How to Regenerate and Protect Desert Riparian Populus euphratica Forest in Arid Areas

We found that the most suitable flooding disturbance model for regenerating Populus euphratica forest was two to three times per year with a duration of 15–20 days and an intensity of 25–30 m3/s. The flooding should take place during the seed emergence to young tree growth stages, and should be based on flooding experiments and data from vegetation quadrats and ecological water conveyance. Furthermore, we found that tree-ring width index for P. euphratica declined as the groundwater depth increased, and ascertained that the minimum groundwater depths for young trees, near-mature trees, mature trees and over-mature trees were 4.0 m, 5.0–5.4 m, 6.9 m and 7.8 m, respectively. These were derived from a quantitative relationship model between groundwater depth and tree-ring width index. The range for ecological water conveyance volume was 311–320 million m3 in the lower reaches of the Tarim River. This study not only provides a technical basis for sustainable ecological water conveyance in the Tarim River Basin, but also offers a theoretical guide and scientific information that could be used in similar areas to regenerate and protect Populus euphratica around the world.

Scientific RepoRts | 5:15418 | DOi: 10.1038/srep15418 flooding disturbance, such as intensity, frequency and duration, fluctuate and are volatile, simulation experiments cannot be conducted in the area 26,[30][31][32] . Hence, previous studies have been unable to identify a suitable river flooding disturbance management plan that will help regenerate damaged ecosystems using artificial regulation (e.g., the EWCE Project in the lower reaches of the Tarim River).
Many studies [33][34][35][36] have indicated that the relationship is close to streamflow and tree-ring data. Streamflow is mainly converted into groundwater to meet the ecological water requirement of vegetation 37,38 . Similarly, the water required by the desert riparian P. euphratica forest is mainly supplied by groundwater 37 . Previous research determined the minimum groundwater depth that will promote P. euphratica growth for the whole forest in terms of plant eco-physiological variables and the tree-ring radial stem growth of P. euphratica etc 8,39,40 . However, diameter at breast height (DBH) for P. euphratica trees is correlated with the different root distributions corresponding to different groundwater depths. Therefore, based on the results of previous studies, the EWCE Project is unable to divide water according to the different DBHs of trees using the present recommendations 8,39,40 . In addition, some studies 39,41 could not accurately describe the variation in P. euphratica growth due to the different groundwater depths because radial stem growth is affected by a self-growth trend. Consequently, the response models based on de-growth-trend tree-ring width indices for P. euphratica tree-ring growth and groundwater depth are rare. According to the above analysis, this study used the ecological water conveyance data from 2000-2012 and the groundwater depth data acquired from 30 monitoring wells, to analyze the effects of different flooding disturbances on P. euphratica seedling growth. Firstly, we produced an optimal disturbance model for seedling growth, based on flood disturbance experiments conducted in the lower reaches of the Tarim River. Secondly, a quantitative relationship model was constructed using different groundwater depths and tree-ring width indices, and then we determined the minimum groundwater depths for different tree DBHs and developed an effective ecological water conveyance management plan to regenerate and protect desert riparian P. euphratica forest in the lower reaches of the Tarim River. The aims of this study were (1) to provide technical assistance that will help achieve sustainable ecological water conveyance in the Tarim River Basin, and (2) to offer scientific guidance that can be used in similar areas to regenerate and protect P. euphratica forest around the world.  16 . The basin wide runoffs are mainly supplied by ice-melt water and precipitation, which amount to 39.83 billion m 3 per year. The total basin area is 1.02 million km 2 , and covers five prefectures, forty-two counties and fifty-five regiments of the Construction and Production Corps of Xinjiang 20 . The population is 11.03 million.

Materials and Methods
The main part of the Tarim River is a closed watershed and has a total length of 1321 km 37 . In the past, there were nine great rivers flowing into the Tarim River. However, since the 1940 s, the Kaxgar, Weigan, Keriya, Kuqa, Dina, Kaidu-Kongque and Qarqan rivers have been cut off from the Tarim River because of climate change and human activities 20 . Only the Hetian, Yarkant, Kongque and Aksu rivers remain tributaries of the Tarim River 37 .
Both sides of the Tarim River contain desert riparian forest, mainly P. euphratica, which is one of the most important arid area forest vegetation communities in Asia 8 . Shrub vegetation mainly consists of Halostachys caspica, Tamarix and Haloxylon ammodendron, and the main herbs are Phragmites australis, Alhagi sparsifolia and Karelinia caspica 37 .
Our research area was located in the lower reaches of the Tarim River, downstream of the Daxihaizi cross-section. It covered a total area of 2191 km 2 (Fig. 1). This area is one of the most arid areas in China with a mean annual precipitation of 40 mm and mean annual potential evaporation of 2590 mm 42 . The cumulative temperature above 10 °C for this area is 4040 to 4300 °C, with a mean diurnal range of 13 °C to 17 °C 38 .
The lower reaches of Tarim River have experienced water cut offs since 1972, which has led to a reduction in the area of the desert riparian forest, a loss of P. euphratica forest and environmental degeneration 37,38 . In order to rescue the eco-environment of the lower reaches of the Tarim River, and to regenerate and protect desert riparian forest, the Chinese government, since 2000, has implemented the EWCE Project, which stretches from the Kongque River to the downstream Tarim River area 24,37,38 . From 2000 to 2006, eight times the normal water conveyance has flowed into Taitema Lake, which amounts to a total of 2.25 billion m 3 and a mean annual amount of 0.32 billion m 3 (Fig. 2). Between 2007 and 2009, twice the usual amount of water was allowed to flow, but because this was far short of the required amount (total amount, 20 million m 3 ; mean annual amount, 10 million m 3 ), the water head only reached the Kardayi Section. However from 2010 to 2012, three times the usual amount of water, which supplied 1.91 billion m 3 , at a mean annual amount of 0.64 billion m 3 , meant that water was able to resupply Taitema Lake again. As the EWCE Project proceeds, river flooding will expand to cover both sides of the river channel, which will lead to a rise in the groundwater table and help repair the damaged desert riparian forest 37 (Table 1) and 434 tree-ring cores from P. euphratica trees with different DBHs that were located in the lower reaches of the Tarim River.
Flooding experiment. Based on the ecological water conveyance data acquired, combined with the actual river flooding conditions recorded by field investigations in the study area, we divided the flooding factors (frequency, duration and intensity) into four grades in the downstream reaches of the Tarim River (Table 2).
According to the grades, we randomly placed 3-5 vegetative quadrats that were 1 m × 1 m, 2 m × 2 m or 5 m × 5 m in size, to investigate the composition, number, height, DBH, crown width, coverage and frequency of vegetation so that we could measure changes in plant community structure under different flooding disturbance regimes. This amounted to 50 sample plots and 467 quadrats in the study area.
The results from the field research enabled us to calculate the important values (IV) for P. euphratica seedlings in each quadrat so that we could determine whether the P. euphratica seedlings would survive or not.
Dr Cr Hr 3 1 where Dr represents relative density; Cr represents relative coverage; and Hr. represents relative height 43,44 . Sampling and processing of tree rings. Four transects where P. euphratica predominated were selected as the study region, i.e., Yingsu (C), Karday (E), Algan (G) and Yiganbujima (I) (Fig. 1). We chose six to ten trees with different DBHs around one monitoring well in order to reduce the deviation. According to the principles of the International Tree-ring Database, we selected 217 trees and 434 tree-ring cores between August, 2009 and August, 2013., Based on the field survey and tree-ring analysis of P. euphratica forest in the lower reaches of Tarim River, the 217 trees were divided into five grades according to DBHs so that we could study the different P. euphratica growth periods. These were 4-10 cm (young trees), 10-30 cm (near-mature trees), 30-40 cm (mature trees) and > 40 cm (over-mature trees) ( Table 3). The tree-ring sampling and cross-dating were completed according to the standards published by Fritts 45 . Two cores were taken from each tree using an increment borer. The samples were handled according to dendro-chronological laboratory procedure (i.e., drying, mounting, sanding and primary cross-dating) 45,46 . Firstly, the ring increments of P. euphratica were measured using the LINTAB TM6 system (precision of 0.001 mm; Rinntech, Heidelberg, Germany). Secondly, cross-dating of increment series data was determined using COFECHA software 47 . Finally, to establish standard and residual chronologies, the growth trend curve for the raw ring increments of P. euphratica was fitted using the negative exponential formula in the ARSTAN software 45,48 . Calculation of the tree-ring width index. In extreme arid lands (e.g., Tarim River Basin), groundwater and tree age are the two most critical factors influencing the tree-ring width of P. euphratica. In order to highlight the effect of groundwater on radial stem growth, we needed to remove the influence of tree age. Therefore, a standardized protocol for radial stem growth data was adopted, i.e., the raw ring growth value (w i ), measured using COFECHA software, was divided by the expected value (y i ) calculated from the growth trend curve by the negative exponential formula in the ARSTAN software. This produced a new sequence for the tree-ring width index (I i ), as shown in formula 2 49    To analyze abrupt change, we first calculated the turning point of the tree-ring width index and groundwater depth using the accumulated anomaly value 41 . Then we checked whether the abrupt changes in the tree-ring width indices either sides of the turning point were salient according to the Mann-Whitney abrupt change test 39,51 . If the statistic (|Z c |) > 1.96, then the abrupt change is significant, otherwise it is not.

Results
The most suitable river flooding disturbance model for P. euphratica seedlings (DBH < 4 cm). The importance value (IV) can show the most suitable habitat for this species 52 . Based on the field trial results, we divided the 467 quadrats into different grades according to the disturbance patterns, calculated the IV value for P. euphratica seedlings (Fig. 3) and identified the most suitable flooding disturbance model for P. euphratica seedling regeneration.
According to Fig. 3A, as the frequency of flooding disturbance increases, the IV value shows a para-curve trend with the highest value appearing in Grade 3 (two to three times per year). When the overflow duration (Fig. 3B) is < 15-20 days (i.e., Grade 3), then the IV value is relatively high and Grade 3 is the highest, but over 20 days, the value drops significantly (P < 0.001). Figure 3C shows that Grade 2 (25-30 m 3 /s) has the highest value and that the value declines significantly when the flood intensity is over 40 m 3 /s (P < 0.001). Therefore, the most suitable flooding disturbance model is two to three times per year with a duration of 15-20 days and an intensity of 25-30 m 3 /s.

The effect of groundwater depth on young P. euphratica trees (DBHs 4-10 cm).
Flooding disturbance (or overflow) is the key factor affecting P. euphratica seedling survival 53,54 . A suitable groundwater depth is another essential environmental factor that ensures that seedlings can grow into young trees 55 . However, trees with different DBHs respond to groundwater depth differently. Therefore, we divided the trees into five growth groups (i.e., seedling, young trees, near-mature trees, mature trees and over-mature trees) according to their DBHs. Based on the tree-ring width and groundwater depth data, we constructed a quantitative model that included tree-ring width index and groundwater depth and determined how tree-ring width index responded to groundwater depth (Fig. 4). Figure 4(1A) shows that the tree-ring width index for young trees declines as the groundwater depth increases (R 2 = 0.62, P < 0.0001), which means that we can use the tree-ring index to deduce the most suitable groundwater depth. In addition, according to the Mann-Kendall monotony trend test (Table S1), when groundwater depth is between 0.6 to 8.8 m, the test statistics for tree-ring width index decrease significantly (Zc is − 13.32 , |Z c | > |Z 0.01 | = 2.58). This shows that P. euphratica growth is very sensitive to variation in groundwater depth. We can also calculate the decrease in tree growth for every 0.1 m increase in groundwater depth. Figure 4(1B) shows that the 4.0 m depth mark divides the curve into two parts: a left part (groundwater depth between 0.6 and 4.0 m) and a right part (groundwater depth between 4.1 and 8.8 m). When the groundwater depth is between 0.6 and 4.0 m, the average tree-ring width index is higher than the total average, which indicates a sharp decrease. When the groundwater depth is between 4.1 and 8.8 m, the average is lower than the total average, which means the tree-ring width index decreases slowly. According to the Mann-Whitney abrupt change test (Table S1), between the two groundwater depth intervals (0.6-4.0 m and 4.1-8.8 m), the mean value for tree-ring width index drops from 0.049 to 0.013, and its abrupt change is significant at the 0.01 test level. (Zc = -7.75, |Z c | > |Z 0.01 | = 2.58). When the groundwater depth is < 4 m, the decrease in P. euphratica radial stem growth is relatively small and the tree-ring width index response to groundwater depth becomes less sensitive, which means that P. euphratica growth is under stress. Thus, the minimum groundwater depth for young trees should be < 4.0 m. The effect of groundwater depth on near-mature P. euphratica trees (DBHs of 10-30 cm).
P. euphratica trees with DBHs of 10-30 cm grew well, with lots of newly bursting branches, and could be described as near-mature trees. In order to efficiently identify the water resource need to protect these near-mature trees, we divided the trees into two grades according to DBH (10-20 cm and 20-30 cm). Figure 4(2A) shows that the tree-ring width index for young trees declines as the groundwater depth increases and this is significantly negatively correlated (P < 0.01). According to the Mann-Kendall monotony trend test (Table S2), the tree-ring width index declines significantly at the 0.01 test level (Zc = − 14.35, |Z c | > |Z 0.01 | = 2.58). Figure 4(2B) shows that the turning point for the decreasing tree-ring width index is 5.0 m. In addition, the Mann-Whitney abrupt change test results (Table S2)   becomes less sensitive. Therefore, the most suitable groundwater depth for near-mature trees (with DBHs 10-20 cm) should be < 5.0 m.
Subsequently, we constructed a response function for tree-ring width index against the variation in groundwater depth based on the data for near-mature tree-rings (trees with DBHs 20-30 cm) and groundwater depth (Figs 4(3A,3B)). Figure 4(3A) shows that there is a significant negative correlation between tree-ring width index for near-mature trees (DBHs 20-30 cm) and groundwater depth (R 2 = 0.62, P < 0.0001). According to the Mann-Kendall monotony trend test (Table S3), increasing groundwater depth leads to a significant decrease in tree-ring width index (Zc = − 13.71, |Z c | > |Z 0.01 | = 2.58). Figure 4(3B) and Table S3 show that the turning point is 5.4 m, and between the two groundwater depth intervals (1.1-5.4 m and 5.5-9.7 m), the mean tree-ring width index value significantly drops (|Z c | = 8.03 > |Z 0.01 | = 2.58) at the 0.01 test level from 0.024 to 0.019. Therefore, we can conclude that the minimum groundwater depth for near-mature trees (DBHs 20-30 cm) is 5.4 m. In brief, the sensitive interval for near-mature tree (DBHs 10-30 cm) response to groundwater depth is between 5.0 to 5.4 m, and the most suitable groundwater depth is < 5.4 m.
The effect of groundwater depth on P. euphratica mature trees (DBHs 30-40 cm). Mature P. euphratica trees are very important components of the P. euphratica community in the Tarim River Basin. Therefore, we studied the growth response of mature trees to different groundwater depths ( Fig. 4(4A,4B) and Table S4).
According to Fig. 4(4A), there is a negative correlation between tree-ring width index for mature trees and groundwater depth (R 2 = 0.51, P < 0.0001). The Mann-Kendall monotony trend test (Table S4) shows that the tree-ring width index holds a significant decrease trend (Zc = − 10.35 > |Z 0.01 | = 2.58). Figure 4(4B) shows that the turning point is 6.9 m. The Mann-Whitney abrupt change test (Table S4) between the two groundwater depth intervals (< 6.9 m and > 6.9 m), shows that the mean value for tree-ring width index significantly declines from 0.038 to 0.017, (test statistic Zc = − 6.12). Based on the results above, we can conclude that the minimum groundwater depth for mature trees (DBHs 30-40 cm) is 6.9 m and the most suitable groundwater depth should be < 6.9 m.
The effect of groundwater depth on P. euphratica over-mature trees (DBHs > 40 cm). In the lower reaches of the Tarim River, P. euphratica over-mature trees are mainly found in areas that are more than 1 km from the river. At this distance, the recharge ability of the river is weak and the groundwater table is low. However, the over-mature trees can resist sandstorms and protect the stability of the natural vegetation ecosystem on both sides of the river. Therefore, we analyzed the correlation between the growth of over-mature trees and groundwater depth (Fig. 4(5A,5B) and Table S5). Figure 4(5A) shows that there is a significant negative correlation between tree-ring width index for over-mature trees and groundwater depth (R 2 = 0.44, P < 0.0001). The Mann-Kendall monotony trend test results (Table S5) show that an increase in groundwater depth leads to a significant decrease in tree-ring width index (|Z c | = 10.24 > |Z 0.01 | = 2.58). Therefore there is a considerable correlation between the two results. Figure 4(5B) shows that the turning point is 7.8 m for the tree-ring width index. The Mann-Whitney abrupt change test results (Table S5) for the two groundwater depth intervals (< 7.8 m and > 7.8 m) show that the mean value for the tree-ring width index significantly drops (test statistic (Zc) is − 6.02) from 0.015 to 0.007. Thus we can conclude that the minimum groundwater depth for over-mature trees (DBHs > 40 cm) is 7.8 m.

Discussion
The regeneration process for P. euphratica seedlings. The intermediate disturbance hypothesis 56,57 is one of the most significant theories in ecology. This theory can be used to guide ecological protection and natural resource planning, and improve the regeneration of P. euphratica forest because river flooding disturbance can activate the soil seed bank and accelerate the regeneration of vegetation 24 .
P. euphratica seeds are smaller than most shrub seeds, and the radicles and plumules after germination are weaker than those of perennial herbs, so they do not compete well for water and have a low water storage capacity. Our results suggest that we should flood the area 2-3 times per year during the seed germination to seedling growth. However, over-frequent flooding is harmful for P. euphratica seedling establishment, and causes the IV value to decline significantly when the frequency of flooding is 4-5 times per year.
Based on the ecological water conveyance data, a minimum flow velocity of 20 m 3 /s can maintain the hydrological process in the lower reaches of the Tarim River, but an intensity below 20 m 3 /s does not lead to overflow. However, when the flow velocity is above 35 m 3 /s, the scouring effect of the overflow on the river bed becomes severe and P. euphratica seedlings are barely able to establish and survive.
Our field research results showed that water-loving plants (such as Phragmites australis and Apluda mutica Linn.) are prevalent in areas where the flooding duration is over 30 days. However, we do not consider that the flora in the lower reaches of the Tarim River represents a climax community, as the ecosystem is constrained by poor water conditions. In contrast, P. euphratica is an important climax species in desert riparian forest, and can be adapted to the arid environment in the Tarim River Basin 58 . In addition, some studies 24,59,60 have shown that long term surface water will raise the groundwater depth and lead to salt accumulation in the surface soil. P. euphratica seedling growth requires a low salt environment, and if the salinity of the soil is high, seedling growth will be suppressed 61 . Therefore, the IV value declines significantly when the surface water duration is over 20 days. Our quantitative results show that the most suitable flood duration is 15-20 days, which is more practicable and achievable when attempting to regenerate P. euphratica forest in downstream areas of the Tarim River.
Minimum groundwater depth for desert riparian P. euphratica forest. The growth of P. euphratica is close correlated with groundwater depth 24 . As the groundwater depth increases, survival and growth rates of P. euphratica decline and the trees become less sensitive to groundwater depth. Previous research 8,62 showed that when the groundwater depth was > 4.5 m, the significance of groundwater to P. euphratica growth becomes less important; the net photosynthetic rate and transpiration rate change, the proline and abscisic acid (ABA) contents become abnormal and the growth of P. euphratica is restricted. When the groundwater depth is > 8 m, the trees begin to seriously decay or even die. According to previous study 39 , the correlation between radial stem growth and groundwater depth shows that the minimum groundwater interval is between 4.71 m and 8.62 m. In reality, our results suggested that the most suitable groundwater depth is < 4 m. The results of this study also indicate that the minimum groundwater range is between 4.0 m and 7.8 m ( i.e., young trees, 4.0 m; near-mature trees, 5.0-5.4 m; mature trees, 6.9 m; over-mature trees, 7-8 m), which is similar to previous research results. However, the most significant result in our study is that the minimum groundwater depths are confirmed for P. euphratica with different DBHs. In the future, the ecological water conveyance, through the use of the EWCE and the ecological water amount, can be defined according to the distribution features of P. euphratica forest and groundwater depth, Water is first transported to P. euphratica forest via deep groundwater. Therefore, this study will provide a theoretical basis and scientific guidance for the rational utilization of water resources and regeneration of P. euphratica forest in arid areas.
Our results show that the decrease in tree-ring width index becomes less sensitive to tree-ring growth as groundwater depth increases. The reason is that when P. euphratica is suffering from water stress, the tree reduces transpiration by secreting ABA to regulate stomatal conductance, which inhibits tree growth 62 .
The best ecological water conveyance levels for P. euphratica protection in the lower reaches of the Tarim River. Based on the best flooding disturbance model (three times a year, duration 20 days and intensity 30 m 3 /s), the calculated ecological water amount needed is 160 million m 3 . The data for ecological water conveyance between 2000 and 2012 in the lower reaches of the Tarim River showed that the annual average water volume was 320 million m 3 . Therefore, it can meet the water requirements of the flooding disturbance model for P. euphratica seedlings.
Our results also show that the groundwater depth should be shallower than the minimum groundwater depth (young trees, 4.0 m; near-mature trees, 5.0-5.4 m, mature trees, 6.9 m and over-mature trees, 7.8 m) to ensure that P. euphratica trees grow well. Figure 5 shows that the groundwater depth in the four transects (Yingsu, Karday, Algan and Yiganbjm) have a strong relationship with the minimum groundwater depth for P. euphratica with different DBHs. The recorded data from the four transects shows that young trees (DBHs 4-10 cm) are mainly found within 50 m of the channel and the majority is located in Algan. The near-mature trees (DBHs 10-30 cm) are mainly found 150 m from the channel and mature trees (DBHs 30-40 cm) and over-mature trees (DBHs > 40 cm) are mainly located 500 m and 1000 m away from the channel. Figure 5 also shows that the actual groundwater depths are higher than the minimum groundwater depth. According to a previous study result 63 , the ecological water requirement for maintaining P. euphratica forest transpiration and minimum groundwater depths is 311 million m 3 . In view of water shortages in this river basin 18,37 , and the requirements of ecological water and river flooding in the lower reaches, the most suitable range of ecological water conveyance is 311-320 million m 3 , which should be sufficient to regenerate and protect P. euphratica in the lower reaches of the Tarim River.

Conclusion
Our results show that.
(1) During the seed to young tree growth period, the most suitable flooding disturbance model for P. euphratica regeneration is 2-3 times per year with a duration of 15-20 days and an intensity of 25-30 m 3 /s. (2) The tree-ring width index for P. euphratica significantly declines as the groundwater depth increases.
(3) The minimum groundwater depths needed for P. euphratica young trees, near-mature trees, mature trees and over-mature trees are 4.0 m, 5.0-5.4 m, 6.9 m and 7.8 m, respectively. In order to protect the desert riparian forest, the actual groundwater depth should be smaller than the minimum groundwater depth for each tree growth stage. (4) The most suitable range for water conveyance amount is 311-320 million m 3 , and this meets the requirements for optimal flooding and ecological water for P. euphratica growth in the lower reaches of Tarim River.