Introduction

Plant’s stress level in densely built-up inner cities, squares, and streets is much higher compared to that of natural sites. Thus, to assure a tree’s vitality, its benefits and longevity in unfavourable environment species should be planted in soil conditions with least possible diminution of water supply, soil gas exchange and nutrient availability for optimal root development and root growth (Sieghardt et al. 2005; Roberts et al. 2006; Korn 2016a).

Nonetheless, many trees species are only less appropriate at highly sealed street tree sites mainly because of their sensitivity to soil drought. Street trees usually have to deal with high VPD and heat loads, which aggravate the problem of drought (e.g. Bhaduri et al. 2001; Sæbø et al. 2003; Roberts et al. 2006; Gillner et al. 2014; Osone et al. 2014). The limited availability of water combined with high temperatures initiate early stomatal closure at the cost of reduced CO2 uptake and CO2 fixation resulting in low primary productivity (Bush et al. 2008; Gallé et al. 2011; Manzoni et al. 2011; Forrai et al. 2012; Flexas et al. 2014; Gillner et al. 2014). Plants with low water use may be adapted under these site conditions and withstand prolonged periods of dry air and soil conditions (Bush et al. 2008; McCarthy et al. 2011; Klein et al. 2013). However, low primary productivity is reflected in reduced growth rate, reduced tree height with smaller canopy and lower LAI, and may counteract climatic benefits as shading and evapotranspirational cooling (McCarthy et al. 2011; Rahman et al. 2011; Armson et al. 2013; Gillner et al. 2015b). Also, this would hamper efforts to achieve the dense, large, and wide crowns that most urban dwellers prefer regarding urban trees (Gerstenberg and Hofmann 2016). A scientific approach for the identification of appropriate tree species is to compare the balance between water loss and carbon uptake on the basis of instantaneous leaf level measurements (water-use efficiency, WUE) (Lösch 2001). It was found that measurements of WUE provide helpful proxies for species trade-off at urban sites (McCarthy et al. 2011), thus for this study an instantaneous approach by calculating the ratio of photosynthesis and transpiration rates of four tree species was used. WUE is strongly influenced by environmental conditions and is species-specific (McCarthy et al. 2011; Flexas et al. 2014; Gillner et al. 2015a). In climate zones with higher heat and drought stress e.g. in the Mediterranean climate zones, identification of adapted species may still be more important to reduce cost for irrigation and maintenance and to maximize ecosystem services. Among the popular urban trees in the Mediterranean climate zones, Brachychiton discolor, Eucalyptus grandis, and Ficus microcarpa were identified as very efficient in balancing water loss and carbon uptake (McCarthy et al. 2011). Increasing VPD and decreasing water availability steepens the gradient for water loss, which theoretically results in an increasing WUE (McCarthy et al. 2011). However, urban trees at highly paved sites show generally low rates of leaf gas exchange and WUE shows only marginal shifts under increasing edaphic and atmospheric drought levels (Gillner et al. 2015a).

Species can be categorized in response to their mechanism to droughts into two groups: isohydric species and anisohydric species (Lösch 2001; Larcher 2003; Klein et al. 2013). Isohydric trees avoid heat and drought-effects by declining stomatal conductance and maintaining water potentials on a constant level during dry periods, whereas anisohydric trees tolerate drought by maintaining stomatal conductance and decreasing water potentials (Klein et al. 2013). The major significance of water potential, leaf gas exchange, and stomatal conductance for assessing anisohydric or isohydric behavior for urban trees is underlined by several studies (Bush et al. 2008; Litvak et al. 2012; Klein et al. 2013, 2014).

Until recently, there have been only a few leaf physiological measurements as gas exchange or leaf water potential on adult urban trees aiming to identify species sensitivities; logistic difficulties and problems in comparing sites with diverging environmental conditions and management regimes may be the main reasons (Ferrini and Baietto 2006; Harris et al. 2008; Leuzinger et al. 2010; Osone et al. 2014; Rahman et al. 2011; Savi et al. 2015). This research therefore aims to study shifts under increasing drought and heat on stomatal regulation strategies, leaf gas exchange including WUE, and leaf water potential.

Accordingly, the overall aim of this work was to evaluate the specific sensitivity of some commonly planted temperate deciduous tree species in urban environments. On the basis of leaf gas exchange, leaf water potential, leaf surface temperature and chlorophyll fluorescence measurements, we attempted to achieve an assessment of four tree species and their ability to cope with dry air and soil conditions and high temperatures.

Materials and methods

Study area and site characteristics

The four chosen sites were located in the city of Dresden, Germany (51° 02′ 55″N, 13° 44′ 29″E) (Table 1). The average annual air temperature at the weather station located approximately 11 km from the study sites was 8.9 °C and the average annual precipitation sum was approximately 640 mm for the period 1961 to 1991 (Bernhofer et al. 2009). However, temperatures and precipitation regimes in 2013 were exceptional compared with long-term means (DWD 2013). Especially in summer months several weeks of high temperatures accompanied by drought were observed (see Fig. 1).

Table 1 Basic site information. Asterisks mark data obtained from the survey conducted during the spring and summer of 2013
Full size table
Fig. 1
figure 1

Climatic condition from 17th June to 31st August 2013. The mean of air temperature was calculated on the basis of hourly records taken every day. Sunshine duration and precipitation sum refer to the meteorological station in Dresden-Klotzsche (220 m.a. s.l.). The means of air temperature and VPD are provided for all urban sites. Both the temperature and VPD of the meteorological station in Dresden-Klotzsche and of the urban sites are indicated

Full size image

Because the definitions of “urban” in scientific research depend on the field of research, the following site conditions defining the term “urban sites” were selected for our study: trees growing in a small tree pit surrounded by a high amount of impervious surface and paved areas. The selected streets Julius-Otto-Straße, Oskarstraße, Beethovenstraße, and Gustav-Adolf-Straße are located near the city centre on even terrain (120 m.a.s.l.). The area comprises typical late nineteenth century buildings with villas and gardens with medium building density in a residential area of the city (Landeshauptstadt Dresden 2013).

The sites are strongly affected by the urban heat island effect, resulting in higher average temperatures from 0.7 to 2.0 K compared to the surrounding rural area in Dresden (Bernhofer et al. 2009). The trees at the sites were only irrigated in the first two years after the initial plantation (Landeshauptstadt Dresden 2013).

To choose sites with comparable site and microclimatic conditions (degree of sealing per area, soil sealing, water reservoir ability, and building density) data were used provided by the environmental office of the city administration of Dresden (Landeshauptstadt Dresden 2013). The percentage of unsealed soil in the crown projection area is low and ranges from a minimum of 6.3% to a maximum of 12.5% (Table 1). The unsealed area of tree pits with average values of 1.19 to 2.71 m2 is very low and, thus, low infiltration rates can be expected (Blume 2000).

Three soil samples at depths of 0.4, 0.6 and 0.8 m were acquired near the trunk of each tree. The dominate type of soil was loamy sand at all sites, as estimated by a finger test (AG Bodenkunde 2005). The C/N-ratio ranged from 15.4 for Liriodendron to 20.8 for Corylus, indicating comparable values at all sites (Table 1). Sites differ in their electrical conductivity from minima of 120.9 μS cm−1 to maxima of 186.0 μS cm−1. The mean pH-value ranges from 7.0 to 7.3 indicating neutral soil conditions. The mean pH-value ranges from 7.0 to 7.3 indicating neutral soil conditions.

Tree characteristics

Four ornamental tree species were used for investigation, including one gymnosperm Ginkgo biloba L., and three angiosperms Corylus corluna L., Liriodendron tulipifera L., and Tilia cordata Mill. ‘Greenspire’. Over recent years Corylus corlurna, Gingko biloba and Tilia cordata were most frequently planted at paved sites in Dresden, the first two due to their heat and drought tolerance and the later one because of its dominating role within the historic tree population (Landeshauptstadt Dresden 2013). There has been little experience of plantations of Liriodendron tulipifera at paved sites in Dresden although it has also been planted frequently in parks, gardens, and on street tree sites over recent years (Landeshauptstadt Dresden 2013).

The deciduous tree species were selected on the basis of age ranging from 13 to 20 years and the tree’s vitality. The tree characteristics age, DBH (Diameter at Breast Height), height, and the LAD (leaf area density, with units of m2 foliage area per m3 canopy volume) are shown in Table 2. LAD was determined with a plant canopy analyser (Licor LAI-2200: Lincoln, Nebraska, USA) for each tree and computed with the model for isolated trees by averaging the values from four files with LAI-2200 software (Licor FV2200: Lincoln, Nebraska, USA). The LAD values ranged from a maximum of 2.56 m2 m−3 for Tilia to a minimum of 0.92 m2 m−3 for Ginkgo.

Table 2 Basic tree characteristics (DBH = diameter at breast height; LAD = leaf area density). Different letters mark significant differences between the tree species for each parameter
Full size table

Leaf gas exchange

Net-photosynthesis (Amax), transpiration (E) and stomatal conductance (gs) were measured during the summer 2013 at saturating light. On the dates of measurements a minimum of four up to six randomly selected trees of one species were measured with a number of four leaves (measurements) per tree. For every tree averages were calculated for further data analysis. Leaf gas exchange rates for Corylus and Liriodendron were measured 3 times and for Ginkgo and Tilia 4 times during the experimental period. The first replicate was done after all four species had been measured once. Measurements were made on mature healthy leaves without any signs of senescence at exposed peripheral twigs in the sun canopy sections in the top third of the crowns using a lift.

Data were measured using an HCM-1000 portable photosynthesis system (Heinz Walz GmbH: Effeltrich, Bavaria, Germany). Leaf gas exchange rates were measured between 7:30 a.m. and 6:00 p.m. (CEST, Central European Summer Time) at two fixed temperatures of i) 25 °C and ii) 30 °C, separately on different leaves, with a relative humidity within the chamber of approximately 40–50% and a constant photosynthetic photon flux density (PPFD) of 1100 μmol m−2 s−1.

The water-use efficiency (WUE) was calculated from the ratio of Amax and E (Larcher 2003).

Leaf water potential

Midday and pre-dawn leaf water potentials (Ψmin, Ψmax ) were measured with two Scholander pressure chambers (PMS Instruments, Corvallis, Ore.), directly at the trees, in the top third of the crowns for every species from the middle of June to the end of August. Ψmin was determined at approximately midday from 12:30 to 02:00 p.m., and Ψmax was measured before sunrise from 03:15 to 04:45 a.m. CEST. Additionally, the diurnal leaf water potentials were determined on two consecutive days (2nd and 3rd August). Leaves were selected following same restriction as for the leaf gas exchange. However, data of Ψmin and Ψmax were measured for four trees per species with five leaves per tree on always the same individuals to demonstrate trends measured with two Scholander pressure chambers (…) ,directly at the trees, in the top third of the crowns for every species.

Leaf surface temperatures

The thermal camera VarioCam hr. inspect 700 (Infratec, Dresden, Germany), with a resolution of 1280 × 960 pixels, was used to record leaf surface temperatures (Tleaf) during two hot, sunny and cloudless days (2nd and 3rd August). The thermal images were taken on single leaves between 08:00 a.m. and 06:15 p.m. CEST from a distance of 1 m from the leaf using a lift. Only healthy, full sun-exposed leaves at the upper crown were used for the investigation. Thermal images were analysed with Irbis professional software (Infratec, Dresden, Germany), using closed polygons to select the surface areas of the full sun-exposed leaves. At an emissivity ranging from 0.94 to 0.99 for the sun-exposed leaves, there are only minor effects of the correction and, therefore, they can be neglected (Leuzinger et al. 2010). Consequently, an emissivity of 1.00 was applied for all images (Siegel et al. 1988).

Chlorophyll fluorescence

Species-specific differences in the potential photochemical efficiency of the photosystem II (respectively optimal quantum yield of photosystems II) were measured by the ratio of variable (Fv = Fm – Fo) to maximal (Fm) fluorescence (i.e., Fv/Fm, where Fo = minimal fluorescence) of dark-adapted leaves (Willits and Peet 2001; Percival 2004). As a stress indicator and as a measure of plant vitality, Fv/Fm value is widely used (Maxwell and Johnson 2000). Data were collected during the 2013 growing season between 08:30 a.m. to 06.00 p.m. on nearby leaves to assure the best possible comparability and accordance of the data sets. Before the measurements were taken, leaves were fixed with a leaf clip holder and covered for dark adaptation for at least 35 min. Then, the fibre optic of the MiniPam was inserted into the clip holder and a saturating pulse of light was applied to the leaf. Initially, dark-adapted leaves were exposed to the modulated measuring light intensity of approximately 0.1 μmol m−2 s−1, followed by exposure to a continuous, actinic white light of 11,000 μmol m−2 s−1 (Genty et al. 1989). For every tree 30 leaves were measured and average was calculated per tree.

Environmental characteristics

Monitoring of climatic conditions was conducted from 17th June to 31st August 2013 by four iButtons® (DS1923) (Maxim integrated, San Jose, California, USA) at each site. iButtons® continuously recorded the air temperature and relative humidity with an 8-bit format in hourly intervals and were installed at the top height of four randomly selected trees per street.

Vapor pressure deficit was calculated (VPD, kPa) on the basis of the Magnus equation (Magnus 1844). Records of the iButtons® were used for correlation analysis of Tleaf, Ψmin, and Ψmax, whereas internal records of air temperature and relative humidity during measurements of leaf gas exchange were used for correlation analysis of leaf gas exchange.

The soil water potential (Ψsoil, hPa) was determined with T1S tensiometers (UP GmbH, Cottbus, Brandenburg, Germany) at depths of 40, 60, and 80 cm and the volumetric soil water content (Θ, Vol.-%) was determined with PR2 soil moisture profile probes (Delta-T Devices Ltd.: Cambridge, Cambridgeshire, United Kingdom) at depths of 20, 30, 40, 60, and 100 cm within the four randomly selected tree pits per site. Ψsoil and Θ were monitored on the same dates when leaf gas exchange was measured.

Statistical analysis

A Shapiro-Wilk-Test (P ≤ 0.05) (Shapiro and Wilk 1965) was applied to test the data for a normal distribution followed by the non-parametric method of Kruskal–Wallis to analyse differences in the distributions. Significant differences were found and in consequence Pairwise Mann–Whitney U rank sum test with a Bonferroni correction was applied to localise the differences. A Spearman correlation was calculated to test the relationships between the variables.

Graphs were created with Xact (Version 8.03, SciLab: Hamburg, Germany) and R software (R Development Core Team 2012) was used for statistical analysis.

Results

Environmental conditions

Average temperatures of March were up to 1.3 K lower in 2013 compared with the long-term means (DWD 2013). During four days from the 1st to 4th June a precipitation sum of a 96.8 mm were recorded exceeding the monthly mean by almost 150%. A long period of heat and drought was recorded afterwards, interrupted by only a few days of cloudy and rainy conditions with a total rainfall amount of only 16.53 mm (Fig. 1). Maximum daily mean air temperatures were recorded from 27th to 29th July 2013 with values up to 32.0 °C. Minimum daily mean air temperatures were found in a period from 25th to 27th June 2013 with minima down to 13.1 °C. VPD increased until a maximum value of 2.8 kPa on the 28th July and then followed a decreasing trend to 31st August (Fig. 1).

At all sites, the time-series of Ψsoil and Θ at a soil depth of 40 cm show a declining trend followed by a recovery after two consecutive rainy days on 28th and 29th July (Fig. 2). Values of both parameters then decline rapidly and the lowest values were reached at the end of August at all sites. The values of Θ ranged from only 11 (Liriodendron) to 19 Vol.-% (Ginkgo) at the end of the measurement period and were at least a third lower than at the beginning of when measurements were taken.

Fig. 2
figure 2

Volumetric soil water content (Θ, Vol.-%) and soil water potential (Ψsoil, hPa) at a soil depth of 40 cm measured from 11th June to 30th August 2013. Θ and Ψsoil were calculated from the four measurements taken at each site

Full size image

Θ was compared for the sites and soil depths (Fig. 3). Θ of the same soil depth did not differ significantly between the sites; e.g., the median Θ at soil depths of 40 cm only ranges from 17.9 Vol.-% for Corylus to 21.9 Vol.-% for Ginkgo. A vertical increase in Θ can be found with significant differences from the Θ at the different soil depths for all sites. This implies a relative comparability of the soil water conditions of the sites and confirms each sites’ suitability for comparing the leaf gas exchange of the species.

Fig. 3
figure 3

Θ of the five different soil depths calculated from all measurements. Different letters indicate significant differences between the Θ at one site at p < 0.05 resulting from the Mann–Whitney U rank sum test

Full size image

Leaf gas exchange

Leaf gas exchange shows distinct differences between the species (Fig. 4). At the baseline temperature at 25 °C highest median rates of E, gs, and Amax were found for Corylus and Tilia reaching almost the double values compared with Liriodendron and Ginkgo and differ always on a significant level. However, the ratio of Amax and E indicated by the median values of WUE, detects only significant differences between Corylus (2.87 μmol mmol−1) and Tilia (3.87 μmol mmol−1).

Fig. 4
figure 4

Transpiration rates (E), stomatal conductance (gs), net-photosynthesis (Amax) and water-use-efficiency (WUE) at 25 °C (left) and 30 °C (right). Significant differences (P < 0.05) among the species are indicated by different letters within the given temperature. White boxes specify significant different values for the parameter and species between 25 °C and 30 °C. n = number of measurements

Full size image

Under an ambient air temperature of 30 °C the leaf gas exchange shifted and again showed species specific responses (Fig. 4). Whereas E rate of Corylus increased significantly by 22 °% compared to the baseline level of 25 °C, rates of gs and Amax of Tilia significantly decreased by 49.8 and 51.4% respectively. Strongest decrease in Amax was observed for Liriodendron with a decline of 72.1%.

WUE for every species differs significantly between the temperature levels of 25 °C and 30 °C with always lower values at 30 °C. Values at 30 °C reached only between 79.0% (Corylus) and 31.9% (Liriodendron) of their WUE-level at 25 °C. Taking into consideration the species responses in E and Amax, the specific differences in WUE are easy to detect, showing a significant higher WUE for Corylus and Tilia compared to Liriodendron.

Leaf water potential and leaf surface temperature

Both, Ψmin and Ψmax varied throughout the season. In the period with an increasing VPD during the approximately four weeks until 28th July (light grey background), a shift is observed in both the Ψmin and Ψmax (Fig. 5). Decreasing values of Ψmin and Ψmax for Corylus and Tilia were measured, accompanied by approximately unchanging differences between Ψmin and Ψmax for Tilia, whereas these differences increased for Corylus. For Ginkgo and Liriodendron, the Ψmin and Ψmax converge under an increasing VPD as a result of the declining values of Ψmax and the rising values of Ψmin. The lowest median values for Ψmin (−2.73 MPa) were measured for Corylus in the period with the highest VPD, whereas the lowest values for Liriodendron (−1.76 MPa) were recorded at the beginning of the measurement period. After the two rainy days on 28th and 29th July, all species showed increasing values indicating a recovery of the Ψmax (Fig. 5). The significant negative correlation for Corylus indicates a decreasing Ψmin under an increasing VPD, whereas the significant positive correlation for Ginkgo refers to an increasing Ψmin under an increasing VPD.

Fig. 5
figure 5

Seasonal midday (Ψmin, circle) and pre-dawn (Ψmax, triangle) leaf water potentials from 18th June to 30th August. The period to 28th July with increasing VPD is marked in light grey. Spearman correlation coefficients (rs) between Ψleaf and VPD at the 95% significance level are marked in bold italic letters

Full size image

During the diurnal course of Ψleaf, there was a pronounced drop in Ψleaf until minimum values were reached at approximately 01:30 p.m., with a recovery period until 06:00 p.m. (Fig. 6 b). Correlations coefficients of Ψleaf and VPD indicate significant negative values for the daily course of Ψleaf, for Corylus and Tilia, whereas analyses for Ginkgo and Liriodendron indicated no significant relationship.

Fig. 6
figure 6

Daily Ψleaf (b) and Tleaf (c) surface leaf temperature) represented by the measurements from 2nd and 3rd August 2013. VPD values were calculated as the mean of the measurements of four iButtons® for every street tree site (a). Spearman correlation coefficients (rs) between Ψleaf and Tleaf and VPD at the 95% significance level are marked in bold italic letters

Full size image

As expected, the surface leaf temperatures exhibited a characteristic diurnal variation too (Fig. 6 c). The maximum medians of Tleaf ranged from 36.0 °C for Liriodendron up to 38.9 °C for Corylus, which occurred in the afternoon. ΔTleaf became smaller over the diurnal course between the species, starting with a ΔTleaf of 5.1 K at approximately 09:00 a.m. followed by a ΔTleaf of 2.9 K in the early afternoon and a ΔTleaf of 2.8 K in the evening. The four species that were monitored showed very similar pattern in their Tleaf compared to the pattern of VPD. This is confirmed by the strong significant positive correlations of Tleaf and VPD calculated for all species (Fig. 6 c).

Chlorophyll fluorescence

Figure 7 shows the medians of Fv/Fm of the four tree species. Liriodendron represents the significantly lowest values with a median of 0.79. A high value is achieved by Corylus (median of 0.83) and Tilia (0.83) showing a measurable difference in their photosynthetic performance through chlorophyll fluorescence.

Fig. 7
figure 7

Chlorophyll fluorescence (Fv/Fm) during the 2013 study period. Different letters indicate significant differences between the medians of the species. n = number of measured trees (30 measurements of each tree)

Full size image

Influence of climate, leaf water potential, and leaf surface temperature on leaf gas exchange

gs and WUE measurements at 25 °C and 30 °C were used to assess the influence of Ψmin, Ψmax, Tleaf, and VPD (Table 3). All significant correlation coefficients show positive signs between Ψmin, Ψmax and gs. The correlation between Ψmin, Ψmax and WUE shows only significant correlations in three cases, with a negative correlation for Tilia and a positive correlation for Ginkgo.

Table 3 Correlation coefficients between gs, WUE and Ψmin, Ψmax, Tleaf, and VPD. Coefficients meeting the 95% significance level are indicated in bold letters
Full size table

For all species, a strong significant negative correlation between gs and Tleaf exists at both temperature levels (25 °C and 30 °C). Thus, our results confirm a strong negative relation between gs and Tleaf. Additionally, between WUE and Tleaf, strong negative and significant correlations were found for the species. An increasing VPD causes a decrease in gs and WUE for all species. At 25 °C, these relationships are significant for all species. At 30 °C, the correlations for Ginkgo, Liriodendron, and Tilia between WUE and VPD decreased to a non-significant level.

Discussion

Physiological reactions under stress and plant strategies to cope with heat and drought

Plant water status, water loss, and water usage are strongly related to diurnal and seasonal climatic factors, e.g., irradiance, temperature, air humidity as well as edaphic factors, e.g., soil water supply (Roberts et al. 1980; Sperry 2000; Larcher 2003; Gallé et al. 2011; Aranda et al. 2014; Osone et al. 2014). For example, seasonal transpiration shows a rapid increase with leaf development in the spring followed by decreasing rates with the ageing of the leaves (Herrick and Thomas 2003; Larcher 2003). Further, leaf gas exchange shows a typical diurnal variation (Larcher 2003). In consideration of these aspects, we decided to take measurements at fixed temperatures, at a constant photosynthetic photon flux density (PPFD) of 1100 μmol m−2 s−1 and at a relative humidity within the chamber between 40 and 50% to exclude changing leaf gas exchange rates caused by diurnal changes in microclimatic parameters. Showing the diurnal variations would give a better insight of species specific reactions; however our results refer to daily means and enable information about the mean differences of E, gs, and Amax rates.

A tree’s specific capability to withstand heat and drought has become an important issue for urban ecologists (Korn 2016b). The far general statement that heat and drought are important limiting factors for tree physiology at urban sites (Roberts et al. 2006; Fini et al. 2008, 2009; Leuzinger et al. 2010; Swoczyna et al. 2010; Forrai et al. 2012; Korn 2016b), was underlined by the fact that all species of this study responded sensitively under restricted water supply and transpirational demand. For July a drought period was identified that had a cumulative rainfall of only 0.2 mm and the highest VPD (Fig. 1). This is especially important since low water supply in combination with high temperatures lead to stomatal closure, the one determinant for reduced net photosynthesis and further allows detecting inter- and intra-specific differences in the trees water regimes (Medrano et al. 2002). Trees showed different adaptation mechanisms under progressive drought stress caused by an increasing VPD and declining values of Ψsoil and Θ. Considering the trend in Ψleaf within the drought period, the gradient between Ψmin and Ψmax becomes higher with increasing drought stress for Corylus and Tilia referring to anisohydric responses (Klein et al. 2013, 2014) (Fig. 5). A contrasting, isohydric response can be observed for Ginkgo and Liriodendron expressed in declining gradients (Fig. 5). The lowest values of Ψmin and the highest difference were observed for Corylus followed by Tilia. The isohydric strategy to avoid water loss via transpiration, as found for Ginkgo and Liriodendron, is confirmed by lower values of gs and simultaneously increasing values of Ψmin under an increasing level of temperature and VPD (Korn 2016b). These reactions prevent water loss as observed with the significantly lower rates of E for Ginkgo and Liriodendron compared to Corylus and Tilia but are accompanied by reduced rates of photosynthesis in periods of stress (Sperry 2000). This strategy appears advantageous when the water supply is sufficient, as observed in the beginning of our measurement period to ensure rapid leaf expansion, plant growth and turgor-mediated processes (Roberts et al. 1980; Larcher 2003). The declining values of Ψmin of Corylus and Tilia can be interpreted as a tolerance of increasing drought aimed to exploit the soil moisture reservoir under low levels of Ψsoil and Θ and maintaining transpiration (Korn 2016b). Even though categorization of Tilia as an isohydric species is less clear (Whitlow et al. 1992; Köcher et al. 2009), our results (maintaining high transpiration rates with decreasing and low values of Ψmin) clearly underline the anisohydric strategy. To gain a better understanding of what determined Ψleaf, diurnal measurements were carried out. Results confirm the categorization with significant correlation of Ψleaf with VPD for Corylus and Tilia (anisohydric) and no significant relationships for Ginkgo and Liriodendron (isohydric).

These responses imply a strong correlation between Ψleaf and gs to confirm the mechanisms of a species’ adjustment to drought (Medrano et al. 2002). All significant correlations are positive revealing the higher gs the higher is Ψleaf (less negative). However, significant correlations between Ψleaf and gs could not be determined for all studied species (Table 3). Corylus, especially, shows no significant correlation. Stomatal regulation is a complex multi-factorial process and besides Ψleaf, related to many other factors e.g. abscisic acid (ABA). Medrano et al. (2002) reported on the role of gs driven by internal factors (ABA, Rubisco activity, xylem conductivity, leaf water status) as well as external factors (VPD, soil water status). They conclude that non-stomatal limitations play an important role but primarily gs is the dominant mechanism limiting photosynthesis. For example, response of Rubisco activity to drought is overlaid by the determinant influence of gs among a wide range of species and different conditions (Medrano et al. 2002).

The order of absolute values of Ψleaf corresponds well with the order of median values of E and gs, showing the highest rates of gs and E for Corylus and Tilia, and the lowest rates for Ginkgo and Liriodendron. These mechanisms illuminate only one aspect of a tree’s ability to cope with heat and high values of VPD. Therefore, other parameters, such as the efficiency of CO2 uptake and the efficiency of photosynthesis are also considered. Do these values support the given classification of the plants?

Water-use-efficiency changes with alteration in water availability (Cavander-Bares and Bazzaz 2000; Fini et al. 2008; Manzoni et al. 2011; Flexas et al. 2014). Stomata close progressively under increasing drought accompanied by parallel decrease of net photosynthesis (Medrano et al. 2002; Flexas et al. 2014). Depending on the stress level, mild water stress often causes an increase of WUE, whereas severe water stress strongly limits net photosynthesis and thus WUE decreases (Larcher 2003). In our study this fact is supported by the decline in WUE under a temperature level of 30 °C and the negative correlation of WUE to VPD at a temperature level of 25 °C for all species referring to sever water stress. But the degree of the decline differed among species. The WUE of Corylus declined the least (21%) and the WUE for Liriodendron declined the most (68.1%) at 30 °C. In declining order, the species reduced their WUE as follows: Corylus, Ginkgo, Tilia, and Liriodendron.

Chlorophyll fluorescence has been proven to detect the effects of environmental stress prior to any visible damage (Percival 2004). Under severe drought stress plants limit the efficiency of their photosystems resulting in a low chlorophyll fluorescence yield (Swoczyna et al. 2010). The achieved results show no acute stress indicated by the relatively high values of Fv/Fm near the level of the non-stressed leaves at 0.83 (Baker 2008). We found significant differences in the efficiency of the photosystem II, allowing us to classify the species Corylus and Tilia as being less affected by stress compared to the group composed of Liriodendron and Ginkgo, which had significantly lower Fv/Fm values. A broad range of stressors is integrated in the Fv/Fm values (Percival 2004). However, species ranking shows similar results by chlorophyll fluorescence and leaf gas exchange. The response of WUE to a change in VPD differs from the response of gs to VPD changes (Table 2). At a temperature level of 25 °C species showed a decline in WUE if VPD increased. At a temperature level of 30 °C only Corylus exhibited a significant relationship as indicated by the negative correlations. This implies that CO2 fixation for Ginkgo, Liriodendron and Tilia becomes independent of stomatal factors at higher temperatures and non-stomatal factors limit the photosynthesis rate, and internal factors controlling leaf gas exchange may become important (Fini et al. 2009). The correlations analyses confirm the fact that stomata close progressively and WUE decreases as drought progresses (increasing VPD) (Medrano et al. 2002).

Transpiration rates and canopy densities are strong determinants for leaf surface temperatures (Leuzinger and Körner 2007). Since a low LAD results in lower Tleaf close to air temperatures (Leuzinger and Körner 2007), Ginkgo and Liriodendron with lowest LAD should exhibit low leaf temperatures. In our study, Tleaf differs between species with a high LAD and species with low LAD only in one case. At 09:00 a.m. Tleaf, for Ginkgo was higher compared to Tilia and Corylus (cf. Figure 6). This can be attributed to the fact that trees showing low LAD also show low transpiration (Liriodendron and Ginkgo) and trees with a high LAD also have elevated transpiration rates (Tilia and Corylus) (cf. Figure 4), which is in contrast to the results of Leuzinger and Körner (2007).

In many studies a strong relationships between the leaf surface temperature and gs was found (Leuzinger et al. 2010; Lin et al. 2012), which was also confirmed in this study by the significantly high negative correlation (Table 2). Long warm periods may lead to an overheating of leaves causing negative impacts or even irreparable damages at the cellular level (Leuzinger et al. 2010; Matyssek et al. 2010) and a decline in net photosynthesis (Lin et al. 2012). However, woody plants and leaf tissue vary considerably in temperature sensitivities (Kattge and Knorr 2007). Our results are in contrast to Leuzinger and Körner (2007), who found no significant relationship between gs and Tleaf. This may be a result of the data used. In their study data of gs were cited from references and not collected during field work (Leuzinger and Körner 2007). Strong significant correlation for every species was also found between VPD and Tleaf indicating declining Tleaf under increasing VPD. Summarizing the above, it can be said that, Tleaf is not only determined by the microclimatic parameters air temperature and air humidity (involved in VPD), further environmental parameters (e.g. PAR, surface specific radiation, energy absorption, wind speed, instantaneous wind pattern) and leaf specific tissues (e.g. leaf angle, leaf position, leaf boundary resistance) should be involved in data analysis (Leuzinger and Körner 2007; Leuzinger et al. 2010; Matyssek et al. 2010).

Plant specific suitability at the urban sites

The growing conditions and, consequently, the leaf gas exchange, growth, and vitality of urban trees are strongly related to site conditions (Whitlow et al. 1992; Roberts et al. 2006; Conway 2007; Bartens et al. 2009; Cekstere and Osvalde 2013; Korn 2016a, b). All sites showed comparable building densities and soil conditions including C/N-ratios (Table 1). The electrical conductivity of the soil is higher at the site where Ginkgo and Tilia are located, indicating a higher amount of dissolved substances that may negatively influence water uptake by roots leading to intensified drought conditions (Roberts et al. 2006). The surveys of Ψsoil and Θ were used to describe the spatial differences in the soil water supply (Figs. 2 and 3) because both parameters are important determinants for leaf gas exchange rates and Ψleaf (Elfving et al. 1972; Larcher 2003). Θ exhibited similar values for all sites at the same depth and did not differ significantly.

Under these conditions, our results confirm the facts that species differ in their sensitivity to water deficits (Roberts et al. 1980) and that various species-specific mechanisms exist for regulating the water flux. For instance, water use decreased faster in isohydric Pinus halepensis compared with the coexisting anisohydric Quercus calliprinos at an experimental plot near Tel Aviv, Israel (Klein et al. 2013). Under increasing heat during summer, street trees of Prunus yedoensis and Zelkova serrata had more durable leaves than park trees of the same species but park trees showed significant higher values of gs (Osone et al. 2014). Plant transpiration is strongly related to hydraulic architecture and wood anatomy may determine trees response to VPD (Bush et al. 2008; Litvak et al. 2012). The higher sensitivity of gs of ring-porous species can be explained by the lower resistance to cavitations compared with diffuse-porous species (Bush et al. 2008). Our findings for diffuse-porous species Corylus, Liriodendron, and Tilia with a negative correlation of gs to VPD are in contrast to those of Bush et al. (2008) (Table 3), resulting in a linear increase of gs and E to VPD for diffuse-porous species. Even more, there is no homogenous response of these three diffuse-porous species indicated by the results of Ψleaf, WUE, and chlorophyll fluorescence in our study. Main differences between the studies are the irrigation and exposure to higher values of VPD of more than 4.0 kPa for the studied trees in Los Angeles (Bush et al. 2008; Litvak et al. 2012).

The detected differences enable a ranking of tree species according to their leaf gas exchange, leaf water potential and chlorophyll fluorescence to assess suitability as well as drought sensitivity. These results might help planners to select appropriate tree species in urban settings.

The rather low decrease of the WUE during soil drought, high values of VPD and high temperatures for Corylus, as well as the higher efficiency of the photosystem II enables this species to maintain important physiological processes. Additionally, Tilia shows a comparable response. These species can be classified as highly adapted to the defined urban site conditions.

Independent from the sensitivity ranking, the early leaf senescence and abscission of Liriodendron was observed from the middle to the end of July, the period with the highest VPD. This is in accordance with the results in the Woodlands of North Carolina, USA reporting severe leaf senescence and abscission of the species (Roberts et al. 1980). In temperate forests, a species-specific decrease in the leaf area index under progressive drought was observed (Le Dantec et al. 2000; Poyatos et al. 2013). This process is related to a decline in the gross primary productivity (Larcher 2003; Poyatos et al. 2013), referring to the low suitability of Liriodendron under drought conditions. Native ranges of Corylus and Tilia includes regions with a frequent occurrence of heat and drought – a potential proxy for the identification of species adaptation under these conditions (McCarthy et al. 2011).

Conclusions

This study combines measurements of gas exchange, water potential, leaf surface temperature and chlorophyll fluorescence to characterise the suitability of street tree species.

The results refer to clear species-specific differences in the adaptability to the harsh and challenging urban environment.

The microclimatic benefits of trees that provide shelter from direct solar radiation and heat through tree shading and transpiration are especially important in periods of heat and drought and have also be considered in the context of planting recommendations. To maximise these benefits trees that maintain high transpiration rates without showing any symptoms of senescence and leaf fall during periods of summer drought should be planted. Efforts to maintain healthy trees showing high vitality over decades are also important to increase the quality of urban living environments and consequentially increasing life satisfaction of the urban dwellers.

At highly sealed sites the species Corylus and Tilia are able to provide the desired benefits of street trees even during periods of increased heat and drought. The results of the leaf physiological study show that these species are able to maintain a high efficiency that prevents early leaf senescence and leaf fall. However, the physiological reactions of the North American species Liriodendron and the East Asian species Ginkgo have a lower efficiency. Accordingly, Corylus and Tilia are recommended for urban greening at stressful sites, whereas Ginkgo and Liriodendron are limited to less demanding sites.

Furthermore, the study shows that a combination of several physiological indicators enable a more holistic view on a tree’s performance in anthropogenically influenced ecosystems. Nevertheless, not only the water balance influences street trees, but other stressors such as de-icing salt, freezing events, vandalism or alkali tree pit conditions should also be elucidated in future studies.