Tree Shelterbelts as an Element to Improve Water Resource Management in Central Asia

In Central Asia, agriculture, notably irrigated agriculture, is the largest water consumer. Currently, flood and furrow irrigation are the dominant irrigation methods in Central Asia, in particular in the post-Soviet countries. Against the background of current and increasing competition for water—e.g., through reduced river runoffs in the course of climate change—water consumption of agriculture needs to be reduced. On the field plot level, improved irrigation technologies, like drip irrigation or plastic mulch, can reduce water consumption substantially. Alternatively, tree lines as wind breaks (shelterbelts) also can reduce crop water consumption, as shown by research from many drylands around the world. As previous research has concentrated on crop water consumption and not on tree water consumption, this paper brings the two together, in order to approach a more holistic picture, in how far shelterbelt systems, including the trees, may have the potential to save water or not. Crop water consumption was assessed through the Penman–Monteith approach for corn, wheat, potato, barley, and pear under open field conditions and under an assumed influence of a tree shelterbelt. Tree water consumption was investigated through sap flow measurements. Crop water consumption was reduced by 10–12% under influence of a shelterbelt compared to open field conditions. When water consumption of shelterbelts was added, a slight reduction of water consumption of the whole crop-shelterbelt system was found for corn, potato, and pear under the assumption 25 ha (500 × 500 m) field sizes. Under an assumption of 4 ha (200 × 200 m) field size, water consumption of the whole crop-shelterbelt system was higher for all crops investigated except for pear. The results suggest that shelterbelts may play a role in improving water resource management in Central Asia in the context of water demanding crops, like corn or cotton. In further research, other effects of shelterbelts, like increased crop yields and additional income from trees, need to be investigated.


Introduction
Central Asia, which extends from the Caspian Sea into Mongolia, is largely dominated by drylands and mountains [1]. Furthermore, Central Asia is the region with the worldwide highest number of endorheic or closed river basins, i.e., rivers that do not drain into the sea, but in an end-lake or inland in Central Asia in the future. To accomplish that objective, we present a holistic approach to water consumption of shelterbelt systems based upon the integrated analyses of shelterbelt tree water consumption data and crop water consumption data.

Study Region
The study was carried out in the village Karasay  During Soviet Union times, Karasay Batyr was a Kolhoz, which was specialized on fruit production, mainly apple and pear. Field plots were of an average size of 250 × 250 m bordered by shelterbelts of poplars (Populus alba and P. nigra) on all four sides. Thus, the agricultural land in Karasay Batyr was structured by a 250 × 250 m grid of shelterbelts.
Today, most of the field plots are grown with grassy vegetation and are cut once or twice a year for hay-making. The field plots, which are under cultivation, are planted with potato or corn. The field plot and shelterbelt structure of today is illustrated in Figure 1.
Water 2017, 9,842 3 of 11 approach to water consumption of shelterbelt systems based upon the integrated analyses of shelterbelt tree water consumption data and crop water consumption data.

Study Region
The study was carried out in the village Karasay  During Soviet Union times, Karasay Batyr was a Kolhoz, which was specialized on fruit production, mainly apple and pear. Field plots were of an average size of 250 × 250 m bordered by shelterbelts of poplars (Populus alba and P. nigra) on all four sides. Thus, the agricultural land in Karasay Batyr was structured by a 250 × 250 m grid of shelterbelts.
Today, most of the field plots are grown with grassy vegetation and are cut once or twice a year for hay-making. The field plots, which are under cultivation, are planted with potato or corn. The field plot and shelterbelt structure of today is illustrated in Figure 1. The climate is continental and semi-arid with a rainy season in spring and hot, dry summers. The closest climate station is Tokmok (Table 1), which is 34 km from Karasy Batyr.  The climate is continental and semi-arid with a rainy season in spring and hot, dry summers. The closest climate station is Tokmok (Table 1), which is 34 km from Karasy Batyr. In the whole Chui Valley, in Kazakhstan and Kyrgyzstan, the major crops today are wheat, potato, corn, fodder plants, and on smaller areas fruit trees and barley (http://www.fao.org/ag/agp/agpc/ doc/counprof/kyrgi.htm). Shelterbelt trees are poplars (Populus alba and P. nigra) in the more rainy and more intensively irrigated parts, while elm (mostly Ulmus minor) and Acacia species dominate on drier and less irrigated sites (own field observations and [17]).

Water Consumption of Shelterbelt System vs. Non Shelterbelt System
The methodological approach of this study was to assess water consumption of major crops of the Chui Valley under open field conditions with no shelterbelts present versus conditions as inside a shelterbelt system. These results are combined with water consumption of the trees of those shelterbelts, in order to obtain the water consumption of crops without shelterbelts versus corresponding crop-shelterbelt systems. Crop water consumption was assessed on the basis of climate data after [18], while tree water consumption was measured through sap flow measurements.
Karasy Batyr was taken as an example to represent the climate, crops, and water consumption conditions of the upper Chui Valley shared by Kyrgyzstan and Kazakhstan. In Karasay Batyr, a shelterbelt was chosen, which still was in good conditions, i.e., dominated by poplars (P. alba) instead of having a mix of tree species, trees of uniform height, not interrupted by gaps between trees, and continuous water supply to the trees ( Figure 2). alba and P. nigra) in the more rainy and more intensively irrigated parts, while elm (mostly Ulmus minor) and Acacia species dominate on drier and less irrigated sites (own field observations and [17]).

Water Consumption of Shelterbelt System vs. Non Shelterbelt System
The methodological approach of this study was to assess water consumption of major crops of the Chui Valley under open field conditions with no shelterbelts present versus conditions as inside a shelterbelt system. These results are combined with water consumption of the trees of those shelterbelts, in order to obtain the water consumption of crops without shelterbelts versus corresponding crop-shelterbelt systems. Crop water consumption was assessed on the basis of climate data after [18], while tree water consumption was measured through sap flow measurements.
Karasy Batyr was taken as an example to represent the climate, crops, and water consumption conditions of the upper Chui Valley shared by Kyrgyzstan and Kazakhstan. In Karasay Batyr, a shelterbelt was chosen, which still was in good conditions, i.e., dominated by poplars (P. alba) instead of having a mix of tree species, trees of uniform height, not interrupted by gaps between trees, and continuous water supply to the trees ( Figure 2). From this shelterbelt, three representative poplars (P. alba) were chosen (Table 2) for the sap flow measurements [19]. Sap flow was measured through the method of heat dissipation using the system PROSALOG by UP GmbH from Germany [20]. Thereby, a pair of sensor needles (2 cm long) was fixed into each tree at 140 cm and 150 cm tree height, respectively. The top needle released heat pulses into the sapwood every ten minutes. After each heat pulse, the temperature difference between both needles was measured. These temperature differences were calculated into sap flow density as follows [19] u = 0.714 ((dT/dTmax) − 1) 1.231 (1) From this shelterbelt, three representative poplars (P. alba) were chosen (Table 2) for the sap flow measurements [19]. Sap flow was measured through the method of heat dissipation using the system PROSALOG by UP GmbH from Germany [20]. Thereby, a pair of sensor needles (2 cm long) was fixed into each tree at 140 cm and 150 cm tree height, respectively. The top needle released heat pulses into the sapwood every ten minutes. After each heat pulse, the temperature difference between both needles was measured. These temperature differences were calculated into sap flow density as follows [19] u = 0.714 ((dT/dTmax) − 1) 1.231 (1) u-sap flow density (mL cm −2 min −1 ) dT-temperature difference between the two needles of one needle pair dTmax-maximal dT, i.e., dT at zero sap flow dTmax was chosen as the average temperature difference at the lowest water vapor pressure deficit over two-week periods. The average temperature difference was calculated from the temperature differences 20 min before to 20 min after the lowest water vapor pressure deficit of a given two-week time periods.
Sap flow density was calculated into sap flow by using the equation The xylem thickness, from which sap wood area was calculated, was determined visually from tree cores. On each tree two perpendicular cores were taken with a tree corer of 5 mm diameter from Suunto, Finland. Sap flow was recorded from 15 June until 3 August 2016. Leafs started to emerge beginning of April and were fully developed by mid-April. Leafs turned yellow during October and fell completely down by 24 October. Therefore, the growing season for this study was defined as 1 April until 24 October.
Reference and crop evapo-transpiration (ETo and ETc) for conditions with and without shelterbelt were calculated after [18] for this growing season. Thereby, ETc was calculated for wheat, corn, potato, barley, and fruit trees (pear).
Outside of the reach of shelterbelts, a climate station was operated during parts of the growing season, in order not to conflict with farm operations. The climate station was equipped with the following sensors, all from Decagon Devices, USA: Sensor for air temperature and humidity (VP-4), radiation, and wind speed (DS-2 Anemometer). Data were logged by an EM-50 data logger. Climate data were recorded every minute and aggregated to daily values. Data gaps were filled by establishing relationships between air temperature, humidity, and wind speed between data measured at Tomok (https://rp5.ru/Wetterarchiv_in_Tokmok) and Karasay Batyr. Radiation for the data gaps was calculated as suggested by [18]. These filled daily climate data finally yielded ETo and ETc under conditions not influenced by any shelterbelt (open field conditions).
ETo and ETc of the above referred crops, under the influence of a shelterbelt, were assessed under the following assumptions: Assumption I: Shelterbelts reduce wind speed by 20% of open field conditions. Assumption II: Shelterbelts reduce wind speed by 35% of open field conditions, decrease mean air temperature by 1 • C, and increase air humidity by 0.05 compared to open field conditions. Thereby, assumption I is a conservative assumption, which reflects the lower boundary of impacts by shelterbelts on wind speed (cf. [9][10][11]13,14]). Assumption II reflects the average of the impacts by shelterbelts on climatic features taken from [9,10,13,14]. To

Results
Climate data as used for the ETo and further ETc calculations are shown in Table 3. ETo and ETc, for corn as an example, in the course of the growing season are shown in Figure 3 for open field conditions and for the two sets of assumption regarding the impact of shelterbelts on climate data.

Results
Climate data as used for the ETo and further ETc calculations are shown in Table 3. ETo and ETc, for corn as an example, in the course of the growing season are shown in Figure 3 for open field conditions and for the two sets of assumption regarding the impact of shelterbelts on climate data. ETo reaches its maximum in June (7-13 June) with over 7 mm/d. During these days, air humidity within a given day ranges between 0.18 and 0.6, which is substantially lower than before and after with values between 0.3 and 0.8. ETc of corn peaks in July with around 9 mm/d, when ETo is high and corn is it its mid development stage.
Regarding ETc over the whole growing season, corn consumes most water (815.1 mm under open field conditions) followed by potato, wheat, and barley (Table 4). Only pears consume more water than corn, 1093 mm over the growing season under open field conditions (Table 4). ETo reaches its maximum in June (7-13 June) with over 7 mm/d. During these days, air humidity within a given day ranges between 0.18 and 0.6, which is substantially lower than before and after with values between 0.3 and 0.8. ETc of corn peaks in July with around 9 mm/d, when ETo is high and corn is it its mid development stage.
Regarding ETc over the whole growing season, corn consumes most water (815.1 mm under open field conditions) followed by potato, wheat, and barley (Table 4). Only pears consume more water than corn, 1093 mm over the growing season under open field conditions (Table 4). The most pronounced reduction of water consumption through the effect of shelterbelts was observed for pear and corn with a reduction by 12% and 6% under the assumption II and assumption I, respectively. As for barley, there is only a reduction of crop water consumption by 8% under assumption II.
Tree water consumption, as measured from the sap flow sensor trees, is highest for sensor tree 1, which has the highest DBH (46.2 cm), followed by sensor tree 3, and 2 ( Figure 4). Sensor tree 2 has the lowest DBH with 19.2 cm (Table 2). Water consumption follows a very similar trend for all trees, despite the huge differences in absolute values of daily water consumption.  The most pronounced reduction of water consumption through the effect of shelterbelts was observed for pear and corn with a reduction by 12% and 6% under the assumption II and assumption I, respectively. As for barley, there is only a reduction of crop water consumption by 8% under assumption II.
Tree water consumption, as measured from the sap flow sensor trees, is highest for sensor tree 1, which has the highest DBH (46.2 cm), followed by sensor tree 3, and 2 ( Figure 4). Sensor tree 2 has the lowest DBH with 19.2 cm (Table 2). Water consumption follows a very similar trend for all trees, despite the huge differences in absolute values of daily water consumption.  Also ETo and water consumption of the shelterbelt, as shown in Figure 5, follow a similar trend, with water consumption of the shelterbelt mostly being higher than ETo. From the water consumption of the shelterbelt and ETo an average Kc value of 1.43 was obtained. This Kc value was used, in order to calculate the water consumption of the shelterbelt for the whole growing season. Water consumption per 100 m shelterbelt was 1297 m 3 for the whole growing season.
Also ETo and water consumption of the shelterbelt, as shown in Figure 5, follow a similar trend, with water consumption of the shelterbelt mostly being higher than ETo. From the water consumption of the shelterbelt and ETo an average Kc value of 1.43 was obtained. This Kc value was used, in order to calculate the water consumption of the shelterbelt for the whole growing season. Water consumption per 100 m shelterbelt was 1297 m 3 for the whole growing season. A square shaped 25 ha field (500 × 500 m) is bordered by 2000 m of shelterbelt, which would consume 25,940 m 3 per growing season. As shelterbelts are shared by neighboring fields, for such a 25 ha field a tree shelterbelt water consumption of 12,970 m 3 was considered for the growing season. Only corn and pear showed a substantial reduction of water consumption, which overcompensated the tree shelterbelts' water consumption ( Table 5), so that a shelterbelt system of corn and pear on 25 ha field plots, respectively, with poplar shelterbelts consumes less water than corn and pear without a shelterbelt system. Table 5. Water consumption of open field conditions (crop without any shelterbelt) and shelterbelt system (crops and trees) for 4 ha and 25 ha field plot, respectively. Negative differences indicate that the whole shelterbelt system consumes less water than the crop without shelterbelt.   A square shaped 25 ha field (500 × 500 m) is bordered by 2000 m of shelterbelt, which would consume 25,940 m 3 per growing season. As shelterbelts are shared by neighboring fields, for such a 25 ha field a tree shelterbelt water consumption of 12,970 m 3 was considered for the growing season. Only corn and pear showed a substantial reduction of water consumption, which overcompensated the tree shelterbelts' water consumption ( Table 5), so that a shelterbelt system of corn and pear on 25 ha field plots, respectively, with poplar shelterbelts consumes less water than corn and pear without a shelterbelt system. A square shaped field of 4 ha (200 × 200 m) is surrounded by 800 m of shelterbelt. 800 m of shelterbelt would consume 10,376 m 3 of water. Analogous to the 25 ha field, a water consumption of 5188 m 3 for the growing season is considered, as shelterbelts are shared by neighboring fields. As for the 4 ha field plot, the shelterbelt system with crops and trees consumes more water than the crops without shelterbelts, except for pears under the set of assumptions II (Table 5).

Discussion
Tree water consumption was in the same range as values published by [21] for Poplars in NW China, ranging from 20.1 to 161.4 kg water per day and per tree. Water consumption of sensor tree no. 1 in this study was higher than values of [21]. This difference can be explained by the bigger DBH and sap wood area of sensor tree no. 1 compared to trees from [21]. ETc values of this study are in the range of other studies, e.g., water consumption of corn with values between 741 mm and 841 mm measured in Texas with a lysimeter under sprinkler irrigation [22]. Wheat ETc calculated from climate data from Almaty [23] was the same as wheat ETc under open field conditions of this study. Annual corn water consumption of 668 mm was found in the Heihe Basin in NW China [24]. This value is lower than ETc of corn under all assumptions in this study, because corn in the Heihe Basin was grown under plastic mulch, which considerably reduced evaporation from the soil surface during the initial crop development stage.
Under the calculations for a hypothetical 4 ha (200 × 200 m) field, all crops-shelterbelt systems (except pear) consumed more water than the respective crop without the shelterbelt (Table 4). Only under calculation for the 25 ha (500 × 500 m) field plot did corn, potato, and pear show a reduction of water consumption when combined with a shelterbelt. This difference is due to the relatively shorter shelterbelt compared to field area for the 25 ha field plot (2000 m shelterbelt) versus 800 m shelterbelt around the 4 ha field plot. On the other hand, field plots must not be too large, as the effect of the shelterbelts on wind speed decreases with distance from the shelterbelts and thus decreases with field plot size.
Reduction of ETc under shelterbelts in this study was about 10% (maximum reduction of ETc was 12% with pear, cf. Table 3). These values are well below the reduction of crop water consumption as suggested by the literature from Soviet Union times [9,10,12,14], but in the range of [11]. This difference might be explained by rather conservative assumptions with respect to the impact of shelterbelts on the micro climate of adjacent field plots. Additionally, some investigations cited here were undertaken in areas that are windier compared to the study site of this paper so that a wind speed reduction has a greater effect on crop water consumption.
With respect to only the water consumption, shelterbelt systems reduce water consumption of corn, potato, and pear, which are crops with a high crop water demand, under the conditions of a square shaped 25 ha field. Under the conditions of a 4 ha field, all crop-shelterbelt systems consume more water (except for pear) than the crop without shelterbelts. So, it can be concluded that shelterbelts around large field plots may play a role to improve water resource management in the context of crops with a high crop water demand, like corn or cotton. In this study, other effects of shelterbelts, like increases of crop yields and additional income from timber and fuel wood, have not been considered. So, if water productivity, which is total income by total water consumption, was compared of the whole crop-shelterbelt system versus crops without shelterbelt, then again a different picture, a more substantial contribution of shelterbelts to improved water resource management, might turn out.

Acknowledgments:
This study was undertaken with support from the Federal Ministry of Economic Cooperation and Development of Germany (BMZ), through the Advisory Service on Agricultural Research for Development (BEAF) under Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ), Germany. BMZ funded a research grant of 100,000 EUR and a six-month internship for Eva Strenge, in order to collect sap flow data. Niels Thevs' position as an integrated expert at the World Agroforestry Center as well as the sap flow equipment and the climate stations were co-funded by BMZ as part of the Center for International Migration and Development (CIM) program.
Author Contributions: Niels Thevs carried out the processing and analysis of climate data for the ETo and ETc calculations together with Maksat Eraaliev. Eva Strange carried out the analysis of sap flow data with supervision from Petra Lang. Kumar Aliev did the necessary tree measurements and tree data analyses. Kumar Aliev, Maksat Eraaliev, and Azim Baibagysov contributed to the literature review, in particular through review of the Russian literature. Jianchu Xu did some editing.

Conflicts of Interest:
The authors declare no conflict of interest.