This study attempted to compare the performance of local polynomial interpolation, inverse distance weighted interpolation, and ordinary kriging. Cross-validation was used to select the optimum method to get distribution results, and kriging was used for making spatial variability analysis. Data were collected from 87 sampling stations in November of 2015 (autumn) and February (winter), May (spring) and August (summer) of 2016. Results indicate that swimming crabs widely distributed in autumn and summer: in the summer, they were more spatially independent, and resources in each sampling station varied a lot; in the winter and spring, the abundance of crabs was much lower, but the individual crab size was bigger, and they showed the patchy and more concentrative distribution pattern, which means they were more spatially dependent. Distribution patterns were in accordance with ecological migration features of swimming crabs, which were affected by the changing marine environment. This study could infer that it is applicable to study crab fishery or even other crustacean species using geostatistical analysis. It not only helps practitioners have a better understanding of how swimming crabs migrate from season to season, but also assists researchers in carrying out a more comprehensive assessment of the fishery. Therefore, it may facilitate advancing the implementation in the pilot of the quota management program of swimming crabs in northern Zhejiang fishing grounds.
Figure 1. Zhejiang coastal waters ranges from 27°–31°N, 120°15′–123°45′E. In order to cover effective study near shore waters, sampling stations were set every 0.5° of longitudes and latitudes in the designated area. The dashed line regulates that fishing vessels with total length over 12 m are not allowed to fish westward of the line.
Figure 2. Semivariograms of each dataset. Red dots represent the semivariance between any pair of points which are binned values grouped by distance (also known as the lag). Conventionally, the lag size (0.094°) multiply the number of lags (12) should be equal to around half of the largest distance (around 2.26° in this study) among pairs of sampling stations. Blue triangles are the averages of each bin, which are used to fit the exponential model.
Figure 3. Distributions of swimming crabs performed by the ordinary kriging interpolation technique. Catch rate is transformed into logarithm form so that the data approximately in normal distribution. According to distribution patterns, the migration behaviors can be inferred.