Abstract

Many studies have focused on analyzing variation characteristics of the watershed resilience based on different indicators, while few efforts have been made to quantificationally evaluate contributions of climatic and anthropogenic factors to the varied resilience. In this study, we investigate changes in the seasonal runoff resilience of the entire Yangtze River basin during 1961–2014 by using a convex model and a resilience indicator (Pi). The MIKE 11HD model and the regression method were adopted to further differentiate effects of climate variations and human activities. Results show that climate variation (especially droughts and floods) and human activities exert negative and positive effects, respectively, and become primary reasons for falling and increasing trends in entire watershed resilience. These impacts grow with time under the gradually intensified climate variability and human activity.

HIGHTLIGHTS

  • Effects of climatic and anthropogenic factors on the varied watershed runoff resilience are quantificationally estimated.

  • Investigating the changes in the watershed resilience in the entire Yangtze River.

INTRODUCTION

Due to the effects of global warming, climate has markedly changed on the global scale (Arnell & Reynard 1996; McCarthy et al. 2010; Chai et al. 2019b, 2020), which brings huge and even devastating impacts on the watershed resilience that represents a watershed system's capacity to cope with disturbances while retaining essentially the same structure and function (Walker et al. 2009; Wilson & Browning 2012). For example, extreme climate events, including floods, droughts and hurricanes, have become more frequent and intense over the past decades, dramatically affecting the basic ecosystem and causing a low watershed resilience (Diez et al. 2012; Lloret et al. 2012). Meanwhile, human activities, under the rising population and the developing economy, have also caused tremendous changes in global rivers’ resilience, in which the construction and operation of large dams play the major role (Iii & Richmond 2016). Against such a backdrop, assessment on the health of the watershed system has become a hot topic of great interest to both hydrologists and ecologists (Hoque et al. 2012).

Existing studies have proposed effective indicators for evaluating spatio-temporal variations of the watershed resilience (Sadeghi & Hazbavi 2017). For instances, Hoque et al. (2012) introduced a method by applying R–R–V indicators (i.e. reliability, resilience and vulnerability), and using these indicators to analyze the risk of existing water supply systems (Mondal et al. 2010). Ahn & Kim (2019) appraised the watershed vulnerability of the Han River basin under artificial stressors by combining the Soil and Water Assessment Tool (SWAT) and six representative indicators (watershed landscape, stream geomorphology, hydrology, water quality, aquatic habitat and biology). Qi et al. (2016) first put forward an indicator (pi) that reflects streamflow autocorrelation to identify the resilience changes of the annual river discharge in the Yellow River and the Yangtze River basins. Above indicators behaved well in describing the variation trends in the watershed resilience triggered by climatic and anthropogenic factors, but failed to separate the contributions of the two factors on the altered resilience due to their complex interactions.

As the longest river of the Asia, the Yangtze River extends about 6,300 km from the Qinghai-Tibet Plateau to the East China Sea, experiencing consecutive climatic and anthropogenic interferences (Yi et al. 2003). In recent years, extreme climate events (e.g. floods in 1991, 1996, 1998 and 2016, and droughts in 1978, 1986, 2006 and 2011) occurred frequently in the Yangtze River, which not only caused huge damage to the ecosystem and economy, but also altered the watershed resilience, particularly in flood (from May to October) and dry (from November to April) seasons (Dai et al. 2008; Chai et al. 2019a). Today, more than 50,000 dams, including the world's largest dam, the Three Gorges Dam (TGD), have been constructed within the river basin, with the total storage capacity of the reservoirs up to 200 × 109 m3 (Yang et al. 2005, 2011). Dam operation has significantly flattened the seasonal hydrological processes by reducing flood peak during flood seasons and releasing the stored water during dry seasons. This human-induced external disturbance on runoff discharge has led to changes in the water resilience during flood and dry seasons of the Yangtze River basin. Nevertheless, variation trends in the watershed resilience during the two seasons have not been assessed to date. Highly changed hydrological processes due to dam operations in the Yangtze River since 1949 have brought discriminative effects on the ecological environment during different seasons. During the dry season, negative effects, resulting from water shortage in the river basin, have been mainly exert on aquatic organisms, such as the large-scale death of fish caused by the drought in 2011. During the flood season, floods primarily generate devastating effects on the terrestrial biota, including vegetation, birds and even people's residences on land. Thus, different seasons with discriminative hydrological processes represent diverse systems that have their own disturbance feature and recovery rate. Thereby, the resilience of different seasons needs to be investigated separately.

The present study aimed to analyze the variation characteristics in the watershed resilience of the seasonal runoff in the Yangtze River basin during 1961–2014 using the resilience indicator pi (Qi et al. 2016). The MIKE 11HD model and the regression relationships between precipitation and runoff discharges were applied to reconstruct the natural watershed resilience during 1961–2014 and that under non-TGD conditions during 2003–2014. In this way, the primary driving factor for the changed watershed resilience from the upper Yangtze to the estuary can be identified, with the contributions from climate variability, the TGD operation and other human activities quantifiably estimated.

MATERIALS AND METHODS

Study area and data sources

The Yangtze River originates from the Qinghai-Tibet Plateau and flows into the East China Sea. The whole river basin is separated into the upper, middle and lower sections, based on the locations of the Yichang and Hukou hydrological stations (Figure 1). From the upper reaches to the estuary, nine hydrological stations are distributed along the main stream, namely, Zhutuo, Cuntan, Yichang, Zhicheng, Shashi, Jianli, Luoshan, Hankou and Datong. The largest hydraulic construction, the TGD, is closely upstream of the Yichang station (∼40 km). Downstream of the TGD, both the Dongting Lake and the Poyang Lake have a strong interaction with the main stream, through the three diversion outlets (Songzi, Taiping and Ouchi), Chenglingji confluence and Hukou confluence, respectively (Figure 1). The Huangzhuang station is selected as the control station for the Hanjiang River, which is the largest tributary in the middle and lower Yangtze.

Figure 1

Geographical position of the study area.

Figure 1

Geographical position of the study area.

In order to analyze the changes in water resilience during dry and flood seasons of the Yangtze River and distinguish the effects of climatic and anthropogenic factors, daily runoff discharge time series at 15 hydrological stations (nine stations at the main stream and six stations at the tributaries) and daily precipitation time series at 145 weather stations during 1961–2014 were collected from the Ministry of Water Conservancy of China and the Resource and Environment Data Cloud Platform (http://www.resdc.cn/UserReg.aspx), respectively. To evaluate the impact of the TGD operation, the inflow and outflow discharges of the TGD during 2003–2016 and the topographic data in the reaches between Yichang and Datong stations in 2011 were gathered from the China Three Gorges Corporation and the Changjiang Water Conservancy Commission, separately.

Methods

Resilience indicator (Pi)

The concept of the stability fate of a ball in a landscape of hills and valleys has been widely applied to describe the ‘system resilience’. As shown in Figure 2(a) and 2(b), the stability of a state for this ball is decided by the slope of the landscape at its present position, while its resilience corresponds to the width of the attraction basin that encompasses the current system (Qi et al. 2016). For a high system resilience state, the ball can counterbalance small disturbances of the equilibrium (Figure 2(a)), and hereby the time series of this state is characterized by a low correlation with subsequent values (Figure 2(c) and 2(d)). Thus, if the system resilience can be reflected by the changes in the difference between a variable Qi at the time i and its adjacent values (Qi−n, Qi−n+1Qi−2, Qi−1, Qi+1, Qi+2Qi+n−1, Qi+n), we can find that the larger the difference the higher the system resilience will be. In contrast, for a low system resilience state, the system is more likely to cross the tipping point due to the reduced attraction basin (Figure 2(b)). A small disturbance will cause the system to shift to an alternative basin. Thus, the corresponding time series of the state variable under low resilience will be highly autocorrelated (Figure 2(d) and 2(f)). Therefore, for variable Qi at the time i and its adjacent values (Qi−n, Qi−n+1Qi−2, Qi−1, Qi+1, Qi+2Qi+n−1, Qi+n), if the difference is small, the system resilience will also be low. Many previous studies have also supported the above conclusions that the increasing autocorrelation of a state variable in a system can be a generic leading indicator of low resilience and an early warning signal for critical transitions (Dakos et al. 2012).

Figure 2

Conceptual scheme of the resilience and stability of a complex system.

Figure 2

Conceptual scheme of the resilience and stability of a complex system.

Thus, based on the convex model, an indicator (Pi) that can reflect the streamflow's autocorrelation at a given time is proposed to identify the resilience changes of river basins based on river discharge (Qi et al. 2016). The Pi was calculated using Equations (1)–(4):
formula
(1)
where, Mi is part of the data series of the seasonal runoff (Q), n is the time step (n here we adopt as 2), and i is the label of a certain year:
formula
(2)
formula
(3)
formula
(4)
where, reflects the dispersion degree of the seasonal runoff (Q), and and mean the maximum and the minimum discharges of the data series of seasonal runoff. Based on the previous study (Qi et al. 2016), the value of φ directly defines the boundary of the convex set. here is chosen as 0.75, which means that the area of the convex set is equal to 75% of that of the largest convex set ( = 1). The value of is the system resilience at the time i. It indicates that we can calculate the resilience of the river basin based on the seasonal runoff in dry and flood seasons in each year using Equations (1)–(4).

Evaluation of impacts of climatic and anthropogenic factors on system resilience

The Mann–Kendall Test is widely applied to analyze variation trends in hydro-meteorological time series, including streamflow, precipitation and temperature, and is also used to detect break points of the series. Based on first break points of runoff discharge series, the whole concerned period can be separated into two sub-ones, namely, the ‘natural period’ (1961–1980) and the ‘impacting period’ (1981–2014) (see method in Supplementary Information), as commonly seen in previous studies (Bao et al. 2012; Yuan et al. 2015). During the natural period, human activities have limited effects. In order to estimate the effect of the TGD operation (in 2003), the impacting period was further separated into the pre-TGD (1981–2002) and post-TGD (2003–2014) periods. The detailed method of the evaluation on effects of climatic and anthropogenic factors on the varied watershed resilience is as follows (taking the dry season as an example):

  • (1)

    ‘pre-TGD period’ vs. ‘natural period’:

The observed resilience during the pre-TGD and natural periods are denoted as and respectively. Changes in the observed resilience () between these two periods are calculated using Equation (5), and are separated into two components (Figure 3), namely, variations caused by climate changes () and total human activities ():
formula
(5)
Figure 3

Schematic diagram of the evaluation of effects of climatic and anthropogenic factors on changes in watershed resilience.

Figure 3

Schematic diagram of the evaluation of effects of climatic and anthropogenic factors on changes in watershed resilience.

Precipitation can be used as an index to account for variations in hydrological records caused by climate change. Data of precipitation and discharge during the natural period are used to establish linear regression equations. The average precipitation in dry season during pre-TGD period is put into these regression equations to extrapolate the average discharge during dry season in the pre-TGD period without effects of human activities. According to the reconstructed runoff discharge and Equations (1)–(4), only the response to climate change during the pre-TGD period can be estimated (Figure 3). Therefore, can be calculated as the difference in the reconstructed (pr1) and the observed resilience (pn) (Equation (6)):
formula
(6)
  • (2)

    ‘post-TGD period’ vs. ‘natural period’:

During the post-TGD period, the observed resilience is denoted as . The difference of the observed resilience between the post-TGD period and the natural period () can be calculated using Equation (7), which is separated into three components (Figure 3), namely, those respectively induced by climate variation (), TGD operation () and other human activities ():
formula
(7)
To acquire , the relationship between precipitation and runoff discharge during the natural period is established to restore the normalization runoff without effects of human activities (Figure 3). is then calculated using Equation (8) based on the multi-year average value of the restored normalization runoff (defined as ):
formula
(8)
To identify the influence of TGD operation (), the MIKE 11HD model was applied to restore the discharge process during the post-TGD period by assuming no flow regulation from the TGD (see method in Supplementary Information). Based on the reconstructed discharge during the post-TGD period, the resilience under the condition of non-TGD operation (Figure 3) was evaluated using Equations (1)–(4). The reconstructed resilience (denoted as ) was then used to calculate the using Equation (9), while effects of other human activities () were attained using Equation (7):
formula
(9)

Definitions of technical terms

Table 1 shows definitions of technical terms involved in this study.

Table 1

Definitions of associated technical terms

Technical termDefinition
System resilience System resilience is an ability of the system to withstand a major disruption within acceptable degradation parameters and to recover within an acceptable time. 
Watershed resilience/watershed health/health of the watershed system A watershed system's capacity to cope with disturbances while retaining essentially the same structure and function. 
Water resilience Reflecting alterations in long-term hydrological processes and the stability of river basins. 
Resilience of the annual river discharge Reflecting alterations in annual river discharge and the stability of river basins. 
Resilience of the seasonal runoff Reflecting alterations in seasonal river discharge and the stability of river basins. 
Technical termDefinition
System resilience System resilience is an ability of the system to withstand a major disruption within acceptable degradation parameters and to recover within an acceptable time. 
Watershed resilience/watershed health/health of the watershed system A watershed system's capacity to cope with disturbances while retaining essentially the same structure and function. 
Water resilience Reflecting alterations in long-term hydrological processes and the stability of river basins. 
Resilience of the annual river discharge Reflecting alterations in annual river discharge and the stability of river basins. 
Resilience of the seasonal runoff Reflecting alterations in seasonal river discharge and the stability of river basins. 

RESULTS

Variation trend of resilience

Dry season

Figure 4 shows variation trends in the resilience indicator Pi in the dry season during 1961–2014. We found that the observed multi-year average values of Pi at Zhutuo and Cuntan stations (located in the upper Yangtze) during the natural period (1961–1980) were merely 0.30 and 0.40, and distinctly increased to 0.59 and 0.64 during the pre-TGD period (Table 2). In contrast, the values of Pi from Yichang to Datong stations presented obvious falling trends when comparing the pre-TGD period with the nature period, and the average and the maximum decreased values reached 0.17 and 0.36. In conclusion, the observed watershed resilience in the dry season increased in the upper Yangtze during the pre-TGD period in comparison with the natural period (1961–1980), while that underwent an opposite variation trend in the middle and lower Yangtze.

Table 2

Multi-year average values of the observed resilience (pi) at the nine main stream hydrological stations during the dry season

PeriodResilience indicator (Pi) at hydrological station
Upper reach
Middle reach
Lower reach
ZhutuoCuntanYichangZhichengShashiJianliLuoshanHankouDatong
Natural period (1961–1980) 0.3 0.40 0.60 0.60 0.50 0.29 1.00 1.00 0.90 
Impacting period (1981–2002) 0.59 0.64 0.55 0.45 0.45 0.27 0.82 0.64 0.55 
Impacting period (2003–2014) 1.25 1.5 1.0 0.67 0.50 0.50 0.17 0.33 
PeriodResilience indicator (Pi) at hydrological station
Upper reach
Middle reach
Lower reach
ZhutuoCuntanYichangZhichengShashiJianliLuoshanHankouDatong
Natural period (1961–1980) 0.3 0.40 0.60 0.60 0.50 0.29 1.00 1.00 0.90 
Impacting period (1981–2002) 0.59 0.64 0.55 0.45 0.45 0.27 0.82 0.64 0.55 
Impacting period (2003–2014) 1.25 1.5 1.0 0.67 0.50 0.50 0.17 0.33 
Figure 4

Variation trend of the resilience (pi) at the nine main stream hydrological stations during the dry season.

Figure 4

Variation trend of the resilience (pi) at the nine main stream hydrological stations during the dry season.

During the post-TGD period (2003–2014), the observed multi-year average vales of Pi in dry season from Zhutuo station (at upper reach) to Jianli station (at middle reach) were all higher than 0.5, with the maximum value even up to 1.5 at Cuntan station. However, the maximum indictor Pi during the natural period was only 0.6 at Yichang and Zhicheng stations. By contrast, the indictor Pi from Luoshan station (downstream of the Jianli station) to Datong station (near the estuary) during the post-TGD period was markedly reduced compared with the natural period, with the average reduction up to 0.8. At Datong station which approaches the Yangtze Estuary and controls the whole watershed, the average indictor Pi decreased from 0.90 (during the natural period) to 0 (during the post-TGD period). Consequently, the watershed resilience during the post-TGD period showed an obviously rising trend from the upper Yangtze to the Jianli station and a sharply decreasing trend downstream.

Flood season

Figure 5 presents changes in the resilience indicator Pi in the flood season. It indicates that the multi-year average values of pi at Zhutuo and Cuntan stations (at upper Yangtze) decreased from 0.8 and 0.76 to 0.73 and 0.45, respectively, compared with the pre-TGD period and the natural period (Table 3). In the middle Yangtze, the indicator pi did not present an obvious variation trend along the main stream. The values showed increasing trends at Yichang, Zhicheng, Jianli and Hankou stations, while presented decreasing trends at Shashi and Luoshan stations. At the Datong station representing the lower reach, the indicator pi decreased from 0.90 during the natural period to 0.73 during the pre-TGD period. In conclusion, the resilience in flood season presented downward trends in the upper and lower Yangtze during the pre-TGD period, but an unobvious variation trend along the middle Yangtze.

Table 3

Multi-year average values of the observed resilience (pi) at the nine main stream hydrological stations during the flood season

PeriodResilience indicator (Pi) at hydrological station
Upper reach
Middle reach
Lower reach
ZhutuoCuntanYichangZhichengShashiJianliLuoshanHankouDatong
Natural period (1961–1980) 0.8 0.76 0.20 0.20 0.40 0.14 0.50 0.60 0.90 
Impacting period (1981–2002) 0.73 0.45 0.35 0.45 0.36 0.27 0.45 0.64 0.73 
Impacting period (2003–2014) 1.33 1.17 0.91 0.83 0.83 0.67 0.83 0.83 
PeriodResilience indicator (Pi) at hydrological station
Upper reach
Middle reach
Lower reach
ZhutuoCuntanYichangZhichengShashiJianliLuoshanHankouDatong
Natural period (1961–1980) 0.8 0.76 0.20 0.20 0.40 0.14 0.50 0.60 0.90 
Impacting period (1981–2002) 0.73 0.45 0.35 0.45 0.36 0.27 0.45 0.64 0.73 
Impacting period (2003–2014) 1.33 1.17 0.91 0.83 0.83 0.67 0.83 0.83 
Figure 5

Variation trend of the resilience (pi) at then nine main stream hydrological stations during the flood season.

Figure 5

Variation trend of the resilience (pi) at then nine main stream hydrological stations during the flood season.

During the post-TGD period, the multi-year average values of pi at Zhutuo and Cuntan stations were 1.33 and 1.17, separately, higher than those (0.80 and 0.76) during the natural period. In the middle reach, the increase in the indictor pi was further enhanced, with the average and the maximum increases up to 0.51 and 0.8. Oppositely, the indictor pi at Datong station decreased from 0.9 (during the natural period) to 0.83 (during the post-TGD period). Consequently, the resilience in the upper and middle Yangtze presented markedly rising trends compared with the post-TGD and natural periods, whereas that in the lower reach showed an adverse trend.

Causes for the altered resilience in dry season

Climate change

Based on the regression relationship between precipitation and water discharge, the resilience in response to climate change during the impacting period (1981–2014) is reconstructed (Figure 6), in which the difference between the reconstructed resilience (during impacting period) and the observed one (during natural period) was affected by climate changes. It can be clearly seen that the reconstructed resilience in the dry season presented a marked falling trend during 1981–2014 (Figure 6), meaning that climate variability reduced the resilience during dry season when comparing the impacting period with the natural period. During the pre-TGD period (Table 4), climate changes reduced the resilience in the whole main stream, with the average reduction up to 0.49. In the upper, middle and lower reaches, reductions of resilience caused by climate changes were 0.35, 0.50 and 0.72, implying that effects of climate changes were downstream enhanced along the river course. During the post-TGD period, the reduction of the resilience caused by climate changes in the entire basin further increased to 0.58. It was apparent that the impact of climate variability on the watershed resilience in the dry season has become more severe over time.

Table 4

Evaluation on effects of climate changes and human activities on the changed watershed resilience of the dry season

Impacting factorResilience indicator (Pi) at hydrological station
Upper reach
Middle reach
Lower reach
ZhutuoCuntanYichangZhichengShashiJianliLuoshanHankouDatong
Climate changes (1981–2002) −0.3 −0.4 −0.6 −0.42 −0.41 −0.11 −0.73 −0.73 −0.72 
Climate changes (2003–2014) −0.3 −0.4 −0.6 −0.6 −0.5 −0.29 −1 −1 −0.57 
Total human activities (1981–2002) 0.59 0.64 0.55 0.27 0.36 0.09 0.55 0.37 0.37 
Total human activities (2003–2014) 1.25 1.5 0.67 0.5 0.33 −0.33 
TGD operation (2003–2014)   −0.33 −0.5 −0.17 −0.33 
Other human activities (2003–2014) 1.25 1.5 1.33 1.5 0.84 0.83 0.33 −0.33 
Impacting factorResilience indicator (Pi) at hydrological station
Upper reach
Middle reach
Lower reach
ZhutuoCuntanYichangZhichengShashiJianliLuoshanHankouDatong
Climate changes (1981–2002) −0.3 −0.4 −0.6 −0.42 −0.41 −0.11 −0.73 −0.73 −0.72 
Climate changes (2003–2014) −0.3 −0.4 −0.6 −0.6 −0.5 −0.29 −1 −1 −0.57 
Total human activities (1981–2002) 0.59 0.64 0.55 0.27 0.36 0.09 0.55 0.37 0.37 
Total human activities (2003–2014) 1.25 1.5 0.67 0.5 0.33 −0.33 
TGD operation (2003–2014)   −0.33 −0.5 −0.17 −0.33 
Other human activities (2003–2014) 1.25 1.5 1.33 1.5 0.84 0.83 0.33 −0.33 
Figure 6

Variation trend of the reconstructed and the observed resilience (pi) during the dry season.

Figure 6

Variation trend of the reconstructed and the observed resilience (pi) during the dry season.

Total human activity

The difference between the observed and the reconstructed resilience during the impacting period was caused by the impacts of total human activities. As shown in Figure 6, the observed resilience was significantly larger than the reconstructed ones at the nine hydrological stations, suggesting that human activities increased the watershed resilience of the dry season for the whole river basin during the impacting period. During the pre-TGD period, human activities increased the resilience in the entire main stream, with the average and the maximum effects up to 0.42 and 0.64 (at Cuntan station) (Table 4). During the post-TGD period, impacts of human activities were much more significant, with the average and the maximum effects up to 0.77 and 1.5 (at Cuntan station). In general, human activities exerted high positive impacts on the resilience in the upper and middle reaches, with the influences up to 1.38 and 0.75, respectively, and imposed a negative effect on that in the lower reach, with the effect up to −0.33.

TGD operation and other human activities

Applying the MIKE 11HD model, the resilience under non-TGD conditions was reconstructed during the post-TGD period. Therefore, effects of total human activities during the post-TGD period were separated into two parts, namely, those due to TGD operation (i.e. the difference between the observed and reconstructed resilience under non-TGD condition) and other human activities (the difference between the reconstructed resilience under non-TGD condition and that in response to climate changes). As shown in Figure 6, the reconstructed resilience in the dry season was higher than the observed one during the post-TGD period, indicating that the TGD operation reduced the resilience, with the average influence up to −0.19 (Table 4). Conversely, other human activities increased the resilience, with the average and the maximum impacts up to 0.79 and 1.50 (at Shashi station). It was apparent that the effects of other human activities on the resilience in dry season were significantly greater than that brought by the TGD). As a whole, human activities (other human activities and TGD operation) produced positive effects on the watershed resilience of dry and flood seasons.

Causes for the altered resilience in flood season

Climate change

As shown in Figure 7, the reconstructed resilience in response to climate variability in the flood season during the impacting period showed an obvious downward trend compared with the observed resilience during the natural period, meaning that climate changes reduced the watershed resilience in the entire watershed during the impacting period. During the pre-TGD period, climate changes brought negative effects on the resilience of flood season at the nine hydrological stations, with the reductions caused by climate changes in the upper, middle and lower reaches up to 0.56, 0.16 and 0.26, respectively (Table 5). Negative impacts induced by climate variations were made more severe during the post-TGD period, and corresponding decreases in the three reaches further increased to 0.79, 0.34 and 0.73, separately. As a result, climate changes reduced the watershed resilience in the entire main stream and these negative effects enhanced over time.

Table 5

Evaluation on effects of climate changes and human activities on the changed watershed resilience of the flood season

Resilience indicator (Pi) at hydrological station
Impacting factorUpper reach
Middle reach
Lower reach
ZhutuoCuntanYichangZhichengShashiJianliLuoshanHankouDatong
Climate changes (1981–2002) −0.62 −0.49 −0.02 −0.02 −0.13 −0.05 −0.32 −0.42 −0.26 
Climate changes (2003–2014) −0.8 −0.76 −0.2 −0.2 −0.4 −0.14 −0.5 −0.6 −0.73 
Total human activities (1981–2002) 0.55 0.18 0.17 0.27 0.09 0.18 0.27 0.46 0.09 
Total human activities (2003–2014) 1.33 1.17 0.83 1.17 0.83 0.67 0.67 0.5 
TGD operation (2003–2014)   −0.34 0.16 0.16 0.08 
Other human activities (2003–2014) 1.33 1.17 0.83 1.34 1.17 0.83 0.51 0.51 0.42 
Resilience indicator (Pi) at hydrological station
Impacting factorUpper reach
Middle reach
Lower reach
ZhutuoCuntanYichangZhichengShashiJianliLuoshanHankouDatong
Climate changes (1981–2002) −0.62 −0.49 −0.02 −0.02 −0.13 −0.05 −0.32 −0.42 −0.26 
Climate changes (2003–2014) −0.8 −0.76 −0.2 −0.2 −0.4 −0.14 −0.5 −0.6 −0.73 
Total human activities (1981–2002) 0.55 0.18 0.17 0.27 0.09 0.18 0.27 0.46 0.09 
Total human activities (2003–2014) 1.33 1.17 0.83 1.17 0.83 0.67 0.67 0.5 
TGD operation (2003–2014)   −0.34 0.16 0.16 0.08 
Other human activities (2003–2014) 1.33 1.17 0.83 1.34 1.17 0.83 0.51 0.51 0.42 
Figure 7

Variation trend of the reconstructed and the observed resilience (pi) during the flood season.

Figure 7

Variation trend of the reconstructed and the observed resilience (pi) during the flood season.

Total human activities

The observed resilience during the impacting period was significantly higher than the reconstructed ones in response to climate changes in the same period (Figure 7), implying that human activities increased the watershed resilience of flood season from the upper reach to the estuary. During the pre-TGD period, human activities increased the watershed resilience in the upper, middle and lower reach, with average effects up to 0.37, 0.24 and 0.09, respectively (Table 5). During the post-TGD period, increases of resilience caused by human activities in the previous three reaches further increased to 1.25, 0.86 and 0.50, separately. In conclusion, human activities had produced obviously positive impacts on the watershed resilience since 1980, and the effects enhanced overtime, but alleviated along the main stream.

TGD operation and other human activities

As shown in Figure 7, the reconstructed resilience under non-TGD condition did not change a lot compared with the observed one during the post-TGD period, whereas presented a significant rising trend compared with the reconstructed one in response to climate variability. It was apparent that the TGD operation had limited effects on the watershed resilience of flood season downstream of the TGD, with average impacts merely up to 0.01 (Table 5), while other human activities significantly enhanced the resilience in middle and lower reach, with average impacts up to 0.80. Thus, although the TGD is the largest hydraulic construction, its influence on the watershed resilience of flood season was markedly less than other human activities.

DISCUSSION

Why human activities increased the watershed resilience

Dam operations and water-soil conservation projects are the most notable human activities in the Yangtze River basin, and exert tremendous effects on the watershed resilience of dry and flood seasons. The higher the cumulative storage capacity of the dams, the greater the regulation capability on the watershed resilience. Hence, an index, named the Degree of Regulation (DOR), has been adopted to reflect the impacts of dam operations on watershed resilience. As shown in Figure 8, the DOR in the in the upper, middle and lower reaches have all presented significantly rising trends since 1981, especially during the post-TGD period. With the rise of the DOR spread over the entire river basin, the ability of dam operations to alleviate drought in the dry season and flood pressure in the flood season also significantly increased (Yang et al. 2017; Chai et al. 2019a, 2019b), which promoted an increase in watershed resilience during two seasons.

Figure 8

Changes in the Degree of Regulation (DOR) in the upper (a), middle (b) and lower reaches (c), and the variation of the area of water-soil conservation projects. Note: DOR is the result of the cumulative storage capacity of a sub-region divided by the average annual runoff discharge in this sub-region.

Figure 8

Changes in the Degree of Regulation (DOR) in the upper (a), middle (b) and lower reaches (c), and the variation of the area of water-soil conservation projects. Note: DOR is the result of the cumulative storage capacity of a sub-region divided by the average annual runoff discharge in this sub-region.

Changes in land surface, especially in vegetation cover, alter the sediment and runoff discharges in rivers (Garbrecht et al. 2010; Wang et al. 2015), and thus affect the watershed resilience. To reduce soil erosion and increase the vegetation cover proportion, many water-soil conservation projects have been implemented in the Yangtze River basin, particularly in the upper Yangtze, with the cumulative area of the projects up to 147.3 × 103 km2. As a consequence, the area of surface erosion decreased from 562 × 103 km2 in the 1980s to 520 × 103 km2 in the 1990s and the vegetation cover rate increased by 14% over the past decades (Yang et al. 2015), both of which helped to strengthen the watershed resilience.

Why climate variability decreased the watershed resilience

Under the intensifying global warming, temperature, which is tightly related to evaporation, glacier-melting, vegetation and aquatic animals’ growth, has been significantly altered in the Yangtze River basin (Wang et al. 2011; Sang 2012; Zhang et al. 2015; Cui et al. 2019), disturbing the watershed resilience. Specifically, temperature in the Yangtze River basin presented an obvious upward trend during 1981–2014, with the rising trend after 2003 being significant, at a confidence level over 95% (Figure 9). The increased temperature contributed to reducing the watershed resilience of dry and flood seasons in the basin.

Figure 9

Variation trend of temperature and frequency of extreme climate events in the Yangtze River basin.

Figure 9

Variation trend of temperature and frequency of extreme climate events in the Yangtze River basin.

Extreme climate events (drought and flood disasters) in the Yangtze River basin are more and more frequent over the past several decades, which brought irreparable damages on ecosystem (Zhang et al. 2006; Jiang et al. 2012; Guan et al. 2015). Based on non-parametric trend and wavelet transformation analyses, Gemmer et al. (2008) found that the heavy precipitation was an essential driving factor for the frequent floods in 1990s in the Yangtze River basin and would aggravate its impact in the incoming decades of the 21st century (Gemmer et al. 2008). For example, the Yangtze River basin, particularly the upper Yangtze, witnessed successive alternations of flood and drought events during the period of 1982–2009. If focusing on the period of 2003–2014, the events were more frequent, such as the droughts in the years of 2004, 2006, 2011 and 2013, and the floods in the years of 2003, 2010, 2012 and 2016 (Zhou et al. 2017). These all contributed to a low watershed resilience.

Potential conversion of the watershed resilience in the Yangtze River basin

According to the reconstructed watershed resilience in response to climate variability during 1981–2014, we found that climate changes (especially extreme climate events) not only exerted negative effects on the watershed resilience of dry and flood seasons in the entire Yangtze River basin, but also caused the value of the watershed resilience (pi) to approximately equal zero. The extremely low watershed resilience triggered by climate variability may cause an irreparable conversion of the watershed resilience of dry and flood seasons to a lower state and bring damages on ecosystem in the future (Qi et al. 2016), which deserves high attention.

CONCLUSIONS

Under intensifying disturbances of climate variability and human activities, the watershed resilience of dry and flood seasons in the whole Yangtze River basin has changed markedly during 1961–2014. Based on the proposed reconstruction method and a resilience indictor (pi), the present study evaluated the contributions of climate changes and human activities on the altered watershed resilience of dry and flood seasons. The main conclusions are as follows:

  • (1)

    During dry season, the watershed resilience presented a rising trend in the upper reach when comparing the pre-TGD period (1981–2002) with the natural period, whereas that in the middle and lower reaches showed obvious falling trends. During the post-TGD period (2003–2014), the watershed resilience underwent increasing trends from Zhutuo to Jianli stations, and downward trends downstream of the Jianli station.

  • (2)

    During the flood season, the watershed resilience presented an obvious decreasing trend in the upper reach during the pre-TGD period in comparison with the natural period, whereas that during the post-TGD period increased markedly. In the middle and lower reaches, the watershed resilience showed unobvious variation trends during the pre-TGD period, while that during the post-TGD period still experienced an increasing trend.

  • (3)

    Climate changes produced negative effects on the watershed resilience of dry and flood seasons throughout the Yangtze River basin since 1981, and these effects became more severe over time. It is noteworthy that the reconstructed resilience in response to climate changes was extremely small (nearly zero), indicating that climate variability, especially the more and more frequent extreme climate events (droughts and floods), suppressed the watershed resilience in the Yangtze River basin. The extremely low watershed resilience may lead to a shift to a lower watershed resilience. In contrast, human activities have exerted positive impacts on the watershed resilience of the two seasons since 1981, and these effects also intensified over time.

  • (4)

    Although the TGD is the largest hydraulic construction in the world, its imposed pressure on the watershed resilience in dry and flood seasons was significantly less compared with effects of other human activities or the influence of climate variability.

COMPETING INTERESTS

The authors declare no competing interests.

ACKNOWLEDGEMENTS

This work was supported by the CRSRI Open Research Program (Nos. CKWV2018463/KY, CKWV2019727/KY and CKWV2017502/KY) and the Research Program of the Education Department of Hunan Province (No. 20B021). We also highly appreciate the valuable insights from the reviewers.

AUTHOR CONTRIBUTIONS

C.C.L, Y.F.C and B.Y.Z conceived the project and the theoretical framework. Y.F.C. developed the computational model. Y.P.Y., C.C.L and J.Y.D analyzed the data. B.Y.Z and Y.H contributed to the analysis and interpretation on the data. C.C.L and B.Y.Z wrote the manuscript.

DATA AVAILABILITY STATEMENT

All relevant data are included in the paper or its Supplementary Information.

REFERENCES

Ahn
S. R.
&
Kim
S. J.
2019
Assessment of watershed health, vulnerability and resilience for determining protection and restoration priorities
.
Environ. Modell. Software
122
, 103926,
https://doi.org/10.1016/j.envsoft.2017.03.014
Arnell
N. W.
&
Reynard
N. S.
1996
The effects of climate change due to global warming on river flows in Great Britain
.
J. Hydrol.
183
(
3–4
),
397
424
.
https://doi.org/10.1016/0022-1694(95)02950-8
.
Bao
Z. X.
,
Zhang
J. Y.
,
Wang
G. Q.
,
Fu
G. B.
,
He
R. M.
,
Yan
X. L.
,
Jin
J. L.
,
Liu
Y. L.
&
Zhang
A. J.
2012
Attribution for decreasing streamflow of the Haihe River Basin, Northern China: climate variability or human activities?
J. Hydrol.
460–461
,
117
129
.
https://doi.org/10.1016/j.jhydrol.2012.06.054
.
Chai
Y.
,
Li
Y.
,
Yang
Y.
,
Li
S.
,
Zhang
W.
,
Ren
J.
&
Xiong
H.
2019a
Water level variation characteristics under the impacts of extreme drought and the operation of the three gorges dam
.
Front. Earth Sci.
13
,
510
522
.
https://doi.org/10.1007/s11707-018-0739-3
.
Chai
Y.
,
Yue
Y.
,
Zhang
L.
,
Miao
C.
,
Borthwick
A. G. L.
,
Zhu
B.
,
Li
Y.
&
Dolman
A. J.
2019b
Homogenization and polarization of the seasonal water discharge of global rivers in response to climatic and anthropogenic effects
.
Sci. Total Environ.
709
,
136062
.
https://doi.org/10.1016/j.scitotenv.2019.136062
.
Chai
Y.
,
Zhu
B.
,
Yue
Y.
,
Yang
Y.
,
Li
S.
,
Ren
J.
,
Xiong
H.
,
Cui
X.
,
Yan
X.
&
Li
Y.
2020
Reasons for the homogenization of the seasonal discharges in the Yangtze River
.
Hydrol. Res.
51
(
3
),
470
483
.
http://dx.doi.org/10.2166/nh.2020.143
.
Cui
L.
,
Wang
L.
,
Qu
S.
,
Singh
R. P.
,
Lai
Z.
&
Rui
Y.
2019
Spatiotemporal extremes of temperature and precipitation during 1960–2015 in the Yangtze River basin (China) and impacts on vegetation dynamics
.
Theor. Appl. Climatol.
136
,
675
692
.
http://doi.org/10.1007/s00704-018-2519-0
.
Dai
Z.
,
Du
J.
,
Li
J.
,
Li
W.
&
Chen
J.
2008
Runoff characteristics of the Changjiang River during 2006: effect of extreme drought and the impounding of the Three Gorges Dam
.
Geophys. Res. Lett.
35
(
7
),
521
539
.
https://doi.org/10.1029/2008GL033456
.
Dakos
V.
,
Van Nes
E. H.
,
d'Odorico
P.
&
Scheffer
M.
2012
Robustness of variance and autocorrelation as indicators of critical slowing down
.
Ecology
93
(
2
),
264
271
.
http://dx.doi.org/10.1890/11-0889.1
.
Diez
J. M.
,
D'Antonio
C. M.
,
Dukes
J. S.
,
Grosholz
E. D.
,
Olden
J. D.
,
Sorte
C. J.
,
Blumenthal
D. M.
,
Bradley
B. A.
,
Early
R.
,
Ibáñez
I.
,
Jones
S. J.
,
Lawler
J. J.
&
Miller
L. P.
2012
Will extreme climatic events facilitate biological invasions?
Front. Ecol. Environ.
10
(
5
),
249
257
.
http:// doi.org/10.1890/110137
.
Garbrecht
J. D.
,
Starks
P. J.
,
Langendoen
E. J.
,
Shields
F. D. J.
&
Römkens
M. J. M.
2010
Watershed sediment yield reduction through soil conservation in a west-central Oklahoma watershed
.
Ecohydrology
2
(
3
),
313
320
.
http:// doi.org/10.1002/eco.55
.
Gemmer
M.
,
Jiang
T.
&
Su
B.
2008
Spatial and temporal variation of extreme precipitation over the Yangtze River Basin
.
Quat. Int.
186
(
1
),
22
31
.
http:// doi.org/10.1016/j.quaint.2007.09.001
.
Guan
Y.
,
Zhang
X.
,
Zheng
F.
&
Wang
B.
2015
Trends and variability of daily temperature extremes during 1960–2012 in the Yangtze River Basin, China
.
Int. J. Climatol.
124
(
3
),
79
94
.
http:// doi.org/10.1016/j.gloplacha.2014.11.008
.
Hoque
Y. M.
,
Tripathi
S.
,
Hantush
M. M.
&
Govindaraju
R. S.
2012
Watershed reliability, resilience and vulnerability analysis under uncertainty using water quality data
.
J. Environ. Manage.
109
,
101
112
.
https://doi.org/10.1016/j.jenvman.2012.05.010
.
Iii
A. J. S.
&
Richmond
R. H.
2016
Watershed restoration as a tool for improving coral reef resilience against climate change and other human impacts
.
Estuar. Coast. Shelf Sci.
183
,
S0272771416302013
.
http://doi.org/10.1016/j.ecss.2016.06.027
.
Jiang
Z.
,
Song
J.
,
Li
L.
,
Chen
W.
,
Wang
Z.
&
Wang
J.
2012
Extreme climate events in China: ipcc-ar4 model evaluation and projection
.
Clim. Change
110
(
1–2
),
385
401
.
http:// doi.org/10.1007/s10584-011-0090-0
.
Lloret
F.
,
Escudero
A.
,
Iriondo
J. M.
,
Martínez-Vilalta
J.
&
Valladares
F.
2012
Extreme climatic events and vegetation: the role of stabilizing processes
.
Glob. Change Biol.
18
(
3
),
797
805
.
http://doi.org/10.1111/j.1365-2486.2011.02624.x
.
McCarthy
M. P.
,
Best
M. J.
&
Betts
R. A.
2010
Climate change in cities due to global warming and urban effects
.
Geophys. Res. Lett.
37
(
9
),
232
256
.
https://doi.org/10.1029/2010GL042845
.
Mondal
M. S.
,
Chowdhury
J. U.
&
Ferdous
M. R.
2010
Risk-based evaluation for meeting future water demand of the Brahmaputra floodplain within Bangladesh
.
Water Resour. Manage.
24
(
5
),
853
869
.
https://doi.org/10.1007/s11269-009-9475-5
.
Qi
M.
,
Feng
M.
,
Sun
T.
&
Yang
W.
2016
Resilience changes in watershed systems: a new perspective to quantify long-term hydrological shifts under perturbations
.
J. Hydrol.
539
,
281
289
.
https://doi.org/10.1016/j.jhydrol.2016.05.039
.
Sadeghi
S. H.
&
Hazbavi
Z.
2017
Spatiotemporal variation of watershed health propensity through reliability-resilience-vulnerability based drought index (case study: Shazand Watershed in Iran)
.
Sci. Total Environ.
587–588
,
168
176
.
https://doi.org/10.1016/j.scitotenv.2017.02.098
.
Sang
Y. F.
2012
Spatial and temporal variability of daily temperature in the Yangtze River Delta, China
.
Atmos. Res.
112
,
12
24
.
https://doi.org/10.1016/j.atmosres.2012.04.006
.
Walker
B. H.
,
Abel
N.
,
Anderies
J. M.
&
Ryan
P.
2009
Resilience, adaptability, and transformability in the Goulburn-Broken Catchment, Australia
.
Ecol. Soc.
14
(
1
),
12
.
https://doi.org/10.5751/ES-02824-140112
.
Wang
Y.
,
Liu
B.
,
Su
B.
,
Zhai
J.
&
Gemmer
M.
2011
Trends of calculated and simulated actual evaporation in the Yangtze River Basin
.
J. Clim.
24
(
16
),
4494
4507
.
http://doi.org/10.1175/2011jcli3933.1
.
Wang
H.
,
Gao
J.
,
Hong
W.
&
Li
L.
2015
Impact of Soil and Water Conservation Measures on Runoff and Sediment Environment in Wei River Basin
.
Nature Environment and Pollution Technology
.
Wilson
M. A.
&
Browning
C. J.
2012
Investing in natural infrastructure: restoring watershed resilience and capacity in the face of a changing climate
.
Ecol. Restorat.
30
(
30
),
96
98
.
https://doi.org/10.1353/ecr.2012.0026
.
Yang
S. L.
,
Milliman
J. D.
,
Li
P.
&
Xu
K.
2011
50,000 dams later: erosion of the Yangtze River and its delta
.
Global Planet. Change
75
(
1
),
14
20
.
https://doi.org/10.11821/dlxb201403013
.
Yang
S. L.
,
Xu
K. H.
,
Milliman
J. D.
,
Yang
H. F.
&
Wu
C. S.
2015
Decline of Yangtze river water and sediment discharge: impact from natural and anthropogenic changes
.
Sci. Rep.
5
,
12581
.
https://doi.org/10.1038/srep12581
.
Yang
S. L.
,
Zhang
J.
,
Zhu
J.
,
Smith
J. P.
,
Dai
S. B.
&
Gao
A.
2005
Impact of dams on Yangtze River sediment supply to the sea and delta intertidal wetland response
.
J. Geophys. Res.-Earth Surf.
110
(
F3
).
https://doi.org/10.1029/2004JF000271
Yang
Y. P.
,
Zhang
M. J.
,
Zhu
L. L.
,
Liu
W. L.
,
Han
J. Q.
&
Yang
Y. H.
2017
Influence of large reservoir operation on water-levels and flows in reaches below dam: case study of the three gorges reservoir
.
Sci. Rep.
7
(
1
),
15640
.
doi: 10.1038/s41598-017-15677-y
.
Yi
S.
,
Saito
Y.
,
Zhao
Q.
&
Wang
P.
2003
Vegetation and climate changes in the Changjiang (Yangtze River) delta, China, during the past 13,000 years inferred from pollen records
.
Quat. Sci. Rev.
22
(
14
),
1501
1519
.
https://doi.org/10.1016/S0277-3791(03)00080-5
.
Yuan
Y. J.
,
Zeng
G. M.
,
Liang
J.
,
Huang
L.
,
Hua
S. S.
,
Li
F.
,
Zhu
Y.
,
Wu
H. P.
,
Liu
J. Y.
,
He
X. X.
&
He
Y.
2015
Variation of water level in Dongting Lake over a 50-year period: implications for the impacts of anthropogenic and climatic factors
.
J. Hydrol.
525
(
10
),
450
456
.
https://doi.org/10.1016/j.jhydrol.2015.04.010
.
Zhang
Y.
,
Xu
Y.
,
Dong
W.
,
Cao
L.
&
Sparrow
M.
2006
A future climate scenario of regional changes in extreme climate events over China using the PRECIS climate model
.
Geophys. Res. Lett.
33
(
24
),
194
199
.
http:// doi.org/10.1029/2006gl027229
.
Zhang
Y.
,
Hirabayashi
Y.
,
Liu
Q.
&
Liu
S.
2015
Glacier runoff and its impact in a highly glacierized catchment in the Southeastern Tibetan Plateau: past and future trends
.
J. Glaciol.
61
(
228
),
713
730
.
http://doi.org/10.3189/2015jog14j188
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY-NC-ND 4.0), which permits copying and redistribution for non-commercial purposes with no derivatives, provided the original work is properly cited (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Supplementary data