To analyze the relationship between the three sample plots (cattle grazed, watershed, and abandoned farmland) and the Poaceae species found in them, you can use a multivariate statistical technique such as canonical correspondence analysis (CCA) or redundancy analysis (RDA).

CCA and RDA are both constrained ordination techniques that can be used to explore the relationships between species composition and environmental variables (in this case, the three sample plots). These techniques can help you determine which environmental variables are most strongly associated with the variation in species composition and identify any underlying patterns or gradients in the data.

Additionally, you can use statistical tests such as PERMANOVA or ANOVA to test for significant differences in species composition between the different sample plots. These tests can help determine whether there are significant differences in the species composition between the sample plots, and if so, which sample plots are different from each other.

Regarding the soil samples collected for the seed bank survey, you can use a variety of statistical techniques to analyze the data, depending on your research question. For example, you could use non-metric multidimensional scaling (NMDS) to explore the similarity between the soil samples based on their seed bank composition, or use permutational multivariate analysis of variance (PERMANOVA) to test for differences in seed bank composition between the different sample plots.

Overall, there are several statistical techniques that can be used to analyze the data collected in this study, and the choice of technique will depend on the specific research question and data structure.

Yes, I agree with Sumin Sumin - PCA, followed by PERMANOVA and then post hoc tests if there is significance :)

It would help if you were clearer with your research question here, but if you are just looking for an exploratory analysis of how the grass species differ between sites you could visualise the communities using NMDS/DCA/CA, test for differences using PERMANOVA or ANOSIM, look at which species are most responsible for differences found using SIMPER, or try indicator species analysis to see which species best characterise your sites (e.g. INDVAL). All of the above can be done in R using the Vegan package: https://vegandevs.github.io/vegan/articles/FAQ-vegan.html

There's a nice intro tutorial to ordination methods here https://ourcodingclub.github.io/tutorials/ordination/

Finally you can also answer questions about multivariate abundance data like yours using model based ordinations e.g. the Mvabund package https://fontikar.github.io/mvabund/index.html