Hydrological recovery in two large forested watersheds of southeastern China: the importance of watershed properties in determining hydrological responses to reforestation

Abstract

Understanding hydrological responses to reforestation is an important subject in watershed management, particularly in large forested watersheds ( > 1000km²). In this study, we selected two large forested watersheds (Pingjiang and Xiangshui) located in the upper reach of the Poyang Lake watershed, southeastern China (with an area of 3261.4 and 1458km², respectively), along with long-term data on climate and hydrology (1954-2006) to assess the effects of large-scale reforestation on streamflow. Both watersheds have similar climate and experienced comparable and dramatic forest changes during the past decades, but with different watershed properties (e.g., the topography is much steeper in Xiangshui than in Pingjiang), which provides us with a unique opportunity to compare the differences in hydrological recovery in two contrasted watersheds. Streamflow at different percentiles (e.g., 5, 10, 50 and 95%) were compared using a combination of statistical analysis with a year-wise method for each watershed. The results showed that forest recovery had no significant effects on median flows (Q_{50<mspace width="0.125em" linebreak="nobreak"/>%}) in both watersheds. However, reforestation significantly reduced high flows in Pingjiang, but had limited influence in Xiangshui. Similarly, reforestation had significant and positive effects on low flows (Q_{95<mspace linebreak="nobreak" width="0.125em"/>%}) in Pingjiang, while it did not significantly change low flows in Xiangshui. Thus, hydrological recovery is limited and slower in the steeper Xiangshui watershed, highlighting that watershed properties are also important for determining hydrological responses to reforestation. This finding has important implications for designing reforestation and watershed management strategies in the context of hydrological recovery.

Hydrol. Earth Syst. Sci., 20, 47474756, 2016 www.hydrol-earth-syst-sci.net/20/4747/2016/ doi:10.5194/hess-20-4747-2016 Author(s) 2016. CC Attribution 3.0 License.

Wenfei Liu1, Xiaohua Wei2, Qiang Li2, Houbao Fan1, Honglang Duan1, Jianping Wu1, Krysta Giles-Hansen2, and Hao Zhang1

1Institute of Ecology and Environmental Science, Nanchang Institute of Technology, Nanchang, China

2Department of Earth and Environmental Sciences, University of British Columbia (Okanagan campus), 1177 Research Road, Kelowna, British Columbia, V1V 1V7, Canada

Correspondence to: Xiaohua Wei ([email protected])

Received: 29 June 2016 Published in Hydrol. Earth Syst. Sci. Discuss.: 4 July 2016 Revised: 27 September 2016 Accepted: 30 October 2016 Published: 1 December 2016

Abstract. Understanding hydrological responses to reforestation is an important subject in watershed management, particularly in large forested watersheds ( > 1000 km2). In this study, we selected two large forested watersheds (Pingjiang and Xiangshui) located in the upper reach of the Poyang Lake watershed, southeastern China (with an area of 3261.4 and 1458 km2, respectively), along with long-term data on climate and hydrology (19542006) to assess the effects of large-scale reforestation on streamow. Both water-sheds have similar climate and experienced comparable and dramatic forest changes during the past decades, but with different watershed properties (e.g., the topography is much steeper in Xiangshui than in Pingjiang), which provides us with a unique opportunity to compare the differences in hydrological recovery in two contrasted watersheds. Stream-ow at different percentiles (e.g., 5, 10, 50 and 95 %) were compared using a combination of statistical analysis with a year-wise method for each watershed. The results showed that forest recovery had no signicant effects on median ows (Q50%) in both watersheds. However, reforestation signicantly reduced high ows in Pingjiang, but had limited inuence in Xiangshui. Similarly, reforestation had signicant and positive effects on low ows (Q95%) in Pingjiang, while it did not signicantly change low ows in Xiangshui. Thus, hydrological recovery is limited and slower in the steeper Xiangshui watershed, highlighting that watershed properties are also important for determining hydrological responses to

Hydrological recovery in two large forested watersheds of southeastern China: the importance of watershed properties in determining hydrological responses to reforestation

reforestation. This nding has important implications for designing reforestation and watershed management strategies in the context of hydrological recovery.

1 Introduction

Water quantity is of the utmost importance for ecosystem functions, and economic and social development. In forested watersheds, forests play an important role in hydrological processes and their associated ecological functions. Numerous studies have indicated that forest changes (e.g., reforestation or deforestation) can signicantly affect hydrological processes (Jackson et al., 2005; Clinton, 2011; Ford et al., 2011; Iroum and Palacios, 2013; Liu et al., 2015a). However, there are large variations in hydrological responses to forest changes, probably depending on climate and watershed characteristics. Understanding those variations can greatly improve our understanding of the possible mechanisms responsible for hydrological responses and support our management decisions on water and watershed protections.

In large forested watersheds, various factors including climate, land cover, forest changes, and watershed properties can inuence streamow (Anderson and Kneale, 1982). While previous research mainly focused on how climate and forest cover change affect hydrology, limited research has been conducted to examine the role of watershed proper-

Published by Copernicus Publications on behalf of the European Geosciences Union.

4748 W. Liu et al.: Hydrological recovery in two large forested watersheds of southeastern China

ties in hydrological responses. However, watershed properties can be an important factor in determining hydrological responses (Allan, 2004; Poff et al., 2006a, b; Price et al., 2011; Troch et al., 2013; Zhou et al., 2015). For example, Zhang and Wei (2014a) studied two neighboring watersheds (3420 km2) and Willow (2860 km2) in British Columbia, Canada, and found that their contrasted hydrological responses to forest harvesting are mainly related to the difference in their topography and landform complexities. Zhou et al. (2015) also found that watershed characteristics such as watershed slope and size play an important role in hydro-logical responses in their metadata analysis from 168 global studies on large forested watersheds. Clearly, more case studies are needed to assess how watershed properties affect hydrological responses in the context of the other key drivers (e.g., climate and forest changes).

Poyang Lake of Jiangxi Province, which directly ows into Yangtze River, is the largest freshwater lake (3500 km2) in China. It is fed by ve rivers including Gan, Xin, Xiu, Rao and Fu. Poyang Lake provides signicant water resources, wildlife habitats (especially for migratory birds), and economic value (Guo et al., 2008; Huang et al., 2012; Schmalz et al., 2014). However, Poyang Lake Basin experienced severe forest disturbance from the 1960s to the 1980s. Such intense land-use changes resulted in severe environmental degradation. To restore the degraded environment, several ecological restoration and protection programs (e.g., large-scale reforestation) have been implemented since 1980s (Wei et al., 2008). As a result, the forest coverage has increased signicantly in the past few decades. Because Poyang Lake Basin plays a strategic role in environmental protection and economic development in the province as well as in the lower reach of Yangtze River Basin, assessing the ecological effects of those large-scale stewardship programs would be crucial for determining the effectiveness of ecological recovery and for guiding future program design. To our knowledge, several studies had been conducted to assess how large-scale reforestation programs might affect soil erosion and forest carbon processes, but no research has been conducted to assess hydrological recovery under those large-scale stewardship programs.

Two large neighboring watersheds including Pingjiang watershed (2689.20 km2) and Xiangshui watershed (1758 km2), which have similar forest change levels but different watershed properties in the upper reach of the Poyang Lake watershed, were chosen for the study. Hydro-logical variables such as streamow at different percentiles (e.g., high ows and low ows) were examined for each watershed, and their differences were then compared. The objectives of this study were: (1) to assess how stream ows (high and low ows) respond to forest changes at each watershed; (2) to compare their hydrological responses between two different watersheds; and (3) to discuss implications for watershed management.

Figure 1. The location of the Pingjiang and Xiangshui watersheds.

2 Watershed descriptions and data

2.1 Watershed characteristics

The Pingjiang and Xiangshui watersheds feed into the Gan River, the largest tributary of the Poyang Lake watershed (Fig. 1). The drainage areas of the two are 2689 and 1758 km2, respectively. The two watersheds are located in the hilly region of Jiangxi Province, China. The Xiangshui watershed is characterized by a steeper topography than the Pingjiang watershed, with the former having higher slopes covering 23.9 % of the watershed area (from 30 to 50 ) while the latter has only 4.6 % for the same slope class (Table 1).Soils are mountain red soil and yellow-red soil with sandy loam texture in both watersheds. The main characteristics of two watersheds are presented in Table 2.

The two watersheds studied are within the subtropical monsoon zone and have a similar precipitation regime. The average annual precipitations are 1575 and 1611 mm in Piangjiang and Xiangshui watersheds, respectively, of which most falls from April to June (the wet season, about 50 %) and less from September to November (the dry season, about 12 %). The average annual temperatures are 18.9 and

Hydrol. Earth Syst. Sci., 20, 47474756, 2016 www.hydrol-earth-syst-sci.net/20/4747/2016/

W. Liu et al.: Hydrological recovery in two large forested watersheds of southeastern China 4749

010203040

Temperature (o C)

350

(a)

(b)

300

Streamflow and precipitation (mm)

Average monthly streamflow

Average monthly precipitation

Average maximum temperature

Average minimum temperature

Month

250

200

150

100

50

0

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

300

Streamflow and precipitation (mm)

Average monthly streamflow

Average monthly precipitation

Average maximum temperature

Average minimum temperature

Month

Figure 2 Average monthly streamflow, precipitation, minimum temperature and maximum temperature from 1957

to 2006 for the Pingjiang watershed (a) and the Xiangshui watershed (b)

150

010203040

o C

250

200

100

50

0

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

Figure 2. Average monthly streamow, precipitation, minimum temperature and maximum temperature from 1957 to 2006 for the Pingjiang watershed (a) and the Xiangshui watershed (b).

Table 1. Averaged slopes in two watersheds studied (Pingjiang and Xiangshui).

Watershed Percentage of watershed area (%)

Slope 2535 1525 815 38 < 3 > 35

Pingjiang 4.60 52.82 2.40 29.44 6.63 4.11 Xiangshui 23.85 26.99 9.20 33.05 6.17 6.91

2

19.2 C, respectively. The maximum temperature in summer and the minimum temperature in winter are 37 and 0 C, respectively (Fig. 2).

The majority of annual peak ows correspond to rainfall events in two watersheds. In Pingjiang watershed, annual peak ows are between 137 and 870 m3 s1 per 1000 km2, while they are between 108 and 728 m3 s1 per 1000 km2 in Xiangshui watershed. Annual minimum ows range from2.0 to 11.6 m3 s1 per 1000 km2 in Pingjiang watershed and are 0.9 to 11.4 m3 s1 per 1000 km2 in the Xiangshui water-shed. Average annual mean ows are 848 and 858 m3 s1, respectively.

The major land cover types include forest, agriculture, grass, and urban and construction land. Subtropical evergreen broad-leaved forest is the major climax vegetation

type in the watersheds studied, including Castanopsis fabri, Castanopsis sclerophylla, Schima superba, Sassafras tzumu and Castanopsis ssa. In contrast, major plantation forests are Pinus massoniana, Cunninghamia lanceolata, Camellia oleifera Abel and Phyllostachys heterocycla.

2.2 Data

Stream ow data area available from 1957 to 2006 for both watersheds. The hydrometric stations for data collection are part of the Chinese National Hydrometric Network (Fig. 1).Climate data are also available for the same length (1957 2014) for each watershed (ve climate stations for Pingjiang and three for Xiangshui), and include the records of daily maximum, mean, and minimum temperatures and daily precipitation. The averaged watershed-based precipitation estimates were derived by the Thiessen polygon method.

3 Methods

3.1 Leaf area index (LAI) and forest coverage

The Global Land Surface Satellite (GLASS) LAI data were used as the proxy of forest coverage in the watersheds studied. The GLASS LAI product provides the global LAI at the spatial resolution of 0.05 and temporal resolution of 8 days

www.hydrol-earth-syst-sci.net/20/4747/2016/ Hydrol. Earth Syst. Sci., 20, 47474756, 2016

4750 W. Liu et al.: Hydrological recovery in two large forested watersheds of southeastern China

Table 2. A summary of watershed characteristics for the Pingjiang and Xiangshui watersheds.

Metrics Pingjiang Xiangshui

Drainage area (km2) 2689.20 1758Average elevation (m) 298 429Soil type Mountain red soil and yellow-red soil Mountain red soil and yellow-red soil Annual mean precipitation (mm) 1575 1611Annual mean temperature ( C) 18.9 19.2

Annual mean ET (mm) 879.2 936.8 Annual mean ow (mm) 848 858 Runoff coefcient 0.54 0.53 Maximum ow (m3 s1) 1530 1280

Minimum ow (m3 s1) 5.5 2.3 BioGeoClimatic zone Subtropic monsoon Subtropic monsoon

Forest type Subtropical evergreen broadleaf forest and conifer forest Subtropical evergreen broadleaf forest and conifer forest Dominant disturbance type Logging LoggingHydrometric station Hanlinqiao Mazhou

for the period of 1981 to 2014 (http://www.bnu-datacenter.com/

Web End =http://www.bnu-datacenter. http://www.bnu-datacenter.com/

Web End =com/ ). The GLASS LAI data have been validated through the eld measurements to ensure data quality for long-term studies in vegetation changes (Liang and Xiao, 2012; Xiao et al., 2014). The growing season LAI values were based on the LAI values from April to October for each year. The watershed-based LAI values were derived by averaging the LAI data for the pixels where more than 50 % of their pixel areas falls inside the watershed boundaries.

Forest change is the main type of land-use change in the watersheds we studied. Because the complete records of annual deforestation and reforestation areas are unavailable, forest coverage and LAI data were used to indicate historic forest changes during the study period (19572006). As shown in Fig. 3, forest cover was greatly reduced in the period 19651984 due to large-scale forest disturbance (e.g., deforestation). Since then, forest cover was significantly increased from about 30 % in the 1980s to 70 % in 2006 in both watersheds due to implementation of the reforestation projects (19902006) (Fig. 3). Thus, the entire study period was divided into the forest disturbance period (19571985) and the forest recovery period (1990 2006).

3.2 Median, high, and low ows

In this study, FDCs (ow-duration curves) were applied to dene high, median, and low ows. FDCs represent the percent of time streamow for any given value exceeded or equaled in a period of record (Vogel and Fennessey, 1994). In this study, median ows are dened as the ows that exceed or are equal to Q50%. High ows are dened as the ows that exceed or are equal to Q5% and Q10% (Q5%: ows exceeding 5 % of the time in a given year and Q10%: ows exceeding 10 % of the time in a given year), while low ows are dened as the ows that are equal to or less than Q95%

(Q95%: ows exceeding 95 % of the time in a given year) (Zhang and Wei, 2014b; Liu et al., 2015b).

3

80

Year vs. Pingjiang watershed Year vs. Xiangshui watershed

(a)

70

60

Forest coverage %

50

40

30

20

1940 1950 1960 1970 1980 1990 2000 2010

5

(b)

4

3

2

LAI

1

0

-1

-2

1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

Figure 3. Forest cover (%) (a) and Leaf Area Index (LAI) (b) from 1982 to 2006 in the Pingjiang and Xiangshui watershed.

Figure 3 Forest cover (%) (a) and Leaf Area Index (LAI) (b) from 1982 to 2006 in the Pingjiang and Xiangshui watersheda

Hydrol. Earth Syst. Sci., 20, 47474756, 2016 www.hydrol-earth-syst-sci.net/20/4747/2016/

W. Liu et al.: Hydrological recovery in two large forested watersheds of southeastern China 4751

In order to assess the impacts of forest changes on high, median, and low ows, the effect of climate variability must be eliminated. For a single watershed, pair-wise comparisons can be used to address this issue (Levy, 1975; Broomell et al., 2011; Zhang and Wei, 2014b; Liu et al., 2015b; Eastwood et al., 2016). Because high ows are mainly caused by rainfall events, we can nd some similar and comparable rainfall events between the reforestation and deforestation periods with similar R5% and R10%, respectively (R5%: rainfall exceeding 5 % of the time in a given year and R10%: rainfall exceeding 10 % of the time in a given year). However, low and median ows are signicantly correlated with annual rainfall, annual maximum temperature, and annual mean temperature (Tables 3 and 4). Therefore, paired years between the reforestation and deforestation periods were selected for analysis of low and median ows (Tables S1 and S2 in the Supplement). More details about this method can be found in Zhang and Wei (2014a) and Liu et al. (2015b).

3.3 Estimation of recession constants

Recession constant is a useful indicator reecting the characteristics of the study basin (Barnes, 1939; Ge et al., 2014).For a watershed, the difference in recession constants of streamow with similar climate conditions between different periods can be ascribed to the effect of land-cover change, while the difference in recession constants of streamow between the two watersheds studied, under similar climate conditions, can be ascribed to the effect of different water properties on streamow.

In this paper, the classical recession curve based on a genetic algorithm (GA) was adopted to study and analyze the daily runoff (Eqs. 1 and 2).

Qt = Q0e t (1) = (lnQ0 lnQt) (2)

Here, Q0 is the initial discharge (t = 0), Qt is the discharge at

a later time (usually in days), and is the recession constant.

The paired-wise approach was also used to assess the effects of forest changes on recession constants. Because high ows are mainly caused by rainfall events (e.g., storm events) in the study area, we can select similar and comparable rainfall events between the reforestation and the disturbance periods (Table S3 in the Supplement).

4 Results

4.1 High ows response to forest changes

As shown in Fig. 4a, the average magnitude of high ows (Q5%) in the reforestation period (327.7 m3 s1) was significantly lower (p < 0.01) than that in the deforestation period (534.9 m3 s1) in the Pingjiang watershed. Similarly, the average magnitude of high ows (Q10%) in the reforestation

4

7.5

7.5

(a)

(b)

7.0

7.0

6.5

6.5

6.0

6.0

Ln Q

Ln Q

5.5

5.5

5.0

5.0

4.5

4.5

4.0

4.0

6.0

(c)

(d)

5.5

5.5

5.0

5.0

Ln Q

Ln Q

4.5

4.5

4.0

4.0

3.5

3.5

(e)

(f)

8

8

7

7

6

6

Ln Q

Ln Q

5

5

4

4

3

3

2

2

The deforestation period The reforestation period The deforestation period The reforestation period

Figure 4. High ows and median ows for the selected pairs in the deforestation and reforestation periods: (a) high ows (Q5%)

for the Pingjiang watershed; (b) high ows (Q5%) for the Xiangshui watershed; (c) high ows (Q10%) for the Pingjiang water-shed; (d) high ows (Q10%) for the Xiangshui watershed; (e) Median ows (Q50%) for the Pingjiang watershed; and (f) Median ows (Q50%) for the Xiangshui watershed.

period (164.4 m3 s1) was also signicantly lower (p < 0.01) than that in the deforestation period (198.7 m3 s1) in the

Pingjiang watershed (Fig. 4c).

For the Xiangshui watershed, the average magnitude of high ows (Q5%) in the reforestation period (233.0 m3 s1)

was lower than that in the deforestation period (251.4 m3 s1) (Fig. 4b), but their difference was not statistically signicant (p = 0.46). The average magnitude of high ows

(Q10%) in the reforestation period (118.0 m3 s1) was significantly lower (p < 0.05) than that in the deforestation period (127.9 m3 s1) (Fig. 4d). Thus, reforestation signicantly decreased high ows in the Pingjiang watershed, while such an effect is relatively limited in the Xiangshui watershed.

4.2 Median ows response to forest changes

As shown in Fig. 4e and f, the averaged magnitudes of median ows in the reforestation period (43.1 and 41.5 m3 s1, respectively) showed no signicant difference (p = 0.21

www.hydrol-earth-syst-sci.net/20/4747/2016/ Hydrol. Earth Syst. Sci., 20, 47474756, 2016

Figure 4 High flows and median flows for the selected pairs in the deforestation and reforestation periods: (a) high flows (Q5%) for the Pingjiang watershed; (b) high flows (Q5%) for the Xiangshui watershed; (c) high flows (Q10%) for the Pingjiang watershed; (d) high flows (Q10%) for the Xiangshui watershed; (e) Median flows (Q50%) for the Pingjiang watershed; and (f) Median flows (Q50%) for the Xiangshui watershed

4752 W. Liu et al.: Hydrological recovery in two large forested watersheds of southeastern China

Table3.CorrelationanalysesbetweenlowowsandclimaticvariablesinthePingjiangandXiangshuiwatersheds.

Table 4. Canonical correlation analyses between hydrological variables (median and low ows) and climatic variables in the Pingjiang and Xiangshui watersheds.

Watersheds Canonical correlation Canonical R Signicant analysis

Pingjiang Precipitation, 0.88 p < 0.01 Xiangshui Tave and Tmax 0.89 p < 0.01

and 0.27, respectively) to those in the deforestation period in Pingjiang and Xiangshui watersheds (40.3 and 38.4 m3 s1, respectively), indicating that reforestation had no signicant effects on median ows (Q50%) in both watersheds.

4.3 Low ows response to forest changes

As shown in Fig. 5a, the average magnitude of low ows in the reforestation period (12.3 m3 s1) was signicantly higher (p < 0.01) than that in the deforestation period(8.7 m3 s1) in the Pingjiang watershed. In contrast, the average magnitude of low ows in the deforestation period did not signicantly differ from that in the reforestation period (Fig. 5b) in the Xiangshui watershed. Thus, reforestation signicantly increased low ows in the Pingxiang watershed but not in the Xiangshui watershed.

4.4 Responses of recession constants to forest changes

As shown in Fig. 5c and d, the averaged recession constant of streamow in the reforestation period was signicantly lower (p = 0.049) than that in the deforestation period in the

Pingjiang watershed, while the difference was not signicant (p = 0.52) in the Xiangshui watershed, suggesting that hy

drological responses to reforestation is more sensitive in the Pingjiang watershed than in the Xiangshui watershed.

5 Discussion

Although the effects of reforestation on peak ows are still controversial (Gafur et al., 2003; Nadal-Romero et al., 2016;Liu et al., 2015a), a general conclusion is that increased forest coverage through reforestation can reduce high ows (Llorens et al., 1997; Gebrehiwot et al., 2010; Nadal-Romero et al., 2016). Our study found that reforestation can signicantly decrease high ows, which can thereby reduce ood risks. Thus, our results are consistent with the general results found in other regions (e.g., Gafur et al., 2003; Bahremand et al., 2007; Tran et al., 2010). Our results are also supported by another study in a neighboring watershed (Meijiang) of the same region (Liu et al., 2015b) where the historic forest change is similar to those in our study. A common reason for reduced high ows after reforestation is that reforestation increases forest coverage and slowly improves soil conditions,

Hydrol. Earth Syst. Sci., 20, 47474756, 2016 www.hydrol-earth-syst-sci.net/20/4747/2016/

Tave, Tmaxand Tmin refer to annual mean, maximum, and minimum temperatures, respectively. Statistical difference at p < 0.05. Statistical difference at p < 0.01.

Xiangshui0.57 0.39 0.44 0.61 0.04 0.04 0.23 0.05 0.07 0.11

Pingjiang0.48 0.68 0.34 0.48 0.14 0.18 0.21 0.32 0.12 0.21

MannaKendallSpearmanMannaKendallSpearmanMannaKendallSpearmanMannaKendallSpearmanManna-KendallSpearman

WatershedsPrecipitationTmax Tmin Tave Wind speed

W. Liu et al.: Hydrological recovery in two large forested watersheds of southeastern China 4753

forestation projects (Gao et al., 2015). The results from a paired watershed experiment in South Africa showed that low ows were reduced by half due to reforestation (Smith and Scott, 1992). A study analyzing the responses of stream-ow to forest plantation expansion in six large river water-sheds (from 94 to 1545 km2) of central-southern Chile indicated that reforestation had less effect on low water ows (Q80% to Q90%) in relatively drier soils (Iroum and Palacios, 2013). However, in humid regions, increases in vegetation cover often lead to greater inltration of rainfall into the soil, and as a result, increase water storages and low ows (Zhou et al., 2010). More case studies are needed before a general comparison between reforestation and low ows can be developed.

Although reforestation generally played a positive role in streamow in our study area, there are large differences in the hydrological responses between the two study watersheds.As shown above, there are more signicant effects on both high and low ows in the Pingjiang watershed than in the Xiangshui watershed. Since both watersheds have experienced similar historic forest change and climate, we believe that the difference in the responses of high and low ows were mainly due to the difference in their watershed properties. A close examination of their watershed properties shows that the main differences in their properties are to do with water-shed slopes and sizes. Many studies show that watershed size can be an important factor affecting hydrological responses to land-cover changes (Buttle and Metcalfe, 2000; Blschl et al., 2007; Zhang and Wei, 2014a; Zhou et al., 2015). A smaller-sized watershed often has less buffering capacity as it may contain fewer heterogeneous landscape components (e.g., wetlands, lakes) and complexities, and as a result, is more sensitive to land-cover changes. In our study, Xiangshui watershed is much smaller than Pingjiang watershed, so a quicker hydrological response should be expected in Xiangshui watershed. The limited and slower hydrological response in Xiangshui watershed after reforestation as compared with Pingjiang watershed suggests that a factor other than watershed size came into play. Thus, we reasonably judge that the difference in watershed slope between two watersheds is the major factor determining the variations of their hydrological responses. The Xiangshui watershed has a much larger area percentage (23.9 %) with the slope class (3050 %) as compared to that (4.6 %) in the Pingjiang watershed (Table 1). In southern China where a monsoon climate is dominant, a steeper watershed often has more severe soil erosion if deforestation occurs, and consequently it would take a much longer time to recover through the reforestation process once severe soil erosion occurred (Chen et al., 2002; Zheng et al., 2015).

The importance of watershed properties in hydrological responses to land-cover or forest changes is gradually being recognized in scientic communities. This is particularly relevant for larger watersheds (e.g., > 1000 km2) where there are more landforms (e.g., wetlands, lakes), more land

www.hydrol-earth-syst-sci.net/20/4747/2016/ Hydrol. Earth Syst. Sci., 20, 47474756, 2016

30

30

(a)

(b)

25

25

20

20

1

1

ms

3

15

15

10

10

5

5

0

0

(c)

(d)

.8

.8

ms

.6

.6

Recession constant

Recession constant

.4

.4

.2

.2

0.0

0.0

The deforestation period The reforestation period The deforestation period The reforestation period

Figure 5. Low flows and recession constants of streamflow for the selected pairs in the deforestation and

reforestation periods: (a) low flows for the Pingjiang watershed; (b) low flows for the Xiangshui watershed; (c)

recession constants for the Pingjiang watershed; and (d) recession constants for the Xiangshui watershed

Figure 5. Low ows and recession constants of streamow for the selected pairs in the deforestation and reforestation periods: (a) low ows for the Pingjiang watershed; (b) low ows for the Xiangshui watershed; (c) recession constants for the Pingjiang watershed; and(d) recession constants for the Xiangshui watershed.

which consequently enhances soil inltration capacity and reduces high ows.

Our study showed that reforestation signicantly increased low ows in the Pingjiang watershed. Although not statistically signicant (p = 0.084), the low ows after reforestation

in the Xiangshui watershed were also improved (Fig. 5b). Thus, reforestation had a positive role in low ows in the study watersheds. Our results are consistent with various reforestation studies, particularly in higher humidity environments (Buttle, 2011; Yao et al., 2012; Liu et al., 2015b). For example, Zhou et al. (2010) studied the effects of large-scale reforestation on hydrology in the whole Guangdong province, and found that an increase of 30 % forest cover played a positive role in redistributing water from the wet season to the dry season and, consequently, in increasing water yield in the dry season. The main reason for enhancing low ows from reforestation is that reforestation improves vegetation and soil conditions, and consequently improves soil inltration and groundwater recharging, which have positive effects on low ows.

The responses of low ows to reforestation are inconsistent across different climate regimes. Lu et al. (2016) rst estimated the effects of reforestation on groundwater resources using seven evapotranspiration models and suggested that Chinas unprecedented reforestation program would result in greatly decreased depth of groundwater in the arid and semi-arid areas of northern China. A similar study conducted in the Loess Plateau of China also found a statistically signi-cant (p < 0.1) reduction of 0.03 mm of groundwater per year from 1955 to 2010 due to implementation of large-scale re-

5

4754 W. Liu et al.: Hydrological recovery in two large forested watersheds of southeastern China

cover types, and thus more interactions and complexities between various watershed properties. Several studies on forest changes and hydrology in large forested watersheds in British Columbia, Canada, conclude that the effects of forest changes on water are likely watershed-specic (Lin and Wei, 2008; Zhang and Wei, 2014b), which clearly demonstrates the importance of watershed properties in determining the relationship between forest changes and water. However, assessing how watershed properties affect hydrological responses among other key drivers, such as forest change and climate, is a challenging subject. Some studies have applied integrative indicators such as topographic indices (Woods et al., 1997; Hjerdt et al., 2004; Liu et al., 2012) or ow paths and transit time (McGuire and McDonnell, 2006; Soulsby et al., 2009) to assess watershed behaviors or functions, while other studies used a landscape approach (Poff et al., 2006a, b; Price et al., 2011). Nevertheless, more case studies are needed in this direction.

Our results from this study have important management implications. The Pingjiang and Xiangshui watersheds are very important headwater systems to Poyang Lake, the largest freshwater lake in China, which is crucial to sustaining aquatic ecological functions (Guo et al., 2008). Many studies had demonstrated alteration of ow regimes (especially for low and high ows) may be one of the most serious and ongoing threats to the integrity of river ecosystems (Ward et al., 1999; Bunn and Arthington, 2002; Poff and Zimmerman, 2010; Liu et al., 2015b). Therefore, it is highly important to manage ow regimes for sustainable watershed ecosystems in Poyang Lake Basin.

Our results demonstrate a positive effect of reforestation on high and low ows in both Pingjiang and Xiangshui watersheds. This conrms that our reforestation programs implemented over the last decades provide important benets to restoration of watershed functions in terms of hydrology.More importantly, our study found that hydrological recovery of a steeper watershed likely takes much longer time once it is deforested or damaged, suggesting that we must take extra care when we design management strategies in more sensitive watersheds.

6 Conclusion

We found that reforestation decreased high ows, but increased low ows in the watersheds we studied, which is benecial to the maintenance of aquatic functions and water supply. We also found that there are large variations in hydro-logical responses to similar reforestation levels, likely due to the difference in watershed properties (e.g., watershed slope).Thus, we conclude that hydrological recovery through reforestation is largely dependant on watershed properties when forest change and climate are similar and comparable.

7 Data availability

The climatic and hydrological data used for this paper are not publicly available due to the constraints of governmental policy in China. The data were obtained through a purchasing agreement for this study.

The Supplement related to this article is available online at http://dx.doi.org/10.5194/hess-20-4747-2016-supplement

Web End =doi:10.5194/hess-20-4747-2016-supplement .

Acknowledgements. Funding was provided by Jiangxi Education Department (No. GJJ151141), the National Science Foundation of China (No. 31170665 and No. 31660234), Jiangxi Education Department (KJLD12097 and KJLD14095), Gan-Po 555 Talent Project, and Funding of Jiangxi Province, Scientic Funding by Jiangxi Province (No. 20142BAB214006 and 20161BBH80049).

Edited by: L. WangReviewed by: two anonymous referees

References

Allan, J. D.: Landscapes and riverscapes: the inuence of land use on stream ecosystems, Annu. Rev. Ecol. Evol. Syst., 35, 257 284, 2004.

Anderson, M. G. and Kneale, P. E.: The inuence of low-angled topography on hillslope soil-water convergence and stream discharge, J. Hydrol., 57, 6580, 1982.

Bahremand, A., De Smedt, F., Corluy, J., Liu, Y., Poorova, J., Velcicka, L., and Kunikova, E.: WetSpa model application for assessing reforestation impacts on oods in MargecanyHornad water-shed, Slovakia, Water Resour. Manage., 21, 1373-1391, 2007.

Barnes, B. S.: The structure of discharge-recession curves, Eos T. Am. Geophys. Un., 20, 721725, doi:http://dx.doi.org/10.1029/TR020i004p00721

Web End =10.1029/TR020i004p00721 http://dx.doi.org/10.1029/TR020i004p00721

Web End = , 1939.

Blschl, G., Ardoin-Bardin, S., Bonell, M., Dorninger, M., Goodrich, D., Gutknecht, D., Matamoros, D., Merz, B., Shand,P., and Szolgay, J.: At what scales do climate variability and land cover change impact on ooding and low ows?, Hydrol. Process., 21, 12411247, 2007.

Broomell, S. B., David, V., and Por, H.-H.: Pair-wise comparisons of multiple models, Judgment Decis. Mak., 6, 821831, 2011.Bunn, S. E. and Arthington, A. H.: Basic principles and ecological consequences of altered ow regimes for aquatic biodiversity, Environ. Manage., 30, 492507, 2002.

Buttle, J.: Streamow response to headwater reforestation in the Ganaraska River basin, southern Ontario, Canada, Hydrol. Process., 25, 30303041, 2011.

Buttle, J. and Metcalfe, R.: Boreal forest disturbance and stream-ow response, northeastern Ontario, Can. J. Fish. Aquat. Sci., 57, 518, 2000.

Chen, S., Hua, L., He, Z., Wei, D., Xiahou, G., and Li, Y.: Effect of soil erosion on soil properties in deep cultivated hill slope in the Loess Plateau, Agro-Environ. Protect., 21, 289292, 2002.

Hydrol. Earth Syst. Sci., 20, 47474756, 2016 www.hydrol-earth-syst-sci.net/20/4747/2016/

W. Liu et al.: Hydrological recovery in two large forested watersheds of southeastern China 4755

Clinton, B. D.: Stream water responses to timber harvest: Riparian buffer width effectiveness, Forest Ecol. Manage., 261, 979988, 2011.

Eastwood, A., Brooker, R., Irvine, R., Artz, R., Norton, L., Bullock,J., Ross, L., Fielding, D., Ramsay, S., and Roberts, J.: Does nature conservation enhance ecosystem services delivery?, Ecosyst.Serv., 17, 152162, 2016.

Ford, C. R., Laseter, S. H., Swank, W. T., and Vose, J. M.: Can forest management be used to sustain water-based ecosystem services in the face of climate change?, Ecol. Appl., 21, 20492067, 2011.Gafur, A., Jensen, J. R., Borggaard, O. K., and Petersen, L.: Runoff and losses of soil and nutrients from small watersheds under shifting cultivation (Jhum) in the Chittagong Hill Tracts of Bangladesh, J. Hydrol., 274, 3046, 2003.

Gao, Z., Zhang, L., Cheng, L., Zhang, X., Cowan, T., Cai, W., and Brutsaert, W.: Groundwater storage trends in the Loess Plateau of China estimated from streamow records, J. Hydrol., 530, 281 290, 2015.

Ge, Y., Jing, T., Liang, C., Wang, X., and Dong, X.: Study on Correlation between Recession Coefcients and Catchment Characteristics in Pearl River, J. China Hydrol., 34, 7277, 2014.Gebrehiwot, S. G., Taye, A., and Bishop, K.: Forest cover and stream ow in a headwater of the Blue Nile: complementing observational data analysis with community perception, Ambio, 39, 284294, 2010.

Guo, H., Hu, Q., and Jiang, T.: Annual and seasonal streamow responses to climate and land-cover changes in the Poyang Lake basin, China, J. Hydrol., 355, 106122, 2008.

Hjerdt, K., McDonnell, J., Seibert, J., and Rodhe, A.: A new topo-graphic index to quantify downslope controls on local drainage, Water Resour. Res., 40, W05602, doi:http://dx.doi.org/10.1029/2004WR003130

Web End =10.1029/2004WR003130 http://dx.doi.org/10.1029/2004WR003130

Web End = , 2004.

Huang, L., Shao, Q., and Liu, J.: Forest restoration to achieve both ecological and economic progress, Poyang Lake basin, China, Ecol. Eng., 44, 5360, 2012.

Iroum, A. and Palacios, H.: Afforestation and changes in forest composition affect runoff in large river basins with pluvial regime and Mediterranean climate, Chile, J. Hydrol., 505, 113 125, 2013.

Jackson, R. B., Jobbgy, E. G., Avissar, R., Roy, S. B., Barrett, D. J., Cook, C. W., Farley, K. A., Le Maitre, D. C., McCarl, B. A., and Murray, B. C.: Trading water for carbon with biological carbon sequestration, Science, 310, 19441947, 2005.

Levy, K. J.: Large-sample pair-wise comparisons involving correlations, proportions, or variances, Psychol. Bull., 82, 177179, 1975.

Liang, S. and Xiao, Z.: Global land surface products: Leaf area index product data collection (19852010), Beijing Normal University, Beijing, 2012.

Lin, Y. and Wei, X.: The impact of large-scale forest harvesting on hydrology in the Willow watershed of Central British Columbia,J. Hydrol., 359, 141149, 2008.

Liu, J., Chen, X., Wu, J., Zhang, X., Feng, D., and Xu, C.-Y.: Grid parameterization of a conceptual distributed hydrological model through integration of a sub-grid topographic index: necessity and practicability, Hydrolog. Sci. J., 57, 282297, 2012.

Liu, W., Wei, X., Liu, S., Liu, Y., Fan, H., Zhang, M., Yin, J., and Zhan, M.: How do climate and forest changes affect long-erm

streamow dynamics? A case study in the upper reach of Poyang River basin, Ecohydrology, 8, 4657, 2015a.

Liu, W., Wei, X., Fan, H., Guo, X., Liu, Y., Zhang, M., and Li, Q.: Response of ow regimes to deforestation and reforestation in a rain-dominated large watershed of subtropical China, Hydrol. Process., 29, 50035015, 2015b.

Llorens, P., Poch, R., Latron, J., and Gallart, F.: Rainfall interception by a Pinus sylvestris forest patch overgrown in a Mediterranean mountainous abandoned area I. Monitoring design and results down to the event scale, J. Hydrol., 199, 331345, 1997.

Lu, C., Zhao, T., Shi, X., and Cao, S.: Ecological restoration by afforestation may increase groundwater depth and create potentially large ecological and water opportunity costs in arid and semiarid China, J. Clean. Product., doi:http://dx.doi.org/10.1016/j.jclepro2016.03.046

Web End =10.1016/j.jclepro2016.03.046 http://dx.doi.org/10.1016/j.jclepro2016.03.046

Web End = , online rst, 2016.

McGuire, K. J. and McDonnell, J. J.: A review and evaluation of catchment transit time modeling, J. Hydrol., 330, 543563, 2006. Nadal-Romero, E., Cammeraat, E., Serrano-Muela, M. P., Lana-

Renault, N., and Regs, D.: Hydrological response of an af-forested catchment in a Mediterranean humid mountain area: a comparative study with a natural forest, Hydrol. Process., 30, 27172733, doi:http://dx.doi.org/10.1002/hyp.10820

Web End =10.1002/hyp.10820 http://dx.doi.org/10.1002/hyp.10820

Web End = , 2016.

Poff, N. L. and Zimmerman, J. K.: Ecological responses to altered ow regimes: a literature review to inform the science and management of environmental ows, Freshwater Biol., 55, 194205, 2010.

Poff, N. L., Olden, J. D., Pepin, D. M., and Bledsoe, B. P.: Placing global stream ow variability in geographic and geomorphic contexts, River Res. Appl., 22, 149166, 2006a.

Poff, N. L., Bledsoe, B. P., and Cuhaciyan, C. O.: Hydrologic variation with land use across the contiguous United States: geomorphic and ecological consequences for stream ecosystems, Geo-morphology, 79, 264285, 2006b.

Price, K., Jackson, C. R., Parker, A. J., Reitan, T., Dowd, J., and Cyterski, M.: Effects of watershed land use and geo-morphology on stream low ows during severe drought conditions in the southern Blue Ridge Mountains, Georgia and North Carolina, United States, Water Resour. Res., 47, W02516, doi:http://dx.doi.org/10.1029/2010WR009340

Web End =10.1029/2010WR009340 http://dx.doi.org/10.1029/2010WR009340

Web End = , 2011.

Schmalz, B., Kuemmerlen, M., Kiesel, J., Cai, Q., Jhnig, S., and Fohrer, N.: Impacts of land use changes on hydrological components and macroinvertebrate distributions in the Poyang lake area, Ecohydrology, 8, 11191136, 2014.

Smith, R. and Scott, D.: The effects of afforestation on low ows in various regions of South Africa, Water SA, 18, 185194, 1992. Soulsby, C., Tetzlaff, D., and Hrachowitz, M.: Tracers and transit times: windows for viewing catchment scale storage?, Hydrol. Process., 23, 35033507, 2009.

Tran, P., Marincioni, F., and Shaw, R.: Catastrophic ood and forest cover change in the Huong river basin, central Viet Nam: A gap between common perceptions and facts, J. Environ. Manage., 91, 21862200, 2010.

Troch, P. A., Carrillo, G., Sivapalan, M., Wagener, T., and Sawicz, K.: Climatevegetationsoil interactions and long-term hydrologic partitioning: signatures of catchment co-evolution, Hydrol. Earth Syst. Sci., 17, 22092217, doi:http://dx.doi.org/10.5194/hess-17-2209-2013

Web End =10.5194/hess-17-2209- http://dx.doi.org/10.5194/hess-17-2209-2013

Web End =2013 , 2013.

www.hydrol-earth-syst-sci.net/20/4747/2016/ Hydrol. Earth Syst. Sci., 20, 47474756, 2016

4756 W. Liu et al.: Hydrological recovery in two large forested watersheds of southeastern China

Vogel, R. M. and Fennessey, N. M.: Flow-duration curves. I: New interpretation and condence intervals, J. Water Resour. Pl. Manage., 120, 485504, 1994.

Ward, J., Tockner, K., and Schiemer, F.: Biodiversity of oodplain river ecosystems: ecotones and connectivity, Regul. Rivers Res.Manage., 15, 125139, 1999.

Wei, X., Sun, G., Liu, S., Jiang, H., Zhou, G., and Dai, L.: The

ForestStreamow Relationship in China: A 40-Year Retrospect1, J. Am. Water Resour. Assoc., 44, 10761085, 2008.Woods, R. A., Sivapalan, M., and Robinson, J. S.: Modeling the spatial variability of subsurface runoff using a topographic index, Water Resour. Res., 33, 10611073, 1997.

Xiao, Z., Liang, S., Wang, J., Chen, P., Yin, X., Zhang, L., and Song,J.: Use of general regression neural networks for generating the GLASS leaf area index product from time-series MODIS surface reectance, IEEE T. Geosci. Remote, 52, 209223, 2014.

Yao, Y., Cai, T., Wei, X., Zhang, M., and Ju, C.: Effect of forest recovery on summer streamow in small forested watersheds, Northeastern China, Hydrol. Process., 26, 12081214, 2012.

Zhang, M. and Wei, X.: Contrasted hydrological responses to forest harvesting in two large neighbouring watersheds in snow hydrology dominant environment: implications for forest management and future forest hydrology studies, Hydrol. Process., 28, 6183 6195, 2014a.

Zhang, M. and Wei, X.: Alteration of ow regimes caused by large-scale forest disturbance: a case study from a large watershed in the interior of British Columbia, Canada, Ecohydrology, 7, 544 556, 2014b.

Zheng, L., Zhou, Y., Zhou, X., Li, D., Wu, Z., and Zhou, C.: Recovery Dynamics after Cutting of Forest Ecosystem Services in Mountain, Issues Forest. Econ., 35, 16, 2015.

Zhou, G., Wei, X., Luo, Y., Zhang, M., Li, Y., Qiao, Y., Liu, H., and Wang, C.: Forest recovery and river discharge at the regional scale of Guangdong Province, China, Water Resour. Res., 46, W09503, doi:http://dx.doi.org/10.1029/2009WR008829

Web End =10.1029/2009WR008829 http://dx.doi.org/10.1029/2009WR008829

Web End = , 2010.

Zhou, G., Wei, X., Chen, X., Zhou, P., Liu, X., Xiao, Y., Sun, G., Scott, D. F., Zhou, S., and Han, L.: Global pattern for the effect of climate and land cover on water yield, Nat. Commun., 6, 5918, doi:http://dx.doi.org/10.1038/ncomms6918

Web End =10.1038/ncomms6918 http://dx.doi.org/10.1038/ncomms6918

Web End = , 2015.

Hydrol. Earth Syst. Sci., 20, 47474756, 2016 www.hydrol-earth-syst-sci.net/20/4747/2016/