Hi John
An approach I have used is to reclassify the land cover into quantitative scores of suitability, ranging from 1 to 100, and then transform to values between 0 and 1. Getting suitable quantitative values on the suitability of certain land cover classes for your bird species can be tricky.
If your observations are not heavily biased to sampling locations that may skew land cover counts, one approach would be to first use the Extract Value to Points tool, using your land cover as input raster. There is an option to append all the input raster attributes to the output features, so you will get the raster value as well as land cover class name. You will need the latter.
Open attribute table and summarise on the land cover filed name. The Case field with be land cover name and under Field name again select the land cover class name and count as statistic type.
This will provide you with a count of the number of times your bird species was observed in each of the land cover types. You can now convert these counts to a value out of a 100 so that you can use the Reclassify tool to reclassify your land cover raster into a new raster with scores ranging from 1-100. One cant use decimals in the reclassify tool, so once you have done this, add a column as float data type and divide by 100 to covert to values between 0 and 1.
To counteract the effect of land cover class size, for example where there is twice as much cropland to forest, then it is best to also determine the size of the different land cover classes and then normalize your count data against the size of the land cover class sizes. This will give you a rate of suitability or preference, which would be a better measure to use in your model.
