Tree taxa selected for the study represented a broad range of leaf-and growth characteristics, including evergreen/deciduous, needle-leaved, scale-leaved and broadleaved trees. The selected tree taxa had growth characteristics that represent commonly grown trees in UK gardens, and are moderately drought tolerant, without known “disservices” (e.g., allergenicity, emission of volatile organic compounds, or high susceptibility to invasive pests and diseases at the time of selection), according to peer-reviewed literature. An online tool (“My Garden”), managed by the Royal Horticultural Society (RHS), holds the data on over five thousand plant taxa registered by garden owners as being grown in their UK private gardens, and was used to obtain an overview of frequently planted tree taxa in domestic gardens. Advice on taxa was also sought from Frank P. Matthews’s nursery (Tenbury Wells, UK), one of the UK’s largest and oldest tree nurseries. This approach identified taxa within the genera Magnolia, Malus and Prunus, as important garden tree representatives in the UK. Additionally, Pyrus calleryana was chosen as it has been studied previously in an urban context for its cooling capacity (Rahman et al., 2014) and was considered a good reference. Picea and Cupressus were selected to represent two types of evergreen conifers with different morphological traits, one needle-leaved and another scale-leaved. The final selection comprised the following taxa: Crataegus persimilis ‘Prunifolia Splendens’, Cupressus sempervirens ‘Garda’, Ilex aquifolium ‘J.C. van Tol’, Magnolia ‘Galaxy’, Malus ‘Scarlett’, Picea pungens ‘Erich Frahm’, Prunus ‘Fugenzo’, P. calleryana ‘Chanticleer’ and Sorbus ulleungensis ‘Olympic Flame’ (Table 1). Because these garden trees are either cultivars or hybrids, none of them are classified as native.

As this study included nine different genera, and Magnolia ‘Galaxy’, Malus ‘Scarlett’ and Prunus ‘Fugenzo’ do not have species epithets, as this is not always included in hybrid taxa (Cubey et al., 2023; Cultivated Plant Code, 2016), each taxon will henceforth be referred to by the genus.

Fifty-four trees (six replicates of each of the nine chosen taxa), were obtained from FP Matthews’ nursery (Tenbury Wells, United Kingdom) in January 2021 and subsequently transplanted into 130 L containers with Sylvamix bark-based growing medium with 10% loam added in (Melcourt, United Kingdom). All trees were 4–5 years old at the time of transplanting. During the study period, an organic fertiliser (Ecor 5 Organic Fertiliser 8-5-6, Fargro, United Kingdom) was added as a top fertiliser every sixth month.

The trees were grown outdoors in the period February 2021 – January 2023 on geotextile-covered ground (Mypex, Don and Low, United Kingdom) in a randomised block design (3 × 3, 3 m apart) at the RHS’s Field Research Facility (51°19′21“ N, 0°28.23” W). The site has a northern temperate climate, with a mean annual maximum temperature of 15.4°C, a monthly average temperature ranging from 5.3°C (January) to 17.9°C (July) and mean annual precipitation of 667.9 mm (Met Office, 2025b). Trees were irrigated by means of naturally occurring rainfall, an automated drip irrigation system (4 drippers per container), and supplementary manual watering when required, to avoid any water stress. To aid with the maintenance of optimal root-zone moisture levels, starting at planting time, continuous measurements of substrate moisture were obtained with substrate moisture probes (SM150T, Delta-T Devices, United Kingdom) in two containers per taxon. Additionally, weekly point measurements were taken 5 cm from the bottom of the container the second year (by inserting probes through two custom made holes in the containers which fit the probe’s sensors), to determine whether additional irrigation was necessary (Supplementary Figure A1). Substrate moisture content was maintained above 0.2 m³ m⁻³ to avoid water stress, while drippers were removed when the moisture concentrations exceeded 0.6 m³ m⁻³, to avoid waterlogging.

As the focus of this experiment was to quantify the trees’ response to high ambient temperature and the subsequent cooling potential, we selected July, the warmest month of both 2021 and 2022, as the period of transpiration and cooling assessment for this study.

A weather station (WS-GP2 Advanced Automatic Weather Station System, Delta-T Devices, United Kingdom) was installed adjacent to the experimental plot (<20 m). Hourly totals of rainfall (mm) near ground-level, and hourly values of air temperature (°C), relative humidity (%), wind speed (m s⁻¹), radiation (W m⁻²), soil temperature (in the surrounding ground, not the substrate of the pots, °C) and atmospheric pressure (kPa) at 2 m height were recorded throughout the duration of the experiment. Hourly values were collated into daily and monthly values and were used to calculate vapour pressure deficit (VPD, kPa) (Allen et al., 1998).

Heat pulse velocities (cm h⁻¹) were measured continuously using the SFM1 sensors (SFM1, ICT, Australia). One SFM1 sensor was installed on the northern side of the trunk of three trees per taxa, at 20–30 cm height above substrate surface (below the first branch) to allow uniform comparison between the trees, with some taxa (i.e., Cupressus) having branches growing closer to the ground than others. A plastic spacer of around 0.8–1.0 cm, depending on the thickness of the bark of the tree, was fitted over the probe so that the measuring point of the probe was positioned at 0.2 cm and 2 cm into the wood. However, on inspection, the data collected from the outer part of the wood was less noisy and more consistent. This could have been due to the size of the trees, with some of the smaller trees (i.e., Ilex aquifolium ‘J.C. van Tol’), not being larger than 2 cm in radius. Consequently, the data used for this study was taken from the outer 0.2 cm part of the sapwood. All probes were covered with reflective foil to minimise the effect of solar radiation on the measurements.

Total transpiration per tree was converted into grams per second and multiplied by the latent heat of vaporization factor (2.45 kJ g⁻¹ at 292K and atmospheric pressure) (Wagner and Pruß, 2002; Moss et al., 2019; Rahman et al., 2019) to estimate energy absorption, Q (kJ) associated with tree transpiration (e) (Equation 6): Q = s a p   f l o w   g s − 1 λ t (6) Where λ is the latent heat of vaporization per unit time (s). Transpirational cooling effect (Tr_c, °C) was further obtained following Equation 7, from Nakazato and Inagaki (2012) and Huang et al. (2022): Tr c = Q ρ ∗ c p ∗ c v o l (7) Where ρ   i s   t h e   d e n s i t y   o f air (kg m^-3), c_p is the specific heat capacity of air (kJ kg⁻¹ C⁻¹), and C_vol (m³) is the crown volume calculated, using a paraboloid volume equation, using measurements of canopy length, depth and height. Volume of the main stem was calculated using the equation of a cone, using the radius at ground level and total height as input measurements. The same was done with each primary branch: girth and length was measured for each individual branch. Branches smaller than 10 cm length and 5 mm diameter were excluded.

All statistical analyses were performed using R studio (R Core Team, 2021). To assess the differences between biomass growth and crown volume, a Two Sample t-test was used (p < 0.05). A post hoc Tukey test was performed to detect differences between taxa.

To investigate the relationship between transpiration and meteorological drivers, a linear mixed model was fitted to the sap flux measurement of all nine taxa using the “brms” package to account for autocorrelation (Bürkner, 2017). The “brms” uses a Bayesian method and gives a posterior distribution and a 95% confidence interval for the estimated outputs (Korner-Nievergelt et al., 2015). Taxa and their interactions with each driver were used as predictors. Because of the high temporal autocorrelation in the data, i.e., data points close together in time displayed the same results, we decided to select 3 h, and 5 h apart, to best understand the daily influence of the environmental drivers on transpiration. Three different models were fitted, each representing one specific hour of measurements: 08:00, 13:00 and 18:00 BST, in July during both years. Using both years allowed sufficient replicates (Supplementary Tables A1, 3, 4). The 08:00 data was used to assess sap flux at a period when rates were increasing rapidly, the 13:00 data to assess transpiration at peak air temperatures, and the 18:00 data to assess transpiration in the evening when rates were declining.

Each of the three models used taxa and block as random effects. Fixed effects were air temperature (AT), squared expression of air temp (AT²), relative humidity (RH), soil temperature (ST), year (YR - 2 levels, 2021 and 2022), wind speed (WS), radiation (RAD) and a squared expression of radiation (RAD²). The squared expressions were included because of the non-linear relationship between AT and RAD and sap flux and including a squared relationship of these parameters improved the model fit.

All predictors were standardised before the model fit, thus all variables had a mean of zero and a standard deviation of one. Standardisation was applied to make effect sizes comparable among variables with different units. After inspection of the residual distributions of each parameter, RH was transformed by calculating the arcsine of the mean. This transformation avoids that a single observation has too much influence on the result (Stahel, 2002). However, we show the effect plots with the raw data for an easier interpretation of the results (Figure 1).

For the predictive model check, Shinystan (Stan Development Team, 2017) was used and thereby, standard deviation and the mean, min and max of the data were compared with those statistics in replicated data from the model. Using the posterior distribution from the “brms” package, predictive plots were generated to show how the average sap flux changed with a specific meteorological driver (relative humidity, solar radiation and temperature), while all other variables in the model were held constant (Figure 1).

Overall, the growing season (May–October) was warmer and drier in 2022 than in 2021 (Supplementary Table A5). July had an average daily maximum air temperature of 22.8°C in 2021, compared to 25.5°C in 2022. Total rainfall at the experimental site in July 2021 was 92.4 mm, and only 7.6 mm in 2022. At 08:00, the VPD did not surpass 1.9 kPa during July for either year, while at 13:00 it reached 4.9 kPa and at 18:00, 3.9 kPa (both during the second year). The highest air temperatures were generally recorded after midday and reached a maximum of 37.1°C on one occasion at 13:00 in July 2022, while corresponding radiation was 1,019.0 W m⁻². The morning (08:00) and evening (18:00) radiation averaged 404 W m⁻² and 429 W m⁻² respectively. Wind speed was routinely highest in the afternoons. Mornings had the highest minimum RH (78.7% in 2021, 48.5% in 2022) compared to a minimum of 22.7% at 13:00% and 26.0% at 18:00. Correspondingly, mean sap flux across taxa reached 18.6, 24.4 and 19.4 cm³ cm⁻² h⁻¹ at 8:00, 13:00 and 18:00, respectively; highest transpiration across taxa was found at 13:00, when air temperature, RH and radiation were highest (Supplementary Figure A1; Supplementary Table A5).

Low relative humidity was the strongest driver for sap flux across taxa and time of day, and the effect was double as high at 13:00 compared to morning and evening (Table 2). Air temperature and radiation had an increasing effect on sap flux up to a threshold, beyond which sap flux plateaued, although this threshold depended on the time of day and taxa (Figure 1; Tables 2, 3).

Across both years, at 08:00, Malus, Crataegus and Pyrus, respectively, had significantly higher sap flux rates than the remaining taxa, while Cupressus and Picea had significantly lower sap flux rate than the remaining taxa (Table 3; Figures 1, 2). An increase in radiation had a smaller effect on the sap flux of Cupressus, Ilex and Picea, respectively, and a greater effect on the sap flux of Malus and Pyrus (Supplementary Table A2). At 13:00 and 18:00, only Crataegus and Picea significantly deviated from the mean sap flux in opposite directions; Crataegus transpiring more than the average, and Picea transpiring less (Supplementary Figure A2; Supplementary Tables A3, 4).

At 08:00, all other drivers being equal, the rise in air temperature had a smaller effect on Cupressus and Picea compared to remaining taxa (Supplementary Figure A1; Supplementary Table A2). In the morning, sap flux of Malus and Magnolia started to plateau at air temperatures above 25°C, while the rest of the taxa kept increasing their sap flux in response to increases in air temperature at this hour (Figure 1; Table 4). An increase in RH decreased sap flux across taxa, although it had a smaller effect on Cupressus and Picea, and a greater effect on Malus at 8:00 and Crataegus at 18:00 (Supplementary Tables A2, 3).

At 13:00, peak sap flux occurred for eight of the nine taxa when air temperatures were between 28°C–30°C (Figure 1; Table 4). Only Prunus had peak sap flux at a higher temperature (around 35°C, but with no upper limit detected, Table 4). At 13:00, radiation had a slightly stronger increasing effect on Prunus and a slightly smaller effect on Cupressus (Supplementary Figure A1; Supplementary Table A3).

Air temperature that corresponded to maximum sap flux was higher in the evening than at midday, averaging 33°C at 18:00, with no upper limit detected across taxa, and with no significant differences detected between taxa (Figure 1; Table 4). An increase in RH however, had a stronger negative effect on the sap flux of Crataegus, which had the highest transpiration at this hour, and a less negative effect on Picea, with lowest transpiration at this hour (Supplementary Figures A1, 2; Supplementary Table A4). The effect of the wind speed or “year” was not assessed for individual taxa, as no general difference was traced for these variables.

Scaling up sap flux density to whole-tree transpiration (L tree⁻¹), the taxon with the highest water-use alternated throughout the month both years (Figures 3, 4). In 2021, Prunus had the highest daily water-use average (1.6 L), Crataegus 2nd (1.5 L) and Pyrus 3rd (1.2 L). In 2022, Pyrus had the highest water-use with a daily average of 3.7 L, Prunus 2nd (3.5 L) and Malus 3rd (3.1 L). In 2021, Magnolia had the lowest water-use (0.4 L), while the second year both Ilex and Picea had the lowest of 1.1 L (Table 5). Cupressus however, had by far the highest water-use per crown area (5.4 L m⁻²) second by Pyrus (2.4 L m⁻²), and consequently Cupressus had the highest transpirational cooling effect (Figure 7). Due to its compact shape, Ilex was the next most effective at cooling, while Pyrus also had a large cooling effect, due to a combination of a high water-use and a compact canopy shape (Figures 6, 7).

In general, sap flux densities were higher in July 2022 compared to July 2021, due to 2022 being both warmer and drier, and consequently exhibiting higher VPDs. Sap flux response to environmental drivers differed significantly between taxa and time of day, resulting in differences in water-use and transpirational cooling effect (Table 5; Figure 7). Overall, the differences in sap flux response between taxa were greatest in the morning; when Malus, Crataegus and Pyrus, had a higher sap flux rate than the average, and Cupressus and Picea lower (Table 3). Although Malus had the highest sap flux density at 8:00, it also had the steepest decrease in response to peak-air temperatures and increasing RH at this hour. This may be associated with its darker scarlet leaf colour (Supplementary Figure A2), leading to overheating of the leaf surface and a closing of stomata to prevent excessive water loss. Leaf colour or leaf lightness is linked to the leaves’ reflectance, with a darker leaf colour decreasing short-wave reflectance, and therefore increasing temperature, while the opposite is true for a lighter colour (Ferguson et al., 1973; Vaz Monteiro et al., 2016). Malus also had the steepest increase in sap flux in response to solar radiation in the morning, a trait that could be important for soil moisture reduction, with this taxon effectively utilising spells of sunshine between rainfall, common in the temperate climate of the UK. This taxon can arguably grow well in most conditions in the UK; effectively using radiation when available, while reducing water-use during peak-temperatures. However, a cultivar with similar habit, but greener/more lightly coloured leaves would be better suited for warm urban areas avoiding inhibition of photosynthetic efficiency (and consequent reduction in growth) due to overheating.

In addition to Malus, Crataegus had the strongest response in sap flux to a change in RH and air temperature. This strong response translated into high water-use during favourable conditions, effective removal of soil moisture and higher transpirational cooling, while reducing water-loss during peak-temperatures. Neither Crataegus nor Malus had the largest woody biomass accumulation, but were superseded in these characteristics by Prunus, Pyrus and Sorbus (Figure 5). Due to the higher woody growth rates, and consequently larger sapwood areas, both Pyrus and Prunus had higher total transpiration volumes than Crataegus and Malus, with Pyrus having a higher transpirational cooling potential (Figures 3, 4, 7). However, the large crown volume of Malus, show that this taxon put their resources into growing the crown rather than the main stem, with a large crown being key for cooling through shading.

In the afternoon and evening, wind speed was generally higher than in the morning and was positively correlated with sap flux. In the morning, with wind speed averaging <2.5 m s⁻¹, no effect was detected on sap flux rates, demonstrating that the effect of wind was only relevant at higher speeds (Tables 2). This could be important under future climates, with an expected increase in storms and higher average windspeeds. However, no difference was found in how the sap flux of the taxa fluctuated with the wind but rather needs to be seen together with other environmental drivers, as having an additional increasing effect on sap flux when values reach 3–4 m s⁻¹ (Supplementary Tables A2, 5).

In temperate regions, cooling through transpiration has been estimated to have a significant impact on human thermal comfort (Leuzinger et al., 2010; Zou and Zhang, 2021). Cupressus had the highest transpirational cooling effect relative to size (Figure 7). This was due to Cupressus having a very narrow canopy shape (Figure 5), however sap flux density of Cupressus was lower than the deciduous taxa (Figure 2). In addition, there was a smaller change in sap flux for Cupressus and Picea in response to increases in air temperature, decreases in RH and increases in radiation, compared to the rest of the taxa. This could be due to the characteristics of scale- and needle-leaves: sunken stomata and a tough leaf surface, which gives a stronger protection from the surrounding environment and, to some extent, reduced water loss (Hirons and Thomas, 2018). This could put these taxa at a disadvantage when attempting to respond to low and changeable levels of solar radiation (Davies et al., 2009; Goddard et al., 2010). Together with Ilex, Pyrus also showed a high transpirational cooling potential, from the combination of high sap flux and a compact crown shape (Figures 2, 7). Pyrus would be a more efficient choice due to having a larger canopy, which could also provide more direct shading than Cupressus and Ilex (Figure 5). Pyrus calleryana ‘Chanticleer’, grows to be a relatively larger garden tree (12 m high and 4–8 m wide) (RHS, Pyrus, 2025a), compared for instance to Ilex ‘J.C. van Tol’, which is estimated to grow to 4–8 m height (2.5–4 m wide) (RHS, Ilex, 2025b), and Cupressus, also estimated to grow 12 m high, but as with the Ilex, only 2.5–4 m wide (RHS, Cupressus, 2025c). This is important in a garden setting, as Ilex would fit into smaller places, and would be an efficient cooler relative to size (Figure 7). Ilex had the smallest stature and lowest growth rate of all the taxa, so a high sap flux rate was somewhat unexpected. However, it is well adapted to a temperate climate and due to its evergreen nature would be a fitting plant choice for smaller gardens, where a less vigorous, but high water-using species is sought.

The second year of the study included a period of unusually high temperatures (4 days of maximum air temperatures above 30°C), providing insight into the water use of each species under heat events expected to become more frequent in the future (IPCC, 2023). At 13:00, Crataegus, Cupressus, Ilex, Magnolia, Malus and Sorbus, all had their peak transpiration rate when ambient temperature was around 29°C. For Picea and Pyrus it was around 30°C, and for Prunus 35°C. While in the evening, peak sap flux of all taxa except Prunus, corresponded to a higher air temperature than in the daytime. This is likely due to less evaporative demand together with lower light intensity in the evening, allowing for less stomatal control of water loss than during the day (Moore et al., 2021). This is relevant in an urban context, due to the urban heat island (UHI) effect, which is particularly prevalent in the evening and into the night, when built-up structures store more heat from the sun than green infrastructure. Trees, in this context, can make an important contribution to ambient cooling.

Rahman et al. (2014) compared transpirational cooling potential of five commonly planted tree species and found it driven by both species’ growth rates and stress tolerance. Similarly, Leuzinger et al. (2010), found species-specific cooling differences, with trees with cooler canopies having a higher cooling effect. Due to increased shortwave and longwave radiation from the urban fabric, trees in urban environments are particularly exposed to heat stress (Leuzinger et al., 2010; Stanley et al., 2019). It is therefore important to monitor trees’ responses to high temperatures to understand how they might affect trees’ capacity to cool and to what extent careful planting choices might help. In addition, due to the UHI effect, evaporative demand is likely to increase, leading to greater water requirements and this may be a problem in an urban context where soil water deficit is a common problem, and it will in return constrain transpiration and growth of urban trees (Bialecki et al., 2018). Previous studies have shown how the cooling effect of P. calleryana was severely reduced by urban stressors such as soil compaction (Rahman et al., 2014; Rahman et al., 2015). In our study, P. calleryana ‘Chanticleer’ had an increasing sap flux rate under high air temperatures in the morning and evening, while at 13:00, it peaked around 30°C, demonstrating a capacity to reduce water-loss under high temperatures and high evaporative demand. This would lead to a reduction of cooling capacity, but also a better strategy to avoid dehydration and to grow in areas prone to soil drought.

Richter et al. (2022) found that the most water-conservative species in their study, Quercus robur and Acer pseudoplatanus, had the largest reduction in sap flux during a decline in soil moisture compared to the non-conservative Fraxinus excelsior. In our study, Prunus ‘Fugenzo’ is an example of a non-conservative taxon which, when soil moisture is abundant, did not reduce sap flux values under high evaporative demand. Rahman et al. (2020) found that the Tilia cordata, with high water use, was more efficient at cooling asphalt-covered ground, where evapotranspiration from grass was absent, compared to the more conservative but water-use-efficient R. pseudoacacia. Over grass surfaces however, the R. pseudoacacia had a better cooling potential, possibly due to the grass absorbing more of the latent heat. Thus, the optimum species for cooling potential also depends on the surface under/around it and the type of soil/substrate which they are planted in. The capacity of a tree to regulate air temperature through transpiration is often restricted during water-limited periods, which frequently correspond with when cooling is most needed, i.e., during summer months. Trees which are most likely to be able to provide any cooling service are trees which will grow well under hot conditions and exhibit a less conservative regulation of the stomata (Buckley, 2005; Pataki and Oren, 2003). An example of such taxon in our study is the Prunus ‘Fugenzo’. However, this is only true if the water stress is not too severe, i.e., causing irreversible damage to the plant. Instead, we would recommend trees with high water-use rate under favourable conditions (such as Crataegus ‘Prunifolia Splendens’ in our study), which can regulate water-use during peak temperatures in the day.

Our work considers the context of 1. Climate changing and (peri)urban areas (where most people now live, globally) being warmer and suffering with other associated environmental concerns; 2. Green areas, and particularly trees within them being able to alleviate some of those issues; 3. Decreasing green areas due to pressure from development (certainly the case for domestic gardens in the UK). Hence this quest for trees that fit into those smaller green areas, while being more “efficient” in the delivery of services, and being well-matched to the local environmental conditions. From the synergy of having plants which excel in service delivery in the weather conditions they most often experience, a greater level of overall benefit is gained by all occupants of that space.

Design of green spaces and domestic gardens will always, in large part, be driven by non-horticultural factors such people’s preferences for certain species/cultivars, shapes, smells and colours (Patino et al., 2023). Increasingly, however, the consideration of a chosen plant’s environmental impact (such as the capacity to cool or increase air quality), along with the environmental conditions a plant can grow in, is also part of the decision-making process (Tomatis et al., 2023).

The focus of this paper was on understanding the cooling potential of small garden trees, by monitoring sap flux and canopy characteristics, and on their capacity for carbon drawdown and storage by monitoring growth over time. While outside of the scope of this paper, we acknowledge the contribution of other services these trees are providing (e.g., the ability to capture pollution, provide habitat and food sources for wildlife; Steinparzer et al., 2023).

Sap flow measurements are prone to some uncertainties when it comes to scaling up to absolute tree transpiration volume (Larsen et al., 2020); radial differences of sap flow rates within the sapwood and seasonal differences of wood-water content were not accounted for when scaling up. Instead, these parameters were measured once for minimum intrusiveness (i.e., to minimise drilling of wood cores and limiting installation of probes into the sapwood). However, due to the age and uniformness of the individual trees within a taxon, we are confident that the relationship between meteorological drivers and sap flux densities, and consequently trees’ transpirational cooling potential, are well represented in our study although uncertainties remain with the total volumes of water demand. Although the trees were kept under optimal conditions, further studies in a wider context (i.e., other soil conditions, trees planted in the ground, etc.) would help to understand if these water-use trends would be the same under different conditions.