I want to assess the impact of diving activities on a coral reef. To do this, I grouped the diving sites into four categories (sites with high diving intensity, medium, low and no divers). Each category doesn't have the same number of sites. Each site was sampled using 10 transects of 10 m where stony corals species composition was recorded, as well as coral colonies sizes, number of colonies per species, type of diseases observed, number of colonies with mechanical damage, etc. If the sites are spatially dependent (they are separated by hundreds to thousands of meters) what kind of statistical tool can I use to make comparisons among both sites and levels of diving intensity regarding the aforementioned stony coral variables without falling into spatial pseudoreplication?
Whatr are your observations of the 10 transects suggestings are there any trends visible. The site with no diving activity shows highest biodiversity and numbers? is this the case?
I assume your unit of replication is the diving site. Since they are widely separated, I think its safe to assume they are independent in terms of diver impacts, so pseudoreplication should not be a problem. That being the case, you could use a straightforward parametric statistic like ANOVA or linear regression to look for differences or trends.
Assuming that the categories are independent, I think you could use a mixed model design including category/site as a fixed factor and transect as random factor. When the response variable is normally distributed, you could use an ANOVA, but for other variables such as number of colonies I think you should use a Generalized Linear Mixed Model with Negative Binomial distribution.
You might want to look at adaptations of the Regional Risk Model to look at the relative magnitude of various impact pathways at different scales
- Badges
- Science method
- Similar topics
- Biological Science
- Ecology