Zhipeng Xu·Wenfei Liu·Qiang Li·Jianping Wu·Honglang Duan·Guomin Huang·Yizao Ge

Abstract Forest recovery plays a critical role in regulating eco-hydrological processes in forested watersheds.However,characteristics of the intra-annual runoffvariation associated with different forest recovery patterns remain poorly understood.In this study,three forest change periods were identified,the baseline period (1961–1985),reforestation period(1986 −2000) and fruit tree planting period (2001–2016).We selected the magnitude of seasonal runoff(wet and dry seasons) and distribution characteristics,i.e.,non-uniformity coefficient (C v),complete accommodation coefficient (C r),concentration ratio (C n),concentration period (C d),absolute variation ratio (Δ R) and relative variation ratio (C max).The pair-wise approach evaluated the intra-annual runoffvariation characteristics between forest change periods.Results indicate that reforestation decreased wet season runoffand increased dry season runoff.In contrast,fruit tree planting increased wet season runoffand had no significant effect on dry season runoff.For intra-annual runoffdistribution characteristics,reforestation significantly reduced the C v,C r,C n and C max .Distribution of the intra-annual runoffin the fruit tree planting period was not significantly different from the baseline.We concluded that reforestation reduced the occurance of extreme water conditions in wet and dry seasons and effectively increased the stability of the intra-annual runoff.In contrast,fruit tree planting increased instability and fluctuation of the intra-annual runoffafter reforestation.The characteristics of the intra-annual runoffto fruit tree planting was similar to those of the baseline.Therefore,adopting fruit tree planting practices to regulate intra-annual runoffcharacteristics may not be a practical approach,and impacts of different reforestation practices should be ascertained in our study region.The implications of this study should guide regional land–water management,and this study adds to the understanding of the impacts gained in forest cover on hydrology.

Keywords Intra-annual runoffvariation·Seasonal runoff·Forest recovery·Reforestation·Fruit tree planting

Introduction

Water resources play a critically important role in the development of human society,ecosystems health,and economic development (Wei et al.2016).Most studies of water resources are focused on the variations of annual water yields in response to changes in forest cover and species composition structures (David et al.1994;Liu et al.2016;Duan et al.2017;Xu et al.2019).On a global scale,water yields and their interactions with land use become more complex due to greater intra-annual runoffvariations resulting from utilization patterns (e.g.,agricultural irrigation,recreation water),land use changes (e.g.,reforestation,agriculture and urbanization) and climate (e.g.,precipitation and rainfall patterns) (Gerten et al.2008;Zhang et al.2016;Qin et al.2018).Thus,understanding variability in the intraannual runoffcharacteristics is of considerable interest to researchers for sustainable water management.According to the literature,variations in intra-annual runoffcharacteristics have direct effects on the integrity of riverine ecosystems and native biodiversity such as physicochemical environment of estuary salinity and sediment,species composition and richness (Richter et al.1996;Alber 2002).Numerous hydrologic indicators and methods have been developed for quantifying variations in intra-annual runoffcharacteristics.For example,the magnitude of runoff,such as the mean runoffat different time scales (annual,seasonal and monthly),has been evaluated in several studies (Lin et al.2015;Hou et al.2018;Geng et al.2020).In addition,the distribution characteristics of intra-annual runoffsuch as the non-uniformity coefficient,the complete accommodation coefficient,and concentration degree are also important hydrological variables (Lin et al.2017;Ren et al.2018),and have often been ignored by preliminary assessments,leading to incomplete evaluations of characteristics of runoffvariations.Therefore,there is a critical need to consider both the magnitude and distribution of characteristic variations in intra-annual runoffto better understand and assess of these ecosystem characteristics under specific management regimes.

Previous studies explored the characteristics of the intra-annual runoffvariations across different watersheds,while most focused on the impacts of climate change or human activities (e.g.,dam and urbanization construction)on annual streamflow variation (Ma et al.2008;Gu et al.2012).For instance,Wang et al.(2011) studied the annual dynamics of runoffdistributions of four rivers within Yulin county,China from 1961 to 2000 under climate change and human activities.Similarly,Gu et al.(2012)investigated runoffpattern variations of several major Chinese river systems under different climates.During the past decades,increasing forest cover has been an important strategy for China to address global climate change (Feng et al.2013).Although several studies have investigated the effects of forest recovery on hydrologic alterations (Liu et al.2016;Hou et al.2018;Li et al.2018a,b),conclusions are inconsistent,possibly due to differences in forest structure,species composition and watershed topography (Liu et al.2016;Xu et al.2019;Yu et al.2019).In particular,information of intra-annual runoffcharacteristics under forest cover dynamics remains limited.As such,more case studies are needed under the global climate change context to provide insights into water resource management and the development of sustainable forest management strategies at a regional scale.

Jiujushui watershed is located in the upper reach of the Poyang Lake basin in Jiangxi Province,China and has experienced large-scale reforestation and agroforestry programs (i.e.,fruit tree planting) over the past 50 years.It is,therefore,a suitable site for an intra-annual assessment of ecosystem hydrological dynamics under different management scenarios.In this study,two questions will be answered:(1) How large is the difference in intra-annual runoffcharacteristics under reforestation and fruit tree planting? (2) Are any similarities in the characteristics of the intra-annual runoffbetween reforestation and fruit tree planting? To answer these questions,the intra-annual runoffresponse of the catchment was compared under different forest recovery (reforestation and fruit tree planting)practices.

Materials and methods

Study area

Jiujushui watershed with a drainage area of 261.4 km2has the main tributaries of the Poyang Lake basin,which is located in the Jiangxi Province in the southeast of China.Average elevation is 231 ma.s.l.(Fig.1).The watershed has a subtropical monsoon climate with an average precipitation of 1780 mm over the years 1961 −2016 and a mean annual temperature of 18.4°C over 1961–2011.The area is mountainous with slopes up to 50°,and red soil,yellow–red soil and mountain yellow soil are dominant soils.Based on historical land use data,forest cover varied from 36.4% in 1961 to 77.1% in 2016,and changes in farmland and urbanization accounted for ＜ 3.5% (1962–2006) and 0.2% (1996–2005),respectively (Xu et al.2019).Thus,forest cover change was the dominant land-use change in the watershed.The forest types are dominated by timber forests,protection forests and economic forests,includingPinus massoniana,Citrusspp.andCunninghamia lanceolata.Detailed descriptions can be found in Xu et al.(2019).

Fig.1 Location of the study watershed

Data on hydrological and forest changes

Daily streamflow data from 1961 to 2016 was observed at the outlet of the Jiujushui watershed (i.e.,Shuangtian hydrometric station,ID:62406200) and provided by the Hydrology Bureau of Jiangxi Province.Climatic data,including daily precipitation from 1961 to 2016 and daily temperature data (daily maximum,mean and minimum) from 1961 to 2011 was obtained from the Climate Center of Jiangxi Province.Forest–related data,including forest cover and area of fruit tree planting were extracted from forest resource inventories compiled by the Forestry Department of Jiangxi Province.

Defining the periods of forest changes

Forest cover in the study watershed changed significantly in the past 50 years.From 1961 to 1985,forest cover slightly decreased but stimulated by several large-scale reforestation programs such as the Grain for Green program and the Mountain-River–Lake Ecological Project,forest cover increased 40.7% from 1986 to 2016,especially with fruit tree planting since 2000.As shown in Fig.2,forest recovery patterns after disturbance were different between natural forest and fruit tree orchards.Three forest change periods can be identified,including a forest degradation period which was treated as the baseline period (1961–1985),the reforestation period (1986 -2000),and the fruit tree planting period(2001–2016) (Xu et al.2019).The rational for period separation of forest changes can be found in Xu et al.(2019).

Fig.2 Forest cover change history in the watershed

Trends analysis

The Kendall tau and Spearman rho (Spearman 1907;Mann 1945;Kendall 1975) tests were used to detect whether significant trends in the intra-annual hydroclimatic data existed over the whole study period.Wet and dry seasons runoff(QWandQD),precipitation in wet and dry seasons (PWandPD),maximum,mean average and minimum temperatures in wet and dry seasons (TWmax,TWave,TWmin,TDmax,TDaveandTDmin) were examined.

Definition of intra-annual characteristics variations in runoff

The changes in seasonal runoff reflect the dynamics of intra-annual water resources in a watershed.In this study,we defined wet (QW:April-June) and dry (QD:December to February) runoffseasons to analyze hydrological responsesof intra-annual runoffto different forest recovery patterns.To further understand these variations fully,six distribution characteristic indicators were used:non-uniformity coefficient(Cv),complete accommodation coefficient (Cr),concentration ratio (Cn),concentration period (Cd),absolute variation ratio(ΔR),and relative variation ratio (Cmax).

Variation ranges ofCvandCr reflect non-uniformity and degree of stability in the runoffdistribution.A higher value ofCvandCrindicates higher instability and fluctuation of intra-annual runoffdistribution characteristics.CvandCrare expressed as:

where,δis the standard deviation,ithe month,R(t) andare monthly and average monthly runoffs in a given year.

CnandCdrepresent the degree of concentration in the intraannual runoffdistribution (Lin et al.2017;Ren et al.2018).A higher value ofCnandCdindicates higher instability and fluctuation of intra-annual runoffdistribution characteristics.Cncan be calculated as:

where,θt=2πt/12 is the angle of vector for month (t) from January to December,R xandR yare the two resultant vectors ofR(t).

Two indicators were used to reflect the variation ratio:absolute variation ratio (ΔR) and relative variation ratio(Cmax).A higher value of ΔRandCmaxindicates higher instability and fluctuation of intra-annual runoffdistribution characteristics.The ΔRandCmaxcan be calculated as:

where,Rmax andRmin are the maximum and minimum monthly runoffin a given year,respectively.

Isolating the effects of climate on intra-annual variations in runoff

To examine variations of intra-annual runoffunder different forest recovery practices,the effects of climate conditions should be minimized.The widely-used pair-wised approach was used which is a statistically sound method suitable for medium and large watersheds (Liu et al.2015b;Duan and Cai 2018;Xu et al.2019).The correlation between climate factor and intra-annual runoffcharacteristic variables as first tested with Spearman’s rho and Kendall’s tau correlation analysis (Appendix Tables S1 and S5).The highest correlation between climate factor and runoffcharacteristic variables can then be identified based on canonical correlation analyses (Appendix Tables S2 and S6).Finally,the pairedyears were selected based on the most relevant climate factors (e.g.,Appendix Tables S3,S4 and S7).A detailed description of the pair-wise approach can be found in Liu et al.(2015b) and Xu et al.(2019).

Results

Precipitation and runoffvariations

Average annual precipitation and runoffwere 1780 mm and 1080 mm from 1961 to 2016,respectively (Fig.3).The wet season precipitation in April to June (893 mm or 51.2% of the annual mean) in the baseline period was higher than that in the reforestation period (719 mm or 40.1% of the annual mean) and the fruit tree planting period (852 mm or 46.6% of the annual mean).Similarly,the wet season runoffshowed the same trends as precipitation.The dry season (December to February) precipitation and runoffin the baseline period were 225 mm (12.9%of the annual mean) and 132 mm (12.9% of the annual mean),respectively.Although dry season precipitations in the reforestation period (243 mm,13.6% of the annual mean) and fruit tree planting period (247 mm,13.5% of the annual mean) were similar,the dry season runoffin the reforestation period (162 mm,14.5% of the annual mean) was higher than that in the fruit tree planting period(158 mm,13.9% of the annual mean).The preliminary assessment,therefore,implies that different forest recovery patterns might be the cause for inconsistent changes in runoff.

Fig.3 a Annual and average monthly precipitation and b annual and average monthly runofffor different periods on the watershed

Trend analysis of intra-annual hydroclimatic variables

Trend analysis indicated that wet and dry season runoffand precipitation,i.e.,QW,QD,PWandPD(Table 1) had no significant trends,while temperature data showed significant increasing trends in wet and dry seasons,includingTWmax,TWave,TWmin,TDaveandTDmin,except forTDmaxover the study period.

Table 1 Trend analysis of intra-annual hydroclimatic variables in the study watershed from 1961 to 2016

Seasonal variations in runoff

Wet season runoff(QW)

The paired-years climate data selected for the baseline-reforestation and baseline-fruit tree planting periods are listed in Appendix Table S3.Runoffbased on the periods was selected and is shown in Fig.4.The results show that wet season runoff(520.6 mm) in the BL (baseline) period was significantly higher (p=0.015) than in the RE (reforestation)period (453.6 mm) (Fig.4 a).TheQwin the FTP (fruit tree planting) period was significantly higher (p=0.026) than those in the baseline (Fig.4 b).The results indicate that different forest recovery practices have distinct impacts on wet season runoff,specifically,reforestation decreased the wet season runoffwhile fruit tree planting increased it.

Fig.4 Variations in Q W:a between BL and RE;b between BL and FTP

Dry season runoff(Q D )

Similar to the wet season runoff,the same analysis was carried out on dry season runoff(Appendix Table S4 and Fig.5).Dry season runoff(QD) in the BL (baseline) period was significantly lower (p=0.001) than in the period of RE (reforestation).Compared with the baseline,FTP (fruit tree planting) decreased theQD,although no statistical significance was found (p=0.406),indicating that reforestation reduced the occurrence of extreme water conditions more than fruit tree planting.

Fig.5 Variations in Q D:a between BL and RE;b between BL and FTP

Distribution characteristics of the intra-annual runoff

Non-uniformity (C v and C r)

Based on the results of the pair-wise analysis (Appendix Table S7,Fig.6 a),theCvandCrwere 0.67 and 0.26 for the BL (baseline) period and 0.58 and 0.24 for the RE(reforestation) period,respectively.Compared with the BL(baseline),theCvandCrfor the RE period was reduced by 13.4% (p=0.004) and 7.7% (p=0.037),respectively.TheCv andCr for the FTP (fruit tree planting) period was,however,similar to the baseline (p=0.828 andp=0.715)(Fig.6 b).

Fig.6 Variations of C v and C r:a between BL and RE;b between BL and FTP

Degree of concentration (C n and C d)

The results show that theCnwas significantly higher in the BL (baseline) than in the RE (reforestation) by 12.8%(p=0.024),but no statistical difference was identified forCd(p=0.689).In contrast,neither theCnnorCdshowed significant difference between BL and FTP (fruit tree planting) periods (p=0.368 andp=0.782).TheCnandCdwere 0.34 and 150.02 for the BL period and 0.36 and 151.48 for the FTP period,respectively (Fig.7).

Fig.7 Variations of C n and C d:a between BL and RE;b between BL and FTP

Variation range (ΔR and C max )

TheCmaxfor the RE (reforestation) decreased by 12.4%(p=0.007) from the BL (baseline) but there was no statistically significant change in ΔR(p=0.203) (Fig.8 a).Compared with the BL (Fig.8 b),change trends of the ΔRandCmaxin the FTP (fruit tree planting) period were similar to those of theCvandCrbut no statistically significant difference was found (p=0.523 andp=0.413).

Fig.8 Variations of Δ R and C max:a between BL and RE;b between BL and FTP

Discussion

Responses of the intra-annual runoffvariation characteristics to reforestation

In this study,the results indicated that the reforestation period significantly decreased the magnitude of wet season runoffand increased the magnitude of dry season runoffcompared with the baseline period.For intra-annual runoffdistribution characteristics,reforestation significantly decreased theCv(non-uniformity coefficient),theCr(complete accommodation coefficient),theCn(concentration ratio),and theCmax(relative variation ratio) of the intraannual runoffover the baseline.This indicates that the variation in intra-annual runoffcharacteristics in the reforestation period reduced the occurrence of extreme water conditions in wet and dry seasons.In addition,the runoffin the reforestation period was more stable and less variable than in the baseline period.Other studies have concluded that reforestation regulates runoffand reserve water due to the structure of forest ecosystems (Wang et al.2013;Yu et al.2019).For example,surface roughness plays a key role in partitioning of rainfall and runoffgeneration as in forested watersheds(Balota et al.2016;Zhao et al.2018).Abundant understory vegetation and litter layers increased surface roughness in the reforestation period,which resulted in reduction in runoffand in flow velocity,and consequently increased soil infiltration capacity compared with the degraded forest (Shen et al.2001;Ding and Huang 2017).At the same time,during massive rainfall events usually occurring in the rainy season in subtropical China,leaf area and canopy interception under reforestation are higher than in degraded forests (Dung et al.2012).In addition,forest vegetation has well-developed root systems,and forest soils are often highly porous which can effectively absorb and store water (Robinson and Dupeyrat 2010).For instance,Zhang et al.(2012) suggested that in a rain-dominated watershed,rain-driven groundwater recharge and water storage in the wet season is used to maintain runoffin dry seasons,so that the change in intra-annual runoffwas more evenly distributed under forest cover.Therefore,forests store more water in aquifers from wet seasons and then slowly discharge stored water into river systems in the dry season,resulting in an increase of runoffin the dry season (Zhou et al.2010;Liu et al.2015a).Based on the above,our results demonstrated that reforestation has a positive effect on regulating variations in intra-annual runoff.

Responses of the intra-annual runoffvariation characteristics to fruit tree planting

Our results show that fruit tree planting significantly increased the magnitude of wet season runoff while no significant change was found in dry season runoff.Moreover,theCv,Cr,Cn,Cd(concentration period),ΔR(absolute variation ratio) andCmaxof the intra-annual runoffwere not significantly different from the baseline.This indicates that variations in the intra-annual runoffwere more uneven under fruit trees than for the reforestation.The distribution of runoffvariations is not only affected by vegetation cover but also by factors such as understory vegetation and litter layer,topography,management methods,and soil texture and properties (Liu et al.2016;Li et al.2018a,b;Yu et al.2018;Du et al.2019).In this study,forest litter,root,and soil conditions are different between reforestation and fruit tree planting and are key factors regulating hydrological response over two recovery periods.Fruit tree planting is cultured differently than reforestation.Anthropogenic factors such as site preparation and removal of understory vegetation and litter layers results in a smoother surface than for the reforestation.Planting practices alter soil texture and hydrological processes and subsequently,runoffand soil and water loss increase due to the reduction of infiltration.In contrast,in the reforestation period,there were no disturbances to natural soils (Cerdà 1997;Sileshi et al.2012;Zhao et al.2018;Xu et al.2019).Toohey et al.(2018) found that soil characteristics were different among four land-use types,and that interactions among different root systems,soil compaction,and soil moisture dynamics affected water storage and infiltration dynamics.However,for fruit tree plantations,to ensure the quality of the fruit,management measures,including long-term fertilizer applications,can compact soil and deteriorate texture,which leads to the cycle of soil water–gas-heat cannot be effectively coordinated (Liu et al.1998;Shui et al.2003).In addition,with an increase in fruit tree planting areas,irrigation demands are also increase in the growing season.Therefore,streamflow depletion due to irrigation is more severe in fruit tree planting than in reforestation.Therefore,the characteristics of variation in the intra-annual runoffare more uneven and the degree of fluctuation higher in the fruit tree period than in the reforestation period.

Limitations of forest recovery on hydrological influence

Our study shows that both reforestation and fruit tree planting are a gain in forest cover,but they have different impacts on water yield and flow variability.Different forest management programs should be distinguished for regional or even global water yield prediction.In addition,the findings add to the literature on the impacts of reforestation on hydrology and help to understand the interaction of water yield and forest change.It is generally recognized that forest recovery,such as improving forest cover,has a positive effect on hydrological restoration (David et al.1994;Li et al.2017).However,the role of forest cover in hydrology may be overestimated.Several studies were based only on static,quasistatic or single factor comparative analysis,for example,only considering the effects of quantity changes of forest cover on watershed hydrology (Stednick 1996;Costa et al.2003;Wei et al.2018).More importantly,many methods ignore the feedback of soil hydrological processes and forest structure change (Hou et al.2018),as a watershed ecosystem is an interactive and interrelated system.Therefore,more case studies are needed to enhance our understanding in the effects of forest recovery on runoffvariation characteristics by changing other hydrological processes (e.g.,soil hydrology).

Conclusions

This study analyses the responses of intra-annual runoffvariation characteristics to different forest recovery patterns(reforestation and fruit tree planting).The results show that reforestation effectively increases the stability of the distribution of the intra-annual runoffand reduces the occurrence of extreme water events in wet and dry seasons,while fruit tree planting increased fluctuation in intra-annual runoff.Furthermore,this study also demonstrated that the characteristics of the intra-annual runoffto fruit tree planting were similar to the baseline,suggesting that the increase in forest cover has limitations in regulating intra-annual runoff.Therefore,it is recommended that the impacts of reforestation and fruit tree planting be considered differently in the watershed to facilitate effective water management.

AcknowledgementsThis work was supported by the Education Department of Jiangxi Provincial (GJJ151141),National Natural Science Foundation of China (31660234),Jiangxi Province Department of Science and Technology (20161BBH80049) and the Outstanding Young Scholar of Jiangxi Science and Technology Innovation(20192BCBL23016).We also thank Dr.Zisheng Xing for improving the English and thoughtful comments on this paper.