Dongting Lake (111°40'‒113°10'E, 28°38'‒29°45'N, Fig. 1), China’s second largest freshwater lake, is one of two large lakes that are freely connected with the Yangtze River. It is an important biodiversity hot spot (Fang et al., 2006), providing foraging ground for hundreds of thousands of wintering migratory waterbirds (Guan et al., 2016). Due to sedimentation and intensive lake reclamation, the whole lake is divided into three sub-lakes, which are hydrologically connected through main river channels. Our study site, the West Dongting Lake (WDTL) (111°57'‒112°17'E, 28°47'‒29°07'N) is the most upstream section (Fig. 1). Dongting Lake hydrology is largely influenced by the prevailing subtropical monsoon climate, with large intra-annual variations in water level and flow. However, the Three Gorges Dam has greatly changed the relationship between Dongting Lake and the Yangtze River and influenced the Dongting Lake hydrological regime (Lu et al., 2018), resulting in great annual and seasonal hydrology variations (Fig. 2). According to the recorded data of Nanzui Hydrological Station in West Dongting Lake, the average annual water levels in lake in 2017, 2018 and 2019 were 30.4±1.5 m, 29.7±1.0 m, and 30.4±1.0 m, respectively; and the average annual flows in 2017, 2018 and 2019 were 2630.55± 1341.45 m³ s^-1, 1975.26±1347.22 m³ s^-1, and 2338.76± 1475.66 m³ s^-1, respectively. In 2017, the average annual lake water level and flow were the highest among the three study years (Fig. 2).

In the late 1980s, a planting program in WDTL was started using P. nigra, an introduced fast-growing tree for wood production. To ensure that the young trees would survive the summer flooding, high diches were dig by pushing up the lakebed sediments, leaving a network of artificial ditches with depths varying from 1.0 to 2.5 m. These ditches had substantially reduced the hydrological connectivity in comparison with other habitats (Dong et al., 2021), providing suitable habitat for P. clarkii (Li et al., 2020).

P. clarkii crayfish were sampled using fyke nets (1200 cm in length, 30 cm in width and 30 cm in depth, mesh size 1 cm, with two doors at both ends) during March to November from 2017 to 2019, except for the winter when P. clarkii stay in holes for wintering. Samples were collected at 24 locations across the WDTL, 17 of which were from the heavily modified ditches at P. nigra plantations (referred to as modified hereafter) and seven were from natural open water areas near the wet meadows (referred to as natural hereafter), with water depths of less than 2.5 m according to the crayfish habitat preferences (Fig. 1). We set three repetitions of nets at each sample site. The nets were set at 9:00-10:00 am for approximately 24 hours and retrieved the following morning. All samples were transferred to the laboratory for identification, counting, and weighing. Crayfish relative abundance (catch per unit effort, CPUE) was calculated for each site.

At each sample site, the water depth, water transparency, pH, water temperature, and redox potential (ORP) were measured in situ. Water depth was measured with a handheld sonar depth sounder SM-5A. Water transparency was measured using a Secchi disk. Water pH, temperature, and ORP were recorded using a YSI multiprobe (YSI Professional plus). Water samples of 5L were also taken for laboratory analysis of total nitrogen (TN) and total phosphorus (TP) following the standard methods of APHA (1998).

In this study, the t-test was used to analyze P. clarkii CPUE and environmental factor variation between natural and modified habitats. To analyze the relationships between habitat environmental factors and P. clarkii CPUE, we firstly used the Vegan package in R (4.0.4) to perform Principal Component Analysis (PCA) and select the key environmental gradients. Then, the relationships between the principal components and log-transformed crayfish CPUE during the three investigation years at the two types of habitats were analyzed using the Generalized Linear Model (GLM) (with the habitat type as a fixed factor, and the sampling time as a random factor). This analysis used the lmer function in the lme4 package, and the Bayesian mixed models were built using the brm function in the brms package to explore how crayfish stocks responded to each of the environmental factors.

P. clarkii was active from April to November, and its abundance was high from May to August, after spring mating and hatching (Fig. 3). The average P. clarkii CPUE for the three-years (2017-2019) was 13.45±26.94 (ind net^-1). Crayfish abundance and average CPUE was largest in 2017 at 23.87±35.87 (ind net^-1), and smallest in 2018 at 3.37±3.30 (ind net^-1). The CPUE in 2017 was significantly higher than in 2018 and 2019 (Table 1), however, the differences in CPUE for different habitat types were insignificant (Fig. 4).

The PCA analysis results of nine environmental factors of WDTL are shown in Table 2 and Table 3. The first three axis interpretations show that PC1 explained 22.2%, PC2 explained 20.6%, and PC3 explained 13.8%, for a total of 56.6% of the variance of the environmental factors (Table 2).

The main determinants of PC1 were TP and TN, which were positively correlated with PC1, and PC1 was negatively correlated with ORP and water temperature. PC2 was mainly determined by water depth and pH; while PC2 was negatively correlated with water depth, transparency, and water temperature, and positively correlated with pH (Fig. 5). PC3 was mainly determined by DO and water transparency, and was negatively correlated with these two factors (Table 3).

A GLM model was established to analyze the influences of PC1-3, habitat type, survey year, and survey month on P. clarkii CPUE. The analysis found that PC1, PC2, PC3 and survey time all significantly affected crayfish CPUE (Table 4).

Water column TN and TP were positively correlated with the P. clarkii CPUE, because as nutrient contents increase, the primary productivity increases correspondingly. On the other hand, P. clarkii activities such as burrowing and feeding would destroy the growth of large aquatic plants and increase the resuspension of sediments, so an increase in its abundance would aggravate the transition of shallow lakes from clear water dominated by large aquatic plants to turbid water dominated by phytoplankton, thus aggravating the eutrophication of water bodies (Matsuzaki et al., 2009). P. clarkii CPUE was negatively related with the DO content, which was consistent with studies on the rivers of Louisiana, USA, the hometown of P. clarkii, where the individual size and average CPUE of crayfish in a hypoxic environment are both smaller than in a normal dissolved oxygen environment (Bonvillain et al., 2015). Water ORP determines the forms of water nutrients and plays an important role in nutrient circulation. Its value can affect the oxidation degree of sediment, and thus affect the survival rate of macrobenthos which are also important food resources of P. clarkii. The results of this study indicated that P. clarkii CPUE was positively correlated with temperature. A review of P. clarkii global population temperature tolerance found that its optimal growth temperature range is 21.8-26.7 ℃, and its highest temperature tolerated is 35 ℃ (Westhoff and Rosenberger, 2016). P. clarkii is a native of tropical regions while the middle-low Yangtze River is a subtropical region, therefore its abundance showed an increasing trend with temperature. This study also found that the P. clarkii CPUE was positively correlated with water depth, which was consistent with the findings in the Pacific Northwest lakes of the United States (Larson and Olden, 2013), probably because a deep water depth could provide a more diversified habitat for P. clarkii as refuge. Submerged plant abundance at WDTL was positively related with water transparency (Guo et al., 2020), and since such plants provide shelter for living benthic animals, P. clarkii CPUE was also positively related to water transparency. Although P. clarkii can tolerate a low pH environment through its hard exoskeleton, as the pH decreases, phytoplankton, zooplankton and macrobenthos that are sensitive to water acidification would disappear, resulting in insufficient food resources for P. clarkii. For example, snails disappear when the pH drops to 5.2 due to an insufficient availability of Ca(HCO₃)₂. A survey of 54 American lakes found that average pH of lakes with P. clarkii survival was 7.2 (Larson and Olden, 2013).

P. clarkii prefers to live in waters with abundant vegetation, high nutrient content and slow water flow, while it is less inclined to inhabit areas with fast flow rates and wide water surfaces (Cai, 2011). The reduced sediment deposition due to fast flow leads to a lack of potential food resources for P. clarkii, and prevents P. clarkii from resisting the adverse effects of environmental changes through burrowing.

In this study, the P. clarkii CPUE varied annually along with the fluctuations in the water level. The water level of WDTL from March to April of 2017 was more consistent with the multi-year average water level, while the water levels during the P. clarkii breeding seasons in spring of 2018 and 2019 were lower than the multi-year average water level. A low water level in the breeding season was detrimental to the survival of the larvae of P. clarkii which hatched that year. Because the water depth was relatively shallow, sufficient refuge was not available, which would lead to fierce competition among species. The water levels in the summer and autumn of 2017 were higher than those in 2018 and 2019 at the same time. During the highwater time, the turbidity of the water is high, and the continuous highwater level causes large numbers of plants to die, thereby reducing the number of adult crayfish that successfully overwinter. Therefore, the abundance of P. clarkii in 2018 decreased. As the water level in the summer and autumn seasons of 2018 decreased, the number of adult P. clarkii which successfully overwintered increased. In 2019, the number of P. clarkii increased, but it was still lower than in 2017. Studies on the control strategy for P. clarkii in Italian ditches also found that increasing water level and flow velocity can help control the numbers of P. clarkii (Gavioli et al., 2018).