I am studying the effect of the land-use surrounding a location on the abundance of aphids on this location. To do this I fit a linear model with the land-use as independent variable and the abundance of aphids as the dependent variable. To check for spatial autocorrelation I plot the correlogram with the Moran I of the model residuals in function of the lag distance.
However I have multiple years of data: where the aphids have been observed each year together with the surrounding land-use. How can I account for this temporal effect? Should I incorporate a 'Year' variable in the linear model and can I then just look at the correlogram of the whole dataset?
Thanks in advance.
Hi Renik,
I'm not sure how useful my answer on this may be, as the data structure is somewhat different than what I am used to. However, what may be beneficial is to work out the rates of change in aphid population for a chosen interval and then fit a model with the rates of change (dependent) and land-use as a categorical independent variable.
You can then get a Global Moran I for the model as a whole or examine the geography of residual values. I have found LISA (Anselin, 1995) very beneficial for getting a more nuanced idea of clustering and outlier areas.
I am assuming for the above that you can fit data to polygon areal units and that it is a contiguous area.
As I say, it may/may not be useful for you. I have had success examining rates of voter turnout change (%) alongside a categorical urban/rural variable (6 categories) in this manner before.
Hopefully this is of some use, and very interested to see other suggestions.
Kind Regards,
William
Hi,
There are so many ways, yet I would prefer R (Rstudio) software for this.
More methods could be found here: -
https://doi.org/10.1111/cobi.13522
doi: 10.1111/j.2007.0906-7590.05171.x
Cheers,
- Badges
- Science topic
- Similar topics
- Psychology
- Mental Processes
- Learning