The protection of biodiversity is important for ecosystem service, but in most regions, biodiversity shows a trend of decline, and the reasons are not fully explained by available data, therefore many researchers have focused on the protection of the biodiversity. Arthropods are important components of terrestrial ecosystems, so both domestic and foreign researchers have been carrying out extensive research on the arthropods. Different semi-natural habitats are important for biodiversity protection because of the formation of heterogeneity in agricultural landscape. Arthropods are hotspot in the indicator species research, accounting for 15% of the total indicator species. Due to that the biodiversity indicator can reflect biodiversity of the region and monitor the change of biodiversity, those indicators can be studied instead of whole community to understand the biodiversity when condition or time is limited. By comparing the dominant groups of ground arthropods across different habitats (hedgerow, woodland, farmland or ditch) and farmland spatial scales (3.6 hm², 14 hm² and 28 hm²), we tried to find the biodiversity indicator in the study area, and then we used the indicator groups instead of ground arthropods community to conduct biodiversity research. The data from the first scale was taken as the average abundance of the ground arthropods in farmland A and farmland B; the data from the second scales was taken as the abundance of the ground arthropods in farmland A and farmland B; the data from the third scales was taken as the sum abundance of the ground arthropods in farmland A and farmland B. Data was done with ln transformation and statistical analysis was conducted with SPSS. Pearson analysis in R was applied (if the assumption of normality was violated, we used Spearman rank correlation instead of Pearson) to analyze the correlations between dominant groups and ground arthropod community with coefficient of 95% confidence interval to compare different habitats and different scales. The results showed that the ground arthropod dominant groups in the study area were Hymenoptera, Coleoptera and Araneae. For both hedgerow habitat and woodland habitat, the indicator group was Hymenoptera, Coleoptera was for farmland habitat, and Araneae was for ditch habitat. The three dominant groups had no correlation between each other (P > 0.05) in farmland habitat. Coleoptera and Araneae had a significant positive correlation in woodland habitat, ditch habitat and hedgerow habitat (P < 0.01). In woodland habitat and ditch habitat, Coleoptera and Hymenoptera also had a positive correlation (P < 0.05). Hymenoptera and Coleoptera existed a positive correlation on the second scale and the third scale in farmland habitat (P < 0.05), and the correlation presented a tendency of increasing with the increase scale. The study has shown that the biodiversity indicator groups have a great difference among different habitats, but the hedgerow habitat and woodland habitat have high similarity, while in farmland habitat the dominant group correlation showed scale-dependent. Land use intensity has a significant effect on the ground arthropod community structure, the more human disturbance, the weaker correlation between the ground arthropods. Increasing semi-natural habitats in the agricultural landscape would contribute to the number of predatory arthropod, and improve the abundance of predatory arthropod conducive to protect biodiversity in agricultural landscape and the controlling of biological in the terrestrial ecosystem. Suggestion for further research in the study area is that we can make use of correlation between the species to build a food chain for the biodiversity protection and ecosystem service.