Friday, March 1, 2024

March 2024 science summary



This month I have two articles on wildlife connectivity, one on global groundwater depletion, one on scientific reproducibility, and one on organizational behavior change.

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

Scientists often complain policy makers don't follow our recommendations (or even read them). But Brisco et al. 2023 finds that recommendations from meta-analyses tend to change over time as research continues. They looked at 79 papers (121 meta-analyses) and found that over time 93% of analyses either had a big change in effect size (+- 50% or more, see Fig 2 for examples) or change in statistical significance (see Fig 3). The key results are in Fig 5, which I find pretty confusing. Results staying consistently statistically significant w/ each study is rare, and ~25% of analyses showed a reversal in effect (from positive to negative or vice versa). Those reversals are the change we care the most about since it means action can backfire. BUT if you ignore the studies that were never statistically significant (seems safe) it's only 12% of analyses that flip, and if you also ignore the ones that lost significance (meaning the reversal is less meaningful) it goes down to 11%. That's still pretty bad though - 1 time out of 9 the scientific recommendation might lead to the opposite outcome we intend. They recommend scientists use cumulative meta-analyses to look for trends, and the reminder that we're often wrong reinforces the need for adaptive management and an empirical approach to seeing what works in a given context.

Jasechko et al. 2024 is a global assessment of groundwater levels since 2000 (using 170,000 wells and 1,700 aquifers) and comparing them to earlier trends for ~1/3 of those aquifers. They found 36% of aquifers were drying up (water level dropping deeper by 0.1 m / yr or more), and 6% of aquifers were improving (water level rising 0.1 m / yr or more), w/ 58% of aquifers not changing quickly in this century. 30% of the aquifers where they had 40 years of data declined faster in the 21st century than the 20 years prior (see Fig 3), but in 49% of those aquifers declines slowed or reversed. Unsurprisingly the trend is worst in drylands w/ farmland, and groundwater deepening is globally correlated w/ low precipitation, high evapotranspiration, and extent of agriculture. See Fig 1 and 2 for global map of trends, highlighting hotpsots of decline in CA, the US high plains, Iran, India, central Chile, and a few others. There's a news article about the study at

Iverson et al. 2024 cautions against assuming that modeled wildlife corridors connecting habitat patches ('linkages') actually receive much heavier use by wildlife. They looked at five linkage models in CA (see Fig 1), and compared them to 1) wildlife-vehicle collisions and 2) modeled wildlife presence (from a large set of wildlife observations). While black bear and puma vehicle collisions were slightly more likely in linkages, racoon collisions were LESS likely in linkages, and the other five species assessed were mixed depending on model. Across all eight species no model did consistently well for either wildlife vehicle collisions nor modeled occupancy. The authors note that the linkage models were all built on human disturbance metrics, but that another study found those metrics only significantly drove away about 1/3 of mammal species studied (including big carnivores and omnivores). Since wildlife don't have apps to find optimal travel routes, it's not shocking that they're not heavily using linkages. But this study is a good reminder to be wary of relying on models for citing narrow corridors, and it's a safer bet to assume wildlife presence is not typically highly concentrated.

Thurman et al. 2024 argues that it's important to consider disease when doing conservation planning and wildlife management. One key point is that in some cases improving connectivity can be net harmful for some species. The case of prairie dogs and black-footed ferrets on the top of page 3 is fairly compelling (the impact of plague is high enough that mitigating its spread should be a priority). Overall, I think it's fair to say that species and ecosystems need climate-resilient connectivity options to adapt to climate change, even if increased disease transmission offsets the benefits somewhat. But thinking about disease and if/how to incorporate it in planning should always be a good idea. I like the orange questions in Figure 1, but found the longer list in Table 1 to be overwhelming (which could make it harder for planners to act). There's no silver bullet being offered here, but maybe they're warning us to watch out for 'friendly fire' (unintended negative impacts of promoting connectivity w/o thinking about disease).

Ferraro et al. 2019 is a non-peer-reviewed working paper that asks whether behavioral psychology nudges known to influence individuals work on organizations too. They looked at a national organization asking for voluntary membership does from 3,000 nonprofits, and tested 1) crafting a clear and salient ask to emphasize public benefits, 2) publicly sharing who contributed and by how much, and 3) showing quarterly progress towards a national goal. All had no effect (each treatment on average made contributions very slightly lower, but w/o statistical significance). The authors hypothesize (w/ evidence from other studies) that group decision making makes orgs less responsive than individuals to these kinds of interventions. An interesting follow-up study would be to target individuals with the individual authority to make decisions that affect the broader organization and see if that works.


Brisco, E., Kulinskaya, E., & Koricheva, J. (2023). Assessment of temporal instability in the applied ecology and conservation evidence base. Research Synthesis Methods, November, 1–15.

Ferraro, P. J., Weigel, C., An, J., & MESSER, K. D. (2019). Nudging Organizations: Evidence from three large-scale field experiments (Vol. 21211).

Iverson, A. R., Waetjen, D., & Shilling, F. (2024). Functional landscape connectivity for a select few: Linkages do not consistently predict wildlife movement or occupancy. Landscape and Urban Planning, 243(March 2023), 1–12.

Jasechko, S., Seybold, H., Perrone, D., Fan, Y., Shamsudduha, M., Taylor, R. G., Fallatah, O., & Kirchner, J. W. (2024). Rapid groundwater decline and some cases of recovery in aquifers globally. Nature, 625(7996), 715–721.

Thurman, L. L., Alger, K., LeDee, O., Thompson, L. M., Hofmeister, E., Hudson, J. M., Martin, A. M., Melvin, T. A., Olson, S. H., Pruvot, M., Rohr, J. R., Szymanksi, J. A., Aleuy, O. A., & Zuckerberg, B. (2024). Disease‐smart climate adaptation for wildlife management and conservation. Frontiers in Ecology and the Environment, 1–10.

p.s. The photo is of a piece called "Shift" by Lisa Wood Studios, and it cycled between saying "Unconscious consumption" and "Conscious conservation" (presumably what we have now) and then as shown above "conscious consumption" and "unconscious conservation" (which makes less sense to me, but presumably means we're mindful of our choices and conservation happens automatically?