Effect of floods on the δ¹³C values in plant leaves: a study of willows in Northeastern Siberia

Study area

The study site is located in the Indigirka River lowland near Chokurdakh (70°38′N, 147°53′E), Sakha Republic (Yakutia), Russian Federation (Fig. 2). Mean annual air temperature in the region between 1950 and 2016 was −13.7 °C, ranging from −33.9 °C in January (the coldest month) to 10.1 °C in July (the warmest month). Mean annual precipitation between 1950 and 2008 was 209 mm year⁻¹ (Yabuki et al., 2011). The Indigirka River lowland, including rivers, lakes, wetlands, hills, and floodplains, is frequently flooded during spring and summer. Soils in the region are loamy or silty-loamy alluvial soils with black- to grayish-olive color along the riverbanks (Troeva et al., 2010). The average depth of the active layer in soils is approximately 30 cm on land and one m near the river in the summer. The local vegetation consists of aquatics, sphagnums mosses, graminoids, shrubs (mainly the willow Salix sp. and the dwarf birch Betula nana), alders, larches, and pines. Between 1970 and 2016, the average intra-annual water level cycle of the Indigirka River was 70 ± 83 mm for April and May (late winter, pre-flooding), increasing to 600 ± 93 mm for June–August (spring and summer, flooding season); then, gradually receding to 343 ± 146 mm for September and October (autumn and early winter, post-flooding), and declining further to 56 ± 26 mm in winter (after October). Field experiments were approved by Hokkaido University, and Institute for Biological Problems of Cryolithozone, Siberian Branch of Russian Academy of Science, and North-Eastern Federal University.

Willows in the Indigirka River lowland

The common willow species observed in 2015–2017 were Salix boganidensis, S. pulchra, S. glauca, S. richardsonii, S. viminalis, S. alaxensis, S. fuscescens, and S. hastata. Most species were ≤1 m tall, except for a few species such as S. boganidensis, S. alaxensis, and S. fuscescens, which were two to three m in height. Diameter at breast height generally ranged between one and six cm. Maximum root depth was approximately one m at the riverbank, but was highly variable and depended on various factors such as the thickness of the active soil layers and moisture levels where the willows grew. Willows were distributed more densely along the riverbanks than on dry lands.

Observations conducted with a GardenWatchCam time-lapse camera (Brinno, Inc., Taipei City, Taiwan) showed that the buds of willow leaves opened around the first few weeks of June, when the snow had melted and the daily average air temperature had increased to >0 °C. The leaves and stems grew rapidly, within 10 days after bud opening, and were fully developed by mid-July. Willow leaf biomass peaked by the end of July, and this observation was consistent with a normalized difference vegetation index (NDVI) study in Alaska (Boelman et al., 2011). Aboveground net primary production (ANPP, newly formed stems and leaves in each year) and the leaf area index (LAI) of the willows in 2016 were measured using the direct harvesting method (Jonckheere et al., 2004) in three blocks which were predominated by willows. ANPP was 63, 119, and 117 g m⁻²·a in each of the three blocks, and the LAI was 0.59, 0.71, and 1.59 in each of the three blocks (Table A1).

Samples

In the summer of 2015 and 2016, we collected leaves from the locally dominant willows Salix boganidensis, S. glauca, and S. richardsonii on three sets of 20 m transects (SBoydom, SKA, and SKB) from the river. SBoydom is located between the mainstream Indigirka and the wetland; while, SKA and SKB are situated next to a secondary tributary, Kryvaya (Fig. 2; Table A2). Three points, named PA, PB, and PC, were marked on each transect based on their distance to the river. The maximum thaw depth was always found at PA. This layout was designed based on the differences in intra- and inter-annual flooding conditions (Figs. 2 and 3). PAs at SKB and SBoydom were continually waterlogged throughout the growing season in 2015 and continuously waterlogged until July in 2016 (Fig. 3). PB at SKB and PA at SKA were flooded only in 2016 (Fig. 3).

Four current-year top shoots were collected at each point at the end of the growing season (the end of July) in both 2015 and 2016. Current-year shoots were also randomly sampled from willows in local scale on the Indigirka River lowland during the same period. A total of 31 sites with different locations were used in 2015 and 2016 (Fig. 2; Table A2). At least four current-year shoots were collected at each location to obtain representative data for each site.

The details of sampling sites, locations, species, and sampling numbers are shown in Fig. 2 and Table A2. All samples were immediately dried at 60 °C for 48 h after collecting.

Stable carbon isotope analysis

Dried leaves were milled into fine powder with liquid N₂ and dried again at 60 °C for 48 h; each sample was then wrapped in a tin capsule and injected into an elemental analyzer (Flash EA 1112; Thermo Fisher Scientific, Bremen, Germany), connected to an isotope ratio mass spectrometry (IRMS, Delta V; Thermo Fisher Scientific, Bremen, Germany) through a continuous-flow carrier-gas system (Conflo III; Thermo Fisher Scientific, Bremen, Germany). The stable carbon isotopic composition was reported in the standard δ notation relative to VPDB. A laboratory standard was injected after every ten samples to verify that the analytical accuracy was better than 0.1‰. To reduce the effect of sampling heterogeneity in δ¹³C within a single site, four samples were measured and the average isotopic composition was reported for each site, with the standard deviation ranging from 0.0 to 1.4‰ (average, 0.7‰).

Photosynthetic rate and stomatal conductance analyses

Supporting data on the foliar δ¹³C values, the photosynthetic rate and stomatal conductance of willow leaves were monitored in the field in 2017 using a portable porometer (LCpro+; ADC BioScientific Ltd, Hoddesdon, Herts, UK) equipped with a conifer chamber and a lighting system. The photosynthetic rate (A) of S. boganidensis, S. richardsonii, and S. glauca under different light levels (10–955 μmol m⁻² s⁻¹) was measured to obtain light response curves and thus, to identify the saturation light intensity.

Site SPh near Chokurdakh village, was set up in the summer of 2017 to monitor the conditions in former transects SKA, SKB, and SBoydom, since the extremely high flooding caused all these three sites totally submerged for the entire summer of 2017. Under gradient flooding conditions on site SPh (-PA: submerged till July 20; -PB: submerged till July 15; -PC: without submergence during the observation period), temporary changes were measured in the photosynthetic rate (A) and stomatal conductance (gs) of S. richardsonii, S. glauca, and S. boganidensis in response to a single saturated light exposure at 600 μmol m⁻² s⁻¹ around noon, the rest of chamber conditions were set to match ambient conditions. For each measurement at the points (PA, PB, or PC), a total of 12 leaves from four trees were marked for leaf ADC data recording for more than six times on any leaf of them. Average of all records was calculated for each measurement. Measurements were taken five times every 2–3 days between July 13, 2017 and July 27, 2017. The leaves were also collected after whole monitoring period to check the foliar δ¹³C values.

Statistical analysis

Linear Mixed Models (LMMs) were used to clarify differences in the foliar δ¹³C value among willows growing in three transects in 2015 and 2016. Foliar δ¹³C value was set as the response variable, with flooding condition was set as the fixed effect, and species (i.e., S. boganidensis, S. richardsonii, and S. glauca) was set as a random effect. Similar analyses by LMMs were also used to figure out any differences in the foliar δ¹³C value among the willows randomly collected on the Indigirka River lowland in 2015 and 2016. Foliar δ¹³C value was set as the response variable, the flooding condition was assigned as the fixed effect, and the location (along the mainstream or the tributary), and species (Table A2), were set as random effects. Tukey’s test was used as a post hoc analysis for multiple comparisons. The lme4 package (Bates, Maechler & Walker, 2015) of R (R Core Team, 2015) was used to build the LMMs.

Foliar δ¹³C in the transects

Foliar δ¹³C values differed among SKA, SKB, and SBoydom, and between years (Fig. 3). Along transect SKA in 2015, the mean foliar δ¹³C values were −30.3 ± 0.8‰ for PA (close to the river but without submergence), which was much lower than those for PB (−28.0 ± 0.6‰) and PC (−27.5 ± 0.7‰). A similar trend was also observed in 2016, when the water level was high; in this case, the foliar δ¹³C values for PA (−29.4 ± 0.4‰) was lower than those for PB (−29.2 ± 0.3‰) and PC (−28.6 ± 0.9‰); although, the difference was small. For the inter-annual changes from 2015 to 2016 on transect SKA, an increase in foliar δ¹³C value was observed for PA (+0.8 ± 0.6‰); whereas, a decrease was recorded for both, PB (−1.2 ± 0.5‰) and PC (−1.0 ± 0.8‰).

In 2015, the sampling point at PAs along both, the SKB and SBoydom transects, were sometimes waterlogged (“WL”; Fig. 3); whereas, all points PBs and PCs were not. A similar trend for the foliar δ¹³C values was found in both, SKB and SBoydom transects, as the foliar δ¹³C values for PAs (−27.4 ± 1.1‰ and −27.3 ± 0.7‰, respectively) were higher than those for PBs (−28.5 ± 0.6‰ and −27.4 ± 1.0‰, respectively), and PCs (−28.1 ± 0.9‰ and −28.3 ± 0.7‰, respectively) (Fig. 3). In 2016, although PAs were also but always waterlogged (“LWL”; Fig. 3), PBs and PCs were not, trends in the foliar δ¹³C value were apparently different between SKB and SBoydom. In SKB, the foliar δ¹³C values for PA (−29.0 ± 0.6‰) were slightly higher than those for PB (−29.2 ± 0.5‰) and PC (−29.7 ± 0.9‰); whereas, in SBoydom, values for PA (−29.5 ± 0.7‰) were slightly lower than those for PB (−28.3 ± 0.5‰) and PC (−28.0 ± 0.8‰) (Fig. 3). For the inter-annual changes in SKB and SBoydom, decreases in the foliar δ¹³C value were observed at all points in all transects, except for PC in SBoydom. The differences in the foliar δ¹³C value between the PB and PC within each transect ranging from −0.5 to +0.9‰; however, those for the PA significantly deviated (−2.5 to +0.9‰) from the mean value of the PB and PC.

Overall, statistical analysis of the foliar δ¹³C values from transects (in Fig. 3) showed a significant difference (F_3,42 = 42.276, P < 0.01, Fig. 4) among the four major hydrological conditions (i.e., “Dry,” “Wet,” “WL,” and “LWL,” in Figs. 3 and 4). The foliar δ¹³C values were high in dry and waterlogged continually conditions. Conversely, values under wet and continuous long period waterlogging were consistently low.

Spatial distribution in the foliar δ¹³C of willows

Neither significant nor large differences were detected in the foliar δ¹³C value in any of the willows growing in the 31 randomly selected sampling sites in local scale (with different locations) in Indigirka River lowland during the same sampling periods at the end of the growing seasons of 2015 and 2016 (Fig. 5). The willow foliar δ¹³C values ranged from −31.1 to −25.3‰ in 2015, and from −31.6 to −25.7‰ in 2016. Statistical analysis shows that there were no significant differences between the four environmental conditions (flooding, flooded, on land, and wetland) in 2015 or 2016. Moreover, the δ¹³C values measured for willow leaves collected in a larch forest in 2016 (−27.6 ± 1.2‰) were significantly higher than those sampled anywhere else in the same year (F_4,168 = 2.58, P = 0.039).

Photosynthetic rate and stomatal conductance

Photosynthetic rate (A) gradually increased asymptotically and reached to about six μmol m⁻² s⁻¹ for S. richardsonii, about six to eight μmol m⁻² s⁻¹ for S. glauca, and about 8–12 μmol m⁻² s⁻¹ for S. boganidensis, under 400–600 μmol m⁻² s⁻¹ (irradiation scanning covered the range from 10 to 955 μmol m⁻² s⁻¹) (Fig. A1). Therefore, the saturating light intensity for willow leaves in the Indigirka River lowland was found in the range from 400 to 600 μmol m⁻² s⁻¹. Maximum photosynthetic rate A recorded in leaves of S. boganidensis was the highest among all three species (Fig. A1).

In the summer of 2017, willows at the SPh-PA were continuously submerged until July 20; willows at the SPh-PB had just come out of the water when monitoring began on July 15; while, willows at the SPh-PC were not submerged during the monitoring period (Fig. 6A), although all three points at transect SPh were within 20 m from the river. The largest decrease (−1.6 ± 1.4 μmol m⁻² s⁻¹) in A at SPh-PA during the monitoring period, was registered on July 21 when the waterlogging just finished; whereas, after flooding A was observed to follow a slow recovery over the last few days (Fig. 6B). A similar decrease-increase trend in A was also detected in the willow leaves at SPh-PB, and also, in this case, the lowest A was recorded soon after waterlogging finished on July 18 (−0.7 ± 2.6 μmol m⁻² s⁻¹ lower than first measurement) (Fig. 6B). However, at SPh-PC, A continuously increased, compared to the initial measurement. These values corresponded with the waterlogging gradients in SPh, that A was reduced under waterlogging and could recover if waterlogging was over. On the other hand, among points, compared to SPh-PC, the lowest gs values were found at SPh-PA; while, intermediate values were recorded in SPh-PB (Fig. 6C). Thus, gs, apparently correlated to the degree of waterlogging among SPh-PA, -PB, and -PC.

River water level and leaf formation

As mentioned before, the willow leaves began opening after the first week in June and finished growing by the end of July. This suggests that the foliar δ¹³C values of willow leaves recorded hydrological conditions experienced from early mid-June to the date of collection, which is a longer period than that of the in situ observation and monitoring period for hydrological conditions. However, it is known that the river water level is gradually decreased but by within approximately one m during the term of leaf growing (Fig. A2). Moreover, only small differences were found in the foliar δ¹³C values between top and bottom of a single current-year shoot, approximately 0.5 ± 0.1‰ (Fig. A3), which may have experienced leaf formation over different periods. Thus, in present field observation study, the hydrological conditions observed during July are assumed to be almost the same or very similar to those for the early part of the growing season.

Foliar δ¹³C values under different hydrological conditions

Foliar δ¹³C values under continuous long period waterlogging

Sporadic waterlogging (“WL”; Fig. 3) related to medium flooding increased the willow foliar δ¹³C values. In 2016, however, the foliar δ¹³C values at SKB-PA (waterlogging) were only slightly higher (approximately 0.4‰) than those at SKB-PB and SKB-PC. The foliar δ¹³C values at SBoydom-PA (waterlogging) were even lower (approximately 1‰) than those at SBoydom-PB. Moreover, between 2015 and 2016 the foliar δ¹³C values at PAs on SKB and SBoydom decreased by 1.7 ± 0.6‰ and 2.1 ± 0.9‰, respectively, despite waterlogging occurring in both years. To date, very few studies have investigated the reasons for the lack of changes or negative shifts in foliar δ¹³C value under waterlogging conditions. We propose that, under the long period waterlogging (“LWL”; Fig. 3) as caused by large flooding observed at PAs on SKB and SBoydom in 2016, the changes in the A are also important factors controlling the foliar δ¹³C values, besides gs. In the present study, low A was observed during submergence, before July 20 and more flooded SPh-PA (Fig. 6C). It has been suggested that waterlogging induces low carboxylation rate by reducing the amount or activity of Rubisco enzyme (Vu & Yelenosky, 1992; Islam & Macdonald, 2004). According to Eq. (3), the foliar δ¹³C values should be dependent on both A and gs; thus, low A may have caused the negative shifts in the foliar δ¹³C value in LWL compared to WL, as the foliar δ¹³C values at SPh-PA were slightly lighter than at SPh-PB, albeit insignificantly so (Fig. 6A). Therefore, the low foliar δ¹³C values observed in the willows at the PAs on SKB and SBoydom in 2016, can be explained by this continuously low photosynthetic activity under long period waterlogging caused by large flooding. The low δ¹³C values, A, and gs under large flooding were previously reported for a pot experiment involving the invasive wetland grass Phalaris arundinacea (Waring & Maricle, 2012). The aforementioned findings together suggest the hypothesis that long period waterlogging (or large flooding) significantly reduces the foliar δ¹³C values compared to sporadic waterlogging (or small flooding), and that the contribution of A and gs is highly dependent on the frequency and magnitude of waterlogging events.

Thus, the large differences (Fig. 4) in hydrological conditions (i.e., “Dry,” “Wet,” “WL,” and “LWL”; Fig. 3) found in transects suggest that the possible foliar δ¹³C values can correspond to scenario 3 in Fig. 1 (i.e., reduced foliar δ¹³C values in long period waterlogging), which can be well interpreted by that both A and gs are affected by hydrological gradients (Fig. 7).

However, we note that scenario 3 in this study is a very simplified, schematic hypothesis, without quantitative meaning. There are other several potential factors for controlling the foliar δ¹³C values, for example mesophyll conductance (gm) (Evans et al., 1986). As gs and gm were found tightly coupled (Vrábl et al., 2009), and both controlled the limitation of CO₂ diffusion. Therefore, in our field study of the determining factors of foliar δ¹³C values, we mainly focused on gs which can be directly monitored. Detailed changes in the δ¹³C value particularly between and within the four hydrological conditions will be illustrated in further studies with determination of these potential factors, including but not limited to gm changes under different water regimes, the quantitative meaning of gs and gm on foliar δ¹³C values, and the features of gs and gm in species with different water tolerance.

Spatial difference in the willow foliar δ¹³C value

Linear Mixed Models analysis of the local scale random sampling indicated that only the willows in the dry larch forest were slightly but statistically enriched in ¹³C (F_4,168 = 2.58, P = 0.039), compared to the other conditions, in 2016. It is accepted that the foliar δ¹³C values are higher in dry than in wet conditions, due to stomatal regulation. In contrast, there was no statistical difference in the foliar δ¹³C value for either year or among the hydrological conditions tested here (flooding, flooded, on land, and wetland) (Fig. 5). The first three conditions were situated near the river and at different levels of flooding, whereas the fourth was never affected by flooding, although it was still abundant in water. These results are likely consistent with the lack of difference in the foliar δ¹³C value among various hydrological conditions in mesic regions or periods, as reported previously (Garten & Taylor, 1992; Alstad et al., 1999). Nevertheless, relatively minor variations in the δ¹³C value in willows growing under these hydrological conditions cannot be explained by the common dry-wet stomatal regulation theory mentioned above. On the other hand, our transect data in this study can explain why there was a slight variation in the foliar δ¹³C value among random sampling sites in local scale in response to hydrological gradients as follows.

In 2015, the water level in the river was low. Consequently, the hydrological status of the willows growing nearest the river in the “Flooded” zone, ranged from slight flooding (similar to “Wet” in Figs. 3 and 4; i.e., “small flooding” in Fig. 7) to continual flooding (results in continual waterlogging, e.g., “WL” in Figs. 3 and 4; i.e., “medium flooding” in Fig. 7). Slight flooding near the wet condition zone caused the stomata to open and low foliar δ¹³C values. In contrast, medium flooding resulted in stomatal closure and high foliar δ¹³C values. Therefore, under the “Flooded” condition, the δ¹³C values varied between low and high. This behavior resembles the positive-negative shift in the foliar δ¹³C value for the “On land” zone under dry-wet conditions. The same interpretation applies to the foliar δ¹³C values measured in the “Wetland” (Fig. 5).

In contrast, in 2016, the water level in the river was high. Therefore, the hydrological status of the willows growing nearest the river in the “Flooding” zone varied between continual flooding (results in continual waterlogging, e.g., “WL” in Figs. 3 and 4; i.e., “medium flooding” in Fig. 7) and continuous flooding (leads to long period waterlogging, e.g., “LWL” in Figs. 3 and 4; i.e., “large flooding” in Fig. 7). Large flooding reduced both, A and foliar δ¹³C values. As was the case for 2015, the conditions in the “Flooded” and “Wetland” zones of 2016 ranged between slight and continual waterlogging (similar to “Wet” and “WL,” respectively in Figs. 3 and 4), although only waterlogging in “Flooded” zones was caused by floods (i.e., “small flooding” and “medium flooding”; Fig. 7). Therefore, the foliar δ¹³C values ranged between low and high in the “Flooding,” “Flooded,” and “Wetland” areas. These responses resemble the positive-negative shift in the foliar δ¹³C value observed for “On land” under dry-wet conditions.

In previous studies (Garten & Taylor, 1992; Alstad et al., 1999), very small differences in the foliar δ¹³C value of plants growing near rivers were detected among the diverse hydrological conditions, where waterlogging frequently occurred. These minor differences can also be explained by the physiological responses of willows related to the different hydrological conditions (Fig. 7). If the δ¹³C values in other organs correlate with those determined by the leaves, then historical records of the wide swings in hydrological conditions could be reconstructed using the δ¹³C records, such as those obtained from tree ring cellulose.