What test to determine if 5 years of population data has a trend?

[dlk2143]

Hello.

I have population data for 5 years for several species of birds and I would like to know whether the populations are growing, staying stable, or decreasing. I am using R.

So far, what I have been doing is creating a linear model of the count data (or log of the count data) by year and looking at the slope to determine if it is different from 0 and then at whether it is statistically significant (p-value less than 0.05). I think this is reasonable, but since it is technically a time series is there another test that would be more appropriate?

I am not positive that the population counts each year really are autocorrelated so maybe what I'm doing is fine? It is for birds migrating through an area and the number seen each year does not necessarily depend on the number that were seen the year before.

If anyone could provide some advice and if there is another test I should be performing, how to do it in R would be greatly appreciated!

Thanks.

 

[noetsi]

It depends if the trend is determanistic or stocastic. If it is stocastic, and this receives the majority of the attention in the literature, an augmented dickey fuller test is usually recommended although it has low power. Determining if it has a determanistic trend is more complex, if I can find how in my notes I will post it. You should also eye the data. None of these methods tells you if the trend is positive or negative, but I would think that would be obvious simply by looking at the trend in your data.