Introduction

Defoliation of eastern white pine (Pinus strobus L.) by pathogens associated with white pine needle damage (WPND) has become a chronic forest health issue throughout the northeastern United States in recent years (Broders et al. 2015; Costanza et al. 2018; Munck et al. 2012). The four ascomycete fungal pathogens which are presently associated with WPND are Lecanosticta acicola Thümen, Lophophacidium dooksii Corlett and Shoemaker, Bifusella linearis Peck, and Septorioides strobi Wyka and Broders. Each of these causal agents are considered to be native to the northeastern US and induce symptoms of chlorosis in mature needles in the late spring, followed by a defoliation of infected foliage. The establishment and persistence of WPND in the region is thought to be tightly linked with climate; notably the trend of warmer-than-average springs coupled with the increasing frequency of precipitation (Wyka et al. 2017, 2018a). Needle casting occurs primarily in the months of June through July, resulting in a 25–73% decline in stem basal area increment compared to pre-outbreak growth trends (McIntire et al. 2018a; Wyka et al. 2018b). Presently, there are no published data on the physiological impacts on WPND-induced defoliation on eastern white pine as it relates to carbon (C) assimilation, C allocation, and water use within infected stands.

Water used through transpiration is necessary for physiological processes such as photosynthesis and nutrient uptake that are essential for maintaining plant life. The recurrent summer defoliations induced by WPND in recent years significantly reduce total leaf area of infected trees and have been observed to coincide with the premature dieback of branches within the lower crown, resulting in a reduction of live crown ratios (Mcintire et al. 2018b). Trees defoliated by WPND are lacking much of their mature second- and third-year foliage for the remainder of the growing season, subsisting principally on the current-year needles for maintaining crown conductance. Due to this loss of functional foliar area, diseased individuals will presumably experience a reduction in water use and gross carbon assimilation corresponding to the severity of defoliation. Such a response has been measured in Tamarix species defoliated by the saltcedar beetle (Diorhabda carinulata), in which episodic partial defoliation resulted in a 16% reduction in mean annual sap flux density (Hultine 2010). However, the relationship between foliar area, leaf gas exchange, and sap flux is complex, as leaf-level compensatory responses have shown to increase stomatal conductance and transpiration in some cases (Damascos et al. 2005; Meinzer and Grantz 1991; Pataki et al. 1998; Reich et al. 1993; Williams et al. 2016). This upregulation of gas exchange and transpiration is not consistent across tree species or among different classes of defoliator (e.g. insect, pathogen, and artificial) and, therefore, warrants additional research to better understand how reductions in leaf area influence water use and productivity over the course of a growing season. Moreover, defoliation-induced alterations in photosynthesis and stomatal conductance may also have significant implications for water use efficiency (WUE) and, in turn, the ability to respond to climate variability.

The dynamics of non-structural carbohydrates (NSC) within the context of defoliation are also highly complex. Trees utilize NSC readily for metabolism and growth, and can potentially store them for years (Richardson et al. 2015). It has been suggested that C allocation priority is first designated to facilitate short-term survival, and then utilized for processes that support long-term fitness (i.e., growth and reproduction), with any excess NSC either actively or passively stored (Dietze et al. 2014; Wiley and Helliker 2012). While many studies have investigated NSC allocation and storage within trees subject to either natural (Hudgeons et al. 2007; Jacquet et al. 2013; Palacio et al. 2012; Saffell et al. 2014a, b; Wargo et al. 2002) or artificial defoliation (Gómez-Gallego et al. 2020; Handa et al. 2005; Jacquet et al. 2014; Li et al. 2002; Palacio et al. 2008; Puri et al. 2015a; Wiley et al. 2013, 2016), there is no consistent response that can be assumed to occur for an emerging pathogen complex such as WPND. A subsequent decline or increase of NSC following needle loss is likely a function of defoliation severity, duration, or repetition of the defoliation event, compounded with other abiotic stressors (e.g., drought) and functional traits of the host species. In P. strobus, NSC have been found to have relatively fast rates of turnover and actively mix with new photosynthate (Richardson et al. 2015), and, thus, may be allocated rapidly in response to defoliation if needed. As NSC are highly mobile, the current-year foliage of conifers may be an important location for photosynthate storage, particularly when previous-year mature foliage is lost to pathogen induced defoliation. Current-year foliage has been found to have significantly lower NSC content compared to mature second-year foliage among the evergreen conifers Abies georgei and Juniperus saltuaria (Li et al. 2009), as well as Pinus koraiensis (Yan et al. 2012). If this relationship holds true for P. strobus, WPND-infected trees may be at a distinct disadvantage for the later part of the growing season once the majority of mature foliage has been lost, compromising the trees’ water uptake and ability to assimilate and store carbon (Thomas and Sadras 2001).

The aims of this study were to quantify changes in tree water use, C assimilation, and NSC allocation within mature trees impacted by natural defoliation from WPND. Specifically, we (1) evaluated how tree sap flux varied between defoliation severity classes over the course of a single growing season in response to climatic drivers of transpiration; (2) compared measurements of post-defoliation leaf gas exchange (photosynthesis and instantaneous water use efficiency) in residual foliage to pre-defoliation levels; and (3) quantified NSC concentrations in healthy and symptomatic trees in five different storage pools over the course of a growing season.

Methods

Study region and site description

Physiological measurements of P. strobus were conducted at three field sites in Maine and New Hampshire: the Massabesic Experimental Forest in Lyman, ME (MEF; 43.567, − 70.640), the Fox State Forest in Hillsborough, NH (FOX; 43.127, − 71.908), and the University of New Hampshire’s Foss Farm in Durham, NH (DUR, 43.108, − 70.952). Sites MEF and FOX are included in a regional permanent plot network in which assessments of WPND symptoms and defoliation severity are surveyed annually by the US Forest Service and state cooperators (Broders et al. 2015). Both MEF and FOX have been documented to experience WPND-induced defoliations since 2012. A nearby study site at DUR was evaluated in an initial survey of WPND in 2012 and was determined to be asymptomatic; however, sub-annual litterfall measurements from 2014 through 2016 revealed a relatively small degree of defoliation occurring in the summer months (McIntire et al. 2018a). Within each site, trees exhibit varying degrees of disease severity. Severity of WPND infection was measured in mid-June on a discrete scale using an established protocol consistent with the US Forest Service’s annual survey (Broders et al. 2015). The total crown area is visually segmented into thirds, and then, chlorosis and needle retention are assessed within each segment. Trees are rated on a 0–3 scale, where 0 corresponds to a healthy crown free of signs and symptoms of WPND, 1 indicates that ≤ \({\raise0.5ex\hbox{$\scriptstyle 1$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}}\) of the crown is affected, 2 indicates that \({\raise0.5ex\hbox{$\scriptstyle 1$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}} - {\raise0.5ex\hbox{$\scriptstyle 2$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}}\) of the crown is affected, and 3 infers that ≥ \({\raise0.5ex\hbox{$\scriptstyle 2$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}}\) of the crown exhibits infection. Trees are referred to herein as exhibiting low and high disease severity, where low-severity trees received a rating of 0 and 1, while high-severity trees are those rated as 2 and 3. For the purposes of this study, observations of chlorosis are factored into the defoliation severity rating, as chlorotic needles are ultimately cast. It is common to find healthy/low-severity trees growing alongside diseased/high-severity trees within a stand. During the initial WPND site surveys in 2012, fungal pathogens were confirmed (L. dooksii, B. linearis, and L. acicola) and trees were classified in low- and high-severity pairs (Broders et al. 2015). Over time, some low-severity trees have achieved a high-severity rating due to progression of the disease based on repeated annual surveys, though no trees have recovered. The severity classifications used in this study reflect the ratings given to each tree in the current growth season. Trees at the DUR stand consist of low-severity (≤ 1/3 observed crown damage) and asymptotic individuals, while trees at the MEF and FOX stands are composed of both low- and high-severity infection. Since DUR did not exhibit any symptomatic trees during initial surveys in 2012, repeated annual assessments were not conducted, only for the selection of asymptomatic trees in 2014 for NSC sampling where trees at this site re-evaluated. Comparison of NSC and growth were conducted at the DUR and MEF stands, while comparisons of sapflux and gas exchange were conducted at the MEF and FOX stands.

In 2014, the mean defoliation severity rating for low- and high-severity trees was 0.6 and 2.4, respectively, at FOX, while at MEF low- and high-severity trees received ratings of 0.9 and 2.4, respectively (based on the categorical scale described above). In 2013, the year preceding measurements of sapflux and gas exchange, low- and high-severity trees at FOX exhibited ratings of 0.6 and 2.3 respectively, while, at MEF, low- and high-severity trees were rated 1.0 and 2.4 respectively. Generally, defoliation severity has gradually increased in the region since 2012 (Broders et al. 2015).

Sap flow and meteorological measurements

Transpiration was estimated using the heat ratio method (HRM), a heat pulse-based sap flow measurement technique that is well suited for resolving low flows and sap velocities up to approximately 50 cm h−1 (Burgess et al. 2001; Steppe et al. 2010). Sap flow probes were constructed at the University of New Hampshire using a protocol adapted from Davis et al. (2012). Probes consist of three thermocouple junctions along the length of a 1.0 mm-diameter steel needle, allowing for temperature measurements at depths of 10, 22, and 35 mm within the sapwood. Trees at sites MEF and FOX were equipped with sap flow probes in May 2014. At each site, a total of eight trees were selected for measurement. Trees were stratified into groups of four low- and high-severity for WPND based on preliminary ocular measurements of the crowns. Tree diameter at breast height (DBH, 1.3 m above ground level) ranged 21.9–55.1 cm (mean 38.0 SD ± 11.8 cm) at MEF and 27.8–43.3 cm (mean 36.0 SD ± 5.1) at FOX. Two probe sets were installed at breast height on opposite sides of the tree in the north–south orientation. Prior to installation, bark and cambial tissue were carefully removed from the measurement point to ensure that probes were in direct contact with the xylem. A drill guide was placed onto the exposed area allowing for precise spacing of the sensor probes and to ensure vertical alignment. Thermocouple probes were coated in petroleum jelly and positioned at a distance 0.6 cm upstream and downstream of a 37 mm nichrome line heater (17–20 Ω). A reflective radiant barrier was fixed around the sensors to inhibit potential heating via direct sunlight. Sensors were connected to a datalogger and multiplexor (CR1000 and AM16/32; Campbell Scientific Inc., Logan, UT, USA) powered by an external 12 V battery. A heat pulse of 2.5 s duration was sent to the heating probe every 15-min and the change in temperature 60 s following the heat pulse was recorded for the upstream and downstream thermocouples. Measurements were conducted from mid-May through the end of August 2014. Sapwood density (ρd) and moisture content (mc) were determined gravimetrically for each tree from the average of two increment core samples collected at the start and end of the measurement period. Sapwood thermal diffusivity (D) was calculated according to Vandegehuchte and Steppe (2012), for which we report a mean value of 0.0023 cm2 s−1 across all core samples (n = 32), which is close to the nominal value of 0.0025 cm2 s−1 that is often used across HRM studies (Looker et al. 2016; Marshall 1958). The heat pulse velocity (Vh) is calculated as:

$$V_{{\text{h}}} = \frac{D}{x}\ln \left( {\frac{{v_{1} }}{{v_{2} }}} \right)3600,$$
(1)

where x is the distance from the heat source (cm), and v1 and v2 are the changes in temperature (°C) of the upstream and downstream thermocouples, respectively. Using the methodology of Burgess et al. (2001), corrections to Vh for probe misalignment and wounding were applied. Zero-flow conditions are often determined using low vapor pressure deficit (VPD) at night and/or periods of 100% relative humidity during the day (Ambrose et al. 2010; Gotsch et al. 2014). Continuous measurements of temperature and relative humidity of the canopy (~ 20 m above ground level) were obtained on a 15-min interval using HOBO U23 Pro v2 sensors (Onset Computer Corp., Bourne, MA), from which VPD was derived. Zero-flow velocities for this study were resolved with Baseliner v3.0.10 (developed by Yavor Parashkevov, Orin Lab, Duke University). Using the Baseliner software, a VPD threshold was defined (< 0.05 kPa) in conjunction with a 4-h nighttime interval in which Vh values remain stable. This zeroing method provides dynamic adjustment of the zero-flow rate, while allowing for nighttime transpiration and without destructive sampling at the probe location (Oishi et al. 2016). Following determination of Vh zero flow, sap flux density was then calculated as:

$$J_{{\text{s}}} = \frac{{\rho_{d} }}{{\rho_{{\text{s}}} }}\left( {m_{{\text{c}}} + \frac{{c_{{{\text{d}}w}} }}{{c_{{\text{s}}} }}} \right)V_{{\text{h}}} ,$$
(2)

where Js is the sap flux density (cm3 cm−2 h−1), ρs is the density of sap (1000 kg m−3), cdw is the specific heat capacity of the wood matrix (1200 J kg−1 K−1), and cs is the heat capacity of water (4186 J kg−1 K−1). As Js is known to decline from the outer xylem towards the sapwood, the radial profile must be accounted for to ensure robust estimates of total sap flux (Alvarado-Barrientos et al. 2013; Gebauer et al. 2008; Wullschleger and King 2000). A weighted average of Js for each tree’s entire sapwood area was calculated using Js at each measurement depth and the assumed area in concentric circles bounded by the mid-point at each thermocouple extending toward the heartwood interface. The sapwood area (cm2) was estimated assuming a circular bole using the average sapwood thickness measured on two increment core samples for each tree. Gap filling of data was conducted on a sensor-by-sensor and tree-by-tree basis using simple linear regressions with available data. Hence, data gaps from a single depth within an individual were amended by creating a regression equation with an adjacent depth on the sample probe set. If data gaps existed in all depths within a sensor due to failure of the heating element, then gap filling was conducted via a regression equation with the probe set situated on the opposite side of the bole. All gap-fill regression equations used no less than 2000 data points (equivalent to 20.8 days of point-measurements) and an R2 > 0.9 for all linear fits. Overall, sap flow data from MEF and FOX were exceptionally complete and gap filling was required for < 3% of the final data set. The average Js of the two probe sets for each tree is reported throughout.

Leaf gas exchange

Leaf gas exchange measurements were collected during the summer of 2014 at FOX and MEF field sites using a portable photosynthesis system (LI-6400XT, Li Cor, Inc., Lincoln, NE). Measurements were conducted in July and August on warm days with low cloud cover, between the hours of 10:00 and 14:00. Twig segments containing intact fascicles were acquired from the upper \({\raise0.5ex\hbox{$\scriptstyle 1$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}}\) of canopies (approx. 20–30 m) exposed to full light from four co-dominant trees of each WPND-severity class (the same individuals sampled for sap flux measurements) using a 12-gauge shotgun. Only twigs bearing both current-year and second-year foliage were selected for measurement, so that potential variability across segments was minimized for comparisons of needle age class. Excised twig samples were immediately submerged and cut under water to reduce the presence of embolism within the xylem (Venturas et al. 2015). Two complete fascicles consisting of ten intact needles were placed parallel to the long side of the 6.0 cm2 cuvette for gas exchange measurement. Samples were allowed to stabilize within the chamber for approximately 2 min while holding constant the environmental parameters influencing photosynthesis. Preliminary light-response curves were conducted on a subset of needle samples to determine a light saturation point of 1000 µmol m2 s−1 which was used for all subsequent measurements. Other environmental parameters controlled for included the ambient CO2 concentration (400 ppm), temperature (26 °C), relative humidity (30–40%), and airflow rate (400 µmol s−1). Once photosynthesis and stomatal conductance stabilized for a given sample, five instantaneous measurements were taken over a 30 s period and the average of those measurements are reported throughout. Since the leaf chamber assumes a uniform leaf area and needles are irregular within the cuvette, leaf area was adjusted by measuring the length and width of each fascicle with digital calipers and the total surface area calculated assuming a cylindrical shape (Johnson 1984, Eq. 2).

NSC sampling and processing

Sampling for NSC occurred at three intervals in the months of June, August, and October, at sites MEF and DUR. These two sites were selected in the interest of comparing the greatest differences in WPND severity, as MEF had exhibited chronic defoliations in the years prior to this study, while, at DUR, it was possible to identify and sample for asymptomatic trees. All trees sampled at MEF exhibited high-severity crown damage. Six mature P. strobus per site were selected for repeat NSC sampling, with a mean DBH of 50.3 cm (SD ± 13.8) and 53.4 cm (SD ± 14.5) at MEF and DUR, respectively. Mean live crown ratio was also similar among sampled trees between sites: 40.1% (SD ± 4.7) and 43.6% (SD ± 9.6) at MEF and DUR, respectively. These relatively high values result from an open grown habit of the individuals, where trees at MEF are generally established on the edge of a recent clear-cut, and trees at DUR are on the boundary of a wide recreational trail exposed to high light conditions. Since NSC are allocated to different storage pools over time, it was crucial to assess sugar and starch content across root, stem, and foliar tissues to derive a robust understanding of NSC throughout the tree. Carbon supply within various tissue types vary depending on the time of year that sampling occurs, such that higher NSC concentration is typically found within foliage and stems of the upper canopy early in the growing season (i.e., near the source location) and moves to storage pools within the bole and root system later in the growing season (Landhausser and Lieffers 2012). Increment core samples were obtained at two locations using a 5.15 mm increment borer: from breast height on the main stem (bole), and from a primary coarse root excavated < 1.0 m away from the trunk. Radial growth for the current-year and previous year (2015 and 2014, respectively) was assessed from core samples obtained at DBH and converted to units of basal area increment (BAI, cm2). Additionally, current-year branch segments, multiple-year branch segments (integrated 2–4 years), current-year needles, as well as second-year needles were collected. Stem samples were considered from two positions in the canopy, upper and lower, defined here as the top \({\raise0.5ex\hbox{$\scriptstyle 1$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}}\) and bottom \({\raise0.5ex\hbox{$\scriptstyle 1$} \kern-0.1em/\kern-0.15em \lower0.25ex\hbox{$\scriptstyle 3$}}\) of the live crown. All stem and needle samples were acquired using a 12-ga shotgun with steel birdshot. Upon collection, samples were immediately placed on dry ice. All tissue samples were then relocated to a − 80 °C freezer until carbohydrate extractions could be performed.

Sugar and starch assays were conducted at Bates College in Lewiston, ME. Tissue samples were freeze-dried (Freezone 6; Labconco Corporation, Kansas City, MO, USA) before being ground to 20-mesh (Wiley Mini Mill; Thomas Scientific, Swedesboro, NJ, USA) and stored in 20 mL scintillation vials with airtight seals. Determination of total soluble sugar concentrations followed the protocol of Chow and Landhäusser (2004). Using this method, a ground and homogenized tissue sample is subjected to heated ethanol extraction followed by colorimetric analysis with phenol–sulfuric acid. The final extract was measured using a spectrophotometer (Spectronic 20 Genesys; Thermo Fisher Scientific, Waltham, MA, USA) at a wavelength of 490 nm with sugar concentration calculated from a standard curve of 1:1:1 glucose–fructose–galactose. For starch determination, residual tissue was subsequently solubilized by sodium hydroxide, hydrolyzed to glucose by an enzyme mixture of α-amylase and amyloglucosidase, and then measured colorimetrically using a peroxidase-glucose oxidase-o-dianisidine solution. Solution absorbance was read at 525 nm and starch concentration calculated based on a glucose standard.

Needle expansion

Morphological development of current-year needles was monitored at the DUR field site for the 2015 growing season. Weekly measurements of needle length were conducted on three fascicles of three white pine saplings (n = 9) exposed to full light conditions using digital calipers. The purpose of these measurements was to determine the start and end time of the current-year needle expansion and its developmental stage relative to WPND-induced defoliation.

Analyses of physiological data

As VPD is known to be a primary driver to tree transpiration, we assessed the relationship between VPD and sap flux density as a function of disease severity. The mean daily VPD is assessed with three different metrics of sap flux density: the mean daily Js (24 h mean), peak daily Js (the mean Js between the hours of 10:00–18:00), and maximum daily Js. Non-linear curves were fit to each metric of Js against mean daily VPD using a five-parameter log-logistic model (Ritz et al. 2015). These model fits were evaluated using a likelihood ratio test to determine if differences between severity classes were significant. To assess the impacts of defoliation on tree sap flux, continuous measurements were segmented into three discrete periods. These periods are defined as: pre-defoliation (May 21–Jun 14), during defoliation (Jun 15–Jul 14), and post-defoliation (Jul 15–Aug 27). These periods are based on repeated seasonal observations (~ 7-day interval) of the needle-cast event and supported in part by quantitative litterfall trapping measurements (McIntire et al. 2018a) and measured spore dispersal of the WPND-associated pathogen Lecanosticta acicola at MEF (Wyka et al. 2018a). Differences in Js (daily mean, peak, and maximum) were analyzed using linear mixed models, with defoliation severity class and time period evaluated as fixed effects, and sampled trees set as a random factor to account for multiple observations among individuals nested within the two sites. The mixed model fits were conducted using the R packages lme4 (Bates et al. 2015) and lmerTest (Kuznetsova et al. 2017), followed by post hoc comparisons (Tukey’s method) using the emmeans package (Lenth et al. 2018). Additionally, for each site, the ratio of mean daily Js between low- (n = 4) and high-severity trees (n = 4) was computed for each day of the study period, excluding days with mean VPD < 0.3 kPa. This daily Js ratio provides a metric of water uptake for defoliated trees relative to adjacent healthier trees within each stand. Since individual low- and high-severity individuals cannot be directly paired, a site-level Js ratio is computed for each day and sites MEF and FOX are analyzed separately. ANOVA was applied to test for the overall effect of time period and a post hoc Tukey-HSD was used to test for significant differences (α = 0.05) in the daily Js ratio means aggregated between the pre-, during, and post-defoliation periods.

For analysis of leaf gas exchange, two dependent variables were considered: photosynthesis (A, µmol CO2 m−2 s−1), and instantaneous water use efficiency (iWUE, μmol CO2 mmol−1 H2O), calculated as the ratio of photosynthesis to transpiration. Since gas exchange measurements could not be conducted on the same dates for both study sites, measurements were grouped into two time periods for analysis (July vs. August), which represent the post-defoliation period. Gas exchange was measured on 8 July (MEF), 21 July (FOX), 14 August (MEF), and 21 August (FOX), during which times the July ambient VPD ranged 2.0–2.6 kPa and August ambient VPD ranged 0.9–1.0 kPa. A three-way factorial mixed model used the independent variables of needle age, defoliation severity class, and month to assess their relative impact and interactions on A and iWUE, while the individual trees for which repeat measurements were conducted was classified as a random effect within the model. Values for A met the normal distribution criteria; however, iWUE values were skewed on the lower end and a log transformation was used to meet assumptions of variance homogeneity. A post hoc analysis was conducted using linear contrast (sub-design ANOVA), testing the overall effect of one factor at a single level of another factor to determine sample means that were significantly different (α = 0.05) between disease severity within a given needle age class and sampling period for each dependent variable.

To assess the effects of time, tissue type, and WPND severity on NSC content, a repeated-measures MANOVA using a full factorial (three-way) design was conducted using log-transformed values of sugar and starch to adhere to assumptions of normality. Significance for MANOVA was tested using the Wilk’s lambda test at an alpha level of 0.05. This analysis allowed for a test of whether sugar or starch concentrations changed over time (June–Aug–Oct), differed between WPND-severity classes, differed across tissue types, and also included interactions among these three factors in a full factorial design. Thus, a within-subject factor of time is considered, while two conditions (WPND severity, tissue type) are considered between trees on the dependent variables of sugar and starch. Whole-tree NSC content (kg) were also assessed relative to the current-year and previous-year basal area increment of each tree. Measured concentrations of soluble sugar and starch were scaled to the whole-tree level using established allometric relationships of roots, foliage, and stem as a function of DBH (Furze et al. 2019; Jenkins et al. 2003). This ratio of carbohydrate content to annual woody growth (kg NSC cm−2 BAI) provides insight to a potential trade-off between secondary growth and carbon allocation among tissue types. In this analysis, data from all sampling dates were pooled and a t test was used to infer differences within each tissue type (α = 0.05) between disease severity using log-normalized data. Statistical analyses were performed using R v3.5.1 (R Core Team 2020).

Results

Tree sap flux

Metrics of daily sap flux were highly correlated with the mean daily VPD. Trees with high-severity infection consistently exhibited lower values of daily mean Js, daily maximum Js, and daily peak Js than trees of low-severity infection (Fig. 1). Likelihood ratio analyses found that the low- and high-severity model fits were significantly different from the common model fits pooling both severity classes for Js-daily [high-severity X2 (1 df) = 900.30, p < 0.01; low-severity X2 (1 df) = 835.83, p < 0.01], Js-peak [high-severity X2 (1 df) = 1215.79, p < 0.01; low-severity X2 (1 df) = 1093.57, p < 0.01], and Js-max [high-severity X2 (1 df) = 1122.61, p < 0.01; low-severity X2 (1 df) = 1026.15, p < 0.01]. Each of the metrics of Js begin to converge for the two severity classes at lower values of VPD (≤ 0.3 kPa). Mean daily Js did not differ significantly between sites for either the low-severity [t test: t (194) = − 1.869, p = 0.631] or high-severity trees [t (193) = 0.732, p = 0.47]. Maximum values of Js were typically achieved on days where VPD exceeded 0.4 kPa (Fig. 1c). The average maximum Js for the low- and high-severity trees above 0.4 kPa was found to be 24.7 and 19.8 cm3 cm−2 h−1, respectively, an apparent 20% reduction of Js in the more diseased trees. Similarly, daily mean Js (Fig. 1a) was found to be 21% lower and daily peak Js (Fig. 1b) 22% lower among the high-severity trees, though these differences tended to increase with VPD. Trees of either severity class exhibited nighttime transpiration when VPD remained elevated (> 0.1 kPa) after sunset (Fig. 2a), and thus, daily mean values of Js incorporate nocturnal flow rates when applicable. While high-severity trees tended to exhibit similar diurnal patterns of sap flux than their low-severity counterparts, the main differences occurred around peak hours of flow (10:00–18:00), such that high-severity trees appeared to be inhibited when VPD exceeded 2.0 kPa (Fig. 2a). On days with relatively low VPD, sap flux of low- and high-severity trees exhibited more similar diurnal patterns and the mean daily Js ratio approached 1.0 (Fig. 2b).

Fig. 1
figure 1

Mean daily (a), peak daily (b), and maximum daily sap flux density (c) as a function of daily average VPD. Each point represents the average of individuals (n = 4) within each severity class for a given day, circles and squares denote trees at FOX and MEF, respectively, and open and filled shapes denote the low-severity and high-severity trees respectively. Non-linear curve fits show the 95% confidence interval in the shaded ribbon for low- and high-severity trees

Full size image
Fig. 2
figure 2

Diurnal patterns of sap flux density (Js) during 2 days of contrasting VPD on July 9 (a, b) and August 21 2014 (c, d) at MEF and FOX. Daily mean VPD for each panel is 0.9 (a), 0.5 (b), 0.2 (c), and 0.1 kPa (d). Solid line indicates VPD derived from canopy temperature and humidity on site. Open and filled symbols denote low- and high-severity trees, respectively, and shaded areas show ± 1 SE (n = 4)

Full size image

We found the relative time period (pre-, during, and post-defoliation) which had a significant effect on daily mean Js [F(2,28) = 43.7, p < 0.001], peak Js [F(2,28) = 19.1, p < 0.001], and maximum Js [F(2,28) = 14.2, p < 0.001]. While the mean Js of high-severity trees was consistently reduced compared to low-severity trees, this factor did not meet our criteria for statistical significance for daily mean Js (p = 0.082), peak Js (p = 0.059), or maximum Js (p = 0.056) (Table 1). The interaction between WPND severity and time period was not significant, suggesting that the summer defoliation event did not result in an appreciable decline in tree water flux between severity classes, in part rejecting our first hypothesis. However, the daily Js ratio was found to differ significantly at MEF [F(2,69) = 7.64, p = 0.001] between the pre-defoliation and post-defoliation periods, indicating a reduction in sap flux density on the order of 5.3% (Fig. 3). The post-defoliation Js ratio was reduced by 3.9% at FOX (compared to pre-defoliation), though this change over time was not significant [F(2,75) = 3.07, p = 0.052]. Interestingly, trees at FOX were found to have a lower pre-defoliation Js ratio than compared to MEF (p < 0.001), suggesting that high-severity trees at the FOX site may have been more pre-disposed to infection and to the influence of chronic defoliation. During this early season pre-defoliation period, high-severity trees at FOX exhibited daily mean Js equivalent to 76.2% the Js of low-severity trees, while trees at MEF exhibited a Js ratio of 87.5%.

Table 1 Linear mixed model results for the effects of time (pre-, during, and post-defoliation) and defoliation severity on daily mean, daily peak (10:00–18:00), and maximum sap flux density (Js)
Full size table
Fig. 3
figure 3

Ratio of mean daily sap flux density (high-severity/low-severity trees) at each study site for the pre-defoliation (MEF n = 16, FOX n = 14 days), defoliation (MEF n = 29, FOX n = 28 days), and post-defoliation (MEF n = 39, FOX n = 39 days) periods. Data exclude days with mean VPD < 0.3 kPa. A difference in letters between time periods indicates a significant difference via a Tukey-HSD test, and p values between each time period are shown in connecting brackets

Full size image

Leaf gas exchange and expansion

A three-factor mixed model was used to test for effects of needle age, defoliation severity, month, and their interactions for instantaneous measurements of photosynthesis (A) and iWUE. The significant interactive terms for A were found to be month × needle age [F(1,42) = 16.15, p < 0.001] and severity × needle age [F(1,42) = 4.91, p = 0.030] (Table 2). Post hoc analyses found that photosynthesis of current-year needles during the month of July was 25.9% lower (p = 0.002) than the mature second-year needles (pooling severity classes). Additionally, photosynthesis of current-year needles in high-severity trees increased by 24.0% from July to August (p = 0.043) (Fig. 4b), while second-year needles among low-severity trees declined by 24.8% over this same time interval (p = 0.013) (Fig. 4a). For iWUE only, the fixed effect of month was found to be significant [F(1,42) = 118.52, p < 0.001]; post hoc tests indicate iWUE of the current-year needles in August increased by 54.1% (p < 0.001) and 78.2% (p < 0.001) in low-severity and high-severity trees, respectively (Fig. 4c), while second-year needles exhibited an iWUE increase of 42.1% (p < 0.001) and 63.8% (p < 0.001) in low-severity and high-severity trees, respectively, compared to measurements conducted in the previous month of July (Fig. 4d).

Table 2 Parameters of the three-factor mixed model of A and iWUE by month (July, August), defoliation severity (high, low), and needle age class (current-year, second-year) with random effect of sampled tree nested within site (FOX, MEF)
Full size table
Fig. 4
figure 4

Photosynthesis rates (a, b) and instantaneous water use efficiency (c, d) derived from gas exchange measurements conducted in July (grey) and August (black) by defoliation severity class and needle age pooled from sites FOX and MEF. Significant differences between means (n = 8 ± SE) are determined with Tukey post hoc comparisons and indicated with brackets

Full size image

Current-year fascicles at DUR were found to initiate development in early May and reach maturity by the first week of August. Growth between June 14 and July 25 advanced at a rate of 0.9 mm day−1, with fully expanded needles measuring 59.3 mm (SD ± 15.0) in length. At the time of WPND-induced defoliation beginning in mid-June, current-year needles measured 23.2 mm in length; only 39% of their mature length.

Non-structural carbohydrates

Multivariate analysis of sugar concentrations (mg g−1) found significant differences between trees for tissue type [F(7,68) = 126.02, p < 0.001], but not for disease severity [F(1,68) = 0.15, p = 0.697]. The within-tree interaction of time and tissue type was also significant for sugar concentration [F(14,134) = 6.43, p < 0.001]. Starch concentration contrasted with sugars in that the interaction of tissue × disease severity was significant [F(7,72) = 4.6, p < 0.001]. The three-way interaction of time × tissue × disease severity was also found to be significant within trees for starch concentration [F(14,142) = 1.70, p = 0.038]. Within each tissue type and across sampling dates, NSC in the two disease severity classes were generally well correlated with each other, though some significant differences are noted for starch and total NSC among stem and needles tissues in July and August (Fig. 5). When considering total NSC (the sum of sugar and starch) fixed effect tests between trees found that differences in concentration were marginally significant for the interaction of tissue × disease severity [F(7,68) = 2.18, p = 0.047], as well as the within trees interaction of tissue × time [F(14,134) = 12.42, p < 0.001]. Full MANOVA results for sugar, starch, and total NSC by disease severity, tissue type, and time are reported in Table 3.

Fig. 5
figure 5

Mean starch and sugar concentration (n = 6 ± SD) for the asymptomatic (white squares) and high-severity WPND-infected (black circles) sites across the eight tissue types measured at DUR (asymptomatic/low severity) and MEF (high severity). Points denoted with * are significantly different (p < 0.05) between severity class for a given sampling period. Tissue-type guide: CYSLC current-year stem from the lower canopy, CYSUC current-year stem from the upper canopy, MYSLC multi-year stem from the lower canopy, MYSUC multi-year stem from the upper canopy, CYN current-year needle, SYN second-year needle

Full size image
Table 3 Between-subject and within-subject results of the three-factor repeated-measures MANOVA for sugar, starch, and total NSC concentrations (mg g−1) comparing low-severity (DUR) and high-severity (MEF) stands
Full size table

Examination of allometric-scaled NSC-to-BAI of the current-year (2015) and previous year (2014) revealed large differences between severity class for the five different tissues (Fig. 6). The mean BAI among the low-severity trees was 45.1 and 37.0 cm2 in 2014 and 2015, respectively, while high-severity trees had a mean BAI of 20.4 and 21.9 cm2 in 2014 and 2015. This is a difference in radial increment between high- and low-severity trees in the previous-year and current-year of 41 and 55% respectively. The ratio of total NSC to BAI (kg cm−2) was significantly higher (t test, p < 0.05) within high-severity trees for multi-year stem segments, current-year needles, and second-year needles when comparing the NSC-to-BAI ratio for both the current (2015) and previous year (2014). Interestingly, this ratio was also significantly elevated in high-severity individuals for root tissue when comparing the previous-year growth [t (18) = 2.32, p = 0.032], but not for the current-year growth [t (18) = 1.79, p = 0.090]. Bole content of NSC did not differ as a function of defoliation severity class.

Fig. 6
figure 6

The NSC content as a ratio of the previous year (2014) and current-year (2015) BAI, for each tissue type. Low-severity/asymptomatic (DUR) and high-severity (MEF) trees are denoted in white and black bars, respectively. Means (n = 18 ± SE) are significantly different at p < 0.05 (*) and p < 0.001 (**) via a paired t test

Full size image

Discussion

Defoliated trees show decreased water use

Across both sites where sap flux was measured (FOX and MEF), high-severity P. strobus were found to have lower values of Js throughout the 2014 growing season. On average, mean daily sap flux density of high-severity trees was 20% lower than healthy/low-severity counterparts. At low VPD (< 0.3 kPa), high-severity defoliated trees were found to have values of Js more similar to that of low-severity individuals. In a study of defoliated Populus tremuloides, it was found that trees subject to leaf area reductions of 50% and 98% showed a significant increase in stomatal conductance (gs) under low-VPD conditions (0.64 kPa) compared to control trees, but the magnitude of this increase in gs was muted at high-VPD (1.9 kPa) (Hart et al. 2000). Similarly, defoliated Eucalyptus globulus demonstrated higher transpiration rates and canopy conductance to control trees at lower values of VPD, while defoliated trees experienced a greater reduction in canopy conductance with increasing VPD compared to controls (Quentin et al. 2011). This suggests that while residual foliage may be compensating for the reduction in canopy leaf area by increasing gs to optimize for C gain, it is not able to do so when evaporative demand is high. This is likely due to constraints induced by a reduction in leaf water potential, which may limit transpiration when VPD is high. However, others have shown an increase in branch water potential (less negative) of P. strobus in response to artificial defoliation of saplings (Vanderklein and Reich 2000). Physiological studies of Pseudotsuga menziesii defoliated by Phaeocryptopus gaeumannii have shown significant reductions in branch leaf specific conductance as a result of leaf area reduction (Manter and Kavanagh 2003). Similarly, Arceuthobium tsugense infection of Tsuga heteriohylla has been found to induce substantial reductions in canopy leaf area, resulting in reduced sapwood area, declines in whole-tree water use, and lower hydraulic conductivity (Meinzer et al. 2004). Though it is well established that VPD, leaf-stem water potential, and conductance are tightly linked to transpiration (Simonin et al. 2015), it is interesting to note the feedback effect in which defoliation appears to alter the influence of VPD on whole tree water use. At high midday VPD, high-severity P. strobus appear inhibited, showing a marked reduction in Js when VPD exceeds 2.0 kPa, while low-severity trees plateau and maintain relatively high values of midday Js (Fig. 2a).

Despite significant differences in Js between severity classes as a function of VPD, the defoliation event which started in mid-June did not influence a substantial further decline in the water use of high-severity trees, as the Js ratio differed between the pre- and post-defoliation periods by only 3.9% and 5.3% at FOX and MEF, respectively. This suggests that the initial defoliation has long-term consequences for tree water use, since it is known that trees sampled at both sites have suffered from chronic pathogen-induced defoliation since at least 2012, 2 years prior to this study. As healthy P. strobus typically retain 2–3 years of foliage throughout the growing season, severely defoliated trees often possess only current-year foliage from July to the subsequent May when the new cohort of needles begins to form. Given chronic annual defoliations, high-severity trees exhibit thinned crowns for multiple years as the older mature foliage is continually reduced. Therefore, the already thinned crowns of diseased trees at the start of a growing season are likely a more important factor than the subsequent defoliation within a given year in terms of tree water use within stands that have been infected previously. All sampled trees were in the co-dominant canopy stratum; thus, it is unlikely that overstory shading played a role in the reduced sap flux rates measured in the high-severity trees. It is also plausible that the impacts of defoliation are subtle due to the concurrent expansion of the current-year foliage during the time at which mature second- and third-year needles are being cast. During the month of June, the current-year needles are only 39% of full expansion at the start of the defoliation event. Thus, gross photosynthesis (and transpiration) would be expected to increase as development of current-year needles progress, despite the loss of much of the mature foliage. As the current-year needles continue to develop post-defoliation, they act to offset the C lost from second- and third-year needles.

Regulation of gas exchange in residual needles

Analyses of photosynthesis (A) revealed significant effects of needle age over time within both severity classes (Table 2). Conifer foliar expansion occurs over a time span of several months (Cuny et al. 2012; Rossi et al. 2009; Sampson et al. 2003), which differs from deciduous broadleaf species that typically expand and mature rapidly. Second-year needles were found to have significantly higher rates of A than current-year needles during the month of July (Fig. 4); however, by the time current-year foliage reached full expansion in August, there was no difference in assimilation rates between age classes. Our linear mixed model did not find a consistent effect of defoliation severity on gas exchange across all age classes and time periods, though it is worth noting that the interaction of defoliation severity over time was found to be significant, indicating a trend of A increasing over time among trees of high defoliation severity. This finding suggests a compensation response occurring within high-severity crowns, in which residual needles of high-severity trees are upregulating photosynthesis due to a lack of mature foliage later in the season. Photosynthesis of current-year foliage within high-severity trees was significantly higher than that of low-severity trees in the month of August. We found the opposite response within the second-year foliage of low-severity trees—a significant downregulation of photosynthesis in August, presumably due to the fact that the current-year foliage is able to contribute a higher proportion of gross carbon assimilation now that the needles have reached full expansion. This compensatory response has been confirmed in other Pinus species subjected to defoliation (Eyles et al. 2011).

The effects of month on iWUE were highly significant, increasing between July and August across both needle age classes and defoliation severity classes (Table 2, Fig. 4c, d). This effect may in part explain the unexpectedly subtle decline in post-defoliation sap flux, since it appears that as the growing season progresses, trees are able to use less water through transpiration per unit carbon assimilated.

Dynamics of NSC allocation and storage

Despite the substantial reduction in total leaf area and similar leaf-level C assimilation rates, we observed that severely defoliated trees are storing NSC at nearly the same concentration across tissue types as low-severity trees (Fig. 5). With the exception of starch content in the month of August, we observed no differences in NSC concentrations between the current-year and second-year needles, contrary to the findings of Li et al. (2009) and Yan et al. (2012). Additionally, we found only a modest effect of disease severity between tissue types’ total NSC (Table 3). This finding is interesting in light of clear declines in basal area increment due to WPND-induced defoliations since the onset of outbreaks in the northeastern US. McIntire et al. (2018a) report declines in BAI that ranged from 25 to 73%, with post-outbreak declines at FOX and MEF of 35% and 26%, respectively. However, research has shown that stored NSC and growth are not always tightly linked. In a foliar fungal disease similar to the WPND complex, Douglas-fir (Pseudotsuga menziesii) defoliation via Swiss needle-cast (Phaeocryptopus gaeumannii) was found to significantly reduce wood growth ahead of observed declines in NSC reserves (Saffell et al. 2014a, b), suggesting that trees will store NSC in times of defoliation stress to allocate towards new foliar material, maintenance respiration, and defense, at the expense of radial growth. Other studies have demonstrated similar results in which a reduction in wood growth is reported with no change in NSC concentration of defoliated trees, including defoliation of Pinus pinaster by Thaumetopoea pityocampa (Jacquet et al. 2013) and repeated artificial defoliation of Pinus radiata genotypes (Gómez-Gallego et al. 2020).

In some cases, the time elapsed since the initial defoliation event can influence NSC storage within various tissue pools. Several studies have reported only short-term declines in NSC concentration, with no long-term difference in sugar or starch concentrations between healthy and defoliated trees (Gómez-Gallego et al. 2020; Palacio et al. 2008, 2012; Puri et al. 2015b; Saffell et al. 2014a, b; Wiley et al. 2013). It is plausible that the impacts of WPND have yet to reach a critical threshold of reduced gross carbon assimilation, at which point it may be expected to observe declines in stored NSC. This assertation is supported in part by the low levels of observed mortality attributed to WPND (Constanza et al. 2018). As depletion of NSC is hypothesized as a pathway of plant mortality (McDowell 2011; McDowell et al. 2008), it is reasonable to expect low soluble sugar and starch stores just prior to tree death (Dickman et al. 2015). Interestingly, we found that starch concentration was significantly higher among trees of high severity, contrary to depleted reserves as we had expected to observe following multiple years of defoliation. However, this was not the case for sugar concentrations, nor statistically significant for each tissue type when considering total NSC.

The literature on the effects of defoliation on NSC concentration in trees is mixed; there are many examples in which a decline or increase in NSC is found to occur in some tissues types and not others, only at certain times in the growing season, in conjunction with other stressors, and at variable levels of relative defoliation intensity (Deslauriers et al. 2015; Handa et al. 2005; Li et al. 2002; Palacio et al. 2008; Piper et al. 2015; Puri et al. 2015b). Each of these studies support the finding that the relative importance of stem growth is a lower priority than the carbon requirement for other physiological processes within trees, though the order hierarchy of such processes is complex and generally unresolved.

Alternatively, the defoliation itself may not induce carbon limitation, such that growth is limited by factors other than available carbon. The influence of shaded needles in the lower crown of conifer species has been known to serve as nutrient storage, which may be resorbed and utilized by newer foliage in the event of a defoliation (Hirose 2012; Millard and Grelet 2010). In a regional assessment of WPND defoliation that utilizes the same sites as this study (DUR, FOX, and MEF), it was found that defoliated P. strobus foliage contained higher levels of nitrogen in the month of June than in October, but June nitrogen content being lower than that of reported values for green foliage, implying that at least a partial resorption had occurred (McIntire et al. 2018a). Therefore, trees subject to pathogenic defoliations may be more nutrient-limited than carbon-limited.

When considering the secondary growth of each tree via the ratio of total sugar and starch content to current-year and previous-year stem basal area increment (kg NSC cm−2 BAI), we found significant differences between trees of high and low severity across needle tissues (Fig. 6), suggesting that diseased trees may be prioritizing NSC for storage to be used in the following years to perpetually bolster lost foliage. We have shown that mature P. strobus have the ability to survive consecutive years of defoliation without a significant effect in total NSC despite reductions in secondary growth. However, this may not be the case for younger trees of seedling and sapling size classes, in which growth is more important for ensuring survival. As this study did not address carbohydrate dynamics of young trees, further research will be necessary to evaluate the long-term survival and competitiveness of P. strobus regeneration within stands affected by WPND.

Conclusions

Pathogen-induced defoliation of mature P. strobus caused by fungi associated with WPND has important physiological consequences for seasonal water and carbon dynamics. Trees that had been defoliated for multiple years prior to this study were found to have significantly reduced transpiration rates, with mean daily sap flux density 20% less than that of healthy/low-severity individuals. We found that high-severity trees reduced sap flux density an additional ~ 5% relative to low-severity trees following the 2014 defoliation event. Trees infected for multiple years of summer defoliations have not yet been observed returning to a healthy state with crowns retaining multiple years of mature foliage. The increase in litterfall to infected stands has cascading influences on the flux of nutrients to the forest floor, as litter deposited in the months of June and July has significantly higher nitrogen content than litter that is released during natural needle senescence in October (McIntire et al. 2018a). This may alter decomposition and soil microbial activity and, consequently, impact the availability of tree nutrients, both for mature P. strobus and co-occurring species such as Quercus rubra and Tsuga canadensis which are often in competition. A thicker litter layer due to chronic defoliations may also impact the ability of precipitation to infiltrate the soil and alter runoff (Neris et al. 2013). While a reduction in sap flux measured among diseased trees could represent greater soil water availability, the reduction in canopy leaf area also modifies the light regime of the understory, such that advanced regeneration could benefit and, perhaps, represent increased competition for resources.

There is evidence that residual current-year needles of diseased trees may partially compensate for the loss of foliar area by upregulating photosynthesis in the late growing season. However, there was a clear difference in C allocation to radial growth between high- and low-severity trees sampled in this study, with severely defoliated growing 55% less basal area than low-severity trees in 2015. Despite clear reductions in water use and C allocation to secondary growth, we found limited effect on total NSC across tissue types throughout the 2015 growing season between our high- and low-severity stand. These findings suggest that mature diseased trees may be prioritizing NSC storage over radial growth and the defoliations caused by WPND to date have not stressed trees to the point at which stored NSC is needed to sustain metabolic processes for survival. Additionally, there exists the possibility that reduced growth resulted due to a scarcity of available nitrogen, lost from the lower crown through repeated summer defoliations. In this case, C-limitation may not occur, yet nutrient limitation may be responsible for declines in secondary growth. Further research is necessary to parse out the precise mechanism of growth reductions, and additional attention should be given to the productivity and survivability of P. strobus subject to chronic pathogenic defoliation, particularly among juvenile and regenerating stands.

Author contribution statement

Conceptualization: CM, IM, BH, and HA. Field measurements: CM, BH, and IM. Lab processing of carbohydrates: BH, and ED. Data analysis: CM. Supervision: IM, HA, BH, and MV. Writing: CM. All authors read and commented on this manuscript.