Urbanization drastically changes the natural environment with significant land-use changes (Berry, 2008). Urban ecosystems are also exposed to pollutants, high CO₂, nitrogen deposition, and elevated temperatures, which affect urban vegetation (Gregg et al., 2003, Zhao et al., 2016). Urban tree growth, for example, can be stimulated by warmer temperatures, particularly in temperate climates. However, high temperatures can also accelerate senescence and aging (Moser et al., 2017; Smith et al., 2019) and additional challenging conditions in cities, such as limited soil volume, modified soils, and reduced water and nutrient availability affect tree growth and performance (Day and Bassuk, 1994; Hauer et al., 2020).

The imposition of large concrete and asphalt slab restricts natural evapotranspiration in cities and considerably increases heat storage (Gui et al., 2007; Göbel et al., 2013). This redistribution of energy results in the urban heat island (UHI) effect. The UHI effect consists of an increase in the air temperature of the urban area with respect to the rural surroundings (Oke, 1995). The UHI in Mexico City, for example, can be as high as 7°C on average, but in some instances can reach intensities of up to 10°C (Ballinas, 2011).

Global climate change exacerbates urban temperature differences (Brazel and Quattrochi, 2005). In coming decades, heat shocks or waves will increase in frequency and severity and be enhanced by the UHI effect (Luber and McGeehin, 2008; Manoli et al., 2019). These predicted increases in air temperature around the world will affect the thermal balance with consequences on natural and urban ecosystems (IPCC, 2021). Undoubtedly, all organisms living in cities, including human populations, will experience more heat stress, which might decrease productivity and severely affect human health and mortality (Laschewski and Jendritzky, 2002).

With proper selection and strategic planning, urban vegetation can mitigate the UHI because of the cooling potential plants have via transpiration (Barradas, 1991, 2000; Susca et al., 2011; Ballinas and Barradas, 2016a; Kornaska et al., 2016). Some trees, for example, can transpire up to 5 L day^–1 tree^–1, which is equivalent to a heat dissipation of up to 12.62 W day^–1 tree^–1 (Ballinas and Barradas, 2016b). For this reason, urban vegetation has gained recognition for management of the UHI effect, along with other mechanisms that contribute to the improvement of the urban environment, such as the capture of atmospheric particulate material and improvement of human health (García-Sánchez et al., 2019; Keeler et al., 2019).

Plants, however, are exposed not only to global but also urban climate change, affecting key physiological functions. As a result, there are three possible plant responses to changes in climate: migration, adaptation, and extinction (Aitken et al., 2008). As cities are fragmented landscapes, species’ movement is limited (Dobbs et al., 2017); therefore, adaptation (i.e., evolutionary changes and/or physiological acclimatization) becomes crucial for species survival (Atkins and Travis, 2010).

Stomatal conductance (g_S) is a key plant trait in photosynthesis and plant productivity (Jones, 1992; Jarvis and Davies, 1998). This trait also responds rapidly to climate and environmental changes (Jones, 1992; Ackerly, 2004; Buckley, 2005). Some climatic changes, such as elevated temperatures, can negatively affect plant functions by reducing transpiration, nutrient assimilation and photosynthesis (Farquhar and Sharkey, 1982; Atkin and Tjoelker, 2003; Mathur et al., 2014). Therefore, understanding how climatic and environmental changes affect species functional responses (e.g., g_S response) becomes necessary in face of global and urban climate change.

Global climate change represents a threat to the persistence of urban forests globally (Esperon-Rodriguez et al., 2021) and these changes increase species’ vulnerability—i.e., the extent to which climate change (urban and global) can damage a system depending on its sensitivity and its ability to adapt itself to new climatic conditions (IPCC, 2001). Further, when these climate changes affect how species respond physiologically to the environment, their ecophysiological vulnerability increases, limiting their ability to adapt physiological functions to environmental changes (Esperon-Rodriguez and Barradas, 2015a). Thus, identifying vulnerable species in urban environments can help to improve species selection while ensuring the provision of desired ecosystem services toward sustainable development.

Here, we aimed to determine and evaluate the ecophysiological vulnerability of Mexico City’s urban forest to climate change. For this, we used g_S as a vulnerability indicator and four environmental variables were selected to assess the species ecophysiological vulnerability: air temperature, photosynthetically active radiation, vapor pressure deficit, and water availability. The objectives of this study were to (1) identify more and less vulnerable species to environmental changes and (2) develop a model summarizing the g_S response to the environment (i.e., the four aforementioned variables). We used the boundary-line analysis to predict stomatal responses. This analysis has great precision and uses meteorological data to assess ecophysiological responses to the environment (Jarvis, 1976; Lambers et al., 1998; Barradas et al., 2011; Ballinas and Barradas, 2016b) and can be used to identify vulnerable species (Esperon-Rodriguez and Barradas, 2015a,b).

The highest g_S was recorded for L. lucidum, followed by E. camaldulensis, whereas the lowest g_S corresponded to D. viscosa and B. cordata. The lowest ψ (i.e., more negative) belonged to B. cordata Q. rugosa and A. acuminata, and the highest ψ corresponded to P. deltoides and L. styraciflua. All species had the lowest ψ, values between 14:00 and 15:00 h (Table 2). Similarly, all environmental variables (i.e., T_A, PAR and VPD) varied among species (Table 2). We found significant differences in air temperature when we compared species among sites, finding higher T_A in the parks EP and MP (average of 24.8°C) and lower T_A in CU and VP (22.1°C) (H = 14.85, df = 7, p = 0.05). This result probably reflects the effect of the urban heat island. In contrast, we found no significant differences when comparing VPD (H = 6.350, df = 7, p = 0.15) and PAR (H = 1.356, df = 7, p = 0.153) among parks.

The parameters for each curve and each variable among the 15 species were compared (one case study is shown in Figure 2 and the parameters of the boundary-line analysis for g_S vs. T_A are presented in Supplementary Table 1). All species presented different T_O (ranging from 18.0 to 28.5°C), optimal thermal interval, and different T_I range. Populus deltoides had the lowest T_O, whereas E. camaldulensis had the highest T_O. Optimal temperature on average and across all species was 22.5 ± 3.2°C (standard deviation). Regarding T_I, P. alba (20.3°C), D. viscosa (19.9°C) and B. cordata (19.3°C) had the widest thermal range, whereas C. occidentalis (11.0°C) and E. camaldulensis (11.2°C) had the narrowest intervals across all species (Table 3).

For VPD, P. deltoides and F. uhdei reached the highest rates of g_S change, while the lowest rates were registered for B. cordata and D. viscosa (Supplementary Table 2). As for VPD_50%, for all species, the critical temperature and relative humidity values were above 33.0°C and less than 30% of relative humidity, where D. viscosa and P. deltoides presented the lowest VPD_50% and B. cordata, E. camaldulensis and U. parvifolia had the highest VPD_50% (Table 4). Concerning PAR, the highest and lowest values were recorded in A. acuminata (119.7 ± 155 μmol m^–2 s^–1) and P. alba (410.5 ± 235 μmol m^–2 s^–1) (Table 2).

Regarding vulnerability, for some species, we found wide thermal ranges (e.g., P. alba and D. viscosa); however, we highlight that although a species interval can be wide, this does not necessarily mean the species is not vulnerable. In the case of the increase in T_A, the temperature at which stomatal conductance decreases by 50% is more useful to assess species vulnerability. Here, we identified P. deltoides as the most vulnerable species, since it reaches its T_50% at 27.1°C (Table 3). We considered less vulnerable, those species that present their T_50% at least at 30°C. The species that presented the highest T_50% were E. camaldulensis and F. uhdei (Table 3).

Leaf water potential was used to analyze tree vulnerability to water availability. Based on the ψ when stomatal conductance decreased by 50% (ψ_50%), the most vulnerable species were P. deltoides, L. styraciflua and C. occidentalis, as these species had the highest (i.e., less negative) values. In contrast, the least vulnerable species were B. cordata, Q. rugosa, and E. camaldulensis, which also coincide with their lowest ψ values (Table 5). Concerning VPD and PAR, vulnerability was assessed using the slope of the hyperbola (or sensitivity) given by the constant β and B, respectively, in which the more vulnerable species were those with the highest β and B values. For VPD, the most vulnerable species was P. deltoides (β = 1367.42), whereas B. cordata (β = 136.01) was the least vulnerable. As for PAR, we identified B. cordata (B = 40.21) and L. lucidum (B = 2.08) as the most and least vulnerable species, respectively (Supplementary Table 2).

The performance of the g_S response to all variables is presented in Table 6. For U. parvifolia, the proportionality kT_A was higher than 1, probably caused by recording high g_S at elevated T_A. We found P. deltoides and F. uhdei presented negative proportionalities (kψ), indicating high vulnerability. Further, P. deltoides presented the lowest proportionality values in all environmental variables, except PAR, indicating that this is the most vulnerable species to climate change. According to the predictions of g_S using the multiplicative model, survival of P. deltoides and F. uhdei might be not possible in urban environments. From our results, it became evident that the effect of ψ was restrictive in all species, except for Q. rugosa, which showed good performance (Table 6).

We used the boundary-line analysis to identify vulnerable species to environmental changes based on their stomatal response. The approximation of the boundary line functions of each variable through the determination coefficient (r²), indicated that these functions were adequate for determining the environmental effect on the g_S response. This response showed a higher g_S sensitivity to VPD and ψ compared to T_A and PAR. Differences in stomatal sensitivity found here may be caused by different microclimates at each study site, which could cause acclimatization of stomata to the urban environment (Barradas et al., 2016). Stomatal responsiveness could also be an adaptation (i.e., protective of water loss) related to different management practices (e.g., irrigation) at each site.

The multiplicative model allowed us to identify the most (i.e., P. deltoides and F. uhdei) and least (i.e., E. camaldulensis, U. parvifolia and A. acuminata) vulnerable species to environmental changes. It is worth noting that P. alba had a better g_S performance than P. deltoides, showing that other factors rather than phylogeny, may improve species performance, such as plasticity and climate of origin (Nicotra and Davidson, 2010; Du et al., 2019). However, studies in tropical areas showed that phenomenological models, such as the boundary line analysis, are site-specific (Barradas et al., 2004). Also, for the multiplicative model, we used input data derived from meteorological stations from Mexico City; a different approach might be using modeled meteorological data (e.g., PRISM). Our results, therefore, are site- and species-specific, and caution must be taken if the parameters are extrapolated to other sites and individuals under different conditions.

As a response to climate change, global warming will continue toward the end of the century increasing air temperature in urban areas (Esperon-Rodriguez et al., 2021; IPCC, 2021). The average air temperature in Mexico City is expected to be 24.5°C by 2050 (Ballinas, 2011) and daytime temperatures will likely exceed the species’ temperature ranges found here. While the annual average air temperature in Mexico City keeps increasing, extreme temperatures will be aggravated by the UHI effect negatively affecting some species by limiting their stomatal function. In this study, we found that temperatures above 27°C limited the stomatal opening, although some species could maintain stomata open (T_50%) at higher temperatures. This was the case of E. camaldulensis and F. uhdei, which had g_S associated with T_50% of up to 39°C. In contrast, species such as P. deltoides, had the lowest T_50% at 27.1°C. Additionally, changes in precipitation and cloudiness, which affect solar radiation, may also affect the stomatal function with effects on species growth and performance.

A direct effect of the increase in air temperature is the increase in VPD when water is limited and relative humidity is low. The response of g_S to high temperatures can be uncertain, particularly when these temperatures are coupled with low relative humidity and high VPD (Medlyn et al., 2002; Esperon-Rodriguez and Barradas, 2014). The increase in VPD restricts g_S, thus, species capable of maintaining stomata open when VPD is high are less vulnerable (i.e., Q. rugosa, E. camaldulensis and B. cordata). In contrast, species with low g_S under high VPD conditions are more vulnerable (i.e., E. americana, A. negundo, L. styraciflua, and P. deltoides). From our analysis, we found all species were vulnerable to changes in VPD and ψ indicating sensitivity to water stress conditions. However, caution must be exercised when interpreting the proportionalities in the model because if two proportionalities are negative, for example, the value of g_SMOD is positive. As for PAR, its effect was less significant compared to the other variables (i.e., VPD, ψ) because C₃ plants become saturated toward 300 μmol m^–2 s^–1 (Jones, 1992), hence, g_S would be limited by PAR below the saturation point; that, however, did not occur in our study.

Finding a high vulnerability to VPD and ψ, two variables directly related to water availability, highlights the importance of water in urban environments to keep stomatal function. Water is crucial to maintain g_S for leaf cooling via evapotranspiration, and therefore is associated with the cooling benefits of trees even under extreme high temperatures (Drake et al., 2018). Leaf temperature is an important parameter; a decrease in g_S can be associated with increases in leaf temperatures (Lu et al., 1994) and leaf temperatures can exceed air temperature by several degrees (Miller et al., 2021) leading to leaf damage (Hueve et al., 2011). Thus, identifying resistant species to high temperatures in water-limited environments is beneficial in urban environments because canopy loss can have negative environmental and social impacts (e.g., decrease in cooling benefits) (Sanusi and Livesley, 2020).

Here, we used leaf water potential as an indicator of plant resistance to drought (Fanjul and Barradas, 1987; Jones, 1992). Furthermore, decreases in ψ can reduce g_S (Fanjul and Barradas, 1985, 1987; Esperon-Rodriguez and Barradas, 2015b). When water is limited, species that respond by decreasing their ψ are less vulnerable. This was the case of B. cordata, Q. rugosa, A. acuminata, and E. camaldulensis, which presented ψ lower than –2.0 MPa. In contrast, the most vulnerable species were P. deltoides and L. styraciflua. For L. styraciflua, previous research showed its vulnerability to water-limited natural environments (Esperon-Rodriguez and Barradas, 2015b), and here, we confirm its vulnerability also in urban environments. To confirm the species vulnerability to ψ, we recommend using the leaf turgor loss point as an additional trait. The less vulnerable species would be those that are able to maintain ψ above their turgor loss point under water stress conditions, allowing gas exchange and evaporative cooling to continue.

Our results can aid species selection in a changing climate to minimize economic losses related to urban tree failures. By working together, scientists, policymakers and urban planners can develop and implement strategies and management actions that not only guarantee the preservation of urban forests but also benefit local communities.

The datasets presented in this article are not readily available because when data is generated for an investigation, by law it is reserved for 10 years in Mexico. Requests to access the datasets should be directed to corresponding author.

VB and ME-R designed the research, collected, analyzed the data, and wrote the manuscript. Both authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.