My problem is about the RDA analysis of time-series audio data with environmental data. I have time-series data (Y) - one data frame with audio data (represented by acoustic indices) which is for 20 sites sampled on 4 consecutive days for 24 hours (One value per hour). Then another data frame (X) with environmental parameters (like plants coverage, diversity).
Now, these replicates (4*24) are obviously temporally autocorrelated from each other, so I have to take that into account when doing statistical tests on the data. I averaged all the replicates and I ran the analysis as a fairly simple RDA. But I'm not sure if this way I can solve the autocorrelation issues. And it seems I lost a lot of information in the process of averaging.
So, I wonder is there any better solution to run RDA with time-series/autocorrelated data, maybe add a random factor? Do you know of any other solution? Any ideas or pointers would be greatly appreciated.
