Comparative water relations of co-occurring trees in a mixed podocarp-broadleaf forest

Abstract

Aims

As extreme climatic events including droughts and heat waves become more common in a changing climate, tree mortality has increased across the globe. In order to determine whether certain species have a competitive advantage over others, we explored the water-relations and leaf-gas exchange of four co-occurring species in a forest in northern Aotearoa-New Zealand. We studied the ecologically and culturally significant foundation species, Agathis australis (a conifer), two additional conifers, Phyllocladus trichomanoides and Podocarpus totara and the angiosperm Knightia excelsa.

Methods

We measured sap flow, leaf-gas exchange and xylem water potentials of leaves and terminal branches with concurrent measures of micrometeorological data on days with very few clouds. We derived whole tree hydraulic conductance and instantaneous water-use efficiency (WUE_i) at our remnant forest in west Auckland during February 2015 (southern hemisphere summer).

Important Findings

The four species behaved similarly in their diurnal curves of gas exchange and water potential. Rates of assimilation, stomatal conductance and WUE_i were similar among trees of different species. Whole tree hydraulic conductance was also similar among species. These results indicate functional convergence in water relations, possibly driven by low nutrient soils at the site. Our results suggest that there is no species with a clear adaptive advantage over the others in the context of climate change.

southern conifers, sap flow, water potential, hydraulic conductance, leaf-gas exchange, functional convergence

INTRODUCTION

Foundation species can shape and define entire ecosystems, due to their influence on hydrology, nutrient cycling and other ecosystem processes (Dayton 1973; Ellison et al. 2005). These locally abundant and dominant trees (Barbour et al. 2009) contribute substantially to ecosystem processes (Martin and Goebel 2013). They influence the local microclimate because of their canopy architecture and physiology (Ellison et al. 2005). Foundation species control the abundance and distribution of plants in their vicinity and the interplay of facilitation and competition within the ecosystem (Callaway and Walker 1997; Ellison et al. 2005).

The southern conifer Agathis australis (D. Don) Lindl. (Araucariaceae, kauri) is one of the largest and longest-lived trees in the world (Ecroyd 1982) and forms the species-rich ‘kauri forest’ (Ogden 1995; Wyse 2012), a temperate rainforest ecosystem, home of many endemic plant species. As a foundation species, A. australis shapes the plant assemblage in its vicinity (Wyse et al. 2013,, 2014). Agathis australis litter decomposes slowly (Enright and Ogden 1987) and creates very acidic, podsolised soil (Verkaik and Braakhekke 2007; Verkaik et al. 2007; Wyse 2013). Agathis australis seems to be responsible for stalling of the nitrogen cycle beneath and close to larger individuals (Enright 2001; Silvester 2000; Wyse et al. 2014). The litter layer beneath A. australis is often considerably drier than the soil of nearby angiosperm dominated forests because of the physical and chemical properties of the litter (Verkaik and Braakhekke 2007). Additionally, A. australis dominated forests frequently occur on ridges and slopes (Steward and Beveridge 2010), which further intensifies soil dryness.

Interspecific functional convergence has been observed in several biomes (Meinzer 2003). Under selective pressure, appropriate resource allocation is vital to achieve optimum plant fitness (Bucci et al. 2004). The enhancement of certain functions always comes at the expense of others, due to universal physiological constraints (Meinzer 2003). There is only a limited number of combinations of functional traits (Bucci et al. 2004) and they all fall on a universal trade-off surface (Díaz et al. 2016; Reich et al. 1997). Patterns in functional convergence of leaf traits are comparatively well-known (Reich et al. 1997, 2003; Wright and Westoby 2002). A prominent example is the relationship between leaf nitrogen and photosynthetic assimilation rate (Wright et al. 2004), but there are fewer studies addressing convergence in plant water relations (Macinnis-Ng et al. 2004; Zeppel 2013).

In temperate and boreal forests, nitrogen availability is the most common limitation of soil fertility (Corlett 2016). It is often related to individual foundation species, mainly conifers (Persson and Wirén 1995; Ste-Marie and Paré 1999; Wyse et al. 2014) affecting the local nitrogen cycle. Nitrogen depletion is also a particular feature of old growth forests, as macronutrients (including N) are locked in established biomass and soil organic matter over time (Gower et al. 1996; Ryan and Yoder 1997).

Overall, our knowledge about the ecophysiology of New Zealand native trees is still limited. However, we do know that all four study species (A. australis, Phyllocladus trichomanoides, Podocarpus totara, Knightia excelsa) can grow on infertile soils (Baylis et al. 1963; Keng 1978; Thomas and Spurway 1987). Jager et al. (2015) measured leaf traits of all species in an A. australis dominated forest and all four showed similarly low concentrations of leaf nitrogen of 0.6 to 0.8%. The xylem embolism vulnerability of our study species varies considerably. Agathisaustralis is highly susceptible to xylem embolism loosing conductivity at relatively low xylem pressure (P₅₀ measured by Pittermann et al. (2006): −2.3 MPa). Phyllocladus trichomanoides and P. totara on the other hand, showed much lower vulnerability with a P₅₀ of −6.6 MPa (Pittermann et al. 2006) and −6.3 MPa (MJ Clearwater, unpublished data), respectively. Recent work shows that A. australis is equipped to protect its hydraulic integrity. Agathis australis seedlings (Wyse 2012) as well as young forest trees (Macinnis-Ng et al. 2017) exhibited conservative stomatal control and mature trees reduced sap flow during drought conditions (Macinnis-Ng et al. 2016). Temperate podocarps are generally tolerant of dryer conditions (Coomes and Bellingham 2011) and there is some evidence that P. totara is a drought-avoiding species (Innes and Kelly 1992). We have very little quantitative ecophysiological knowledge about K. excelsa, but Metcalf (2011) suggests tolerance of dry conditions, in potted saplings.

We investigated whole tree water-use and leaf water relations of A. australis and three co-occurring species to evaluate differences in a forest in West Auckland, in order to gain insight into the broad patterns and functions of ‘kauri forest’. We explored the diurnal patterns of sap flow, leaf water potential (Ψ) and leaf-level gas exchange including stomatal conductance (g_s) and assimilation rate (A) and their responses to environmental variables. We hypothesized, that A. australis would exhibit more conservative water use, lower A and g_s and therefore less negative Ψ. We expected P. totara, P. trichomanoides and K. excelsa to have less conservative water-use behaviour based on previously reported P₅₀ values.

METHODS

In this study, we explored the water relations of A. australis and three co-occurring species, the two podocarps P. trichomanoides D. Don (Phyllocladaceae, tanekaha) and P. totara G. Benn. (Podocarpaceae, totara) and the angiosperm K. excelsa R. Br. (Proteaceae, rewarewa). All four species are endemic and typically occur with A. australis (Burns and Smale 1990; McKelvey and Nicholls 1959; Nicholls 1976; Pook 1978), but A. australis has a much smaller geographical distribution than the other three species.

Study site

The study was carried out at the University of Auckland Huapai Scientific reserve, situated in the Northern foothills of the Waitakere ranges, West of Auckland (−36.796, 174.492, 90 m a.s.l.). The 15 ha reserve consists of mature podocarp-broadleaf forest (Thomas and Ogden 1983), that is dominated by A. australis along its two main ridges, with this species making up 80% of the basal area (Wunder et al. 2010) of the study plot. The other three study species contribute 12% of the basal area (P. trichomanoides 6%, P. totara 4%, K. excelsa 2%) (Wunder et al. 2010). The individual study trees were chosen to cover a range of sizes, but also for their accessibility for tree climbers (see Table 1 for more details of sample trees). We conducted leaf scale measurements on two consecutive days, measuring different trees each day.

Table 1:

overview of the studied trees

Tree	DBH (cm)	Canopy dominance	Sap wood depth (cm)	Fd	ΨL	A/gs	Leaf N (%)
AGau1	79.5	Dominant	14.5	x	x	x	0.73
AGau2	128.8	Dominant	17.8	x	x
AGau3	176.5	Dominant	17.8	x	x	x
PHtr1	33.1	Co-dominant	6.5	x			0.81
PHtr2	35.1	Co-dominant	7.3		x	x
POto1	54.3	Dominant	5.8	x	x		0.63
POto2	72.8	Co-dominant	7.9	x
KNex	19.6	Subordinate	8.4		x	x	0.64

AGau: Agathis australis, POto: Podocarpus totara, PHtr: Pyllocladus trichomanoides, KNex: Knightia excelsa. Canopy dominance taken from Wunder et al. (2010). Leaf nitrogen (N) content in % of dry weight for mature trees in Puketi kauri forest (Northland) from Jager et al. (2015). The x indicates on which trees the measurements of F_d (sap flux density), Ψ_L (leaf water potential) and A/g_s (assimilation rate and stomatal conductance) were taken.

Meteorological and soil moisture conditions

Meteorological data were recorded as 30-min averages on a nearby open paddock (<1 km from the forest site). Precipitation was measured using a tipping bucket (RIM8020, Measurement Engineering Australia (MEA), Magill, S.A.) with a resolution of 0.2 mm. A quantum light sensor (Li190, Licor, Lincoln, NE, USA) was used to measure photosynthetically active radiation (PAR, µmol m⁻² s⁻¹) and a cup anemometer and wind vane (WMS301, MEA) for wind speed (m s⁻¹) and wind direction (°). All three instruments were installed on top of a weather station tripod (MEA) ~2 m above ground. The probe to measure air temperature (T, °C) and humidity (RH, %) (HMP155, Vaisala HUMICAP, Vaanta, Finland) was housed in a radiation shield that was also attached to the tripod 1.5 m above ground. Vapour pressure deficit (VPD) was later calculated from T and RH. All meteorological data were recorded with a data logger (CR 10x, Campbell Scientific, Logan UT, USA).

Soil volumetric moisture content was recorded at two locations at four depths (soil-litter boundary, 10, 30 and 60 cm from the soil-litter-boundary) within the forest plot, in close proximity to all study trees. At both locations, 30-cm-long soil moisture probes were horizontally installed; water content reflectometers (CS616, Campbell Scientific) and measurements were recorded every 30 min using a data logger (CR10x). The Campbell Scientific calibration equation for litter and loam/clay (soil type described as Parau clay in Thomas and Ogden 1983) was used to calculate soil volumetric water content (%).

Sap flow

Sap flux density (F_d) was measured using customized Granier-type (Granier 1985) sap flow probes constructed according to James et al. (2002). Modifications and construction details are described in Macinnis-Ng et al. (2013) and Macinnis-Ng et al. (2016). Pairs of sensor probes were inserted into holes 10 cm apart on the vertical plane. In each pair, the upper probe was heated by four chip resistors with 660 Ω total resistance instead of nichrome wire used by James et al. (2002).The temperature difference between the two probes was measured with copper-constantan thermocouples every minute along a 1-cm-long aluminium tip. A Campbell CR10x logger (Campbell Scientific) recorded the mean temperature difference between the upper and lower probes every 15 min. After installation, all sensors were protected from sunlight with silver foil bubble-wrap. A minimum of two (up to five on the largest tree) sap flow sensors were installed on the north facing side of six A. australis trees in July 2011. Sensors in three P. trichomanoides and three P. totara trees were installed in June 2013. The system was powered with three 40 Ah, 12 V batteries that were exchanged every 2 weeks.

Sap wood depth for each tree was determined following the dye injection technique described by Meinzer et al. (2001) using a 0.1% carmine blue solution. A 5 mm increment borer (Haglöf Sweden AB, Långsele, Sweden) was used to drill a hole up to 30 cm deep and filled with dye solution. We topped up the dye every hour. After 4 h, a second core was taken 5 cm above the initial hole. The depth of the stained part of the new core was measured to indicate active sapwood. The dying experiment was conducted on sunny days when F_d is highest.

Water potential

Water potential (Ψ_l) was measured on short terminal branches for six trees across 2 days (see Table 1). The branches were harvested from the upper crown and for two A. australis (AGau2, AGau3) additionally from the mid and lower crown area (15 and 10 m below the upper crown sample collection). Special care was taken to always choose branches of similar length and width, as well as from the same height and location within the tree, to ensure minimal variability in hydrostatic water potential. Our samples were cut by professional climbers, immediately sealed in plastic bags to avoid further transpiration and then sent down to the forest floor where a runner collected the samples and delivered them to the pressure bomb operator. We used a custom-built Scholander-type pressure chamber to measure Ψ within minutes after harvest (Macinnis-Ng et al. 2016). For A. australis, the bark of the cut ends was carefully removed to prevent resin covering the xylem and the sap wood was freshly cut with a razor blade before insertion into the pressure bomb. Measurements were done on three branches per study tree and repeated in a 90-min rotation between 6 am and 6 pm.

Because of safety regulations restricting climbing before dawn, we measured Ψ_predawn on three terminal branches (three per study tree), sealed in aluminium foil and a plastic bag on the evening of the previous day. The measurements were done just after sunrise. Previous studies have shown that this technique produces accurate estimates of Ψ_predawn and can be used as a reliable surrogate for soil water potential (Ψ_soil) (O’Grady et al. 2006; Zeppel et al. 2008b).

Stomatal-density and leaf-gas exchange

Stomatal density counts were carried out on the abaxial and adaxial surface of 15 mature sun leaves (phyllodes for of P. trichomanoides) from every tree. Peels of each leaf were created using clear nail polish and attached to microscope slides. For each peel, eight random 1 mm² areas were photographed under a microscope (100× magnification, Leica MZ 16, Leica Microsystems, Wetzlar, Germany) and the stomata were counted. The stomatal density was later used to calculate the ratio of stomata on the abaxial and adaxial surfaces which is needed as an input for the CIRAS gas exchange system.

We measured diurnal leaf-gas exchange in-situ with a portable photosynthesis system (CIRAS-2, PP Systems, Boston, MA, USA) attached to a PLC6 automatic universal cuvette with the 25 × 7 mm insert. Ambient levels of temperature and humidity were maintained in the cuvette chamber. We also used ambient light for the measurements with the clear cuvette head window. The cuvette head was angled toward the sun in order to avoid shading during the measurements and whenever possible fully sunlit leaves were used. Reference CO₂ levels were set to 385 ppm and the flow rate to 200 ml min⁻¹. Five leaves from each tree were measured at least once every 2 h. When closing the cuvette head on the leaf, we waited until the CO₂ differential had stabilized before taking five recordings at 10-s intervals. These five recordings where then averaged to provide an indicative leaf value. No gas exchange measurements were conducted on P. totara, because the leaves were too small for the available cuvette insert.

Data analysis and calculation of whole-plant hydraulic conductance

Sap flux density (F_d) was calculated from the temperature differential using the Granier (1985) equation,

$${\text{F}}_d=119{((\Delta {\text{T}}_m-\Delta \text{T})/\Delta \text{T})}^{1.231}$$

where ΔT is the temperature difference between heated and unheated probe and ΔT_m is the temperature difference when F_d is zero. We determined ΔT_m by using the maximum ΔT from a 10 day window around each measurement. Scaling to whole tree water use for trees with multiple sensors (which was the case for the all A. australis) involved calculating a weighted average of all sensors based on the measured sapwood area. For trees with just one sensor and unknown radial sap flow profiles, we used the R sapflux package (Berdanier et al. 2016) to predict radial profiles.

Multiple regression was used to identify environmental drivers of F_d. First, we examined the variance inflation factor (VIF) to exclude predictors with strong collinearity. We followed the suggested threshold by Quinn and Keough (2002) and removed RH due to strong collinearity with VPD. Soil moisture of all three depths and the litter layer also showed strong collinearity, so we only used the litter layer for the regression analysis. All other predictor values had VIF below the suggested threshold. Second, we looked at the time lag that all trees show in response to environmental drivers, due to the size and volume of the tree boles. We used simple regression analysis to compare F_d and each environmental variable individually for different time lags, ranging from 0 to 270 min and then used an average of the best time lag for VPD and PAR as these were the most important drivers. Third, we examined the relationship of F_d and all predictors with multiple regression. In order to identify significant differences in water potentials between individual trees, we used analysis of variance (one way) and Tukey’s pairwise comparison after testing for homogeneity and normality. We only compared water potential means measured on the same day. Multiple regression for sap flow data and comparison of water potential were conducted in R, version 3.2.0 using the stats (R Core Team 2015) and asbio packages (Aho 2016).

Hydraulic conductance of whole trees was determined by plotting F_d against the difference in water potential between the root surface and leaf (Ψ_l − Ψ_Soil)(O’Grady et al. 2006; Zeppel et al. 2008a). Conductance was then calculated from the negative of the slope obtained by least square regression (O’Grady et al. 2006). The resulting conductance is sapwood area specific and therefore normalized by wood cross-sectional area (Becker et al. 1999). For every Ψ_l measurement, we used the mean of the closest previous and following F_d measurement, but due to the size of the trees we applied a time lag of 2 h between measured Ψ_l and F_d (Macinnis-Ng et al. 2016). We used Ψ_predawn as a surrogate for Ψ_Soil.

RESULTS

Meteorological conditions

The intensive sampling was conducted on the 17 and 18 February 2015. These days were reasonably sunny, with just a few intermittent clouds (Fig. 1). Due to the maritime climate of New Zealand, full sunshine days are rare. The previous months of December 2014 and January 2015 were dryer than usual, receiving ~50% less rainfall than the long-term average in Auckland. The average summer temperature was 0.8°C above normal (NIWA Seasonal Climate Summary). Nevertheless, soil moisture was still at 45.3 ± 0.6 % and 48.4 ± 1.0 % on the first day and 45.2 ± 0.6 % and 48.2 ± 0.9 % on the second day (at 30 cm and 60 cm, respectively, daily mean of two sensor locations ± standard error). Soil moisture decreased by 14% (at 30 cm) from the beginning of the southern summer until a larger rainfall event 10 days before the field campaign which caused an increase of 3%. Soil moisture was around 8% higher than during a drought in summer 2013, which showed little impact on water use of A. australis at this site (Macinnis-Ng et al. 2013). Overall soil moisture was representative of average summer conditions. Both days had comparable weather with maximum temperatures of 25.4 and 24.7°C and minimum temperatures of 7.5 and 7.6°C. Average wind speed and radiation were also similar (data not shown). Maximum VPD was 1.48 and 1.45 kPa on the first and second day, respectively.

Figure 1:

meteorological conditions during diurnal measurement of water potential, gas exchange and whole tree water use on 17 and 18 February 2015. The top panel shows air temperature, PAR and VPD measured on an open paddock around 1 km from the field site. The two bottom panels show whole tree water use on 17 and 18 February for Agathis australis (AGau), Podocarpus totara (POto) and Phyllocladus trichomanoides (PHtr) of different sizes (DBH of each tree indicated in cm).

Sap flux

The time of the onset of sap flow differed between trees (Fig. 1). For all trees, there was a clear time lag between sunrise and the increase of sap flux during late morning hours. The smallest A. australis (AGau1) showed the earliest onset at around 7:30 am, the second largest A. australis (AGau2) followed ~1 h later. This time difference appeared again for the timing of peak sap flow. Both trees showed a slight drop or plateau in sap flux at around 1 pm caused by a decline in PAR and air temperature, due to clouds coming through (Fig. 1). The patterns on the following day showed sap flow onset of A. australis 3 (AGau3) at 9 am, 1 h earlier than P. totara (POto) and P. trichomanoides (PHtr). Peak sap flux showed a similar time difference.

The strongest driver of sap flow (Table 2) overall was PAR explaining between 19 and 44% of variation of sap flow (P < 0.001 for all trees), followed by VPD explaining between 0 to 39% (P < 0.001 for all trees apart from P. trichomanoides). AGau2 was the only tree for which VPD was the stronger driver.

Table 2:

multiple regression analysis for half-hourly F_d (sap flux density) of the outer most sap flow sensor of Agathis australis (AGau), Podocarpus totara (POto) and Phyllocladus trichomanoides (PHtr) for January and February 2015

Tree		Tair	PAR	Wind speed	VPD	Litter
AGau1	0.8372
	Partial slope	-0.6609	0.0028	-0.1003	0.4945	0.0183
	P value	<0.001	<0.001	<0.001	<0.001	<0.001
	R2	0.0519	0.4028	0.0071	0.0395	0.0702
AGau2	0.9126
	Partial slope	0.0169	0.0035	0.0732	1.7590	-0.0023
	P value	<0.001	<0.001	0.0041	<0.001	0.1178
	R2	0.0059	0.4482	0.0029	0.2858	0.0011
AGau3	0.9067
	Partial slope	0.0234	0.0020	0.0514	1.7220	0.0087
	P value	<0.001	<0.001	0.0155	<0.001	<0.001
	R2	0.0162	0.2846	0.0021	0.3558	0.0169
POto1	0.8149
	Partial slope	0.0102	0.0017	0.1900	1.0390	0.0049
	P value	0.0053	<0.001	<0.001	<0.001	<0.001
	R2	0.0027	0.1903	0.0248	0.1527	0.0050
POto2	0.8455
	Partial slope	0.0284	0.0032	0.1124	1.2080	0.0374
	P value	<0.001	<0.001	<0.001	<0.001	<0.001
	R2	0.0114	0.3181	0.0048	0.1151	0.1393
PHtr1	0.8266
	Partial slope	0.0639	0.0020	0.1139	-0.0415	0.0251
	P value	<0.001	<0.001	<0.001	0.2630	<0.001
	R2	0.1454	0.3536	0.0141	0.0004	0.1678

Different time lags between F_d and environmental drivers (T_air = air temperature, PAR, VPD, litter = soil moisture of the litter-to-soil boundary) were applied to each tree: 180 min to AGau2, AGau3, POto 1; 150 min to AGau1 and PHtr; 210 min to POto2.

Water potential

Measured predawn water potentials (Fig. 2) ranged from −0.5 to −0.8 MPa for all species on both days. Knightia excelsa (KNex) had the least negative predawn water potential on the first day and A. australis 2 (AGau2) the most negative, and they were significantly different from each other (as shown by the letters in Fig. 2). On the second day, the largest A. australis (AGau3) showed the least negative water potential which was significantly different from P. totara (POto). Phyllocladus trichomanoides (PHtr) exhibited the most negative value. Over the course of the day, the water potential gradually decreased for all trees (Fig. 3), reaching a minimum between 2 and 4 pm. Knightia excelsa (KNex) showed the highest amplitude, but also the highest variability. The diurnal pattern of P. trichomanoides (PHtr) and P. totara (POto) was similar, with P. trichomanoides’ (PHtr) water potential dropping a little lower than P. totara’s (POto). For the two larger A. australis (AGau2, AGau3), we measured the water potentials in the lower/medium and upper crown. In both cases, the amplitude was higher for measurements from the upper crown, but the daily time-course was similar. Minimum water potentials did not significantly differ among trees on the first day (Fig. 2). On the second day, the values of the lower and medium crown of A. australis 3 (AGau3) were significantly (P < 0.05) less negative than the water potentials measured in the upper crown of the tree.

Figure 2:

predawn and minimum water potential of terminal branches measured on four different tree species (Agathis australis (AGau), Podocarpus totara (POto), Phyllocladus trichomanoides (PHtr) and Knightia excelsa (KNex)). Averages of three measurements are shown. Columns marked with different letters show statistically different water potentials (P < 0.05). For the two A. australis, we measured leaves at different locations within the crown (lower (l), middle (m), upper (u)) where the vertical distance between lower and upper measurements was 10 to 15 m.

Figure 3:

diurnal water potential of terminal branches measured on 17 and 18 February 2015 for Agathis australis (AGau), Podocarpus totara (POto), Phyllocladus trichomanoides (PHtr) and Knightia excelsa (KNex). On the two larger A. australis (AGau2 and AGau3), we measured the water potential in the upper part of the crown (_u), in the lower (_l) and middle (_m) crown.

Hydraulic conductance

Sapwood-specific hydraulic conductance was similar in the two large A. australis trees (AGau2 and 3) with values of 0.010 and 0.011 kg s⁻¹ MPa⁻¹. The two podocarps (POto, PHtr) had slightly lower conductance of around 0.008 kg s⁻¹ MPa⁻¹ and the smaller A. australis (AGau1) showed the lowest conductance with only 0.004 kg s⁻¹ MPa⁻¹ (Fig. 4). Even though the measurements were taken on two different days, we assumed that meteorological and soil moisture conditions were sufficiently similar to compare values from both days with each other.

Figure 4:

relationship of sap flux and leaf water potential (Ψ) for three Agathis australis (AGau 1–3) of different sizes (stem diameter is given in cm), Podocarpus totara (POto 1) and Phyllocladus trichomanoides (PHtr) on 17 and 18 February. The slope of the relationship was used as an estimate for hydraulic conductance. Asterisks in the legend show significance level (***P < 0.001, **P < 0.01).

Gas exchange

The stomata were concentrated on the abaxial side of the leaves with mean densities of 89 ± 13, 120 ± 20 and 199 ± 31 stomata per mm² (±standard error) for A. australis, P. totara and K. excelsa, respectively. There were no stomata on the adaxial surface of P. totara leaves and very few for A. australis and K. excelsa (4 ± 1 and 36 ± 7 stomata per mm²). Stomatal density was similar between phyllode surfaces of P. trichomanoides, with 100 ± 8 stomata per mm² on the abaxial and 83 ± 8 on the adaxial surface.

Overall, all trees showed similar g_s and A (Fig. 5). Apart from the largest A. australis (AGau3), stomatal conductance increased for all trees between 8 and 11 am. Knightia excelsa (KNex) showed the earliest drop and the smallest A. australis (AGau1) the latest. Measured maximum g_s values did not exceed 120 mmol m⁻² s⁻¹and occurred all before midday. Stomatal conductance measured on the largest A. australis AGau3 stayed relatively constant between 40 and 60 mmol m⁻² s⁻¹, with a slight decline during the day. The diurnal course of A for the smaller A. australis (AGau1), K. excelsa (KNex) and P. trichomanoides (PHtr) looked similar, increasing until 11 am to 1 pm and slowly decreasing after. The largest A. australis (AGau3) reached its highest A toward the end of the measurement period at around 6 pm. Instantaneous water-use efficiency (WUE_i) increased for the smallest A. australis (AGau1) and K. excelsa (KNex) throughout the day, peaking at around 0.1 µmol CO₂ per mmol H₂O. WUE_i of P. trichomanoides (PHtr) also increased but only reached levels of 0.5 µmol CO₂ per mmol H₂O. Agathis australis 3 (AGau3) had its highest WUE_i before noon.

Figure 5:

diurnal measurements of stomatal conductance (g_s) and photosynthetic assimilation rate (A) on 17 and 18 February. Single points represent an average of five measurements and include the standard error. The bottom panel shows the instantaneous water-use efficiency (calculated as A/g_s). Three tree species were measured: Agathis australis (AGau), Knightia excelsa (KNex) and Phyllocladus trichomanoides (PHtr).

Responses of g_s and A to environmental conditions are shown in online supplementary Fig. S1. A increased with increasing PAR for all trees, but plateaued at 800 µmol m⁻² s⁻¹ in K. excelsa (KNex) and the two A. australis. Stomatal conductance on the other hand showed a slight decline with increasing light with the exception of P. trichomanoides (PHtr) where no clear trend could be seen. The relationship of A and g_s with VPD was less apparent. A increased with VPD for the two A. australis and P. trichomanoides, but the increase for the larger A. australis (AGau3) was less steep than for the other two trees. With increasing VPD, g_s increased for A. australis 1 and stayed constant for A. australis 3. Knightia excelsa (KNex) and P. trichomanoides (PHtr) did not show visible trends. The relationship of g_s and A did not show any significant trends, but overall A increased with increasing g_s in all four trees on both days (data not shown).

DISCUSSION

This study is the first to look at water-use and leaf-level processes of mature A. australis and three co-occurring ‘kauri forest’ species. To our knowledge, our measurements of sap flow were the first on P. totara and P. trichomanoides as were the leaf-gas exchange measurements on P. trichomanoides and K. excelsa. Our results point to functional convergence of the four co-occurring species (Meinzer 2003).

Leaf-gas exchange

Both, A and g_s were relatively low (Hetherington and Woodward 2003) throughout the day for all measured species. They peaked early (before midday) and saturated light levels were low. The stomatal densities of the four species differed considerably. Despite this, A and g_s were very similar in all species which we regard as an indicator for functional convergence. In comparison with the data set published in Lin et al. (2015), A and g_s are below the average for conifers and the measurements for K. excelsa well below the average for angiosperms. Macinnis-Ng et al. (2017) measured gas exchange on young A. australis and found maximum A and g_s around 30–50% higher than our values. Very similar values were also recorded by Wyse et al. (2013) on potted A. australis during a drought experiment with no significant difference between droughted and non-droughted plants. Both studies also showed closing of stomata early in the day, consistent with our results. Brodribb et al. (2005) documented low leaf conductance, that was linearly related to stomatal conductance in temperate southern conifers in Chile (Araucaria araucana, Austrocedrus chilensis, Prumnopitys andina, Pilgerodendron uviferum, Saxagothea conspicua), suggesting conservative stomatal behaviour in these plant families. Cernusak et al. (2011) notes that photosynthetic rates in southern conifers might be reduced due to their vascular system not being able to supply water to the large leaf area of these broad-leaved podocarps consistent with our measurements.

In direct comparison, the values of photosynthesis for A. australis (AGau2 and 3) are about half those reported by Macinnis-Ng et al. (2017) from a young and open A. australis stand. The differences in tree height and age could explain the lower values at our site. As trees grow taller, the hydraulic path length increases and so do hydrostatic and hydrodynamic tension causing more negative water potentials (Woodruff et al. 2004). Consequently, stomatal apertures will be reduced (Ryan and Yoder 1997) and this in turn will reduce photosynthetic rates (Woodruff et al. 2004). The age of a forest stand can affect nutrient availability as nutrients are increasingly fixed in living biomass and soil organic matter in older forests (Gower et al. 1996).

At our study site, we suggest nutrient limitation is caused by the long-term presence of A. australis and the effect of this foundation species on the litter properties and soil formation. Wyse et al. (2014) demonstrated stalling of the nitrogen cycle and an inhibition of nitrification beneath old A. australis individuals. Puketi forest has similar vegetation characteristics as our site and the same species displayed low SLA and low concentrations of leaf nitrogen (Jager et al. 2015), both traits associated with lower assimilation rates (Wright et al. 2004). Nitrogen and other nutrients like phosphorus are crucial for photosynthetic processes (Lambers et al. 2008). Therefore, it seems likely that a deficiency of nutrients causes a reduction of photosynthesis and consequently lowers the demand for water (Katul et al. 2003).

Water relations

Despite their size, all trees had relatively low daily water use, which is common for conifers. Meinzer et al. (2005) measured water use in Pseudotsuga menziesii with similarly large stem diameters. The total water use of their largest study tree (1.67 m DBH) was around 260 kg d⁻¹ similar to our largest tree (AGau3, 1.77 m DBH) with an average daily water-use of 274 kg during the study period. Pinus ponderosa of similar diameter used around 50% more water and a larger (~1.5 m DBH) Thuja plicata used slightly less than our trees; and angiosperms from the same study (Meinzer et al. 2005) with comparable diameter used up to four times as much water.

We identified PAR as the main driving force for sap flow and dominant trees simply receive more light, compared to shaded, subordinate trees. Less dominant trees will also experience less VPD resulting in less evaporative demand. Our sample size is small and not all tree sizes are covered for all species, which leaves some uncertainty for P. totara and P. trichomanoides. We did, however, sample the largest trees of each species in our sample plot. We found that VPD and PAR were the strongest drivers of sap flow (Table 2), consistent with Jarvis (1976). Due to the stronger influence of PAR on sap flow, we assume partial decoupling of the canopy from the bulk air as described by Wullschleger et al. (2000). The largest A. australis (AGau3) was an exception, with more variation explained by VPD, a pattern associated with tighter stomatal control. This is not surprising because this was the tallest tree and therefore its canopy transpiration was likely more tightly coupled with the surrounding atmosphere. The influence of soil moisture on sap flow was very small for the larger trees and slightly more pronounced for smaller trees (Table 2) but given the relatively small measurement window, the range of soil moistures was limited.

We also see some evidence of nocturnal sap flow in our data, especially for the larger A. australis. This is likely due to greater water storage capacities of large trees (Scholz et al. 2011) and refilling during the night. But another possibility might be actual transpiration at night time in order to enhance nutrient uptake while minimizing dehydration.

The water potential values measured for all four tree species are not very negative, suggesting none of the trees showed signs of elevated water stress. This is not surprising as the soil moisture during the measurement period was well above the permanent wilting point defined for clay. Our measured water potentials are similar to those reported by Macinnis-Ng et al. (2016) at the same site during the 2013 drought summer, when soil moisture declined to 30%. Even though the soil moisture was lower in 2013, our predawn water potentials were similar. Agathis australis are likely drought avoiders, due to a low P₅₀ values in stems and roots (Pittermann et al. 2006) maintaining low water potentials through tight stomatal control to avoid embolism (Wyse et al. 2013). The same was reported for P. totara (Innes and Kelly 1992) despite their high P₅₀ values (MJ Clearwater, unpublished data). In terms of hydraulic safety, P. totara and P. trichomanoides could theoretically risk much lower water potentials, as both species have much lower P₅₀ values than the measured midday water potential. Agathis australis on the other hand has a much smaller safety margin as the difference between P₅₀ and the measured water potential was only 0.6 MPa.

We measured water potential at up to three different locations of the crown in the two biggest A. australis. The minimum water potential showed a significant difference between upper and lower crown location. This could be due to the increased path length within the spreading crown of A. australis, which causes an increase in hydrostatic and hydrodynamic tension (Jarvis 1976; Ryan and Yoder 1997; Woodruff et al. 2004). Different microclimates at the two crown locations might also play a role, as the upper crown is likely more tightly coupled with the surrounding atmosphere. Additionally, leaf anatomy (sun vs. shade leaves) might differ slightly at the two measurement heights leading to differences in stomatal and mesophyll conductance which could also influence the measured water potentials (Cano et al. 2013).

Unsurprisingly, the whole tree hydraulic conductance was low, which is common in conifers, due to their xylem anatomy (Tyree and Ewers 1991). Moreover, high conductance is a condition of high productivity (Tyree 2003) and all of our study species are known to be slow growing in A. australis dominated forests. This is reflected in our measurements. All three conifer species had similarly low hydraulic conductance, lower than angiosperms measured in tropical northern Australian (O’Grady et al. 2006) or savanna and dry woodland trees (David et al. 2004).

Hydraulic safety or nutrient limitation

Our study site is situated on a ridge and all trees were growing on sloped ground. This could lead to increased water runoff, less water retention and potentially limit water accessibility. Due to the measured water potentials, we concluded that the study trees did not actually show signs of water stress. Additionally, the soil moisture was measured at two locations within the study plot and did not drop below 30% in the upper 10 cm, neither during the time of the field campaign, nor during the weeks before. We therefore assume that water availability was not a limiting factor.

For A. australis, we know that they are highly vulnerable to embolism (Pittermann et al. 2006) and therefore very conservative in their water use, governed by a tight stomatal control (Wyse et al. 2013; Macinnis-Ng et al. 2016). This stomatal sensitivity (Wyse et al. 2013) points to an isohydric water-use strategy (Tyree and Ewers 1991) also observed in other Araucariaceae (Zimmer et al. 2016). Agathis australis also influences the soil properties in its vicinity, causing a general limitation of nutrients. The addition of fertilizer enhanced growth of A. australis (Barton and Madgwick 1987) suggesting A. australis is often nitrogen limited itself while also being the cause of nitrogen limitation. We still know little about the three co-occurring species. The two podocarps are not as vulnerable to embolism (Pittermann et al. 2006) and should therefore have an advantage over A. australis on the sites where they grow (mainly dryer ridges). All co-occurring species also have a much wider distribution than A. australis which includes dryer parts of Aotearoa-New Zealand (Wardle 2011). Surprisingly they show low water use and low assimilation rates as well. This could mean that the co-occurring species show signs of nutrient limitation which suppresses photosynthesis and growth. Their demand for water would consequently be limited and that would be the most likely cause of functional convergence of water relations of the studied trees in kauri forest. The relative contributions of nutrient and water limitation for growth of each species could be easily teased apart with fertilization and dry down experiments in pots or in the field. Data for other sites without A. australis present would help properly determine the influence of A. australis on canopy level processes of P. totara, P. trichomanoides and K. excelsa.

It is the first time either water-use or leaf-gas exchange were measured on P. trichomanoides, P. totara and K. excelsa and also the first in-situ measurements of leaf-gas exchange of mature A. australis. All measurements show little difference among the four species and point towards functional convergence, which is likely caused by the very nutrient poor growing conditions. Agathis australis influences its environment and seems to create a competitive advantage for itself, by inhibiting growth and related physiological processes of surrounding species due to nutrient limitation. On the other hand, A. australis also facilitates an environment where the co-occurring species are able to out compete plants that are not suited to the soils created by A. australis (mainly angiosperms), even though A. australis inhibits their physiological processes. The balance of both processes, competition and facilitation (Callaway and Walker 1997) shapes the plant community in the kauri forest. Our measurements also suggest no competitive advantage of any species under the current climate change scenarios for New Zealand (more frequent drought events), at least not in the ‘kauri forest’ ecosystem.

FUNDING

This research was funded by a grant from the Marsden Fund (UOA1207), administered by the Royal Society of New Zealand to C.M.

ACKNOWLEDGEMENTS

We thank Fredrik Hjelm, Scott Forrest and Chrissy Spence for their help climbing the study trees. Daniel Taylor, Jo Peace and Victor Kieffer provided assistance in the field. Derek Eamus provided comments on an early version of this manuscript.

Conflict of interest statement. None declared.