Kathryn J Fiorella is an Assistant Professor in the Master of Public Health Program and Department of Population Medicine and Diagnostic Sciences at Cornell University.
This blog post is based on a recently published paper in Science Advances. Data and analysis code used in our analyses are available on Dataverse.
Extensive climate change modeling has focused on the effects of rising temperatures and changing environmental conditions on our ecosystems. Yet climate change promises to also reshape how people behave.
In response to climate change, people may shift the ways they make a living or get food. They may shift the balance of their activities, alter the strategies they use (e.g., farming different crops, fishing in different ways), or change who in the household does these activities. Ultimately, people may harvest more resources – or fewer. When we fail to account for these dynamics in climate modeling, we risk getting those models wrong.
While aggregate climate change impacts are overwhelmingly projected to be negative, in individual ecosystems the proximate impact of warming temperatures on dynamics like crop productivity or fisheries yields is ambiguous. For example, while warming temperatures from 1930-2010 negatively affected stocks of 19 fish species, they also positively affected 9 others.
Regardless of the direction of climate change impacts, those who rely directly on the harvest of natural resources in low and middle income countries are likely to bear the brunt of climate change—a phenomenon they did little to cause and have limited ability to mitigate. Understanding how these households may respond is a matter not only of accurately predicting climate impacts, but of appreciating the complexity of how climate may impact their livelihoods and food access, assessing risks they face, and best positioning support to vulnerable households and communities. Ultimately, this is a matter of equity.
To examine the question of how temperature affects people and ecosystems, we focused on small-scale fishers and fish in inland Cambodia (see Figure 1). Cambodians rely on fishing for their income, with particularly high reliance in the regions surrounding the Tonle Sap Lake that we study. Cambodians also consume some of the largest quantities of fish per capita. Additionally, Cambodia’s fisheries are among the most biodiverse globally, second only to the Amazon. In examining fishery dynamics in this setting, we found that rising temperatures had little effect on catch size, but made people less likely to fish. We find this pattern for not only fish, but also the harvest of other aquatic animals (like snakes and frogs) and aquatic plants.
We use data collected by WorldFish in the Feed the Future Cambodia – Rice Field Fisheries Phase I project that followed 400 households over 3 years (19 time points). We pair household survey data with remotely sensed data on temperature, and control for rainfall and flooding. We use a multi-part analysis to examine the effects of rising temperatures on people’s behavior and indirectly estimate the effects on availability of fish in the ecosystem. To do this, we analyzed the effects of temperature on 1) fishing participation and effort, 2) fish catch, and 3) fish catch while controlling for fisher behavior (participation and effort). The first analysis was designed to tell us if temperature affects fishing behavior. Then, by comparing the effects on fish catch with and without behavioral controls, we were able to disentangle the behavioral component of fish catch and indirectly measure the ecological component of fish catch.
By using multiple analyses and fixed effects with correlated random effects distributed lag regression models fit to repeated observations, we provide strong controls for time invariant household characteristics and identify the causal effects of exogenous temperature changes. Given the complexity of the social-ecological system, we also control for seasonal, rainfall and flood cover variation (see the full methodological details here).
We first analyzed how temperature affects household fishing participation – meaning whether or not people fish at all. We find that with rising temperatures, participation in fishing falls and that the marginal effect is larger in higher temperature ranges. Fishing participation may fall for many reasons, however, and integrating this finding with fish catch is critical to revealing the real picture of how temperature change affects households.
We next examined households’ fish catch. Catch is a function of both fishing effort, meaning ‘how hard are people trying to catch fish?’, and fish availability, meaning ‘how many fish are there to be caught?’. We did not find evidence of a net effect of temperature on fish catch. However, when controlling for fishing participation, time spent fishing, and gear choice to isolate ecological effects, we found temperature rises slightly increased fish catch. This result is comparable to an increase in catch-per-unit-effort, a metric commonly used to monitor fish stocks.
When we examined the harvest of aquatic plants and other aquatic animals, we saw a strikingly similar pattern. These resources are also widely used for food and income, and households in our study not only caught nearly 5kg of fish/week, they also harvested an average 1.3kg of aquatic animals, such as snakes or frogs, and 1.6kg of aquatic plants each week. When we controlled for harvest participation, the harvest of both plants and animals was greater than without controls, suggesting a consistent ecological and behavioral mechanism at play across aquatic food systems.
The livelihoods of Cambodian households in our study are highly complex, as is the case for many global smallholders. Households are at once engaged in rice farming, fishing, and small businesses. They often have household members who are working in nearby cities and sending money home. Given the relative increases in the productivity of aquatic systems at the same times that households are devoting less effort to harvesting fish, aquatic plants and other aquatic animals suggests that rising temperatures are pulling their attention in another direction. Although we did not test what may be driving these shifts, it is plausible that agricultural demands, such as weed pressure, or migration to urban areas may increase with temperature.
If we failed to account for fisher behavioral responses to higher temperatures, it would appear that temperature had no effect on fish catch. However, we actually find that fewer households were fishing or harvesting aquatic plants and animals. In this setting, aquatic foods are often among the most micronutrient-rich and high-quality foods in local diets. Reduced access to them could impact both incomes and diet quality for fishing families. Appreciating the full extent of how warming temperatures affect people’s behavior could ultimately be critical to understanding the effects of climate on well-being of vulnerable communities.
Our results underscore the importance of integrating behavior into climate change modeling. To accurately predict the impacts of climate change, the effects on ecological systems and the people that use them must be integrated. Incorporating human behavioral responses to changing environmental conditions will be fundamental to determining the feedbacks through which climate change affects rural livelihoods, food production, and food access.
