Photosynthetic capacity of senescent leaves for a subtropical broadleaf deciduous tree species Liquidambar formosana Hance

Photosynthetic capacity and leaf life span generally determine how much carbon a plant assimilates during the growing season. Leaves of deciduous tree species start senescence in late season, but whether the senescent leaves still retain capacity of carbon assimilation remains a question. In this study, we investigated leaf phenology and photosynthesis of a subtropical broadleaf deciduous tree species Liquidambar formosana Hance in the central southern continental China. The results show that L. formosana has extended leaf senescence (more than 2 months) with a substantial number of red leaves persisting on the tree. Leaf photosynthetic capacity decreases over season, but the senescent red leaves still maintain relatively high photosynthetic capacity at 42%, 66% and 66% of the mature leaves for net photosynthesis rate, apparent quantum yield, and quantum yield at the light compensation point, respectively. These results indicate that L. formosana may still contribute to carbon sink during leaf senescence.

Forests form the most important carbon pool of the terrestrial ecosystems, thus play a significant role in global carbon balance. The recent decade has seen an increasing research interest on carbon assimilation and balance in tropical and temperate forests [1][2][3][4][5] . However, studies on ecological process (e.g. carbon uptake, water use) on subtropical forests are still limited until recently [6][7][8][9][10][11] . In the East Asian monsoon region, it distributes a typical subtropical forest ecosystem, which is composed of evergreen broadleaf, deciduous broadleaf and mixed stands 8 . Photosynthesis of forest trees in this region might be limited by temperature in winter, and by solar radiation in summer when other resources are abundant 9 . Nevertheless, the subtropical evergreen forests maintain high rates of photosynthetic activity (around 50% of peak values) in cold winter 7 . Considerably high winter carbon assimilation (5.4 to 8.8 μmol CO 2 m −2 s −1 ) has been documented for ten subtropical evergreen broadleaf tree species 12 . Extended leaf senescence of deciduous species in this climatic zone can promote carbon assimilation in winter 9 . However, the subtropical deciduous tree species have attracted little attention regarding their photosynthetic capacity and potential carbon assimilation during leaf senescence.
Leaf senescence is a primary characteristic of deciduous species, with visual change in leaf pigmentation in autumn and winter. Trees need to store resources to fuel bud flush and shoot growth in the following spring 13 . Nutrients (e.g., carbohydrates) stored in the leaf require a continual supply of photosynthesis to support resorption 14 . Therefore, photosynthesis during leaf senescence is critical to plant physiological processes 14,15 . Timing of leaf senescence can have a significant impact on ecosystem productivity 16 . An extended growing season provides extra time for plant photosynthetic activity, increasing carbon assimilation potential. Goulden et al. 17 has documented that a delay in senescence for 5-10 days in a temperate deciduous forest can result in an increase of about 500 kg C ha −1 in the annual gross production. For subtropical deciduous broadleaf forest in China, a previous study shows that its growing season has prolonged from both an earlier onset of green-up and a delay in dormancy 18 . A good understanding of photosynthetic capacity of senescent leaves is essential to estimate the effects of the prolonged growing season on carbon assimilation for such deciduous species.
Temperature is regarded as an important driving factor on plant phenology. The length of growing season increases due to global warming [19][20][21][22][23] . Delays in vegetation dormancy onset date have been documented recently for a wide range of ecosystems in China 24 . Different deciduous tree species differ in leaf life span and duration of leaf senescence, which may lead to different impacts on late season (autumn and winter) carbon assimilation during leaf senescence. Zhang et al. 9 show that red leaves of a subtropical deciduous tree species during leaf senescence maintain a relatively high photosynthetic rate (about 36% of peak values in summer) and contribute to carbon uptake in the late season. Whether or not such a late season photosynthesis pattern of senescent leaves is common for other deciduous species remains to be investigated.
Lack of observation-based studies on the deciduous species is responsible for the current poor understanding of photosynthesis of senescent leaves. This situation also limits our capacity to model the carbon fluxes of forest ecosystems. It is reported that most of the published models for the terrestrial carbon cycle fail to capture the response of tropical carbon fluxes to climate variability 25 . Such problems may be associated with biases in the estimated response of productivity or ecosystem respiration to climate 5 , or the misrepresentation of photosynthetic process among different plant functional types (e.g., evergreen and deciduous species) in the models.
In this study, we investigate leaf photosynthesis and phenology of a common subtropical deciduous tree species, Liquidambar formosana Hance, in the field in Hunan Province, the central southern China. The primary objectives are to investigate temporal variation of photosynthetic capacity of L. formosana and to examine whether senescent leaves still retain capacity for carbon assimilation.

Results
Leaf development and senescence. The leaves of L. formosana usually sprout in March, with the start of the growing season in the study area. Leaf color changes over time from light green in early spring (March-April) to deep green in summer (June-July), and finally it turns to reddish in autumn, lasting until the end of December. The leaf life span is as long as ~10 months while other deciduous trees (such as Quercus fabri) in the study area are completely leafless at the end of November. L. formosana leaves also sprout earlier than some other deciduous trees. A quick increase in canopy leaf area index (LAI c ) occurs in spring during its early leaf development period (Fig. 1). The LAI c maintains a relatively constant value in and after June, which means that L. formosana leaves are fully expanded around this time. The LAI c decreases rapidly after leaves senescing and shedding in November, but before mid-December, the LAI c value is still relatively high (LAI c > 0.5; approximate 17% of the maximum LAI c value in summer) (Fig. 1). This indicates that a large number of red leaves remain on L. formosana trees after about two months into senescence.
Leaf gas exchange and photosynthetic capacity. L. formosana leaves are most sensitive to photosynthetic photo flux density (PPFD or I) in April when leaves are new. The observed maximum rate of net photosynthesis (P nmax ) is highest in April, and then decreases slightly to a mean value of 6.2 μmol CO 2 m −2 s −1 in July and August when leaves are mature (Fig. 2). The observed P nmax decreases significantly (P = 0.000006) when leaves start senescing and turning red in November. But the red leaves still maintain a positive P n even in the later stage (early December) of leaf senescence, with observed P nmax around 0.9 μmol CO 2 m −2 s −1 . This result indicates that the senescent red leaves maintain positive CO 2 assimilation in late autumn and early winter.
The modeled maximum rate of net photosynthesis, the dark respiration rate (R d ), the apparent quantum yield (α), and the quantum yield at light compensation point (Φ c ) decrease gradually in the leaf life span of L. formosana (Fig. 3a-c). In summer when leaves are mature, P nmax , R d , α and Φ c maintain a relatively constant value (5.98 μmol CO 2 m −2 s −1 , 1.45 μmol CO 2 m −2 s −1 , 0.05 and 0.04 mol (CO 2 ) mol −1 (photon), respectively). In autumn and winter when leaves are in senescence, P nmax , R d , α and Φ c decrease significantly (Fig. 3a-c) to about 42%, 46%, 66% and 66% of that of the mature green leaves in summer, respectively. This indicates that the senescent red leaves still maintain relatively high photosynthetic capacity during this period. The model simulated light compensation point (I c ) and light saturation point (I sat ) remain relatively constant during leaf senescence, but both of them are lower than those of mature leaves in summer. The average I c and I sat of the senescent leaves are 69% and 45% of mature leaves, respectively.   Leaf transpiration rates (T r ), water use efficiency (WUE), and leaf stomatal conductance (g s ) shown in Fig. 4 further demonstrate the differences in these physiological properties between senescent red leaves and mature green leaves. The T r and g s of green leaves show relatively high and, statistically different values (2.98 mmol H 2 O m −2 s −1 and 0.10 mol H 2 O m −2 s −1 , respectively) for mature leaves (July to August) compared to those for senescent leaves (P = 0.00001 for T r , and P = 0.0003 for g s ). The average g s is about 0.061 mol H 2 O m −2 s −1 during the leaf senescence, about 64% that of mature green leaves in summer (Fig. 4c). This relatively high value of g s indicates that the stomata of red leaves are still active (open), which contributes to maintaining a positive P n and T r during the leaf senescence. But the mean WUE of red leaves is close to that of green leaves. There is no significant difference (P = 0.365) in WUE between red leaves (late October to December) and green leaves (July to August).

Discussion
Leaf development and senescence is a regulated metabolic but irreversible process for plants during the course of development 15,[26][27][28] . Time of leaf development and senescence are influenced by environmental and internal factors 15 . Even in one forest ecosystem, different deciduous trees are different in leaf life span. These differences can have a significant impact on ecosystem process such as carbon uptake, water cycle, and annual net ecosystem production 29 . In this study, L. formosana growing season starts in March and has a long leaf span for about 10 months. This leaf span is shorter than that of the same species in Taiwan (about 11 months, growing in further south to this study area of the subtropical region of China) but longer than that of Liquidambar styraciflua L. (less than 8 months) 16 in a temperate region (Fig. 5). Figure 5 shows that subtropical deciduous tree species usually have a longer leaf life span and pronounced extended leaf senescence than the temperate deciduous trees. This leaf phenological characteristic for subtropical deciduous plants is likely a result of long-term climate acclimation of these subtropical species. The prolonged growing season (compared to temperate trees) for subtropical deciduous trees means a longer time of photosynthetic activity in comparison to their temperate counterparts.
With a warming climate, leaf life span of deciduous species likely becomes longer (e.g., earlier onset of green-up, or extended leaf senescence, or both) 15,18 . Yu et al. 18 has reported that temperature is a main influence factor on advances of green-up and delays of leaf senescence in subtropical deciduous broadleaf forest in China. In order to find the relationship between leaf phenology of L. formosana and temperature, we correlate the duration of leaf shedding for L. formosana among different areas in China and the relationship between their phenological parameters (time of leafing, the end of leaf shedding, leaf life span and duration of leaf shedding) with mean annual temperature (Fig. 6). L. formosana trees that grow in lower latitudes show a longer duration of leaf shedding (Fig. 6a). Meanwhile, a negative correlation is showed between mean annual temperature and the time of leafing, and a positive correlation between the end date of leaf shedding and mean annual temperature (Fig. 6b,c). These results indicate that L. formosana trees living in lower latitudes, experiencing higher mean Comparison of leaf transpiration rates (T r ), water use efficiency (WUE), and leaf stomatal conductance (g s ) at light saturation point between mature green leaves (July to August) and senescent red leaves (late October to December). Values are averages ± SE (n = 3 days for mature leaves, n = 9 days for senescent leaves). Different uppercase letters on the top of the bars mean significant differences at P = 0.01 level between the mean values for mature and senescent leaves.
temperatures, have a longer leaf life span than those in higher latitudes. Leaf life span and duration of leaf shedding show a significant positive correlation (P = 0.0004 and 0.0007 respectively) with mean annual temperature. The difference in leaf phenology of L. formosana among different sources results from the long-term environmental acclimation, including temperature acclimation. This indicates that the leaf life span and leaf senescence (shedding) of L. formosana is sensitive to temperature, among other environmental factors (e.g., soil nutrients and water availability) and is likely to, if factored in global warming, extend in the future. In fact, autumn air  temperature has increased at a rate of 0.37 °C/decade from 1970 to 2015 at the study site ( Figure S1), which has possibly influenced leaf senescence of L. formosana.
Leaf senescence is a natural loss of leaf function for deciduous species and generally reflects the acclimation to the environmental conditions. Warmer ambient temperature is expected to delay autumn senescence 23 , and increases plants activity. Keenan et al. 30 and Dragoni et al. 31 have shown that warming-induced late autumn senescence enhances net carbon uptake in temperate evergreen and deciduous forests. But the conflicting results have been reported for boreal forest 32,33 . In addition to temperature, other environmental variables may cause variation in leaf physiological activities. Herrick et al. 16 shows that elevated CO 2 concentration stimulates leaf photosynthesis of a temperate deciduous species (Liquidambar styraciflua) in late season, but does not lengthen its growing season. Leaf senescence can be accelerated or delayed in some species under drought condition 34,35 . Overall, environmental factors are closely related to plants leaf phenology, growing season and photosynthetic function, which crucially defines how much carbon a plant assimilates during the leaf life span. Documentation of leaf phenology and photosynthetic capacity provides baseline data to investigate the possible ecosystem response to future climate change. In this study, L. formosana shows an extended leaf senescence period with positive net photosynthesis rates under the current climate condition. This evidence supports the notion of late season carbon assimilation for this subtropical deciduous species.
The down-regulation in photosynthetic capacity during senescence has been documented for many deciduous tree species 9,16,[36][37][38] . The results of this study show that the photosynthetic capacity (P n , P nmax , α, Φ c ) of L. formosana senescent red leaves is lower than that of mature green leaves, but the red leaves still maintain relatively high photosynthetic capacity (about 42%, 66% and 66% of P nmax , α and Φ c of the mature green leaves, respectively) during leaf senescence (Fig. 3). Figure 7 gives a further demonstration of changes in percentage loss of the photosynthetic capacity of L. formosana, in comparison to some other deciduous trees from summer to winter in the Northern Hemisphere. Of all species included in the comparison, Lyonia ovalifolia (Wall.) Drude and L. formosana maintain relatively high photosynthetic capacity during the senescence period, with a reduction of ca. 65% and 81%, respectively (Fig. 7).
In this study, we focus on photosynthetic capacity of L. formosana during leaf senescence. Although the net carbon gain of L. formosana cannot be accurately estimated due to a lack of night-time respiration measurements, we can conclude that L. formosana trees maintain positive daytime carbon assimilation during the extended leaf senescence. Whether or not this daytime carbon assimilation exceeds carbon loss from night-time respiration remains for future investigation. For subtropical deciduous tree species (including L. formosana), further research, in the context of global warming, is needed to investigate the responses of leaf phenology and photosynthetic capacity to climate changes.

Methods
Study site. This study was performed in Yuelu Mountain, located in Changsha city in the central southern China (28°10′36″N, 112°55′58″E, 190 m above sea level). Changsha is characterized with a humid subtropical monsoon climate, with a mean annual precipitation of 1,447 ± 36 mm (mean ± SE, average of 1970-2015) and mean annual temperature of 17.4 ± 0.1 °C (mean ± SE, average of 1970-2015). Resulting from northerly cold air mass influences in the winter monsoon season, it usually has dry and cold winter. Rainfall mainly occurs in spring to early summer. Figure 8 shows the climate conditions in 2014 and 2015. Both years were wetter under the context of El Niño phenomenon. Especially for 2015, it rained more than normal from September to December. Mean monthly air temperature of 2014-2015 was lower in the summer than that of long-term mean monthly temperature and was higher in autumn and early winter than the long-term average. The soils of this site are Figure 7. Percentage loss of photosynthetic capacity (P nsat (light-saturated net photosynthetic rate) or P nmax ) of L. formosana and other subtropical (solid symbols) and temperate (open symbols) deciduous tree species for comparison. All data are percentage loss from the maximum P nsat in summer season. Data for Lyonia ovalifolia (monthly) are from Zhang et al. 9 , Liquidambar styraciflua from Herrick et al. 16 , Quercus robur from Morecroft et al. 36 , Larix occidentalis from Rosenthal et al. 37 , Populus tremula from Sun et al. 38 , Quercus alba L., Quercus prinus L. and Acer rubrum L. from Wilson et al. 53 .
yellowish red soils. The surface soil layer contains organic matters 39 . Forest in Yuelu Mountain is dominated by typical subtropical evergreen broadleaf trees, mixed with some deciduous trees.
Liquidambar formosana Hance, investigated in this study, is the dominant deciduous tree species interspersed naturally among the dominant evergreen species. This species widely distributes in subtropical deciduous broadleaf forest, and evergreen and deciduous broadleaf mixed forest 40,41 . The leaves of L. formosana usually sprout in spring (March) and begin to turn yellow and/or red in autumn. Senescent leaves of this tree species usually last for about two months in this region. Nevertheless, in northern Taiwan (southern China), L. formosana shows a longer senescence period (more than three months) 42 .
Leaf gas exchange. Leaf gas exchange measurements were performed on six L. formosana trees from October to December 2014, and April to November 2015. Measurements of 3-5 sun-exposed leaves for each individual tree were taken for light response curves between 0900 and 1130 hour local time on typical sunny days, using a LI-6400XT portable photosynthesis system (LI-COR, NE, USA). The photosynthetically active radiation (PAR) was provided by a red/blue light source (Li-6400-02B) connected to the system with a specific gradient of I (PPFD) ranging from 0 to 2000 μmol m −2 s −1 (0 to 1000 μmol m −2 s −1 for measurements from October to December 2014). At each light intensity level, we kept a minimum wait time of 120 s, and a maximum wait time of 200 s before capturing the value. For dark respiration, we kept a wait time of 200 s for a dark adaptation of leaves prior to capturing its value. The air flow rate was set to 500 μmol s −1 , chamber temperature was kept constant at 25 ± 1 °C and chamber CO 2 concentration was kept at 400 ± 2 μmol CO 2 mol −1 .
Responses of net photosynthesis rates (P n ) to I were fitted with a recently published photosynthetic light response model modified from the rectangular hyperbolic model 43 , as shown in where α is apparent quantum yield, R d is dark respiration rate, β and γ are coefficients which are independent of I. Then, the maximum rate of net photosynthesis (P nmax ), light compensation point (I c ), and light saturation point (I sat ) can be calculated from the model. Moreover, this new model also calculates the quantum yield at I c (Φ c ) 43 Measurements in a day were averaged for each light intensity level, and then were fitted with equation (1). The parameters calculated from the light response model provide quantitative representation of the photosynthesis capacity and efficiency 45,46 .
Measurements of leaf gas exchange also include water vapour, which can be used to calculate leaf transpiration rates. Leaf-scale water use efficiency (WUE) is defined as the ratio between leaf net photosynthesis rate (P n ) and transpiration rate (T r ).
Leaf area index for the canopy of an individual tree. Leaf area index (LAI) is a key parameter reflecting the structure of plant canopy. Unfortunately, LAI is difficult to quantify accurately although many methods have been proposed [47][48][49] . For the purpose of this study, leaf area index for the canopy 50 of a single tree (LAI c ) is used to reflect canopy characteristic and number of leaves retained on the studied L. formosana trees. We adopted a regular camera-based method developed by Pekin and Macfarlane 49 . The method estimates crown cover and LAI c based on the digital photography. LAI c is calculated by the following equations: