Abscisic acid can augment, but is not essential for, autumnal leaf senescence

Abstract

Senescence vividly marks the onset of the final stages of the life of a leaf, yet the triggers and drivers of this process are still not fully understood. The hormone abscisic acid (ABA) is an important regulator of leaf senescence in model herbs, but the function of this hormone has not been widely tested in deciduous trees. Here we investigate the importance of ABA as a driver of leaf senescence in winter deciduous trees. In four diverse species we tracked leaf gas exchange, water potential, chlorophyll content, and leaf ABA levels from the end of summer until leaves were abscised or died. We found that no change in ABA levels occurred at the onset of chlorophyll decline or throughout the duration of leaf senescence. To test whether ABA could enhance leaf senescence, we girdled branches to disrupt ABA export in the phloem. Girdling increased leaf ABA levels in two of the species, and this increase triggered an accelerated rate of chlorophyll decline in these species. We conclude that an increase in ABA level may augment leaf senescence in winter deciduous species but that it is not essential for this annual process.

Abscisic acid (ABA), autumn, deciduous, chlorophyll, leaf life span, senescence, stomata

Introduction

Autumn leaf senescence is one of the final metabolic processes of the annual flush of leaves in deciduous temperate forests and annual crops, the timing of which is critical for determining leaf life span, through to influencing global carbon cycles and harvest time and post-harvest quality of agricultural products (Knorr, 2000; Richardson et al., 2010, 2013). Senescence is important for the remobilization of nutrients from leaves to support new growth in the following spring (Lim et al., 2007), or to augment seed maturation (Lim et al., 2007). Despite the importance of this process, from carbon cycles to agriculture, there remains a suite of unpredictable and conflicting hypotheses explaining the physiological and molecular drivers of leaf senescence, particularly in temperate deciduous tree species (Gan and Amasino, 1997; Richardson et al., 2012; Liu et al., 2020). Temperate forests sequester ~30% of global forest carbon, assimilating ~11.1 metric tons of carbon per hectare per year, a process that happens almost exclusively in the spring and summer, ending each autumn with the onset of leaf senescence (Wofsy et al., 1993; Pan et al., 2011). Consequently, being able to accurately model this process is critical for improving predictions of future biotic feedbacks on the changing global climate, yet this requires a comprehensive understanding of the mechanisms driving senescence (Zani et al., 2020).

At the onset of autumnal leaf senescence, the leaf degrades chlorophyll, leading to a dramatic seasonal leaf color change in many species (Lim et al., 2007); throughout this process, the photosynthetic machinery is disassembled, and lipids, proteins, and mobile nutrients are remobilized out of the leaf to be stored in woody tissue (Smart, 1994; Buchanan-Wollaston, 1997). The known triggers and regulators of senescence are diverse, encompassing both internal and external stimuli, including leaf age, colder temperatures, shorter day length, shading, the onset of flowering or seed set, and pathogen attack (Smart, 1994; Buchanan-Wollaston, 1997; Noodén et al., 1997; Woo et al., 2019). The pathway through which plants sense and signal these drivers has not been fully elucidated, but phytohormones are believed to play a vital, coordinating role in translating the diverse environmental and endogenous cues into an initialization of the cascade of metabolic events defining senescence.

Ethylene is well established as a fundamental phytohormone responsible for triggering and accelerating leaf senescence (Aharoni and Lieberman, 1979; Jibran et al., 2013; Iqbal et al., 2017). Abscisic acid (ABA), a hormone primarily associated with drought response, has also long been thought of as another important phytohormone driving leaf senescence (Noodén and Leopold, 1988). ABA triggers and accelerates the rate of leaf senescence in leaf discs and detached leaves (El-Antably et al., 1967; Chin and Beevers, 1970; Gepstein and Thimann, 1980; Philosoph-Hadas et al., 1993; Fan et al., 1997; Hung and Kao, 2003; Gao et al., 2016; Zakari et al., 2020). It may also be involved in the lethal action of auxin-based herbicides (Grossmann, 2010; McCauley et al., 2020). Yet the role of endogenous ABA in leaf senescence in intact plants is less clear. Some evidence suggests that ABA can trigger leaf senescence if large quantities of the hormone are applied (El-Antably et al., 1967; He and Jin, 1999), while others have found that the leaf ABA level increases as leaves age, especially as senescence peaks (Gepstein and Thimann, 1980; He et al., 2005). In contrast, others have found that leaf ABA levels decrease during senescence (Uzelac et al., 2016), which suggests that ABA could be acting both synergistically and antagonistically with other drivers of leaf senescence depending on the timing of ABA increase, the specific tissue, and level of the hormone (Wingler et al., 1998; Pourtau et al., 2004; Rivero et al., 2007; Yang et al., 2011; Kong et al., 2013; Uzelac et al., 2016; Asad et al., 2019).

An abundance of recent molecular work has provided compelling evidence for the importance of the ABA signaling pathway in regulating leaf senescence. Leaf senescence in response to ABA application is reduced in Arabidopsis and rice mutants of single or multiple ABA signaling genes, from receptors through to downstream kinases (Liang et al., 2014; Gao et al., 2016; Liu et al., 2016). Overexpression of some of these genes, particularly the ABA receptor gene PYRABACTIN RESISTANCE1 LIKE 9 (PYR9), as well as downstream regulators of the ABA response, results in an enhanced leaf senescence phenotype in response to ABA application (Zhao et al., 2016; Mao et al., 2017). ABA is believed to enhance leaf senescence and cause early leaf necrosis in Arabidopsis through the regulation of a number of key transcription factors, mediated by reactive oxygen species (ROS) and calcium signals (Asad et al., 2019). When exogenous ABA is used to promote leaf senescence, in many studies across angiosperm species, significant changes to the expression of genes critical for ABA biosynthesis, signaling, and senescence pathways are altered (Fan et al., 2015; Park et al., 2018; Ren et al., 2018; Asad et al., 2019). While this evidence links ABA with senescence, there are some studies which show that ABA application, or mutation to some ABA synthesis genes, conversely, delays leaf senescence in response to environmental stresses, particularly osmotic stress induced by salt or glucose (Wingler et al., 1998; Pourtau et al., 2004; Yang et al., 2011). Furthermore, the effect of ABA on leaf senescence appears to be highly dependent on ontogeny, with both young leaves, which rarely senesce, and old, senescing leaves having similarly high levels of ABA (Powell, 1975; Raschke and Zeevaart, 1976; Weiler, 1980; Cornish and Zeevaart, 1984; Kane et al., 2020). It has also been suggested that the high ABA levels in senescing leaves may be due to a progressive decline in ABA sensitivity in the final stages of senescence, this insensitivity leading to more open stomata driving leaf desiccation in spite of high ABA levels in senescing leaves (Zhang and Gan, 2012; Zhang et al., 2012). In an ABA-deficient sunflower mutant, senescence as leaves age occurs at the same rate as in the wild type, and leaf ABA levels similarly decline as leaves senesce (McAdam et al., 2022).

While molecular evidence provides support for a role for ABA in driving senescence, a potentially confounding aspect of this relationship comes from observations that closed stomata can induce senescence independently of ABA (Thimann and Satler, 1979a, b). These results suggest that the primary pathway by which ABA could promote, or enhance, the rate of leaf senescence might be via ABA-driven stomatal closure, possibly leading to oxidative stress (Hensel et al., 1993; Thimann and Satler, 1979a, b). In support of this idea, work on drought-deciduous tree species indicates that leaf senescence is strongly associated with drops in leaf hydraulic conductance, decreasing leaf water potential (Ψ_L), and stomatal closure, which continue to decline throughout senescence (Brodribb and Holbrook, 2003). Low Ψ_L and reduced hydraulic conductance can also drive senescence in winter deciduous trees, with end-of-season senescence being associated with declining Ψ_L and reduced hydraulic conductance due to accumulating tyloses in the xylem in Castanea sativa Mill. (Salleo et al., 2002). Declines in Ψ_L might implicate ABA, which is synthesized in leaves as cells lose turgor (Pierce and Raschke, 1980; McAdam and Brodribb, 2016), although this has not been measured.

Work in herbaceous models can be hard to apply to deciduous trees because the process of senescence in annual and perennial plants may be divergent (Lim et al., 2007). Many herbaceous and annual plants have linked flowering and seed set with the onset of leaf senescence; the removal of flowers or fruit can substantially delay leaf senescence in many herbs (Leopold et al., 1959; Fan et al., 2020; Bucher and Römermann, 2021). In contrast, woody plants tend to time senescence ­almost exclusively with either, or both, declines in temperature and shortening day length, independent of reproduction (Lang et al., 2019). This further raises questions about whether ABA-triggered leaf senescence in herbaceous model species translates to a mechanism that regulates the senescence of leaves in forests.

A recent study in four deciduous species found that leaf ABA level increases during autumn, but only after senescence was initiated and Chl a content had begun to decline (Zhang et al., 2020). ABA has also been found to increase considerably after the onset of terminal embolism, including in three marcescent temperature species Quercus falcata Michx., Betula nigra L., and Carpinus betulus L. after frost-induced xylem embolism (McAdam et al., 2022). Experiments applying exogenous ABA to apple trees indicate a slight increase in senescence rates after the application of very high levels (3.78 mM, applied twice 7 d apart) of ABA as a foliage spray (Guak and Fuchigami, 2001). Other studies have utilized the girdling of branches, or whole trees, to test whether phloem obstruction enhances leaf senescence (Dann et al., 1984; Lihavainen et al., 2021). ABA levels increase following girdling due to interruption to phloem ABA export and a subsequent accumulation in leaves (Setter et al., 1980; Dann et al., 1984; López et al., 2015; Mitchell et al., 2017; Lihavainen et al., 2021). Girdling experiments in peach and Populus tremula L. indicate that girdled trees, or branches, experience accelerated leaf senescence in autumn (Dann et al., 1984; Lihavainen et al., 2021). In these studies, it is not clear whether ABA might be driving senescence indirectly by closing stomata, similar to the results of earlier work in leaf discs (Thimann and Satler, 1979a, b). Support of a stomatal pathway for the initiation of autumn senescence can be found in the accelerated rate of leaf senescence reported in Populus tremuloides Michx. in which leaves were herbivorized by epidermal leaf miners which fed exclusively on the lower epidermal cells of the hypostomatic leaves (Wagner et al., 2008). In plants with abaxial epidermal damage, including damage to guard cells, photosynthesis is greatly reduced because stomata cannot open; leaf life span is subsequently shortened by up to a month in these heavily herbivorized leaves (Wagner et al., 2008).

Here we sought to test whether: (i) increased endogenous ABA level drives autumn leaf senescence in four diverse winter deciduous trees, and (ii) whether enhanced ABA levels driven by girdling triggers autumn leaf senescence. We chose a phylogenetically and ecologically diverse set of mature, outdoor, specimen trees or shrubs including the gymnosperm Ginkgo biloba L. (Ginkgoaceae), angiosperm tree Phellodendron amurense Rupr. (Rutaceae), semi-frost-tolerant shrub Lonicera×purpusii Rehder. (Caprifoliaceae), and marcescent angiosperm tree Quercus falcata (Fagaceae). We included a marcescent species because these species display a delayed leaf senescence and limited leaf abscission as adaptations to extend leaf life span and maximize late season photosynthetic gains (Abadía et al., 1996) or to deter winter herbivory (Svendsen, 2001). We conducted a branch girdling experiment to determine the role of elevated levels of ABA on leaf senescence rates and physiological function, measuring leaf ABA levels, leaf gas exchange, Ψ_L, and leaf Chl a content during the autumn. Lonicera×purpusii was included in this study because leaves are highly tolerant of mild frosts with a very long life span (Upson and Kerley, 2007), allowing us to determine senescence rates driven by increased ABA independent of temperature. We hypothesize that ABA levels increase during early senescence in winter deciduous species and that girdling enhances the rate of leaf senescence via higher ABA levels.

Materials and methods

Plant material

The trees or shrubs of each species were established individuals growing in the grounds of the campus of Purdue University, West Lafayette, IN, USA, near Lilly Hall of Life Sciences (N 40.4233208, W 86.9167627). All plants received a biannual application of 15:3:3 N:P:K fertilizer; no pesticides were applied to any individual and no pruning was conducted in the 3 years preceding the study. Measurements were made from early September until all leaves had abscised, or were dead on the tree in the case of the marcescent Q. falcata.

Girdling experiment

To test whether increased endogenous ABA levels in leaves could promote senescence, three branches were randomly selected to act as ungirdled controls while three branches were randomly selected to act as girdled treatments. Our sampling approach treated branches as independent replicates, with the hope of minimizing potential confounding factors, including intraspecific variation, from obscuring treatment effects. The harvesting of leaves during the experiment impacted <0.01% of all leaves in the canopy. At 74 days after the summer solstice (DAS) in 2020, three branches from each species were girdled with a razor blade to remove the phloem while avoiding disruption of the xylem. Girdles were located 120 cm from the branch apex in the two deciduous angiosperm tree species, and 40 cm from the apex of the gymnosperm and shrub species. After the phloem was removed, the xylem was covered in petroleum jelly and wrapped in cotton wool and tape to limit desiccation from the girdled xylem. This ensured embolism was not induced at the girdling zone so that stem hydraulic conductance remained unchanged for the duration of the experiment. The first measurement of the girdled branches was made 3 d after girdling. During each collection, a leaf from each of the girdled and control branches was randomly selected and measured for gas exchange, Ψ_L, leaf ABA content, and leaf chlorophyll content. Measurements were made approximately twice a week on sunny days (to ensure a standard maximum rate of leaf gas exchange) until all the leaves had fallen, or died on the tree in the case of the marcescent Q. falcata. In all species, leaves were measured in randomly selected, ungirdled branches from 75 DAS until abscission or death twice per week in the preceding 2 years (2018 and 2019) to populate a photosynthetic rate and leaf chlorophyll content dataset in senescing leaves, as well as to provide a baseline dataset of senescence rates which was used to inform our sampling times in the year branches were girdled.

Measurements

Measurements were made between 11.00 h and 13.00 h. An infrared gas analyzer (LI-6800 Portable Photosynthesis System; LI-COR Biosciences, Lincoln, NE, USA) was used to measure CO₂ assimilation and stomatal conductance; intake air for the gas analyzer was drawn from a buffer drum so that CO₂ in the cuvette was not controlled and matched ambient atmospheric concentrations. Conditions in the cuvette were set to ambient vapor pressure difference between the leaf and atmosphere (VPD) and a saturating light intensity of 1500 µmol quanta m^–2 s^–1. Immediately after recording assimilation and stomatal conductance, the same leaf was detached from the tree and rapidly wrapped in damp paper towel before being double-bagged in Ziploc sandwich bags (SC Johnson, MI, USA) to prevent water loss for later determination of Ψ_L and sampling of tissue for pigment and hormone analysis. The bagged leaves were then placed in a dark bag to allow for Ψ_L equilibration for a minimum of 5 min. Ψ_L was measured using a Scholander pressure chamber (PMS Instrument Company, OR, USA) with a microscope. After slow depressurization of the Scholander chamber, leaf tissue was harvested into two tubes to measure ABA and chlorophyll content.

Leaf ABA was measured by physicochemical methods with an added internal standard. A subsample of tissue from the middle of the lamina, avoiding the midrib, was taken from each leaf, or the terminal pinna of the imparimonopinnate leaves of P. amurense, for hormone and pigment analysis. The mass of the fresh leaf sample was recorded (OHAUS Corporation, Parsippany, NJ, USA), and then the tissue was covered in –20 °C 80% methanol in water (v/v) containing 250 mg l⁻¹ butylated hydroxytoluene (BHT), chopped into fine pieces, and stored in a –20 °C freezer overnight. Methanolic extraction ensures both free and bound ABA is extracted (Georgopoulou and Milborrow, 2012). The leaf tissue was homogenized and 15 ng of deuterium-labeled [²H₆]ABA (OlChemIm, Olmouc, Czech Republic) was added to each sample before extracting overnight at 4 °C. An aliquot of supernatant, ~5 ml, was taken from each sample and dried to completeness in a vacuum sample concentrator (Labconco, MO, USA); the sample was then resuspended in 200 μl of 2% acetic acid in water (v/v), after which centrifugation was carried out at 14 800 rpm for 4 min. A 100 μl aliquot was taken for quantification of ABA and internal standard levels using an Agilent 6460 series triple quadrupole LC/MS (Agilent, CA, USA) according to McAdam (2015). Following quantification, the homogenized leaf samples were dried at 70 °C, and approximate leaf dry mass was determined by subtracting the mass of the clean empty tube from the mass of the tube containing the dried homogenized leaf material.

Fresh mass was also determined for the second leaf sample which was used for chlorophyll quantification. After weighing, the sample was covered in –20 °C acetone with 250 mg l⁻¹ BHT and roughly chopped and stored overnight at –20 °C before being homogenized and extracted overnight at 4 °C. A 100 μl aliquot of supernatant was taken for Chl a quantification using an Agilent 1100 high-performance liquid chromatograph, with an Agilent 1100 G1315B diode array detector for UV-visible spectrophotometric detection of pigments (Agilent, CA, USA). A Zorbax StableBond S8-C18 column (4.6 mm×150 mm with 5 µm particle size) (Agilent) and a quaternary pump (Agilent) were used according to McAdam et al. (2022). The amount of Chl a was calculated in the injected sample volume using a standard curve of known quantities of Chl a. Total extracted Chl a was determined by then multiplying this number by the total volume of the acetone used to extract pigments. The total volume of acetone in the sample was determined gravimetrically by obtaining the mass of the total homogenized sample in acetone, then drying down that sample at 70 °C and determining the mass of the dry sample. The volume of acetone was determined by multiplying the mass change and the density of acetone at standard atmospheric pressure at 180 m above sea level. Dry mass of the sample was then determined by cleaning and redrying the tube to determine the mass of the clean tube which was subtracted from the mass of the dry sample in the tube with no acetone.

Data analysis

Generalized additive models (GAMs) (and SEs) were fitted to Chl a content and Ψ_L data over the season for each species. The timing of the onset of senescence was determined for each species and year as the day that recorded a 25% reduction of Chl a content from the maximum according to the trend line of the GAM. The rate of senescence was taken as the number of days needed for leaf Chl a content to decline from 75% of maximum to 50% of maximum according to the GAM function. Single-factor ANOVAs were used to determine if there was a significant difference in the timing of the onset of senescence and rate of senescence across species and years.

To determine the impact of girdling on foliar ABA levels, the first five collections after girdling (~15 individual leaf samples) ending up to 22 d after girdling were averaged to quantify leaf ABA level after girdling. Significant differences between leaf ABA levels were determined using a Students t-test for each species.

Results

Variation in the onset and duration of senescence does not coincide with a change in ABA levels

In all species measured, autumn triggered considerable declines in Chl a content (Fig. 1). The onset of Chl a content decline varied depending on species (Fig. 1). The onset of Chl a content decline (or senescence) in ungirdled branches occurred 105 ± 3 DAS for P. amurense, 131 ± 25 DAS for L.×purpusii, and 87 ± 9 and ±2 DAS for Q. falcata and G. biloba, respectively. Average leaf ABA levels in ungirdled branches in P. amurense declined from 434 ± 153 ng g^–1 DW at 74 DAS to 296 ± 99 ng g^–1 DW 124 DAS, after which they increased until leaf abscission (Fig. 2). By the time foliar ABA levels began to increase, Chl a had already declined to 40% of initial levels. Quercus falcata showed a similar trend, with ABA levels falling from 243 ± 378 ng g^–1 DW to 176 ± 262 ng g^–1 DW 118 DAS, after which levels began to increase at 118 DAS when Chl a content had already declined to 37% of initial levels (Figs 2, 3). In L.×purpusii we never observed an increase in leaf ABA level, with the average leaf ABA level declining by 273 ng g^–1 DW over the course of the season (177 d) (Figs 2, 3). In G. biloba we observed a rapid increase in leaf ABA content 102 DAS when leaf Chl a had already declined to 47% of initial levels (Figs 2, 3).

Fig. 1.

Leaf Chl a level from mid-September 2020 in leaves from girdled (gray) and ungirdled (black) branches of Phellodendron amurense (A), Lonicera×purpusii (B), Ginkgo biloba (C), and Quercus falcata (D). Girdling took place 74 d after summer solstice. The solid lines depict a general additive model with R² values shown, while the dashed lines demarcate the SE. Each point represents a leaf from an independent branch.

Fig. 2.

Leaf ABA level from mid-September 2020 in leaves from girdled (gray) and ungirdled (black) branches of Phellodendron amurense (A), Lonicera×purpusii (B), Ginkgo biloba (C), and Quercus falcata (D). Girdling took place 74 d after summer solstice. The solid lines depict a general additive model with R² values shown, while the dashed lines demarcate the SE. Each point represents a leaf from an independent branch.

Fig. 3.

(A) Mean (±SE) of foliage abscisic acid levels in girdled (white) and ungirdled (black) branches from 3 d to 20 d after girdling (n=15). * denotes a significant difference in means (P<0.05, t-test). (B) The relationship between mean foliage abscisic acid levels (±SE) and the day after summer solstice when mean leaf Chl a content declined to 50% of maximum Chl a content, in girdled (white) and ungirdled (black) branches of Phellodendron amurense (squares), Lonicera×purpusii (hexagons), Ginkgo biloba (triangles), and Quercus falcata (circles). In species in which ABA levels were significantly different, points are connected by lines. The insert depicts the relationship between the onset date of leaf senescence and mean leaf ABA content, in girdled (white) and ungirdled (black) branches of P. amurense (squares), L.×purpusii (hexagons), G. biloba (triangles), and Q. falcata (circles). In species in which ABA levels were significantly different, points are connected by lines. (C) The relationship between mean foliage abscisic acid levels (±SE) and the rate of senescence or the number of days for Chl a content to decline from 75% to 50% of maximum in girdled (white) and ungirdled (black) branches of P. amurense (squares), L.×purpusii (hexagons), G. biloba (triangles), and Q. falcata (circles). In species in which ABA levels were significantly different, points are connected by lines.

Senescence onset timing and rate are enhanced by high ABA levels induced by girdling

The onset of senescence in leaves from the girdled branches fell within the standard error of the day of the onset of senescence in leaves from the ungirdled branches of Q. falcata (89 ± 3 DAS) and G. biloba (93 ± 2 DAS), but was different in the shrub L.×purpusii and P. amurense (Fig. 1). The onset of senescence in P. amurense was 10 d earlier in leaves from the girdled branches than in leaves from the ungirdled branches. In L.×purpusii, leaves from the girdled branches commenced senescence 33 d earlier than leaves from the ungirdled branches. The speed of senescence was also impacted by girdling in these two species, with the decline from 75% to 50% of maximum Chl a levels taking 30 d in leaves from the ungirdled branches of P. amurense but only 18 d in leaves from the ­girdled branches (Fig. 3). While senescence in L.×purpusii leaves on the girdled branches occurred in just 23 d, leaves on the ungirdled branches took 108 d to decline from 75% to 50% of maximum Chl a content. This earlier onset and accelerated senescence rate coincided with a much higher ABA levels in the leaves of the girdled branches of P. amurense and L.×purpusii in the 20 d after girdling. In P. amurense, the mean ABA level in the leaves from the ungirdled branches was 438 ± 50 ng g^–1 DW, while in the leaves from the girdled branches mean ABA level was 1084 ± 136 ng g^–1 DW during the same period (Fig. 3). In L.×purpusii, the mean leaf ABA level in leaves from the ungirdled branches was 483 ± 92 ng g^–1 DW compared with 2415 ± 467 ng g^–1 DW in the leaves from the girdled branches.

This contrasts with G. biloba and Q. falcata which had broadly similar patterns of decline in Chl a content between leaves from girdled and ungirdled branches (Fig. 1), and no significant increase in leaf ABA levels after branch girdling (Fig. 3). Ginkgo biloba leaves started senescing around the same time in both girdled and ungirdled branches at 93 ± 2 and 87 ± 2 DAS, respectively. The time it took for G. biloba leaves to go from 75% to 50% of maximum Chl a content was not influenced by girdling, taking 15 d in leaves from ungirdled branches and 18 d in leaves from girdled branches. This corresponded with no change (one-tailed t-test: P=0.0936) in leaf ABA content after girdling, with mean leaf ABA levels being 518 ± 70 ng g^–1 DW prior to girdling and 673 ± 91 ng g^–1 DW 20 d after girdling (Fig. 3). In Q. falcata, leaves from girdled branches began to senesce at 89 ± 3 DAS while leaves from ungirdled branches began to senesce at 87 ± 9 DAS. The rate of senescence was faster in leaves from girdled branches, taking only 5 d to decline from 75% to 50% of maximum Chl a, while leaves from ungirdled branches took 34 d to decline by the same percentage. The ABA content in leaves did not change after branch girdling in this species either (one-tailed t-test: P=0.1298), with mean leaf ABA level prior to girdling being 319 ± 47 ng g^–1 DW and 247 ± 43 ng g^–1 DW in the 20 d after girdling (Fig. 2).

Senescence is not driven by a loss of hydraulic conductance

Both P. amurense and L.×purpusii displayed similar seasonal trends in Ψ_L between the girdled and ungirdled branches during the autumn, with leaves becoming more hydrated through the season (Fig. 4). Ψ_L at 77 DAS in leaves on girdled branches of P. amurense was around –1.38 ± 0.06 MPa; at the same time, Ψ_L in the leaves on the ungirdled branches was –1.44 ± 0.08 MPa (two-tailed t-test: P=0.56). Leaves of both ungirdled and girdled branches became more hydrated to under –1 MPa on the initiation of abscission (Fig. 4). In L.×purpusii at 77 DAS, Ψ_L in girdled branches was –2.90 ± 0.6 MPa, while Ψ_L in leaves from ungirdled branches was at a similar –2.84 ± 0.55 MPa (two-tailed t-test: P=0.94). After this, Ψ_L trended to more hydrated values until the leaves of the girdled branches had abscised at ~151 DAS when Ψ_L in leaves from girdled branches measured –1.06 ± 0.41 MPa, while Ψ_L in leaves from ungirdled branches was –0.81 ± 0.14 MPa. Ψ_L in leaves from ungirdled branches continued to become more hydrated until all leaves were abscised at ~245 DAS when Ψ_L approached 0 MPa (Fig. 4). This increase in Ψ_L through the autumn means that there was no declining Ψ_L or loss of conductivity that could explain the seasonal trends in ABA level or the observed increase in leaf ABA level in leaves from girdled branches of P. amurense and L.×purpusii (Fig. 5). The only species in which Ψ_L did not increase as the season progressed was the marcescent Q. falcata in which Ψ_L declined dramatically after 134 DAS, when Chl a content had declined to >90% of maximum (Fig. 4).

Fig. 4.

Leaf water potential from mid-September 2020 in leaves taken from girdled (gray) and ungirdled (black) branches of Phellodendron amurense (A), Lonicera×purpusii (B), Ginkgo biloba (C), and Quercus falcata (D). Girdling took place 74 d after summer solstice. The solid lines depict a general additive model (GAM) with R² values shown, while the dashed lines demarcate the SE. Each point represents a leaf from an independent branch.

Fig. 5.

The relationship between leaf ABA level and leaf water potential in leaves taken from girdled (gray) and ungirdled (black) branches of Phellodendron amurense (A), Lonicera×purpusii (B), Ginkgo biloba (C), and Quercus falcata (D) in leaves sampled from 70 d after summer solstice through to leaf abscission or death on the plant.

Senescence reduces assimilation

CO₂ assimilation forms a strong relationship with stomatal conductance across all species, not only in the year measured but also when including data taken at the same time of year in all species in the preceding 2 years [R² of 0.8649, P-value of <0.0001, F-statistic 2490.07, and degrees of freedom (df)=390 (Fig. 6A)]. This relationship rises to a maximum assmilation rate of 19.4316 μmol m^–2 s^–1 of CO₂ and stomatal conductance of 0.5134 mol m^–2 s^–1 (Fig. 6A). CO₂ assimilation forms a slightly weaker but still significant linear relationship with Chl a content in data from all species measured in the autumn over 3 years (Fig. 6B). CO₂ assimilation forms a relationship with Chl a content, with an R² of 0.3595, a P-value of <0.0001, F-statistic of 199.84, and df=357 (Fig. 6B). As Chl a content declines, so does CO₂ assimilation (Fig. 6B).

Fig. 6.

(A) The relationship between CO₂ assimilation rate and stomatal conductance measured in leaves from all species from late summer until the end of the autumn. The solid line fits an exponential rise to maximum, single, two parameter {assimilation rate=21.519×[1^{– (–4.6816×stomatal conductance)}]}. The dashed lines represent the 95% confidence interval of the fit curve. (B) The relationship between the percentage of maximum Chl a content in leaves and percentage of maximum CO₂ assimilation rate in all species from late summer until the end of the autumn. The solid line depicts a linear regression (percent maximum assimilation rate=5.7428 + 0.7559×percent maximum Chl a content). The dashed lines represent the 95% confidence interval of the fit curve. Each point represents one leaf measured. Squares represent data collected from Phellodendron amurense, hexagons Lonicera×purpusii, triangles Ginkgo biloba, and diamonds Quercus falcata. Black symbols indicate data collected in 2018, white in 2019, and gray in 2020.

Discussion

ABA can enhance deciduous leaf senescence but does not increase at the onset of senescence

In contrast to some observations in Zea mays L. and Avena sativa L. (Gepstein and Thimann, 1980; He et al., 2005), we did not observe an increase in leaf ABA content at the onset, or for most of the duration of leaf senescence, in winter deciduous species. In our girdling experiments, we did observe that increased foliar ABA content during the autumn can trigger an earlier onset and faster senescence rate in two winter deciduous species (P. amurense and L.×purpusii). Our data are similar to those of four deciduous species measured in China, in which ABA levels were found to increase in deciduous tree leaves but only after Chl a content had begun to decline (Zhang et al., 2020). Our data also concur with many studies in Arabidopsis and other herbaceous species showing that either exogenous application of ABA, soaking leaf segments in ABA solutions, or mutation to the ABA synthesis or signaling pathway can enhance and trigger leaf senescence, or, in the case of mutants, delay senescence (El-Antably et al., 1967; Chin et al., 1970; Gepstein and Thimann, 1980; Philosoph-Hadas et al., 1993; Fan et al., 1997; Hung and Kao, 2003; Liang et al., 2014; Gao et al., 2016; Liu et al., 2016; Zakari et al., 2020).

The absence of an increase in ABA levels prior to the onset of senescence, or during senescence, in leaves from ungirdled branches implies that leaf ABA is not necessary for triggering the onset of autumn senescence in winter deciduous trees. We did observe an increase in leaf ABA level just before abscission or leaf death by heavy frost in the marcescent species Q. falcata. This could be related to phloem disruption during the formation of the abscission zone or frost-induced embolism causing a death-associated spike in leaf ABA level, as has been reported in marcescent species, including Q. falcata (McAdam et al., 2022). Our observations that ABA does not increase at the onset of leaf senescence should trigger further investigation into the nature of ABA dynamics in Arabidopsis leaves, and other herbaceous species during natural aging, to determine if endogenous leaf ABA levels increase at the onset of and during leaf senescence. We also do not know if ABA is likely to trigger leaf senescence if a plant is experiencing stressful conditions that normally augment the levels of ABA in leaves, such as drought.

Even though we did not observe increases in leaf ABA levels during the onset of leaf senescence in any of our observed species, our experimental data indicate that high levels of ABA may be able to enhance leaf senescence in deciduous species. The two species that showed significantly higher levels of leaf ABA after girdling (P. amurense and L.×purpusii) showed a corresponding earlier onset of senescence and, in P. amurense, a more rapid decline in Chl a content through the autumn. This was not observed in G. biloba in which leaf ABA level did not increase significantly following girdling. In the marcescent Q. falcata, no change in leaf ABA levels was observed after girdling, but we did observe a slightly enhanced degradation of Chl a following girdling, with the leaves from the girdled branches appearing to have an early decline in Chl a before matching leaves from the ungirdled branches later in the season. This could be evidence of an additional factor driving girdling-induced changes in chlorophyll degradation, such as changes in source–sink feedbacks triggering faster senescence as suggested by Zani et al. (2020). Our results confirm earlier work in herbaceous species which indicates that ABA can augment leaf senescence (Guak and Fuchigami, 2001; Zhao et al., 2016; Mao et al., 2017; Asad et al., 2019) and other studies that have reported increases in leaf senescence activity after girdling (Dann et al., 1984; Lihavainen et al., 2021).

Leaf dehydration and loss of conductance does not drive annual leaf senescence

In our deciduous species, we did not observe declining Ψ_L during the season. In C. sativa, Salleo et al. (2002) suggested that the gradual accumulation of occlusions in the xylem leads to decreases in hydraulic conductance, which was the primary driver for leaf senescence in this deciduous species. We did not see any evidence of hydraulic failure driving senescence in our species, with Ψ_L becoming more hydrated as stomata closed and leaves senesced. We could hypothesize that potential declines in hydraulic conductance in C. sativa would coincide with higher levels of ABA, which one might expect as ABA levels are closely linked with leaf water status (Pierce and Raschke, 1980; McAdam and Brodribb, 2016), and that this potential accumulated ABA drove the leaf senescence observed by Salleo et al. (2002). Declining pre-dawn Ψ_L values during the autumn have been observed in two cultivars of Prunus dulcis Batsch. grown in southeastern Spain; this decline in pre-dawn Ψ_L did not translate into a decline in mid-day Ψ_L which remained around –2 MPa for the whole season; similar results were observed in pot-grown seedlings of Magnolia grandiflora L. and Liquidambar styraciflua L. (Augé and Stodola, 1989; Ruíz-Sánchez et al., 1993). In contrast, in six evergreen conifer species native to Wyoming, both morning and mid-day stem water potentials increase towards 0 MPa during autumn, similar to what we observed here (Smith et al., 1984).

Stomata close during senescence and may drive autumnal leaf senescence

While ABA levels may play a role in accelerating the onset, and in some cases the rate of, leaf senescence, in deciduous species the interaction between ABA and stomatal closure triggering or accelerating senescence cannot be ruled out. Thimann and Satler (1979a, b) found that stomatal aperture played a critical role in leaf senescence activation and speed, independent of ABA. Further work is required to elucidate if ABA is directly activating and accelerating leaf senescence in winter deciduous trees, or if ABA acts indirectly via reductions of stomatal conductance and photosynthesis to drive leaf senescence. Leaf and whole-plant non-structural carbohydrate status may play a role in triggering senescence independently of ABA as has been suggested by Zani et al. (2020) and Westgeest et al. (2022, Preprint). Westgeest et al. (2022, Preprint) found that stomatal conductance in senescing leaves can be driven by leaf starch status. We found that the relationship between stomatal conductance and CO₂ assimilation stays consistent (Wong et al., 1979), and at no point during the leaf senescence process does CO₂ assimilation and stomatal conductance decouple as might be observed if stomatal opening during the senescence process due to ABA insensitivity occurred, which has been suggested by Zhang and Gan (2012) and Zhang et al. (2012) based on work with Arabidopsis. Our observed seasonal decline in stomatal conductance matches similar trends found in drought deciduous forests and evergreen conifers in temperate climates (Smith et al., 1984; Brodribb and Holbrook, 2003).

Conclusion

We find that ABA enhances leaf senescence in some winter deciduous species, but in unstressed plants ABA levels do not increase until the very end of leaf life after chlorophyll levels have almost completely declined. We observed that in deciduous trees as temperatures drop and days shorten, Chl a begins to decline at the same time that stomatal conductance and photosynthesis decline, while Ψ_L becomes less negative as transpiration declines.

Abbreviations

Abbreviations
 
• ABA

  abscisic acid

 
• DAS

  days after summer solstice

 
• GAM

  generalized additive model

 
• Ψ_L

  leaf water potential

Acknowledgements

We acknowledge the use of the facilities of the Bindley Bioscience Center (National Institutes of Health-funded Indiana Clinical and Translational Sciences Institute), particularly the Metabolite Profiling Facility. We would also like to thank Rodrigo Avila and Kean Kane for their assistance with field work, and two reviewers for their helpful comments.

Author contributions

SM: conceptualization; CK: conducting the experiments, collecting data, data analysis, and preparing the manuscript with input and supervision from SM.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be interpreted as a potential conflict of interest.

Funding

This work was supported by a United States Department of Agriculture (USDA) National Institute of Food and Agriculture Hatch project (1014908) and a National Science Foundation grant (IOS-2140119) to SM, and a Fulbright Fellowship to CK.

Data availability

The data supporting the findings of this report are available upon request to the corresponding author Scott McAdam.