River flow conditions affect its biodiversity and can be examined with hydrological and environmental indicators (Jiang et al. 2014; Sun et al. 2019) and it evaluates many physical and ecological activities (Roy et al. 2006; Usman et al. 2021). (Monico et al. 2022) stated that the use of HI for optimal decision making and planning is common in water resources management. However, river flow components play an important role in the exploitation of water resources, biochemical processes, environment and maintaining the integrity of the natural ecosystem (Kennard et al. 2010; Arthington 2012). Therefore, the assessment of water resources related to dams should include the assessment of the total flow regime in order to determine the more accurate range of changes in water resources and provide a proper understanding of the mechanisms of change in rivers and environmental systems (Zhang et al. 2017). IHA and EFC have been developed to assess the reflection of human manipulations in NRF and the influence of dam construction (Mathews and Richter 2007). Quantification of IHA parameters is a common approach for assessing NRF change that involves comparing flow regimes (The Nature Conservancy 2009). Therefore, based on the findings of (Cardoso 2013), which stated that using these indicators, the basic characteristics of river flow including magnitude, duration, timing, frequency and rate can be examined.
The process of selecting effective indicators in HI and EFC to identify critical watersheds using the main component showed that four groups of components were effective in selecting the most influential indicators. The first and second components explain the 59.96 and 32.00% of the variance in the data. The percent of variance explained by the third and foutrh components were 4.43 and 2.83%, respectively. The eigenvalues of components (Bro and Smilde 2014; Adhami and Sadeghi 2016) 1, 2, 3 and 4 based on their total variance explained in the main components were 74.84, 19.02, 2.99 and 2.37%, respectively (Table 2 and Fig. 3). Based on the results of the rotated component matrix (Foody et al. 2004; Yang et al. 2008; Bucherie et al. 2022), 40 hydrologic indices were selected in the first components. Therefore, 40 indices, the correlation coefficient was significant in 30 indices, the highest correlation coefficient was related to small flood rise rate, high flow frequency, small flood peak and July-low flow with the value of 0.98 (α < 0.01) (Table 3).
In the second components, 29 hydrological indicators were selected in the main components. Accordingly, 19 indices had an acceptable correlation coefficient, with the June and May indices ranking first with a value of 0.99. In general, based on the results of the main components, it can be said that among the 47 hydrological indicators studied, only small flood fall rate, fall rate, small flood frequency and N-zero indicators had not an acceptable correlation and other indicators were selected as effective parameters in the study of susceptible watersheds. Consequently, according to reliable sources, these four physical indicators are the most important parameters in the study of NRF, and therefore these indicators were used for the further analysis. One of the main reasons for selecting these indicators in this study was that according to (Belay et al. 2010; Tegos et al. 2018; Greco et al. 2021), with the occurrence of small floods, aquatic animals and other living organisms are able to move downstream and establish new habitats in floodplain areas. Mathews and Richter, (2007) also stated that the large floods affected the physical and biological structure of rivers, which leads to competition between species, and contribute in maintain key habitats.
Quantitative results of HI and EFC showed that in most indicators, small values in DRGSa were the maximum and confirmed that among the studied watersheds, the condition of this watershed was appropriate. Hence, this condition in this watershed was related to the pre-dam period and the post-dam period (DRGSb), the values of all indicators changed and had a decreasing trend (Ely et al. 2020; Villablanca et al. 2022; Zolfagharpour et al. 2022). In PRGS watershed, which was affected by Yamchi dam, N-zero index in the pre-dam period was 0.033, but in the post-dam period, it increased to 82.11. Also the values N-zero days in DRGS had a significant upward trend in the post-dam periods (Zuo and Liang 2015; Nasiri Khiavi et al. 2019b). Based on Table 4, BFI also had a decreasing trend in the stations affected by the dam (Nasiri Khiavi et al. 2019a). The small flood peak in PRGSa was 14.08 CMS, which was reduced to 3.5 CMS in the post-dam period (Mathews and Richter 2007; Zhang et al. 2015; Sojka et al. 2016; Marcinkowski and Grygoruk 2017; Nasiri Khiavi et al. 2019a; Villablanca et al. 2022).
The correlation matrix between monthly flows in the studied watersheds showed that the values of these variables in PRGSa and ARGS watersheds had a correlation coefficient of 0.91. PRGSb watershed was also ranked second with DRGSa watershed with a correlation coefficient of 0.74 (α < 0.01) (Fig. 4). Rise rate index in watersheds affected by the dam also showed a decreasing trend. In PRGS and DRGS watersheds, the percentage difference of these indicators in the pre- and post-dam periods decreased by -62.96 and − 79.03%, respectively, which showed the effect of dam construction (Zuo and Liang 2015; Nasiri Khiavi et al. 2019b). However, the values of this index was normal in unaffected regimes of upland stations (Table 4). The grading of the studied watersheds based on HI and EFC using BSA based on GT showed that the watersheds affected by the dam won the game. DRGSb and PRGSb watersheds with 216 and 174 points, respectively, gained the first ranks and were selected as the most critical watersheds in study area. However, in these watersheds in the pre-dam periods, BSA based on GT showed the lowest score and no critical condition was observed in these watersheds (Table 8). One of the reasons for choosing this algorithm in the present study was that the BSA is one of the most optimal methods of GT (Sheikhmohammady et al. 2010; Mahjouri and Bizhani-Manzar 2013; Avand et al. 2021) and this algorithm is based on linear comparison and scoring. (Madani 2010; Avand et al. 2021) confirmed that BSA best reflected the behavior of the actors involved in the decision.
The box plot related to BSA points also showed that the highest range of changes (BSA = 5) was observed in the watersheds affected by the dam construction (Du et al. 2020; Villablanca et al. 2022) and even in DRGSb, the range of changes in BSA points was very low and confirmed the critical condition in this watershed. However, BSA scores are lower (1 to 2) in stations that are not affected by the dam (DRGSa and PRGSa) (Fig. 5). The spatial arrangement of HI-EFC approaches and the combination of these two approaches based on BSA scores also showed that in all approaches, the watersheds affected by the dam had the highest value and were in critical condition (Fig. 7). Although, according to Fig. 6, it can be seen that flow regime is normal in stations that are not affected by the dam.
Despite the impact of Yamchi and Sabalan dams construction on flood control and flow regulation along with socio-economic benefits, the destruction of natural ecological services and threatened biodiversity in the Balkhlou-Chay and Qareh-Sou Rivers are the negative impacts of studied dams (Ripl 2003; Gao et al. 2009). Zhang et al., (2015) confirmed that the hydrological changes caused by the construction of the dam and its effect on EF caused great concern for hydrologists and ecologists. In this regard, (Chen et al. 2010) stated that the construction of dams caused significant hydrological changes that severely upset the balance of river flow. In general, it can be concluded that in the study watersheds in Ardabil province, unfortunately, the construction of Yamchi and Sabalan dams has a negative impact on HI and EFC (Zuo and Liang 2015; Wang et al. 2016; Nasiri Khiavi et al. 2019a, b) and in terms of NRF, Balkhlou-Chay and Qareh-Sou Rivers were in a critical state in the post-dam period.