I did not have any difficulty implementing my distance-based simple random sampling strategy, besides it being a bit cumbersome to use the random number generator and then converting it to coordinates. However, I think this was a poor strategy for this data. It worked great on gathering data of the predictor variables, the trees, but worked poorly for data on the response variables, the birds. Some of the birds followed us between different sites, making it hard to distinguish if that would be their normal range. Additionally, the nearest species of tree and the trees that the birds were spotted in were not always the same.
The data was surprising as there were less birds than previously observed. This could be caused due to an error in the time of the observations, as most birds feed in the morning rather than in the afternoon.
I need to modify my approach. For starters, the study needs to be more focused on the response variables. Perhaps having sample units as individual birds rather than individual trees would improve the data. With this approach, I could record how long each individual spends in each type of tree. Additionally, rather than simply recording if they spend more time in a certain species of tree, this should be compared against the portion of trees in the area. For example: if a nuthatch spends 90% of its time in oak trees, but the area is 90% oak trees, then there is no significant preference to oak trees. I would also expand the study area to multiple different replicates around the park rather than in just one forest, as well as gathering observations at dawn. I think these changes would help focus my observations and increase the understanding of the pattern.
yes I was wondering if greater distance between the point counts would remove the bias of having the same birds present
the other variable that could be tracked (relative to tree species) is percent canopy over at a given point location – just a thought – this would reflect age of tree x species
not “fewer” birds (vs less) – fewer is something that can be counted ; )
Hi, this is Janine. I wanted to provide some feedback and ask a question about your study. I really like the idea of your predictor variable because it seems easy to measure in the field, either visually or through other means. However, the response variable (bird behavior) might require more detailed measurements, like observing how much time individual birds spend in different tree species. This could be challenging, but with careful observation and data collection, it’s still possible to obtain the necessary information have a couple of concerns that I’d like to discuss regarding potential confounding variables in your study. Could the presence of birds following you between different sites and the difference in observation times (morning vs. day) potentially contribute to confounding variables in your study?