I tried to keep my sampling strategy relatively straight-forward, so I didn’t run into any major difficulties. One thing I need to clarify for myself is the operational definition of a ‘visit’ by a pollinator. A lot of times, pollinators would mill about close to the flowers, but never actually landed. I counted those who milled about the flower for more than 5 seconds as a visitor, which I’m not sure if that should be included or not. So, that is definitely something that needs revisiting.
After reflecting, I think I will define a visit as a pollinator landing on any part of the flower OR milling about closely to the centre of the flower for more than 3 seconds. My main logic behind this is that both actions demonstrate clear attraction to the flower, even if they don’t land.
Another issue I ran into is the closeness of my 10 replicates in each predictor variable treatment. I’m unfortunately quite limited in this regard, as I am looking at a smaller garden where the flowers I’m sampling are very close together in some cases. I don’t really know if there is a way for me to work around this unless of course I took fewer replicates that were spatially far enough from one another, but I don’t think that’s the most favourable trade-off. I think I would rather have more data to look and follow the rule of ten, rather than worry about the individual differentiation between my replicates.
I can’t expect my replicate samples to not influence one another because they are so close together. So, it is reasonable that a pink flower that is right next to a yellow one has more visitors than a pink flower that is more isolated. Despite some pink and yellow flowers being very close to one another, I do have at least one replicate (i.e., flower) in each treatment (i.e., colour) that is significantly further away from the other treatment’s replicates. I think this is something I will just have to keep in mind and elaborate on in my final report in the limitations section.
I think I’m fairly happy with the allotted time I’ve set for data collection at each replicates, as 5 minutes seems to be enough to observe some bee visits. This is a good time frame because the sun is only out for so many hours in a day that if I were to increase it to say, 10 minutes, I would be spending too much time collecting data and the temperature and sun exposure would probably decrease by the time I got to the final few replicates and create a confound in my study.
On the topic of confounds, when I took data the first time I just moved from the leftmost plant to the rightmost plant. Moving forward, I think it makes most sense to alternate the colour of flower I observe each time so that I’m not looking at the majority of one colour of flower at the beginning of the session and the other colour at the end.
I think I am doing the best I possibly can regarding my approach to data collection. I will have to keep the aforementioned things in mind when collecting data, but I am going to continue to do what I am doing right now with those tweaks to the definition of a pollinator visit as well as alternating observations between the two colour flowers, and collect as much data as possible in my free time before the sun leaves me!