Preliminary evidence for differential habitat selection between bird species of contrasting thermal-tolerance levels

///Preliminary evidence for differential habitat selection between bird species of contrasting thermal-tolerance levels

Preliminary evidence for differential habitat selection between bird species of contrasting thermal-tolerance levels

2023-06-02T16:34:48-07:00 June 3rd, 2023|Biology, Environment|

By Phillips.

Author’s note
: Since coming to college, I have wanted to conduct research on the environmental impacts of agriculture and contribute to efforts to make farming work for both people and nature. In pursuit of this goal, I signed up as an intern with Daniel Karp’s agroecology lab in my freshman year and stayed with them for my entire undergrad. During this internship, I worked alongside several Ph.D. students, such as Katherine Lauck and Cody Pham, who research the cumulative effects of land conversion and climate change on native avifauna at Putah Creek. I was so inspired by their work that I decided to conduct an independent project investigating similar phenomena. Specifically, I was curious about how birds respond to temperature across multiple landscapes, and how this pattern of behavior might influence their choice of habitat. While reading this paper, I would like you to consider the broader implications of the findings as they pertain to species conservation in the context of climate change.


Increasing frequency and severity of temperature spikes caused by climate change will disproportionately impact heat-sensitive species. However, certain types of vegetation may protect animals from temperature spikes. Heat-sensitive species can retreat to shaded microhabitats when temperature increases, allowing them to avoid detrimental effects on fitness. Here, we examined habitat selection and behavioral responses to temperature of Western Bluebirds (Siailia mexicana) and Northern Mockingbirds (Mimus polyglottos). We conducted transect surveys and collected behavioral data on bird movement for two months in riparian forest and perennial cropland in the Central Valley of California, where breeding season temperatures are often above 35°C. Bluebirds were observed more frequently in shaded riparian forest, while mockingbirds were observed more frequently in exposed agricultural fields. Correspondingly, bluebirds became less active at higher temperatures, while mockingbirds exhibited no response. Together, our results imply that heat-sensitive species may be more likely to select natural or semi-natural habitats and change their behaviors when temperatures spike. The results of this study imply that the combined effects of anthropogenic land development and climate change may be more destructive for heat-sensitive species than for heat-tolerant species.


Climate change is increasing the frequency and intensity of temperature spikes across the world [1]. Many species will likely experience increased mortality due to these extreme conditions [2–4], with heat-sensitive species experiencing especially detrimental effects [5,6]. However, thermally-buffered habitats could mitigate the impact of heat spikes on organisms, as certain habitat features, like vegetative cover, have been shown to cool local temperatures through shading and evapotranspiration [7,8]. Landscapes with high amounts of thermally-buffered habitats, such as closed-canopy forests, have been shown to have less dramatic temperature extremes than open habitats [9,10]. Furthermore, it has been shown that animals in these thermally-buffered habitats are less likely to be impacted by rising global temperatures [11,12]. As such, organisms that are sensitive to temperature extremes may preferentially select for these habitats, and therefore may be able to avoid potentially lethal effects. Birds have been observed to retreat to shaded habitats when temperatures spike [13]. However, it is unclear whether heat-sensitive species specifically select for thermally-buffered habitats, or if heat-tolerant species persist in non-buffered habitats. Therefore, we sought to understand how the habitat selection of bird species may be associated with their behavioral responses to temperature.

Bird populations in North America are in rapid decline [14], and are predicted to continue declining with climate change [15]. As such, determining the habitat requirements of birds in response to increasingly extreme temperatures could be crucial to their conservation. We conducted behavioral surveys of birds in the Central Valley of California to address two questions: 1) does habitat selection differ between Western Bluebirds and Northern Mockingbirds, and 2) are behavioral responses to temperature different between these species? We hypothesized that birds species which exhibit significantly different behavior during high temperature will preferentially select habitats with more vegetation cover.


Experimental design

We selected two sites along Putah Creek in the Central Valley of California. In this system, temperatures often reach 35°C during the hottest months of the year. These sites contain a combination of riparian (forest existing along a river bank) and agricultural land and are approximately five miles apart from each other. At each site, the two focal land cover types–riparian forest and perennial agriculture–were present within one half-mile of each other (Figure 1). We obtained observations along four 100 meter (m) transects. In the riparian areas, we placed transects along regions of the sites where vegetation was sparse enough that birds could be observed, as dense vegetation made it difficult to track the individual birds. In the agricultural areas, we placed transects along areas that were close enough to the crops that birds could be spotted. Transects were placed approximately 50 meters apart from each other. 

We focused on Western Bluebirds (Siailia mexicana) and Northern Mockingbirds (Mimus polyglottos) due to their high abundance at Putah Creek. Additionally, we chose these species because they forage on the ground rather than in the air, and therefore were easier to observe with the naked eye.

Figure 1. Our two study sites were in close proximity to both riparian and agricultural habitats along Putah Creek. At each site, we observed birds along a total of 16 transects (depicted in red).

Data collection

We conducted our surveys from late April to early July 2022, the height of the breeding season for our study species. We visited each site at least once a week, in either the morning or early afternoon. During each visit, I would walk along the transects. Once a bird of either target species was spotted, I would track the bird for two minutes and record all behaviors displayed, along with the amount of time spent engaging in each behavior. These behaviors included “foraging” (searching for, chasing, or eating an insect), “moving” (locomotion with wings or legs), “resting” (standing or sitting motionless), “singing” (repetitive vocalization for more than three seconds), “preening” (use of beak to position feathers), and “disputing” (fighting between birds that occurs due to territorial disputes). I recorded temperature and wind speed each hour using a Kestrel 2000 Weather Meter.

Data analysis

We ran Fisher’s exact tests to determine if mockingbirds and bluebirds preferentially selected different landscape types across sites. The variables in this model included ‘species,’ and ‘landscape type,’ which was defined as either “Agriculture” or “Riparian.” We ran the model across both sites and did not distinguish between the two separate sites depicted in Figure 1.

Then, we implemented multiple linear regression models examining the relationship between the time spent engaging in various behaviors and temperature for each species. We considered the time spent engaged in a particular behavior to be the percentage of time during the two-minute observation period in which the individual bird exhibited that behavior (i.e., time spent moving, foraging, resting, preening, singing, or disputing).

To account for the effects of spatial autocorrelation (or the tendency of areas which are close together to provide similar data values), we first included a site covariate in our models. We additionally attempted to control for the effects of a natural circadian rhythm on behavior by including a time-of-day covariate. As temperature and time were highly correlated (r = 0.696 for bluebird observations and 0.548 for mockingbird observations), we included these covariates using a temperature residual approach. Specifically, we regressed time against temperature and obtained residual values, representing whether temperatures were hotter or cooler than the average expected temperature at any given time of day. We then ran a multiple linear regression including the effects of temperature residuals, time of day, and site on bird behavior.


Landscape preference

Bluebirds and mockingbirds exhibited significantly different habitat preferences. Bluebirds preferentially resided in riparian areas, whereas mockingbirds preferentially resided in agricultural landscapes across both sites (Fisher’s exact test, p = 4.583E-15; Figure 2).

Figure 2. Mockingbirds (n=34) are observed to reside in agricultural landscapes more frequently than riparian landscapes. Bluebirds are observed to reside in riparian landscapes more frequently than agricultural landscapes (n=35).

Changes in patterns of behavior

We found that temperature negatively affected the amount of time that bluebirds spent moving (Linear regression, p = 0.0077, F = 8.069, df = 1, 33; Figure 3; Supp. 1). However, temperature did not significantly affect mockingbird movement (Linear regression, p = 0.297, F = 1.125, df = 1, 32; Figure 3; Supp. 1).

Results were broadly similar after including ‘site’ as another effect in the model to account for multiple observations at the same location. Specifically, temperature still did not affect mockingbird movement (Multiple regression, p = 0.635, F = 0.577, df = 3, 30; Supp. 3) and marginally affected bluebird movement (Multiple regression, p = 0.0682, F = 2.622, df = 3, 31; Supp. 3). However, one of the sites had very few bluebird observations (n=4); when this site was removed from the model, temperature again negatively affected bluebird movement (Linear regression, p = 0.0123, F = 7.138, df = 1, 29; Supp. 2).

The last model we ran tested the effects of both temperature residuals and time of day on bird behavior. Using these models, temperature again did not have a significant effect on the behavior of mockingbirds but did have a marginal effect on bluebird movement (p = 0.07; Supp. 4).

For all of the models, resting, foraging, disputing, singing, and preening of bluebirds and mockingbirds exhibited no significant association with any environmental variable (Supp. 1, Supp. 2, Supp. 3, Supp. 4).

Figure 3. Bluebirds (left) are observed to reduce the percentage of time they spend moving as temperature increases. Mockingbird movement (right) did not significantly decline with rising temperature. The black points represent individual bird observations, the solid lines represent the linear model predictions, and the gray bands represent the 95% confidence intervals.


Our results suggest that bluebirds select for shaded riparian habitats, while mockingbirds select for exposed agricultural habitats. Correspondingly, the temperature-altered patterns of movement in bluebirds suggest that they are sensitive to heat and may potentially select for thermally-buffered habitats as a result. In contrast, a lack of observed heat sensitivity in mockingbirds suggests that persistence in open habitats could in part be driven by thermal tolerance. While more data are required to make definitive conclusions, considering only patterns at our site with sufficient data, we found significant evidence for temperature-altered patterns of movement. Together, these results suggest that temperature sensitivity could drive patterns of habitat selection.

Previous research also suggests that habitats with low vegetative cover (i.e., without thermally-buffered microclimates) are likely to contain heat-tolerant species [16,17]. For example, Wilson et al. 2007 demonstrated that populations of leaf-cutter ants (Atta sexdens) residing in cities took 20% longer to succumb to high temperatures than ants dwelling in rural areas. In Brans et al. 2017, it was observed that water fleas (Daphnia magna) from urban areas were more tolerant to high temperatures than rural populations, partially because they had smaller body sizes. Both studies imply that organisms must have high heat tolerance to live in habitats with low vegetative cover. This is similar to our finding that mockingbirds, a heat-tolerant species, were more likely to reside in unvegetated agricultural landscapes than were bluebirds, a heat-sensitive species. However, while the previous studies provide evidence that organisms become heat-tolerant in these landscapes due to natural selection, our findings suggest that behavioral differences between heat-tolerant species and heat-sensitive species may also cause unvegetated landscapes to become dominated by heat-tolerant species.

Additionally, we demonstrate that riparian and other thermally-buffered habitats could be crucial to the persistence of heat-sensitive species. Other studies have shown that vertebrates are more likely to exhibit heat-related mortality in habitats with low vegetation cover [12,18]. For example, Zuckerberg et al. 2018 demonstrated that avian survival in small grassland patches was negatively associated with temperature, while survival in large grassland patches was not. Additionally, Lauck et al. 2023 showed that temperature spikes are associated with a decline in bird reproduction across the continental United States for organisms living in agricultural areas, but not for organisms living in forests. These results suggest that vegetation protects vertebrates from heat stress. Although the mechanisms of this protection are not clear, one potential explanation is that vegetation provides shaded areas that animals can use as refuges to avoid lethal temperatures [7]. Additionally, it has been shown that plants regulate local temperatures through evapotranspirative cooling [8], potentially playing a role in protecting vertebrates from heat spikes.

One caveat of our study is that bluebird responses were only marginally significant under multiple regression models that included time of day as a covariate. Associations between bird behavior and time could either be due to circadian rhythms or temperature shifts; it is difficult to statistically disentangle the effects of temperature and time of day. However, the significant results from the models including only temperature imply that bluebirds do indeed alter their behavior in response to environmental factors that likely include temperature.


Our findings provide preliminary evidence that Western Bluebirds are temperature-sensitive and preferentially select vegetated habitats, while Northern Mockingbirds do not preferentially select vegetated habitats. To obtain enough data to provide definitive evidence of these patterns, the methods could be repeated for several more years and across more sites. Nonetheless, the results from this study suggest that anthropogenic land development will be more destructive for heat-sensitive species than for heat-resistant species. As such, we suggest incorporating thermally-buffered habitats such as groups of trees or hedgerows in working landscapes to mitigate the negative impacts of anthropogenic land development on heat-sensitive organisms.