1. Protected areas are increasingly being promoted as an important means of protecting freshwater biological diversity and ecological processes. A robust assessment of ecological changes in protected areas is fundamental to optimize conservation policies and adaptive management. China's efforts to establish aquatic reserves have attracted worldwide attention, especially the "10-year fishing ban" implemented in the Yangtze River basin.
2. We focused on Liangzi Lake, a freshwater protected area in the Yangtze River basin, to understand the effect of a fishing ban occurring after a short period of overfishing. In this aim, the time series of fish community taxonomic and functional structure encompassing the overfishing period and a post-ban period have been analyzed. Fish community metrics with direct, indirect, and no responses to fishing bans were identified.
3. The results indicate that in the early period of the fishing ban, the trophic level and body size structure of the fish community were in the way of recovery. However, species that prefer benthic habitats did not recover after fishing-induced habitat degradation. Functional traits were more sensitive than taxonomic indices and revealed subtle community changes, such as the recovery of some ecological functions, despite a non-recovering species richness.
4. Synthesis and applications. This study provides a rare case of a freshwater protected area in which the effects of conservation measures are studied with a temporal survey. The effectiveness of the functional trait approach in the application of protected area assessment was demonstrated by revealing the recovery of trophic level and body size structure of fish communities after the implementation of the fishing ban and the inadequacy of habitat restoration efforts. We suggest that in freshwater protected areas, insistence on a fishing ban is necessary but not sufficient for full biodiversity recovery, and other measures are needed, such as habitat restoration and species-focused stocking.

Field surveys were conducted four times a year (April-spring; July-summer; October-autumn; January-winter) from April 2016 to January 2022 for a total of 24 sampling occasions. At each sampling occasion, fish were collected at eight sites along the littoral zone of Liangzi Lake. For each site, we collected fish using two multi-mesh gillnets, one benthic and one pelagic. This standard multi-mesh gillnet measuring 30 m in length and 2.0 m in height (modified from European standard methods, CEN, 2015), has 12 panels (2.5 × 2 m) with respective mesh sizes of 5, 6.25, 8, 10, 12.5, 15.5, 19.5, 24, 30, 35, 43 and 55 mm. The gillnets were set in the afternoon (approx. 18:00) and harvested in the following morning (approx. 6:00) for three days in a row (36 h of sampling in total). All fish collected were individually identified with species (Department of Ichthyology, Institute of Hydrobiology, CAS 1976), counted and weighed at 0.01 g. The scientific names of the 45 species collected (Table S3) were updated according to FishBase (Froese & Pauly 2021). No fieldwork permits and ethics approvals were required.

Functional traits were classified into three trait categories reflecting the life history strategy (11 traits), feeding (5 traits) and habitat use (6 traits). We used six community metrics to describe the fish assemblage on each occasion and at each site of sampling. The metrics used cover three aspects of fish community information, community size, taxonomic diversity, and functional diversity. Community size (total number of individuals, total biomass) is often used to assess recovery of fish community productivity under fishing bans (Babcock et al. 2010; Coleman et al. 2015). Total abundance and biomass of fish were expressed per unit effort, number of individuals m^-2 h^-1 (NPUE) and g m^-2 h^-1 (BPUE), respectively. Two taxonomic metrics, species richness (SRic) and inverse Simpson index (invsimpson) and two functional metrics, function richness (FRic) and Rao’s quadratic entropy (RaoQ) were calculated. Functional richness represents the volume of the minimum convex hull filled by the community without considering the species abundances (Villéger et al. 2008). Rao’s quadratic entropy is an index of functional diversity based on the sum of squared distances among all species in the multidimensional functional space (Botta-Dukát 2005). RaoQ is not affected by species richness but is sensitive to species relative abundances used as weight for pairwise distances into functional space. All analyses in this study were performed in R version 4.2.0 (R Core Team 2022). The function dpFD of package “FD” was used to calculate the functional richness and Rao’s quadratic entropy (Laliberté et al. 2014).