, I’ve compiled the summaries of my favorite 15 papers from 2025 (more or less – I left out some papers I loved to make room for a more well-rounded set). I also wanted to re-share two non-paper tidbits:
- You all sent me lots of great advice for scientists, much of which was similar to a paper I co-authored (and we have lots of companion products for that paper).
- A case study of effective science communications (from this video overview): People struggle to understand the risk of relatively safe medical procedures (like anesthesia). The “micromort” is an activity with a one in a million chance of resulting in death (Howard 1989, “Microrisks for Medical Decision Making”), allowing you to compare risks between activities you understand and those new to you. Table 1 in Keage and Loetscher 2018 has a nice comparison of understandable risks, from 1 micromort (walking 27 miles), to going under anesthesia once (10 μmorts = walking 270 miles), to giving birth (120 μmorts), and even climbing Mt Everest (12,000 μmorts). Doctors can use this to explain risk, as in Sieber and Adams 2017 which note the risk of getting lymphoma from breast implants is lower than a day of skiing or drinking 500 mL of wine. Table 1 in Ahmad et al. 2015 has a table comparing the risk of a range of medical procedures, making a great complement to the “everyday risk” table in Keage and Loetscher (e.g., elective open heart surgery is risky, but 5 times safer than emergency open heart surgery).
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).
DEFORESTATION:
Franco et al. 2025 looks at how deforestation and climate change have affected temperature and precipitation in the Amazon. Losing trees not only increases atmospheric carbon which drives global warming, for tropical forests in particular it also increases LOCAL warming and especially decreases water cycling. They found that deforestation caused 3/4 of the decline in dry season precipitation but only 17% of the increase in temperature. This is similar to other work which has found deforestation has delayed the onset of the rainy season in the Cerrado (Spera et al. 2016) and Pantanal (Lázaro et al. 2020).
Roquette et al. 2025 is ostensibly about the technical details around land use mapping in Mato Grosso, Brazil. But it actually makes a much broader point: boring things like data sources and algorithms used can have an outsize backdoor policy impact. They argue that a shift in how land use was classified could sneakily allow a ton of deforestation. Basically by changing how "forest" is defined and reclassifying some forest as a type of savanna (confusingly, cerrado is an ecosystem but the Cerrado is a biome / region), this could allow deforestation since 80% of forest has to be set aside from development but only 35% of savanna does. In some cases the new map actually does the reverse (classifying what was savanna as forest) but that can be reversed upon request. The worst case scenario is that up to 4,1 million ha could be authorized for deforestation.
FOREST DEGRADATION:
Bourgoin et al. 2024 is a global analysis of degradation of tropical moist forests using LiDAR satellite data (from GEDI) along with Landsat optical imagery. It's worth reading the whole paper, but a few key findings: 1) degradation is hard to sense but is also a) a good predictor of deforestation in addition to b) being inherently meaningful. 2) Forest height is reduced by selective logging (15%) and fire (50%) and recovery is low after 20 years. 3) Expansion of ag and roads leads to a 20-30% drop in forest canopy height and biomass at the edge, with effects up to 1.5km in from the edge. Fig 1 has maps of global results - it's a bit confusing but the x-axis of the charts is height in meters, so lower numbers mean more degradation or younger forests. There's an article about this one at https://phys.org/news/2024-07-reveals-human-degradation-tropical-forests.html#google_vignette
FORESTATION (REFORESTATION AND AFFORESTATION):
Wang et al. 2025 finds that earlier estimates of how much of the world could be forested were much too high. One the one hand - this means we can't count on trees to do as much sequestering as some models hoped for. But as one of the authors (Susan Cook-Patton) points out (in this excellent post) the larger area wasn't feasible anyway, so this smaller estimate gives us more actionable priorities for planting. She also notes that while reducing fossil fuels (and protection of acutely threatened forests) is higher priority than forestation, we absolutely need all of the above. Note: reforestation is restoring trees where they used to be, afforestation is planting trees in what used to be grasslands or other ecosystems, forestation is both.
FIRE:
Ellis et al. 2022 is a global analysis of fire risk over the past 41 years, using estimated moisture trends in fuel (surface litter and dead plant debris) as the key metric. Fig 2a shows the % of days during the fire season where fuel moisture is expected to be below 10% (and thus at high fire risk), while 2b shows the total # of fire season days at high fire risk and Fig 4 shows the trend in high fire risk over time (36% of ecoregions are drying vs. only 4% getting wetter). They note that days w/ dry fuel isn't the only key driver of risk; consistent dry conditions limit fuel accumulation and thus the risk of severe fire. So surprisingly historically wet forests which accumulate lots of plant matter in wet years between fires may be the most at risk (as other research has shown).
Siquiera et al. 2025 has a helpful reminder that finer data isn't always better. They found free MODIS data (which is relatively fast and easy to process due to 500m resolution) actually did a little better at detecting areas that had burned in the Pantanal than Sentinel 2 data (also free but needs more babying and at 10m [2500* finer resolution] it takes a ton more processing time). LANDSAT (at 30m) did worse than either. The only caveat is that their validation for “burned areas” is fire foci which are partly derived from MODIS which likely biases the results somewhat. But anytime we can get away with using coarser data to save time and money it’s good news! One of my papers has a table about when coarser vs. finer data may be better, and that table is also in a blog about how to pick the right spatial resolution of data
Guzmán-Rojo et al. 2025 is an interesting model of how fires in the Bolivian Chiquitano dry forest affect water availability, focusing on soil changes (crusting, ash deposition, and reduced porosity which can decrease infiltration and increase runoff) rather than vegetation loss. They found ~40% lower recharge in the first year after a hypothetical very severe fire, returning to ~10% lower than pre-fire after two years. However, they note that field data showed that moderate to severe fires actually reduced soil porosity by 39%, while their hypothetical very severe fire assumed a 70% reduction which is probably an upper bound for how much soil permeability could be reduced. They also note that if they had accounted for vegetation loss in their model, the recharge may have been lower in some areas where dense vegetation transpires more water than it intercepts. While they had limited data to validate their model they used other studies and proxies where they could.
FRESHWATER
Sayer et al. 2025's headline is that 1/4 of freshwater animals are threatened with extinction. But that's similar to other estimates; to me the new thing here is guidance on how to prioritize sites to conserve with limited data. They found that 1) prioritizing with just threatened freshwater tetrapod data (animals with four legs like some amphibians, reptiles, birds, and mammals) does well for overall freshwater biodiversity (range-size rarity) but 2) prioritizing on abiotic factors alone (low flow / water stress, nitrogen as a proxy for water pollution) does worse than random! Also from Fig 2b permanent rivers are home to almost all of the threatened FW species, while species in other fresh wetlands fare much better.
Petry et al. 2025 has predictions of changing streamflow and flooding across South America by 2100 under a moderate climate change scenario. Figure 4 has the key findings about how much more or less frequent floods may be. Note that “RP” means “return period” as in a “5 year flood” or “100 year flood” (the magnitude of flooding you’d expect on that frequency / rarity, so higher numbers mean more severe flooding). RPCF means how much more or less frequent those floods would be (with negative sign indicating less frequent flooding, e.g. the -2 on the Paraguay river in the Pantanal means half as often). But much more flooding is expected in Peru, Ecuador, Colombia, and Southern Brazil, and parts of the Amazon will see 1/10 as much flooding as they historically have. They find Pantanal floods (in the Paraguay River and some tributaries like Cuiaba and Negro) will be roughly half as frequent and half as severe, they don’t have a clear trend in the Chaco, and in Chile the area from roughly Santiago to Valdivia has some rivers where flooding will be ~2-3 times less frequent while the northern part of Chile will only see slightly less flooding.
CLIMATE CHANGE - CARBON MARKETS AND ALBEDO:
Riley et al. 2025 have an update on an issue I've written about several times: that the albedo of trees (how much they reflect sunlight compared to bare soil) can reduce or even fully negate the climate benefit of trees in some cases. They looked at 172 tree planting projects in the voluntary carbon market to see how carbon credits issued compared to true climate impact once albedo was considered. On average 18% of the issued credits shouldn't have been, and 25% of the projects offered more than double the credits they should have once albedo was considered. 12% of the projects (a subset of that 25%) were even net harmful (representing 30% of total credits issued). In good news, 9% of projects actually had more benefit than estimated, and over half of projects had 0-25% of their issued carbon credits negated by albedo. This is important to get right, but a shortcut is to focus protection and forestation in more tropical places while avoiding areas with heavy snow cover and/or very light-colored soils (the pink places in Fig 2a are good to avoid).
CATTLE AND CLIMATE CHANGE:
Meo-Filho et al. 2024 is an important study (part of a special issue of papers around the sustainability of animal foods and plant alternatives, https://www.pnas.org/topic/561). There's been lots of research with hyperbolic claims about red algae reducing methane production in cattle (mostly in vitro, or tests in a petri dish). I believe this is the first paper to measure methane reductions not only 1) in vivo (a test of what happens in real animals) but 2) in the grazing phase of their life cycle. Most American cattle graze for very roughly 15 months before spending 3 months in a feedlot, and they emit more methane per day when grazing. Only 15% of total cattle production GHGs come from the feedlot phase, so even big reductions then can't touch cattle's high carbon footprint. But this paper found supplementing with seaweed reduced GHGs relative to control animals by 38% over 90 days with no side effects! That is great news and this is work worth following up, BUT a few key caveats: 1) this was a study of only 24 animals, 2) the animals began at 15 months when they would typically go to a feedlot, to show how much this could really reduce the total carbon footprint of cattle (and if any side effects crop up eventually) it would need to be tested on calves from when they are weaned off of milk to when they go to slaughter, and 3) variations in cattle breed, the dominant type of grass they're eating, and climate could all affect the results. So it's very good news, but does not yet mean it's possible to produce a burger w/ 1/3 less carbon. Also, while the numbers are hotly contested, remember that the carbon footprint of beef is very roughly 10* that of pork or chicken (~50* the GHG of beans), so even if the reduction IS scalable, beef will still be a high carbon food.
BIODIVERSITY OFFSETTING:
While Lourival et al. 2025 ponders 10 questions for biodiversity offsetting in the Brazilian Pantanal and its watershed (listed at the top of the 3rd page), it's relevant to biodiversity offsets in general. I especially like Fig 4 with different ways to think about equivalence for offsets (area, economic value, ecosystem value, or a mix). So for example if you clear high-value land in the Cerrado, you could need to protect a much larger lower-value area in the Pantanal. They find 57% of properties in the highlands feeding the Pantanal are out of compliance with the Forest Code (Table 1, Fig 3). Properties WITHIN the Pantanal are only 22% out of compliance, so considering the whole river basin could offer opportunities BUT ecological equivalence may be low and state regulations restrict what is possible. One could argue that's a huge market for offset purchases OR evidence the law is toothless and demand will be low.
GLOBAL THREATS:
Oakleaf et al. 2024 is a long-awaited global analysis of likelihood of short-term (by 2030) habitat conversion (1 km resolution, see Fig 5 for the results). They include not only cropland and city expansion, but also land use change for energy (fossil or renewable) and mining, and considered several suitability factors (like slope and land cover). Their approach assumes recent trends (2000-2015) will continue rather than modeling different scenarios of economic development. Like any global data you can find places where it seems wrong, but it's a great resource nonetheless. You can grab the data or view the data in a web map. There's also a blog about the paper.
WILDLIFE:
Kopf et al 2024 really made me think. They summarize some of the important contributions of old (and often "wise"!) animals, and proposes "longevity conservation" as a strategy to retain them. Old animals are especially important for species who rely on cultural transmission (like migratory species) and those who reproduce more as they age and grow, and the article goes into detail with examples of both. The article also covers some of the impacts of losing large and old animals, like changing ecosystem structure and function, pack instability, and even infanticide. The ability of older animals to help their group adapt to drought or food shortages maybe increasingly important as climate changes (although the individual resilience of older animals is likely to lower to some stresses like disease). With trophy hunting and fishing selecting for larger (typically older) animals, they argue for the importance of better population modeling, setting age class targets for fisheries, restricting hunting of larger and older animals, and going beyond tracking biomass or abundance to watch for "longevity depletion."
CHOCOLATE:
Gopaulchan et al. 2025 advances the frontier of scientific understanding of flavor development in chocolate. It has a few components. First they measured changes in temperature and pH over a week in fermenting Colombian cacao beans to estimate microbial activity (they correlated with the color changes used to assess fermentation end point). Second they sequenced the genome every 24 hours to watch diversity decline for both bacteria (Fig 1e, Acetobacteraceae dominating over time) and fungi (Fig 1f, Saccharomycetaceae dominating over time). But crucially, they third compared how these shifts happened at three farms w/ similar genetics but different microbial mixes. Look at Fig 2e and 2f!!! The two new farms had diversity increase again eventually, and one farm had significant populations of families that were only present in trace amounts before! OK, stay with me. Fourth, they extracted cocoa liquor and compared their three farms to reference chocolates (Fig 3b). One farm tastes "West African" (roasty, dark wood, tobacco), and two taste more "Malagasy" (one emphasizing light / caramelly / tropical flavors, and the other more fruity and bitter). Fifth they figured out which families of microbes were likely most essential for flavor (Fig 4) and sixth put together different inoculant mixes (most had the full diversity of microbes minus a different single strain missing for each mix, plus a random mix). They then (seventh) did controlled cacao bean fermentation w/ each mix plus an uninoculated control (in the lab so it wouldn't have the wild strains), and tasted the results (eighth), finding that the beans inoculated with the full mix of strains had better flavors than those inoculated w/ fewer strains or none at all (Fig 5h). The authors argue that with synthetic starters and controlled fermentation, chocolate flavor could be improved in more industrial-scale chocolate.
REFERENCES:
Bourgoin, C., Ceccherini, G., Girardello, M., Vancutsem, C., Avitabile, V., Beck, P. S. A., Beuchle, R., Blanc, L., Duveiller, G., Migliavacca, M., Vieilledent, G., Cescatti, A., & Achard, F. (2024). Human degradation of tropical moist forests is greater than previously estimated. Nature, 631(8021), 570–576. https://doi.org/10.1038/s41586-024-07629-0
Ellis, T. M., Bowman, D. M. J. S., Jain, P., Flannigan, M. D., & Williamson, G. J. (2022). Global increase in wildfire risk due to climate‐driven declines in fuel moisture. Global Change Biology, 28(4), 1544–1559. https://doi.org/10.1111/gcb.16006
Franco, M. A., Rizzo, L. V., Teixeira, M. J., Artaxo, P., Azevedo, T., Lelieveld, J., Nobre, C. A., Pöhlker, C., Pöschl, U., Shimbo, J., Xu, X., & Machado, L. A. T. (2025). How climate change and deforestation interact in the transformation of the Amazon rainforest. Nature Communications, 16(1), 7944. https://doi.org/10.1038/s41467-025-63156-0
Gopaulchan, D., Moore, C., Ali, N., Sukha, D., Florez González, S. L., Herrera Rocha, F. E., Yang, N., Lim, M., Dew, T. P., González Barrios, A. F., Umaharan, P., Salt, D. E., & Castrillo, G. (2025). A defined microbial community reproduces attributes of fine flavour chocolate fermentation. Nature Microbiology, 10(9), 2130–2152. https://doi.org/10.1038/s41564-025-02077-6
Guzmán-Rojo, M., Silva de Freitas, L., Coritza Taquichiri, E., & Huysmans, M. (2025). Groundwater Vulnerability in the Aftermath of Wildfires at the El Sutó Spring Area: Model-Based Insights and the Proposal of a Post-Fire Vulnerability Index for Dry Tropical Forests. Fire, 8(3), 86. https://doi.org/10.3390/fire8030086
Kopf, R. K., Banks, S., Brent, L. J. N., Humphries, P., Jolly, C. J., Lee, P. C., Luiz, O. J., Nimmo, D., & Winemiller, K. O. (2024). Loss of Earth’s old, wise, and large animals. Science, 2705, 1–19. https://doi.org/10.1126/science.ado2705
Lourival, R. F. F., de Roque, F. de O., Bolzan, F. P., Guerra, A., Nunes, A. P., Lacerda, A. C. R., Nunes, A. V., Alves, A., Filho, A. C. P., Ribeiro, D. B., Eaton, D. P., Brito, E. S., Fischer, E., Neto, F. V., Porfirio, G., Seixas, G. H. F., Pinto, J. O. P., Quintero, J. M. O., Sabino, J., … Tomas, W. M. (2025). Ten relevant questions for applying biodiversity offsetting in the Pantanal wetland. Conservation Science and Practice, July 2022, 1–21. https://doi.org/10.1111/csp2.13274
Meo-Filho, P., Ramirez-Agudelo, J. F., & Kebreab, E. (2024). Mitigating methane emissions in grazing beef cattle with a seaweed-based feed additive: Implications for climate-smart agriculture. Proceedings of the National Academy of Sciences, 121(50), 2–9. https://doi.org/10.1073/pnas.2410863121
Oakleaf, J., Kennedy, C., Wolff, N. H., Terasaki Hart, D. E., Ellis, P., Theobald, D. M., Fariss, B., Burkart, K., & Kiesecker, J. (2024). Mapping global land conversion pressure to support conservation planning. Scientific Data, 11(1), 830. https://doi.org/10.1038/s41597-024-03639-9
Petry, I., Miranda, P. T., Paiva, R. C. D., Collischonn, W., Fan, F. M., Fagundes, H. O., Araujo, A. A., & Souza, S. (2025). Changes in Flood Magnitude and Frequency Projected for Vulnerable Regions and Major Wetlands of South America. Geophysical Research Letters, 52(5). https://doi.org/10.1029/2024GL112436
Riley, L. M., Cook-Patton, S. C., Albert, L. P., Still, C. J., Williams, C. A., & Bukoski, J. J. (2025). Accounting for albedo in carbon market protocols. Nature Communications, 16(1), 8810. https://doi.org/10.1038/s41467-025-64317-x
Roquette, J. G., Vacchiano, M. C., Daher, F. R. G., & Finger, Z. (2025). Pseudo-legal deforestation due to changes in the classification of native vegetation in Mato Grosso, Brazil. Environmental Conservation, 1–7. https://doi.org/10.1017/S037689292510012X
Sayer, C. A., Fernando, E., Jimenez, R. R., Macfarlane, N. B. W., Rapacciuolo, G., Böhm, M., Brooks, T. M., Contreras-MacBeath, T., Cox, N. A., Harrison, I., Hoffmann, M., Jenkins, R., Smith, K. G., Vié, J.-C., Abbott, J. C., Allen, D. J., Allen, G. R., Barrios, V., Boudot, J.-P., … Darwall, W. R. T. (2025). One-quarter of freshwater fauna threatened with extinction. Nature, 3(December 2023). https://doi.org/10.1038/s41586-024-08375-z
Siqueira, R. G., Moquedace, C. M., Silva, L. V., de Oliveira, M. S., Cruz, G. D. B., Francelino, M. R., Schaefer, C. E. G. R., & Fernandes-Filho, E. I. (2025). Do finer-resolution sensors better discriminate burnt areas? A case study with MODIS, Landsat-8 and Sentinel-2 spectral indices for the Pantanal 2020 wildfire detection. International Journal of Remote Sensing, 00(00), 1–24. https://doi.org/10.1080/01431161.2025.2496000
Wang, Y., Zhu, Y., Cook-Patton, S. C., Sun, W., Zhang, W., Ciais, P., Li, T., Smith, P., Yuan, W., Zhu, X., Canadell, J. G., Deng, X., Xu, Y., Xu, H., Yue, C., & Qin, Z. (2025). Land availability and policy commitments limit global climate mitigation from forestation. Science, 389(6763), 931–934. https://doi.org/10.1126/science.adj6841
Sincerely,
Jon
p.s. I found these Christmas decorations at the Carnegie Hotel in New York especially striking b/c of the contrast w/ the dark background, lights, and glittery balls
