Water relations of coast redwood planted in the semi-arid climate of southern California

Authors


E. Litvak. Fax: 1 949 824 3874; e-mail: elitvak@uci.edu

ABSTRACT

Trees planted in urban landscapes in southern California are often exposed to an unusual combination of high atmospheric evaporative demand and moist soil conditions caused by irrigation. The water relations of species transplanted into these conditions are uncertain. We investigated the water relations of coast redwood (Sequoia sempervirens) planted in the urbanized semi-arid Los Angeles Basin, where it often experiences leaf chlorosis and senescence. We measured the sap flux (JO) and hydraulic properties of irrigated trees at three sites in the Los Angeles region. We observed relatively strong stomatal regulation in response to atmospheric vapour pressure deficit (D; JO saturated at D < 1 kPa), and a linear response of JO to photosynthetically active radiation. Total tree water use by coast redwood was relatively low, with plot-level transpiration rates below 1 mm d−1. There was some evidence of xylem cavitation during the summer, which appeared to be reversed in fall and early winter. We conclude that water stress was not a direct factor in causing leaf chlorosis and senescence as has been proposed. Instead, the relatively strong stomatal control that is adaptive in the native habitat of coast redwood may lead to carbon limitation and other stresses in semi-arid, irrigated habitats.

INTRODUCTION

Urbanization of the semi-arid US south-west has been accompanied by increased forest cover in many areas (Dwyer et al. 2000). This afforestation raises questions about the physiological responses of non-native tree species to urban environments, and the possible impacts of afforestation on regional water balances. Plants grown in southern California may be exposed to an unusual combination of atmospheric drought (high vapour pressure deficit, D) coupled with an almost unlimited water supply to roots caused by frequent irrigation. However, few studies have investigated tree physiology under these conditions, and it is difficult to predict the responses of mature trees to irrigated urban environments. Bush et al. (2008) found systematic differences in stomatal responses to high D between ring-porous and diffuse-porous tree species growing in irrigated urban environments. McCarthy & Pataki (2010) reported that Pinus canariensis, which is not native to southern California, used less water than native Platanus racemosa, with site-to-site variability in sap flow rates mostly attributable to water availability.

Coast redwood (Sequoia sempervirens), is a common landscape species that originates from moist, coastal regions of central and northern California. It has been cultivated in many different settings (Wu & Guo 2006) and has cultural significance in California. Although the extent of old growth redwood was greatly reduced by logging and development, there have been many efforts to preserve the trees that remain (Ricketts et al. 1999; Noss 2000). An extreme example is a historic 33.5-m-tall coast redwood in Palo Alto, CA, USA where a pipe has been placed along the trunk to keep the top of the tree moist (Farmer 2010). Coast redwood grows rapidly (Noss 2000) and develops the largest known leaf area index (Westman & Whittaker 1975), which is beneficial for providing ecosystem services such as shade, carbon accumulation and pollution reduction in urban environments. In its natural environment, coast redwood is exposed to a mild humid climate with mean annual temperatures of 10–16 °C and annual precipitation of 60–250 cm (Wu & Guo 2006), which mostly occurs in winter. Summer water inputs are mainly in the form of fog (Byers 1953; Oberlander 1956; Azevedo & Morgan 1974).

In contrast, in the semi-arid climate of southern California, coast redwood is exposed to altered environmental conditions compared to its native habitat, in particular to the lack of abundant marine fog. These conditions may inhibit its performance, as mature individuals of coast redwood in many urban settings of southern California often show leaf senescence and signs of water stress (Barnhill, personal communications; Downer 2004), the causes of which are unknown. Possible reasons for poor performance include presumably weak stomatal regulation of coast redwood in response to high D suggested by Burgess & Dawson (2004). Inadequate stomatal regulation could result in xylem cavitation and failure of the vascular system (Tyree & Sperry 1989). Alternatively, urban coast redwood could be sensitive to declines in soil moisture, which may occur during periods of inadequate irrigation and high evaporative demand. Elevated concentrations of ozone, which has been shown to be damaging to other conifer species in southern California (Miller et al. 1963; Evans & Miller 1972; Takemoto, Bytnerowicz & Fenn 2001), is unlikely to be the cause of poor performance as coast redwood is known to be ozone tolerant (Lefohn 1991).

We investigated the water relations of irrigated, urban coast redwood at three sites in the Los Angeles region to better understand the influence of semi-arid urban conditions on both tree ecophysiology and plot-level water use. We coupled continuous measurements of sap flux with meteorological measurements, laboratory estimates of native embolism and measurements of vulnerability to cavitation. We addressed the following questions: What is the stomatal behaviour of coast redwood in an irrigated semi-arid environment? Do patterns of stomatal conductance lead to the development of xylem cavitation? What is the magnitude of stand transpiration of coast redwood in a semi-arid environment? Following Burgess & Dawson (2004), we hypothesized that urban coast redwood would show weak stomatal regulation in response to D, with subsequent declines in stem hydraulic conductivity caused by cavitation. Given the frequent irrigation of coast redwood in southern California, we did not expect to see a strong effect of soil moisture on transpiration. Based on the visible signs of stress in coast redwood in the region, we hypothesized that stand transpiration rates would be relatively low compared to other irrigated urban tree species.

MATERIALS AND METHODS

Study sites

The three study sites were located in the Los Angeles Basin, California, a coastal plain surrounded by several mountain ranges. The climate is Mediterranean with a mean annual temperature of 17.0–18.3 °C and annual precipitation of 32.6–37.7 cm in the form of rain, mainly in winter (Western Regional Climate Center, http://www.wrcc.dri.edu). The sites were located along a coastal-to-inland climate gradient, and were expected to vary in temperature and humidity as a function of proximity to the coast.

The ‘Los Angeles’ site was located in a recreational rock garden at the Los Angeles Police Revolver and Athletic Club, on a north-western hill slope. This site was referred to as the LA Police Academy site in Pataki et al. (2010). Individuals of coast redwood occurred in a mixed species stand with several other tree species, and an extensive understory of herbaceous plants and woody shrubs. The ‘Fullerton’ site was located at the California State University Fullerton Arboretum. At this site, coast redwood was grown in a single species stand along an artificial stream. The ‘Irvine’ site was located on the campus of the University of California, Irvine, where coast redwood was grown in a mixed species stand with a thick understory of shrubs and grasses. All locations were equipped with automatic sprinklers and received regular irrigation. The details of study sites and trees are summarized in Table 1.

Table 1.  Description of study sites and trees
Study siteCoordinatesAnnual average temperature, annual precipitationArea (m2)Number of treesPlanting density (trees ha−1)Year of plantingMean DBH (cm)Mean sapwood depth (cm)Tree height (m)
  • Values are accompanied by one standard error.

  • a

    Historical climate in Yorba Linda, Western Regional Climate Center (http://www.wrcc.dri.edu).

  • b

    Historical climate in downtown Los Angeles, Western Regional Climate Center (http://www.wrcc.dri.edu).

  • c

    Historical climate in Tustin Irvine Ranch, Western Regional Climate Center (http://www.wrcc.dri.edu).

  • d

    The year when most studied trees were planted.

  • e

    The year when the garden was established.

  • f

    Based on growth rings of felled trees.

Fullerton33°53′12″N
117°53′03″W
Elevation 79 m
17.6 °C, 34.8 cma47391901977d48.4 ± 5.13.6 ± 0.318 ± 1
Los Angeles34°04′57″N
118°14′28″W
Elevation 191 m
18.3 °C, 37.7 cmb6755741935e39.4 ± 6.83.7 ± 0.620 ± 4
Irvine33°38′48″N
117°50′52″W
Elevation 19 m
17.0 °C, 32.6 cmc14418561986–1990f44.0 ± 2.73.7 ± 0.314 ± 1

Environmental parameters

Air temperature and relative humidity were continuously measured at all sites at one-third to one-half of canopy height (HMP45C; Vaisala Inc., Helsinki, Finland). Because measuring overstory photosynthetically active radiation (IO) was not feasible at our study sites, data from the California Irrigation Management Information System weather stations within a 20 km radius were used (http://www.CIMIS.water.ca.gov). The soil water content in the top 30 cm was continuously monitored at six locations within each site using soil water content reflectometers (CS616; Campbell Scientific, Logan, UT, USA). Temperature, relative humidity and soil water content were sampled every 30 s, and averaged every 30 min (CR1000; Campbell Scientific).

We assessed the frequency of fog by assuming that 100% relative humidity corresponds to foggy (or rainy) conditions. Based on the resolution of our humidity sensors (±3% from 90 to 100% relative humidity), we considered 95% humidity as a threshold for possible saturation.

Sap flux measurements

Thermal dissipation probes (Granier 1987) were installed in the outer 2 cm of sapwood on the north side of each study tree. Sensors were placed approximately at breast height (1.35 m from the ground) in Los Angeles and Irvine, and at 2–3.5 m from the ground in Fullerton to deter vandalism. The sensor output was measured every 30 s and stored as 30 min averages (CR1000; Campbell Scientific). Sap flux density in the outer 2 cm of sapwood (JO, g cm−2 s−1) was calculated from the temperature differences between the sensors (ΔT) after Granier (1987):

image(1)

where ΔTmax is the value recorded at night under conditions of zero flow. In order to account for night-time sap flow, maximum values of night-time ΔT were considered ΔTmax only when values of D were less than 0.2 kPa.

Sap flux measurements were made from May 2008 through January 2009. At the end of the measurement period, the trees were cored both at sensor height and at 1.35 m from the ground, and the sapwood depths were determined visually. Sap flux rates in the outer 2 cm of sapwood (JO, g cm−2 s−1) were standardized to breast height (1.35 m from the ground) by multiplying JO by the ratio of sapwood area at sensor height to sapwood area at breast height.

Leaf water potential measurements

We measured midday leaf water potentials (ΨMd) monthly from July through November 2008. We collected one sunlit branch – within 6 m of the ground – from each tree with installed sap flux sensors (except one tree in Los Angeles that did not have accessible branches). Leaf water potentials were measured between noon and 1400 h, immediately after the collection of each branch, with a Scholander-type pressure chamber (PMS Instrument Company, Albany, OR, USA). We report only the midday leaf water potentials because pre-dawn measurements were not logistically possible at our study sites.

Plant hydraulic properties

Hydraulic conductivity

The hydraulic conductivity of lower canopy branches (up to 6 m from the ground) was measured once a month from August 2008 through May 2009. Six branches (≥20 cm length) with fully developed leaves were collected from six trees at each study location and transported to the lab in sealed plastic bags with wet paper towels. At the Los Angeles site, branches could only be accessed on four trees. Thus, to maintain a consistent sample size, branches from two other coast redwood trees were substituted. Hydraulic conductivity was measured the same day as collection with the gravimetric apparatus described by Sperry, Donnelly & Tyree (1988). Branches were immersed in purified water, cut to 14 cm length and cleaned of bark at the ends. The rate of water flow caused by gravimetric pressure was measured using a balance (accuSeries; Fisher Scientific Inc., Waltham, MA, USA). Following Sperry et al. (1988), we measured purified water temperature (varying between 20 and 25 °C) and recalculated the conductivities to 20 °C based on the temperature dependence of the water viscosity. Stem area specific hydraulic conductivity (K, mg s−1 MPa−1 mm−2) was calculated after Sperry et al. (1988).

Our inability to remove native emboli by flushing samples did not allow us to determine % losses of K with respect to maximum conductivity of each sample. This is a common methodological limitation with conifer wood (Burgess, Pittermann & Dawson 2006; and references therein), although Ambrose, Sillett & Dawson (2009) reported that soaking samples overnight in filtered water under vacuum removed emboli. While we did not use this method in our study, we assumed that: (1) the maximum K over the period of the study represented the maximum functional K; and (2) 100% hydraulic conductivity did not vary strongly within sites during the study (Stout & Sala 2003). To evaluate temporal trends in embolism based on these assumptions, we used maximum K measured for each site over the period of the study as Kmax to calculate percent losses.

Vulnerability to cavitation

Vulnerability to cavitation was measured on stems collected in December 2008 using centrifugal force following Alder et al. (1997), with the ‘flushing’ step removed (because of the methodological limitation described earlier). Following Burgess et al. (2006), we collected the stems on a day following ample rain to maximize initial K. Then, we calculated % losses with respect to functional K (assuming that they represent Kmax). The negative pressure at which 50% of maximum K was lost because of induced cavitation (P50) was determined from vulnerability curves.

Tree transpiration

Tree-level sap flux was calculated as the sum of daily sap flux density measured in the outer 2 cm of sapwood (JO, g cm−2 d−1) and the sap flux density in the deeper layer of sapwood (J1, g cm−2 d−1), which was derived using the empirical function for gymnosperms presented by Pataki et al. (2010) as:

image(2)

where d1 is the depth of the inner sapwood layer. Whole-tree transpiration (ET, kg d−1) was then calculated as:

image(3)

where A0 and A1 are the sapwood areas in cm2 corresponding to the outer and inner increments of sapwood.

To determine the plot-level transpiration, ET was summed for all individuals of coast redwood and divided by the plot area. Estimates are reported only for coast redwood and do not include individuals of other species or understory plants.

RESULTS

Environmental conditions

There was no evidence of frequent summer fog in this study. In our record, summertime humidity reached 95% on one occasion only in Irvine (day 175, from 0600 to 0700 h), and the two other sites did not experience relative humidity ≥95% during summer.

To evaluate seasonal changes in D and IO that were unrelated to day length, we normalized daily averaged D and daily sum of IO with day length/12 h (Fig. 1a). Day-length corrected daily sums of IO did not differ between locations until July. Starting in mid-July until November (days ∼200–300), IO was greater in Irvine than the other sites (P < 0.001).

Figure 1.

Daily sum of photosynthetically active radiation IO (a), mean daytime vapour pressure deficit of air D (b) and relative soil water content Θ (c) at each location. IO and D are day-length corrected (see text for details). Error bars for Θ represent one average standard error.

Daily averaged D varied from ∼0.5 to ∼2 kPa and reached ∼3 to ∼4 kPa when weather conditions were influenced by Santa Ana winds (Fig. 1b). These are strong offshore winds associated with low atmospheric humidity. Santa Ana wind conditions were especially pronounced at the end of the record, starting in September. Irvine is located closer to the coast than the other two sites, and the local D in Irvine was significantly lower than at the other sites (P < 0.001).

Soil moisture at all locations remained high for most of the study period because of irrigation inputs. We calculated the relative water content (Θ) by dividing the measured soil water content by the maximum value recorded at each location. At all locations, Θ was in the range of ∼0.6–0.9 during the summer, and increased to ∼0.7–1.0 in late November (Fig. 1c). Soil moisture in Irvine exhibited the smallest day-to-day variability compared to the other two sites. The highest absolute and relative soil water contents were recorded at the Fullerton site (P < 0.001).

Responses of sap flux to the environment

There was a seasonal pattern of JO at each study site (Fig. 2) with the highest values at the beginning and end of the record, and a gradual decline from July through December. JO in Fullerton and Irvine reached a maximum of ∼130 g cm−2 d−1, which was significantly higher than JO in Los Angeles (P = 0.001 with respect to Fullerton, and P = 0.01 with respect to Irvine).

Figure 2.

Time series of the daily sum of sap flux in the outer 2 cm of sapwood (JO) for the year 2008. Gaps in the record are present because of missing data. Error bars represent one standard error.

JO responded to D with a saturating function at all study sites, indicating considerable stomatal control (Fig. 3). We fit a linear function of the logarithm of D to these data:

Figure 3.

The daily sum of sap flux in the outer 2 cm of sapwood (JO) as a function of a day-length corrected daytime vapour pressure deficit (D). Days are numbered for the year 2008, continuing through the mid-January of the year 2009. Fitted curves correspond to the equation JO = a + blnD. Coefficients and adjusted R2 values are shown in Table 2.

image(4)

To evaluate seasonal changes (largely corresponding to different light regimes) in this relationship, we divided the data set into three periods: days 129–235 (May–August), days 236–286 (August–October) and days 287–18 (October–January). The parameters of the model changed from period 1 through period 3 as sap flux declined over time (Fig. 3; Table 2). At each study site, JO was significantly higher in May–August compared to October–January for the same value of D (P = 0.002). The best fit of JO to D was obtained for October–January, probably in response to high D during that time. JO during August–October was marginally higher than October–January (P = 0.09), and did not significantly differ from May–August. A repeated measures analysis of variance (anova) of the slopes b in Eqn 4 showed that JO in Los Angeles was less sensitive to D than in Fullerton (P = 0.026) regardless of time period (Table 2). In Irvine, b was not significantly different from the other two sites.

Table 2.  Average regression coefficients and adjusted R2 values of the linear models JO = a + b · ln D, JO = c · IO, and JO = a + b · ln D + c · IO
EquationDaysFullertonLos AngelesIrvine
abcadj. R2abcadj. R2abcadj. R2
  1. D and IO are day-length corrected. Values are accompanied by one standard error.

JO = a + b ln D129–235102.4 ± 16.745.8 ± 8.30.62 ± 0.0557.4 ± 10.519.5 ± 3.80.25 ± 0.08107.6 ± 15.925.3 ± 3.90.45 ± 0.06
236–28685.8 ± 14.827.2 ± 6.40.44 ± 0.0734.6 ± 8.510.4 ± 3.80.21 ± 0.0679.7 ± 13.010.2 ± 3.00.20 ± 0.08
287–1857.0 ± 9.428.7 ± 4.70.79 ± 0.0322.3 ± 7.211.1 ± 4.10.48 ± 0.1356.1 ± 9.921.3 ± 3.70.56 ± 0.08
JO = c · IO129–182.6 ± 0.40.94 ± 0.011.1 ± 0.30.86 ± 0.032.2 ± 0.30.94 ± 0.01
JO = a + b ln D cIO129–1818.7 ± 10.015.3 ± 4.52.03 ± 0.470.71 ± 0.01−6.84 ± 5.551.85 ± 2.151.34 ± 0.180.51 ± 0.0213.5 ± 7.34.35 ± 2.761.84 ± 0.260.60 ± 0.03

We also observed a strong linear relationship between JO and day-length corrected daily IO (Fig. 4; Table 2). Surprisingly, IO explained more of the temporal variability in JO than D. Because IO and D are correlated with each other, both show significant relationships with JO (Figs 3 & 4), but the relationship with IO does not shift seasonally and has a higher R2 for the entire study period. Analysis of the slopes of this relationship for the three sites indicated that JO in Los Angeles was less sensitive to light than in Fullerton (P = 0.001) and Irvine (P = 0.019). There was no significant difference in the light sensitivities of JO in Fullerton and Irvine.

Figure 4.

The daily sum of sap flux in the outer 2 cm of sapwood (JO) as a function of day-length corrected photosynthetically active radiation (IO). Fitted curves show the linear model JO = a · IO. Coefficients and adjusted R2 values are shown in Table 2.

We combined the observed relationships in the model:

image(5)

This function captured a large part of the variability in JO (R2 = 0.51–0.71 across sites) for each location during the entire study period (Table 2). There were no significant differences in model coefficients among sites. Regression analysis of the residuals of this relationship against Θ showed very weak correlations for some trees (both negative and positive; P values from <0.001 to 0.9, site-averaged R2 from 0.03 ± 0.02 to 0.06 ± 0.02), but Θ generally did not exhibit a large-enough range to result in biologically meaningful relationships.

Leaf water potentials

ΨMd varied between −2.4 ± 0.1 MPa (Irvine, July) and −1.1 ± 0.1 MPa (Los Angeles, November). Overall, ΨMd in Irvine was more negative compared to Fullerton and Los Angeles (P < 0.001), while ΨMd in Fullerton and Los Angeles did not differ (P >> 0.05, Fig. 5). According to a repeated measures anova (α = 0.05), there were no significant differences between monthly ΨMd in Fullerton. In Los Angeles, ΨMd was significantly lower in October compared to November (P << 0.05). ΨMd in Irvine was more negative in July compared to August, October and November (P << 0.05), and also in September compared to August (P = 0.04) and November (P = 0.03). In July, ΨMd in Irvine was lower than both in Fullerton and Los Angeles, and in September and October, ΨMd in Irvine was only lower compared to Fullerton (P << 0.05 for all, Fig. 5).

Figure 5.

Midday leaf water potentials of sunlit branches collected once a month from heights up to 6 m. Asterisks indicate significant differences among sites in July, September and October. Error bars represent one standard error.

Hydraulic properties

There were no statistical differences in vulnerability to cavitation among sites (P > 0.1), although P50 varied from −4.1 ± 0.6 MPa in Los Angeles to −5.9 ± 0.6 MPa in Irvine (Fig. 6). Monthly values of K varied from 2.3 ± 0.5 mg s−1 MPa−1 mm−2 (October, Irvine) to 5.1 ± 0.5 mg s−1 MPa−1 mm−2 (February, Los Angeles), increasing from October through January at each study site (P = 0.003, Fig. 7b). Percent loss of K reached its maximum in October, with site averages of 52.3% ± 9.6 in Irvine, 41.7% ± 18.5 in Fullerton and 26.6% ± 8.0 in Los Angeles, and then declined through January to ≤17% (P = 0.003, not shown). During the entire study period, K in Los Angeles was marginally greater than in Fullerton (P = 0.07) and Irvine (P = 0.05).

Figure 6.

Percent loss of hydraulic conductivity (K) induced by cavitation and corresponding values of stem area-specific K. Sigmoidal curves (y = a{1 + exp[−(x − x0)/b]}−1) are fitted to the data. Error bars represent one standard error.

Figure 7.

(a) Monthly totals of JO from this study and from Burgess & Dawson (2004). (b) Stem area-specific K of stems at the study locations for 2008–2009. Stems were collected once a month. (c) Mean daytime temperatures. Error bars represent one standard error.

Tree and plot-level transpiration

Average whole-tree transpiration (ET) was ∼30 kg d−1 in Fullerton and Irvine, and ∼15 kg d−1 in Los Angeles, with declines at the end of the year (Fig. 8a). Values of tree transpiration were similar to JO because of the shallow sapwood depths (Table 1). Plot-level transpiration (EC, Fig. 8b) had the same temporal trend as ET, but EC in Los Angeles and Irvine was lower than in Fullerton because of the difference in tree density.

Figure 8.

Time series of tree transpiration ET (a) and plot-level transpiration EC (b) for the year 2008. Gaps correspond to missing values. Error bars indicate the standard deviation.

DISCUSSION

Contrary to our initial hypothesis, coast redwood exhibited strong stomatal regulation in response to atmospheric D, and IO was also important in explaining temporal variability of JO. Seasonal changes in K indicated the presence of xylem cavitation and its subsequent reversal. In agreement with our expectations, we did not observe a pronounced response of sap flux to soil moisture, which remained high and unlimiting to transpiration. In addition, plot-level transpiration rates of stands of irrigated coast redwood were relatively low, and did not exceed 1 mm d−1 in our study.

Sap flux and stomatal responses to environmental variables

To compare JO from this study with the reported values in natural redwood forests, we converted the units of sap velocity used by Burgess & Dawson (2004) to mass flow rates (by multiplying by the density of water). In both this study and Burgess & Dawson (2004), the minimum sap flux was measured in December and was 0.5 kg cm−2 month−1 in the natural redwood forest, and 1.1 ± 0.1 kg cm−2 month−1 in this study (cross-site average). The maximum sap flux was 4.6 kg cm−2 month−1 in the natural redwood forest (June), and 2.5 ± 0.3 kg cm−2 month−1 in this study (August, cross-site average). The differences in methodologies prevent direct comparison, but we note that despite contrasting weather regimes, tree sizes and ages, our measured JO values were of the same magnitude as the trees in native forests (Fig. 7a). We observed a pronounced saturation of daily JO starting at D < 1 kPa (Fig. 3), and indicating that leaf stomata of coast redwood were partially closing in response to D. Similar stomatal behaviour, with saturation at D < 1, was reported for transpiration rates near the tops of tall coast redwood trees in their native environment (Ambrose et al. 2010).

In addition to saturation of JO with D, the irrigated coast redwood showed a strong linear increase of JO with IO (Fig. 4), indicating efficient utilization of light. This ability to utilize IO across a range of intensities with a linear relationship between JO and IO has been found in many conifer species [e.g. Picea mariana and Pinus banksiana (Ewers et al. 2005); Tsuga canadensis (Tang et al. 2006)]. The foliage of mature coast redwood is known to be efficient in harvesting light within large, self-shaded canopies (Steinberg 1996; Burgess et al. 2006), although coast redwood seedlings exhibit shade avoidance (Peer, Briggs & Langenheim 1999) and do not usually grow in the understory of old forests (Lorimer et al. 2009). The strong linear response found in this study is consistent with the highly significant linear relationship between treetop transpiration and IO (Ambrose et al. 2010).

Hydraulic conductivity and vulnerability

We found branches of coast redwood to be relatively resistant to cavitation. P50 values of −4.1 ± 0.6 MPa to −5.9 ± 0.6 MPa (Fig. 6) are relatively negative compared to many conifer species (Sperry & Tyree 1990; Cochard 1992; Oliveras et al. 2003; Maherali, Pockman & Jackson 2004; Schoonmaker et al. 2010). In addition, our vulnerability curves were similar to previous reports in natural coast redwood forests (Koch et al. 2004; Burgess et al. 2006; Pittermann et al. 2006). For example, Burgess et al. (2006) reported P50 of −5.8 ± 0.0 MPa for branches at 30 m canopy height, which is similar to our cross-site average of −5.3 ± 0.4 MPa. Resistance to cavitation may be a genetic adaptation to low xylem water potentials that may occur as redwoods reach their mature heights (Koch et al. 2004).

Our measurements of K were somewhat higher, but still close to those reported in Burgess et al. (2006) (0.5–0.6 kg m−1 MPa−1 s−1 in this study; 0.2–0.4 kg m−1 MPa−1 s−1 in the previous study). Similarity of K in contrasting environments has been reported previously in Pinus ponderosa (Maherali & DeLucia 2000) and Platanus racemosa (McCarthy & Pataki 2010). The seasonal changes of K that we observed most likely indicate xylem cavitation (Stout & Sala 2003). Measured values of K were significantly lower in October compared to January (by 1.12 ± 0.74 mg s−1 MPa−1 mm−2 in Los Angeles, by 1.23 ± 1.01 mg s−1 MPa−1 mm−2 in Fullerton and by 2.48 ± 0.67 mg s−1 MPa−1 mm−2 in Irvine; P = 0.003). According to our estimates, percent losses of K declined from ∼27–52% in October to ∼0–17% in January. This indicates that declining JO (P = 0.002, Fig. 7a) coupled with the factors discussed below may have allowed a reversal of earlier cavitation from October through January, causing an increase in K (P = 0.003, Fig. 7b). Assuming that our measurements adequately reflect native embolism, we conclude that in spite of observed stomatal regulation, tree branches experienced cavitation during summer, with subsequent recovery later in the season.

We do not attribute xylem recovery to conduit refilling as a result of increased soil water availability, because soil moisture did not show large seasonal variations. Possibly, the increase in K was caused by new sapwood development (Brodribb & Cochard 2009). However, there has been evidence of cavitation reversal in conifer tracheids under negative pressure (Borghetti et al. 1991; Sobrado, Grace & Jarvis 1992; Edwards et al. 1994; Zwieniecki & Holbrook 1998). When temperatures drop, gas solubility increases allowing gas dissolution into adjacent functional xylem (Schenk & Espino 2010). If transpiration also occurs, there may be embolism repair (Edwards et al. 1994). Earlier, Burgess & Dawson (2004) reported significant night-time transpiration in natural redwood forests driven by very high D. We suggest that coast redwood might have benefited from colder night-time temperatures coupled with high D conditions to refill cavitated tracheids under negative pressure. In this study, we did not observe night-time transpiration, but embolism recovery occurred during the period with the highest D (Fig. 1b), relatively high rates of transpiration (Fig. 2) and decreasing mean daytime temperatures (Fig. 7c). Therefore, we speculate that in addition to new sapwood growth, the seasonal increase in K observed in this study may have resulted from daytime embolism repair facilitated by lower temperatures coupled with appreciable transpiration.

Water relations and tree function

Leaf chlorosis and senescence in irrigated coast redwood could be related to water relations through: (1) vascular failure caused by negative water potentials; and/or (2) inadequate carbon uptake caused by stomatal closure. In our study, because Θ remained high and non-limiting, soil water stress was unlikely to be an issue. While atmospheric water stress is possible, ΨMd reached, but did not exceed, the values corresponding to minor losses of K (Figs 5 & 6). Koch et al. (2004) previously observed the same pattern in coast redwood and attributed it to cavitation avoidance, a strategy beneficial for tall trees. Coast redwood can presumably tolerate cavitation to some extent, but the tops of the tallest trees may be less capable of recovery (Koch et al. 2004). Cavitation avoidance may be the key to the longevity of coast redwood, allowing it to survive through historical droughts (Stine 1994; Noss 2000; Koch et al. 2004), and grow to the greatest tree heights in the world. Although the coast redwoods in this study were not tall, they seem to exhibit similar stomatal behaviour, with strong regulation of ΨMd that did not result in excessive xylem cavitation. Hence, if the monthly changes in K were caused by seasonal cavitation and its reversal, the degree of cavitation was likely not damaging. We conclude that coast redwood might have experienced atmospheric water stress, but without extreme consequences such as severe cavitation and hydraulic system failure.

In addition to water stress, strong stomatal control of transpiration (Fig. 3) may result in carbon limitation (Lange et al. 1971; Bunce 1981; Chaves 1991). To evaluate whether the urban trees that we measured showed suppressed stomatal conductance caused by high D relative to natural forests, we substituted the environmental variables reported in Ambrose et al. (2010) from June to October 2008 into the empirical model for JO (Eqn 5). In that study, daily average D was generally lower than in our study: it varied between ∼0 and ∼2 kPa and rarely reached ∼3 kPa (Ambrose et al. 2010). Nevertheless, as JO saturation starts at D < 1 kPa, the modelled monthly JO values were very similar to our measurements. However, differences in D with similar JO imply differences in canopy conductance. We divided JO by D as a simple proxy of canopy conductance, and in fact, canopy conductance is up to ∼40% higher in the natural forest compared to the Los Angeles Basin. Therefore, trees in the Los Angeles basin appear to have lower canopy conductance than natural coast redwood.

Moreover, in addition to experiencing higher D, the studied trees were not exposed to any summer fog, in contrast to their native habitat where Burgess & Dawson (2004) reported fog during 11 days and 20 nights in summer. Nearly zero daytime JO in coast redwood often occurred during and following foggy conditions in its native habitat (Burgess & Dawson 2004). We never observed such patterns of JO in our study; instead, urban coast redwood exhibited detectable and positive daytime JO. Fog was shown to increase water potential in coast redwood foliage (Burgess & Dawson 2004; Limm et al. 2009; Simonin, Santiago & Dawson 2009), leading to increased stomatal conductance and photosynthetic rates (Simonin et al. 2009). Considering that partial stomatal closure starts at low D, fog episodes may therefore allow coast redwood to reach maximal stomatal conductance and high rates of carbon assimilation. We cannot evaluate the effect of high D and the absence of fog on carbon uptake directly with our measurements; however, this is a plausible mechanism for poor performance in these trees based on our conclusions about their water relations.

Tree and stand transpiration

Our estimates of ET (Fig. 8a) are relatively low compared to other irrigated tree species measured in the Los Angeles metropolitan area (Pataki et al. 2010). Pataki et al. (2010) found that urban ET differs by an order of magnitude among tree species, with the low-transpiring group including Australian species (Brachychiton spp.) and tropical trees (Jacaranda mimosifolia), in addition to coast redwood.

Of the three sites in the current study, the lowest values of ET were found at the Los Angeles site. Trees at this site were much older than those at the other sites (Table 1), which might explain some of the differences. Age-related effects on transpiration that have been found in conifer trees include a smaller sapwood depth [in Picea abies (Alsheimer et al. 1998)], a decline of K[in Pinus sylvestris (Mencuccini & Grace 1996; Cienciala et al. 1997)] and lower stomatal conductance [in Pinus ponderosa (Hubbard, Bond & Ryan 1999)]. However, previous studies of age-related changes in transpiration were conducted in natural conifer forests where older trees were always associated with greater diameter and height. In our study, tree height, sapwood depth and diameter of older coast redwoods in Los Angeles were not significantly different from those in the other locations (P >> 0.5). Furthermore, branch K in Los Angeles was not lower than that in Fullerton and Irvine (K in Los Angeles was even marginally greater, P = 0.05–0.07; Fig. 7b). However, lower ET in Los Angeles may be linked to age-related declines in stomatal sensitivity caused by a potential decrease in whole-tree K with age (Ryan & Yoder 1997). Indeed, the repeated measures anova of slopes of the relationships JO(D) and JO(IO) (Table 2) indicated lower sensitivity of JO to D in Los Angeles compared to Fullerton (P = 0.026, Fig. 3), and to IO compared to the other sites (P = 0.001 with respect to Fullerton, and P = 0.019 with respect to Irvine, Fig. 4). Therefore, the tree ages in Los Angeles could have contributed to the lower ET.

While some of the differences in EC among sites were attributable to site differences in ET (Fig. 8b), this did not explain all of the spatial variability. Much of the remaining variability can be explained by tree density, which was highest at the Fullerton site (Table 1). The urban forests at both the Los Angeles and Irvine sites included other tree species and understory vegetation (not accounted for in the EC estimates), and had a relatively low density of coast redwoods. With a density of 190 trees ha−1 in Fullerton, the relatively low ET of coast redwood produced EC of ∼0.1–0.9 mm d−1 (Fig. 8b).

To our knowledge, there are no published estimates of EC of coast redwood in its natural habitat. Literature values of EC from conifer stands vary from ∼0.1–1.8 mm d−1 for low productivity, old growth temperate coniferous rainforests (Barbour et al. 2005) to 0.95–4.64 mm d−1 for boreal forests (Lundblad & Lindroth 2002). We cannot directly compare our urban EC estimates with EC from natural forests because of the differences in tree species, stand densities, age and environmental conditions. Overall, our EC estimates (Fig. 8b) are at the low range of values reported for boreal and temperate coniferous forests (Granier et al. 1990; Sellers et al. 1995; Loustau et al. 1996; Cienciala et al. 1997; Kelliher et al. 1998; Pataki, Oren & Smith 2000; Zimmermann et al. 2000; Lundblad & Lindroth 2002; Barbour et al. 2005). In addition, our estimates are approximately an order of magnitude lower than rates reported for natural riparian forests, which have similar environmental conditions of both D and high soil water availability, but may have EC exceeding 10 mm d−1 (Pataki et al. 2005).

Implications

Based on the unique characteristics of coast redwood, including large leaf area index (Westman & Whittaker 1975) and fast growth (Noss 2000), this species has the potential to provide significant benefits to urban residents (such as shade and carbon uptake). It is also native and culturally significant for residents of California (Ricketts et al. 1999; Noss 2000). Our results indicate that there is another desirable aspect of planting coast redwood in urban environments: ET is relatively low compared to other urban tree species in Los Angeles (Pataki et al. 2010), which is advantageous for water conservation (Hayhoe et al. 2004; Kundzewicz et al. 2007).

However, irrigated coast redwood often exhibits leaf chlorosis and senescence in the semi-arid region of southern California (Barnhill, personal communications; Downer 2004). Our results suggest that water relations are not the direct cause of poor performance. Coast redwood was not lacking in stomatal control of water loss; instead, we observed relatively strong stomatal regulation (Fig. 3). In addition, coast redwood maintained leaf water potentials that seemed to avoid excessive xylem cavitation of the branches (Figs 5 & 6). On the other hand, partial stomatal closure in response to high D and the absence of coastal fog may cause carbon limitation. A more extreme mechanism that involves complete or almost complete stomatal closure has been widely discussed as one of the possible causes of widespread tree mortality in the southwestern USA (‘carbon starvation hypothesis’, McDowell et al. 2008; Sala, Piper & Hoch 2010 and references therein). In addition to carbon limitation, long-term partial stomatal closure may cause photodamage (Cornic & Massacci 1996) and leaf overheating (Kolb & Robberecht 1996; Martin et al. 1999). Because of their needle-shaped leaves, conifers are usually able to sustain close-to-ambient leaf temperatures and avoid overheating during water stress (Gates 1968; Waring & Franklin 1979). However, the drier and warmer urban environment in Los Angeles may exceed the limits of adaptation to light and temperature, and could result in photo- and thermal damage (Allakhverdiev et al. 2007; Wahid et al. 2007; Takahashi & Murata 2008). Another possible reason for poor performance could be nutrient deficiency (Barnhill, personal communications) or salinity stress. Nutrient deficiency is less likely because conifers from the coastal Pacific region have very low nutrient requirements (Waring & Franklin 1979). Wu & Guo (2006) attributed leaf chlorosis and reduced growth of coast redwood in the San Francisco Bay Area to the salt stress caused by the application of recycled irrigation water. In our study, recycled water was used at the Irvine site, but not at the other two sites, and trees showed signs of chlorosis at all three study sites. Further investigation is needed to determine the mechanisms of chlorosis and leaf senescence of coast redwood in southern California. Our study shows that despite this senescence, the hydraulic system is remaining functional in the local climate.

This study illustrates that a priori assumptions about functioning of planted trees in southern California based on their physiological behaviour in native environments may not be justified (see also Bush et al. 2008; McCarthy & Pataki 2010). Given the diversity of landscape trees in the Los Angeles metropolitan area, where no particular tree species or functional group is dominant (Weller et al., unpublished data), it is essential to study in situ the physiological behaviours of many species from different habitats. Further studies are necessary to better understand the water use of other common urban tree species in areas with scarce water resources. These studies can provide urban planners with the ability to better assess the water requirements of urban forests.

ACKNOWLEDGMENTS

We thank University of California, Irvine Facilities Management, Fullerton Arboretum and the Los Angeles Police Academy for permission to conduct research on their properties. We also thank Chris Barnhill for valuable information, and Neeta Bijoor, Thomas Gocke, Christine Goedhart, Eric Nguyen and Amy Townsend-Small who provided assistance in the field. Michael Goulden provided helpful comments on an earlier version of this manuscript. This study was supported by the National Science Foundation (HSD 0624342) and the Environmental Protection Agency (RD-83336401-0).

Ancillary