Wednesday, February 1, 2023

February 2023 science summary

Gyokuro green tea on vacation

Hi all,

This month I have one article on metascience (problems with reproducibility may be worse than we think), two on people and nature, and one on wildlife connectivity.

As a bonus, I wondered how my summaries would compare to automatically generated ones from Chat GPT - take a look and decide for yourself what each one is good for and what it misses! I was interested how much the length varied (I asked for a 250 word summary for each), and how much text was devoted to describing the topic without really presenting relevant results. I'll have to play around with longer summaries and see how it compares!

I'd love to hear what others think about the ChatGPT summaries, and what your own experience playing with it has been.

Jon's summary:
Breznau et al. 2022 has some scary news about science - not only is it less reproducible than we think, we can't even figure out why results vary so much. To be fair, they note that natural sciences and/or experimental research should have less variation than social science based on existing surveys (what this study looked at). But it's still concerning! Or preliminarily concerning but waiting for many more replicas, to take their message to heart. The models the 73 teams built were: 17% positive (more immigration increases support for social policies), 25% negative (more immigration reduces support for social policies), and 58% did not find a clear effect (the confidence interval included zero, although they may have had a positive or negative average effect). 61% of researchers concluded that immigration does not reduce support for social policies, 26% concluded it DOES reduce support (the text says 28.5% but it's a typo, reinforcing the core message of the paper), and 13% concluded it couldn't be tested w/ the given data. And Fig 2 shows that not only are results and conclusions all over the place, the variation isn't explained by the variables they tracked like expertise or prior beliefs. That means researcher bias is only part of the problem. I have some questions about the metanalysis itself that make me suspect they could have explained more of the variance with different methods (ironically, that is consistent with their core findings about how small method changes can drive results). But the paper reveals two problems: 1) scientists can produce different quantitative results from the same data and hypotheses, and 2) scientists' conclusions are often not well tied to their results (this paper found only ~1/3 of variation in conclusions came from how consistent the set of models each team used were). I see a lot of #2 when I peer review papers. Let's all remain humble and skeptical, and look for more replication in 2023!
ChatGPT's summary:
The article "Observing many researchers using the same data and hypothesis reveals a hidden universe of uncertainty" published in the journal Nature Human Behavior, explores the concept of reproducibility in scientific research and its significance in understanding the reliability of scientific findings. The study conducted an experiment where multiple researchers were given access to the same data set and hypothesis, and were asked to conduct their own independent analysis and interpretation of the data. The study found that even when given the same data and hypothesis, researchers often come to different conclusions, which highlights the importance of reproducibility in scientific research. The article argues that reproducibility is crucial in determining the reliability and validity of scientific findings, and that the lack of reproducibility can lead to false conclusions and misunderstandings. The article also notes that reproducibility is not always easy to achieve, as it requires careful experimental design, sufficient sample sizes, and transparent reporting of methods and results. The article stresses on the need for scientific community to openly acknowledge and address the problem of reproducibility, as it can help in identifying the sources of uncertainty and improving the reliability of scientific research. The article concludes by emphasizing that reproducibility is a crucial aspect of the scientific process, and that it is the responsibility of the scientific community to ensure that the methods and results of scientific studies are transparent, accurate, and reproducible. The study suggests that reproducibility should be an integral part of the scientific process and should be used to identify and reduce uncertainty in scientific findings.

Jon's summary:
Chaplin-Kramer et al. 2022 is a great global summary of 14 ecosystem services I've been waiting to see for years! Their big finding is that 90% of nature's local contributions to people within each country come from a total of 30% of global land area and 24% of coastal waters (EEZs). Globally only 15% of those places are protected. The land and water needed is uneven by country (e.g., the US needs 37% of land and 15% of coastal waters), and protection varies too. See Fig 1 for the areas that provide the most benefit per unit area. The land required would be lower if optimizing globally, but at the cost of less equitable benefit distribution. The 2 global ones (carbon storage and moisture recycling) need 44% of land (optimized globally) to stay at 90% of current levels, mostly overlapping with the 30% (see Fig 3). Roughly 87% of the world population benefits from at least one of the ecosystem services, but benefits are not distributed equally (see Fig 2). The local services include: water quality (regulating nitrogen and sediment), crop pollination, livestock fodder, production of timber and fuelwood, flood regulation, fish harvest (from rivers and oceans), recreation (on land and oceans), and coastal risk reduction. If interested you can get combined GIS data from or one of the authors (Rachel Neugarten) is happy to send individual maps and data.
ChatGPT's summary:
Mapping the planet’s critical natural assets" is an article published in the journal Nature that discusses the importance of mapping and protecting the world's natural assets, such as forests, wetlands, and coral reefs. These natural assets provide a wide range of benefits to human societies, including carbon sequestration, water regulation, and biodiversity conservation. The article emphasizes that these natural assets are under increasing threat from human activities such as deforestation, pollution, and climate change, and that urgent action is needed to protect them. The article also highlights the work of various organizations, such as the Natural Capital Project and the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services, that are working to map and protect these natural assets at a global scale. It also covers the importance of these natural assets for human well-being, such as providing food, medicine, and livelihoods. The article also emphasizes on the fact that these natural assets are inextricably linked with human well-being, as they provide vital ecosystem services, including air and water purification, pollination of crops, and storm protection. The article states that the current rate of loss of natural assets is alarming and calls for immediate action to be taken to protect them. The article concludes by emphasizing the need for collaboration between governments, businesses, and civil society to protect these natural assets, and to invest in sustainable land use and management practices that support the conservation and restoration of these vital resources.

Jon's summary:
Salvatori et al. 2023 used camera traps in the Dolomite mountains of Italy to see how wildlife responded to human recreation (mostly hiking, mountain biking, and driving). They found that all 8 sampled wild species tended to favor visiting sites frequented by humans at times when humans weren't present (see Fig 5 - some of this is natural peak activity times, but more human presence resulted in a stronger temporal effect in wildlife). Bear, deer, and chamois (goat-antelopes) all avoided sites w/ more human visitation, while hares and red fox favored such sites. My main take-aways are that we should keep in mind how people will affect wildlife behavior, and that for big species we should probably look at where human trails are relative to key connectivity corridors to ensure wildlife aren't scared away from places they especially need to visit. Also - humans are looking for wildlife to interact with, while they are trying to avoid us.
ChatGPT's summary:
The article "Crowded mountains: Long-term effects of human outdoor recreation on a community of wild mammals monitored with systematic camera trapping" published in the journal Biological Conservation, explores the effects of human outdoor recreation on wild mammal populations. The study conducted over a period of six years, used camera traps to monitor a community of wild mammals in the Swiss Alps. The study found that human recreational activities, such as hiking, skiing, and mountain biking, had a significant impact on wild mammal populations. The study found that wild mammal populations in areas with high levels of human recreational activity were smaller and less diverse than those in areas with low levels of human recreational activity. The study also found that the presence of human recreational activities caused wild mammal populations to shift their activity patterns, leading to increased competition for resources and reduced survival rates. The study highlights that human outdoor recreation can have a significant impact on wild mammal populations, and calls for more effective management strategies to mitigate these effects. The article emphasizes the importance of monitoring and understanding the effects of human outdoor recreation on wild mammal populations, as it can help in developing effective conservation strategies. The study suggests that creating protected areas and implementing regulations on human activities in these areas could help in reducing the negative effects of human outdoor recreation on wild mammal populations. It also suggests that educating the public about responsible outdoor recreation practices can help in reducing the impact of human activities on wild mammal populations. The study concludes by highlighting the importance of balancing human recreational activities with the protection of wild mammal populations.

Jon's summary:
Newmark et al. 2023
model how linking up protected areas could slow down species loss. They use "relaxation theory" to look at how long it takes for mammal species to go extinct within a given patch of habitat due to insufficiently large and resilient populations. Figure one shows how this works - the smaller the population, the less time it takes for them to go locally extinct. They focus specifically on carnivorans (like cats, canines, bears, weasels, etc.) even-toed ungulates, and lagomorphs like rabbits. Their model assumes that linking protected areas by removing all barriers will create one much larger patch of habitat which helps wildlife survive roughly four times as long. One catch: these aren't small linkages. The proposed linkages between Glacier and Yellowstone have double the area (~2.9 Mha) than the two parks being linked (~1.5 Mha), and would connect to another 2.8 Mha of wilderness in the region. So to reframe: increasing the connected habitat area by a factor of almost 5 helps species last over 4 times as long. The Mt. Rainier North Cascades uses a linkage about 2.5* the core park size. The paper makes a strong argument for the importance of well-designed connectivity work, but the big benefits they propose would also take a really big effort.
ChatGPT's summary:
The article "Enhanced regional connectivity between western North American national parks will increase persistence of mammal species diversity" published in the journal Biological Conservation, examines the potential effects of enhancing regional connectivity between national parks on mammal species diversity in western North America. The study used computer simulations to model the effects of different connectivity scenarios on mammal species diversity, and found that enhanced regional connectivity between national parks would lead to an increase in the persistence of mammal species diversity. The study found that enhanced regional connectivity would allow mammal species to move more freely between national parks, which would increase genetic diversity and reduce the risk of local extinction. The study also found that enhanced regional connectivity would lead to an increase in the number of mammal species that are able to persist in the region, particularly for species that are currently at risk of extinction. The study highlights the importance of connectivity for the persistence of mammal species diversity in western North America and emphasizes the need for conservation efforts to focus on maintaining and enhancing regional connectivity. The article concludes by emphasizing the importance of preserving and enhancing regional connectivity between national parks for the protection of mammal species diversity in western North America. The study suggests that conservation efforts should focus on connecting national parks and other protected areas through corridors and other connectivity measures, to enhance the chances of mammal species survival. It also suggests that the management of these protected areas should be coordinated and integrated to improve the effectiveness of conservation efforts. The study highlights the importance of connectivity conservation as a way to ensure the long-term survival of mammal species diversity in western North America.

Breznau, N., Rinke, E. M., Wuttke, A., Nguyen, H. H. V, Adem, M., Adriaans, J., Alvarez-Benjumea, A., Andersen, H. K., Auer, D., Azevedo, F., Bahnsen, O., Balzer, D., Bauer, G., Bauer, P. C., Baumann, M., Baute, S., Benoit, V., Bernauer, J., Berning, C., … Żółtak, T. (2022). Observing many researchers using the same data and hypothesis reveals a hidden universe of uncertainty. Proceedings of the National Academy of Sciences, 119(44), 1–8.

Chaplin-Kramer, R., Neugarten, R. A., Sharp, R. P., Collins, P. M., Polasky, S., Hole, D., Schuster, R., Strimas-Mackey, M., Mulligan, M., Brandon, C., Diaz, S., Fluet-Chouinard, E., Gorenflo, L. J., Johnson, J. A., Kennedy, C. M., Keys, P. W., Longley-Wood, K., McIntyre, P. B., Noon, M., … Watson, R. A. (2022). Mapping the planet’s critical natural assets. Nature Ecology & Evolution.     

Newmark, W. D., Halley, J. M., Beier, P., Cushman, S. A., McNeally, P. B., & Soulé, M. E. (2023). Enhanced regional connectivity between western North American national parks will increase persistence of mammal species diversity. Scientific Reports, 13(1), 474.

Salvatori, M., Oberosler, V., Rinaldi, M., Franceschini, A., Truschi, S., Pedrini, P., & Rovero, F. (2023). Crowded mountains: Long-term effects of human outdoor recreation on a community of wild mammals monitored with systematic camera trapping. Ambio.

p.s. The photo shows my favorite kind of green tea (gyokuro) which is high in L-theanine, which in turn provides a strong umami taste and reportedly promotes a calm and alert state similar to meditation.

No comments:

Post a Comment

Questions, comments, suggestions, and complaints welcome.