Differential water-use strategies in co-occurring pioneers and late-successional tree species in secondary tropical montane forests of Eastern Himalaya


 Plant-water relations in secondary tropical montane forests (TMFs) are driven by complex interactions between environmental conditions, species composition, and forest structure. We investigated the differential water-use strategies of cooccurring pioneers and late-successional tree species in a secondary TMF of Eastern Himalaya, India. It is the wettest (mean annual precipitation = 4500 mm yr -1 ) high-elevation (> 2000 m) site in the world. The observed maximum daily stand transpiration (5.3 mm) is highest among other tropical montane or lowland forests. Although energy-limited, increased moisture availability allowed the observed sap flux densities from the studied species, Symplocos racemosa, Eurya acuminata, and Castanopsis racemosa , to be 3-9 times higher than their conspecifics from relatively drier TMFs. Interestingly, differential access to solar radiation, a characteristic of the forest canopy in secondary forests, induced significant radial and azimuthal variability in sap flow. Solar radiation was the key driver of transpiration in energy-limited winters and Vapour pressure deficit in energy-abundant summers. Nocturnal (1800-0500h) transpiration was significant (13.8%) part of daily T and was dominated by pre-dawn flux. Shallow-rooted pioneers, S. racemosa and E. acuminata , exhibited strong midday depression in sap flow in response to environmental extremes and soil moisture fluctuations, whereas the deep-rooted late-successional C. hystrix transpired unaffected. The complex interactions between different successional groups for accessing changing energy and moisture conditions are highlighted for prioritized conservation and management of these secondary forests in Eastern Himalaya.


89
In most terrestrial systems, the availability of energy (and light) and water are the primary drivers 90 of water-use by vegetation (Schulze et al. 1987, Heimann andReichstein 2008). Spatiotemporal variability 91 in temperature, precipitation, and soil moisture exert significant controls on the distribution and 92 composition of vegetation (Todd et al. 2010, Piedallu et al. 2013. Over the long term, vegetation adapts to 93 the environmental conditions and biological competition through functional traits and strategies enhancing 94 photosynthesis and water-use efficiency (Ackerly et al. 2000, Troch et al. 2009, Tang et al. 2014. In turn, 95 vegetation also significantly alters the availability of water in their environment through transpiration and 96 hydraulic redistribution (Neumann andCardon 2012, Schlesinger andJasechko 2014). Transpiration taps 97 into subsurface moisture and releases it into the atmosphere altering local moisture balance (Federer 1973, 98 Barros and Lang 2003). This see-saw interaction between vegetation and moisture availability is the focus 99 of studies on the Soil-Plant-Atmosphere-Continuum (SPAC), an active area of research in the field of 100 ecohydrology (Asbjornsen et al. 2011, Manzoni et al. 2013, Mencuccini et al. 2019. The SPAC framework contribute to large biases in estimating stand transpiration, especially in bio-diverse secondary tropical 117 montane forests (TMFs) (Čermák et al. 2004, Forrester 2015, Hernandez-Santana et al. 2015.

129
Similarly, research indicates differential growth rates between dry and wet sites and across elevation 130 gradients, again with soil moisture being the key limiting variable (Poudyal et al. 2004b, Shrestha et al. 131 2015. Others have explored the variability in their ecophysiological responses to seasonal moisture deficit 132 across elevation, species range and canopy structure (Zobel et al. 2001 133 Tewari et al. 2018). They have also pointed that a majority of the broad-leaved TMFs in the Himalaya are 134 secondary in nature due to the long history of human forest use (Ramakrishnan and Kushwaha 2001, Sikkim, representing the Eastern Himalayan eco-climatic zone, is characterized by steep 144 topography, complex geomorphology, high rainfall, and diverse ecosystems (Bookhagen andBurbank 145 2010, Tambe et al. 2011). Broad-leaved forests (800 -2800 masl) form the most dominant vegetation type 146 in Sikkim (Tambe et al. 2011). A third of these are secondary forests modified by decades of anthropogenic 147 pressures and remain understudied (Kanade and John 2018). They provide sanctuary to the principal water 148 resources, the springs and streams, and understanding their plant-water relations is critical to the regional 149 water and ecological security (Tambe et al. 2012). Plant-water relations have not been studied in the wet 150 secondary TMFs of Eastern Himalaya using direct whole-tree methods. To address the knowledge gap, we 151 use the SPAC framework to understand the relative influences of the wet tropical environment and 152 secondary forest structure on plant-water relations in an East-Himalayan broad-leaved evergreen secondary 153 TMF. Specifically, we address two main questions: (a) What are the environmental and ecophysiological 154 drivers of water-use in key broad-leaved evergreen tree species under wet tropical Eastern Himalayan eco-155 climate? and (b) How the plant-water strategies differ between co-occurring pioneers and late-successional 156 species in a secondary forest structure? The study is conducted in an early successional secondary forest annual precipitation between the years 2013-2016 was 4649±120 mm and the mean annual daily 171 temperature ranged from -2⁰C to 24⁰C. The year is divided into three distinct seasons: Summer (MAM) 172 characterized by warm days, cloudy afternoons, significant pre-monsoon precipitation, and high 173 evapotranspiration; Monsoon (JJASO) with concentrated precipitation, high humidity, and low 174 evapotranspiration; and Winter (NDJF) with bright days, cold nights, occasional snowfall events and 175 moderate evapotranspiration (Pandey et al. 2016, Kumar et al. 2021. The forest sloped at 10°-35°, 176 predominantly facing the north-east direction, and the soil was well-drained sandy-loam with an average 177 soil depth of 70 cm. The forest stand represents the East Himalayan broad-leaved wet montane forests 178 (Sudhakar et al. 2008, Kanade andJohn 2018)

194
In the forest stand, the pioneer tree species dominate the short-statured canopy. The older and large-girthed 195 remnant individuals of the Fagaceae family such as C. hystrix standing out as emergent. Table 1 summarises the information on rooting depth, phenology, and other ecophysiological characteristics of the 197 three species collated from available literature and personal observations (Ohsawa et al. 1986, Suzuki et al. 198 1991, Sundriyal and Sharma 1996, Chettri et al. 2002, Li et al. 2013

204
Sap flux was measured in a total of 13 individuals of the three dominant species using Granier's 205 thermal dissipation method-based probes (TDPs) ( Table 2) (Granier 1987, Lu et al. 2004). The sample size 206 of 13 trees (5 trees each of S. racemosa and E. acuminata, and 3 trees of C. hystrix) was at the higher end 207 of the average number of trees sampled in sap flow studies globally (Mackay et al. 2010, Asbjornsen et al. 208 2011. TDPs are known to perform well in cold and low-flow conditions (Lu et al. 2004, Chan andBowling 209 2017). The probes were manufactured locally based on suitable modifications to the original design with 210 only the top one cm measuring the sap flux (for details see Phillips et al. 1996, James et al. 2002, Harmon 211 2009, Davis et al. 2012. The choice of radial and azimuthal installation sites on a tree was carefully done 212 to optimize number of sensors and xylem variability (Guyot et al. 2015). We installed radial probes (1-5 cm 213 in length) spirally at one cm incremental depth from the cambium in one tree per species. The replicate 214 trees were fitted with two 2-cm length probes installed in north and south azimuthal aspects (Shinohara et 215 al. 2013, Komatsu et al. 2016. The length of sapwood was measured using wood cores extracted using an 216 increment borer. C. hystrix showed distinct heartwood formation and only the outer xylem (4-5 cm) was 217 functional. No heartwood was detected in S. racemosa and E. acuminata and the entire xylem was assumed 218 functional, a characteristic of diffuse-porous species (Berdanier et al. 2016). The probe signals (in 219 millivolts) were recorded and converted to temperature difference (∆T in °C) between the heater and cm 2 hr -1 ) was computed using Granier's empirical equation (equation 1) and maximum temperature 222 difference (∆Tmax) was calculated daily (for details see Granier 1987, Lu et al. 2004).

224
The period of data collection was from November 2013 to May 2014. The first two weeks of data were 225 ignored to avoid sap flux underestimation due to installation wounds (Wiedemann et al. 2016). Due to very 226 wet conditions, especially in the summer, it was difficult to maintain the field instrumentation for a long 227 duration and data gaps were inevitable (see supplementary data Figure S2 for time-series plots of the data 228 used in the analysis). After due quality checks, a total of 114 days of sap flow data across all species was 229 used for the final analysis (species-wise break-up provided in Table 2). The data processing, analysis, and 230 visualization were done in R (version 3.6.0) (R Core Team 2021) using RStudio (RStudio Team 2021). Table 2. Biometric details (± standard deviation) along with the total number of data days and number of 233 radial probes installed for the three species.

236
Soil water potential was recorded at 10 cm incremental depths from the topsoil using granular 237 matrix-based (watermark) sensors (Virtual Electronics, Roorkee) and converted to volumetric water content 238 using the site-specific van Genuchten water retention curve (Schaap et al. 2001, Gribovszki 2018. Total 239 soil moisture (S in mm h -1 ) was computed for the topsoil (0-30 cm depth) using the trapezoidal method 240 (Nachabe et al. 2005). S was smoothened using a 3-step moving-average window to gap-fill stray missing 241 values. In-canopy air temperature and relative humidity were recorded (iButton Hygrochrons, Maxim Int., 242 USA). Air temperature (°C), relative humidity (Rh in %), wind speed (u in m s -1 ), and incoming short-wave 243 radiation (Rs in kW m -2 ) were recorded using an automatic weather station (AWS) (Vantage-pro Davis Net, in mm h -1 ) was recorded using an automated tipping-bucket rain gauge. Reference evapotranspiration (ET) 247 was computed based on FAO's Penman-Monteith method (FAO56PM, for detailed steps please see Allen 248 et al. 1998), which was reported to be the suitable method for estimating ET at the nearby Gangtok in 249 Sikkim (Pandey et al. 2016). Leaf area index (LAI) was extracted for the site from MCD15A3Hv006 level- We used a combination of the zero-averaged technique and weighted mean method to estimate 257 whole-tree sap flow (Vrad in kg hr -1 ) in radial trees (for details see Hatton et al. 1990, Pausch et al. 2000.

261
Vrad.2cm showed significant overestimation (15 %) in reference to Vrad suggesting the use of an appropriate 262 scaling mechanism to incorporate the diurnal variability in radial patterns.

439
acuminata followed by C. hystrix and S. racemosa (see supplementary data Table S4).  fraction of (13.8 ± 6 %) of the daily T with a marginally higher proportion of evening than pre-dawn flux.

450
and moderately saturated soil moisture conditions (see supplementary data Figure S8). The MLR models 451 predicting Vnight concurred with the observations (see supplementary data Table S5). S (negative slope) was 452 a significant predictor of Vnight in E. acuminata (r 2 = 0.12, P < 0.05) and VPD (negative slope) was a 453 significant predictor for C. hystrix (r 2 = 0.13, P < 0.009). None of the predictors were significant for S.

533
Further, species composition, in terms of functional groups, canopy structure, and phenology exert a strong 534 influence on differential access of vegetation to solar radiation and soil moisture (Ziegler et al. 2009). We 535 discuss their relative roles in explaining the observed intra and interspecific variability in transpiration.

567
Sapwood area was found to be a better bio-scalar of tree to stand transpiration than tree diameter and basal The observed nocturnal transpiration (Tnight) was similar to studies (~12 %) from China (Siddiq 572 and Cao 2018) and other parts of the globe (Forster 2014, Alvarado-Barrientos et al. 2015 and lower that 573 reports from Mediterranean Europe (35-40 %) (Barbeta et al. 2012). While most of the nocturnal sap flux 574 occurred in the evening, as reported commonly, pre-dawn flux is an uncommon observation from

575
Himalayan TMFs (Forster 2014). The onset timing (0300-0400 h) during days with pre-dawn flux was 576 similar to reports from Mediterranean Europe (Barbeta et al. 2012). The evening sap flux is commonly 577 attributed to stem refilling and thus is a feature of large trees, although not restricted to them (Zeppel et al.

619
canopy Rhododendron arboreum (Zobel et al. 2001, Poudyal et al. 2004a, Tewari et al. 2018. However, 620 the absence of significant midday depression in deep-rooted species like C. hystrix at the peak of the dry 621 season and strong diurnal cycles in soil moisture and streamflow (please refer to Chapter 4 section 3.2.4) 622 provide evidence of vegetation accessing moisture from groundwater (Tanaka et al. 2004, Maeght et al. between co-occurring tree species in Semi-arid China has shown high transpiration rates by pioneer 627 species, especially in the growing season (Lyu et al. 2020). The relatively dominant contribution by small 628 and medium-sized trees to T underlines the importance of conducting more sap flow studies on pioneers, 629 especially in diverse tropical secondary forests (Hornbeck et al. 1997, Nogueira et al. 2004

636
VPD as drivers of T has been reported in other TMFs as well . We observed that solar 637 radiation was a stronger driver of T than VPD and S in energy-limited winters, similar to report from TMF 638 in Southern Andes (Motzer et al. 2010) and Alps (Fiora and Cescatti 2006), but in contrast to warmer 639 Central Himalaya, where VPD was observed as a stronger driver than solar radiation (Ghimire et al. 2014). The observed Gaussian radial profile with increasing sap flux density in the inner xylem in C.

659
hystrix is typical of isolated trees with long crown (Fiora and Cescatti 2006). This is ascribed to higher light 660 availability to the older mid-lower portions of the crown anatomically connected to the inner xylem.

675
2013). The biases induced by the failure to incorporate azimuthal and radial variability were found to be

679
It is still challenging to estimate stand transpiration from sap flow measurements in diverse secondary

680
TMFs because of the variability among different functional groups and diversity in traits that regulate 681 plant-water relations (Huc et al. 1994, Nogueira et al. 2004, Küppers et al. 2008. Radial and azimuthal 682 probes opened a whole new dimension to plant-water use for Himalayan species and their use is highly 683 recommended to reduce errors in estimating stand transpiration (Shinohara et al. 2013, Komatsu et al. 684 2016. We also highlight the importance of incorporating information on stand structure and tree density 685 while scaling from individuals to stand-level (Küppers et al. 2008). Regardless, the stand transpiration of 686 these East Himalayan wet tropical montane broad-leaved forests has been estimated for the first time using within different parts of a tree, and between co-occurring species (Francescantonio et al. 2018). Thus, the 703 secondary broad-leaved forests, with the dominance of shallow-rooted pioneers, are more prone to the However, increased summer precipitation in the future could result in higher cloud cover negatively 707 impacting both vegetation productivity and transpiration leading to increased streamflow in summer (Aston 708 1984, Donohue et al. 2017. Increased summer moisture and cloud cover could also allow the opportunistic 709 fast-growing pioneers to outcompete the late-successional species (Lyu et al. 2020).

710
The overall effect of changes in temperature and precipitation on biodiversity in the region remains 711 complex and requires ecohydrological models specific to the East Himalayan TMFs (Tsering et al. 2010, 712 Asbjornsen et al. 2011). More regional studies inclusive of diurnal and seasonal variability in transpiration 713 will be critical to increasing the accuracy of land-surface interaction models and predicting the impact of 714 climate change on Himalayan ecohydrology Dickinson 2012, Miller et al. 2018). It will be 715 worth investigating the role of soil mycorrhizae and root hair densities in facilitating the high transpiration 716 rates observed in these forests (Breda et al. 2006, Carminati et al. 2017 Table 3. Estimates of sap flux density (J), whole-tree sap flow (V), tree-to-tree coefficient of variation in sap flow (COV) in the studied species and at stand-level. Percentage biases in whole-tree sap flow (V)estimation due to ignoring radial and azimuthal variability (standard deviation in parentheses) are presented across species and in stand transpiration. C. hystrix 13 (5) 23 (16)  8 (7) -21 (41) Stand-level 22 (9) 21 (11) 9 (5) -16 (24) Table 4. Results from Generalized least squares (GLS) linear regression model with corARMA correlational structure for transpiration (*P < 0.05, ** P < 0.01, *** P < 0.001). Predictor variables include