Effects of forest cover on richness of threatened fish species in Japan

Abstract

Estuaries--one of the most vulnerable ecosystems globally--face anthropogenic threats, including biodiversity loss and the collapse of sustainable fisheries. Determining the factors contributing to the maintenance of estuarine biodiversity, especially that of fish, is vital for promoting estuarine conservation and sustainability. We used environmental DNA metabarcoding analysis to determine fish species composition in 22 estuaries around Japan and measured watershed-scale land-use factors (e.g., population size, urban area percentage, and forest area percentage). We sought to test the hypothesis that the richness of the most vulnerable estuarine fish species (i.e., registered by the Japanese Ministry of the Environment in the national species red-list) is determined by watershed-scale land-use factors. The richness of such species was greater, where forest cover was highest; thus, forest cover contributes to their conservation. The proportion of agriculture cover was associated with low species richness of red-listed fishes (redundancy analysis, adjusted R² = 43.9% of total variance, df = 5, F = 5.3843, p = 0.0001). The number of red-listed species increased from 3 to 11 along a watershed land-use gradient ranging from a high proportion of agriculture cover to a large proportion of forest cover. Furthermore, the results showed that throughout Japan all the examined watersheds that were covered by >74.8% forest had more than the average (6.7 species per site) richness of red-listed fish species. This result can be attributed to the already high average forest cover in Japan of 67.2%. Our results demonstrate how the land use of watersheds can affect the coastal sea environment and its biodiversity and suggest that proper forest management in conjunction with land-use management may be of prime importance for threatened fish species and coastal ecosystems in general.

Keywords: actividad antropogénica; anthropogenic activity; coastal water bodies; cuerpos de agua costeros; environmental DNA metabarcoding; estuario; estuary; land use; metasecuenciación de ADN ambiental; uso de suelo.

FIGURE 1

Map of water sampling locations in Japan. The size of circle is proportional to the number of fish species on the Japanese Red Lists detected at each site. Numbers are rivers: 1, Mukawa; 2, Mabechi; 3, Naruse; 4, Naka; 5, Fuji; 6, Yahagi; 7, Miyagawa; 8, Yamato; 9, Asahi; 10, Niyodo; 11, Hijikawa; 12, Onogawa; 13, Oyodo; 14, Sendai; 15, Kuma; 16, Chikugo; 17, Takatsu; 18, Hino; 19, Kuzuryu; 20, Arakawa; 21, Akagawa; 22, Yoneshiro

FIGURE 2

Results of the redundancy analysis (RDA) of the Hellinger‐transformed fish species richness data per Japanese Red List category relative to five explanatory variables for (a) scaling 1 and (b) scaling 2 (point size, proportional to species richness; dotted vectors, red‐list categories; unbroken vectors, explanatory variables; circle, equilibrium contribution; ** p < 0.01; * p < 0.05; ^. p < 0.1; DO_S, surface dissolved oxygen; categories: DD, data deficient; LC, least concern; NT, near threatened; LP, local population; VU, vulnerable; EN, endangered; CR, critically endangered). Numbers, river identity [see Figure 1]. A red‐list category vector that is longer than the equilibrium circle radius makes a higher‐than‐average contribution to the ordination in the observed plane

FIGURE 3

Generalized linear model (GLM) based on the Poisson model with counts of Japanese Red List fish species per estuary as a function of the proportion of watershed area covered by forest (numbers, river identity [see Figure 1]; solid line, regression line; shading, standard error; horizontal dashed line, average number [6.7] of red‐list species; vertical dashed line, threshold of the watershed proportion of forest cover for sites to harbor more red‐listed fish species than average)