Data are collected for roughly for 24 hours, with the session starting at mid day, and being continued untill noon of the next day. Assuming that all the nest visits are associated with feeding, a variation in feeding frequency may be expected during the day, e.g. due to varying food availability related to air temperature/humidity/etc. Even if not all the nest visits are associated with feeding, one can imagine that various factors varying over the day may affect the nest visits. Alternatively, there may be some artifacts due to the onset/ending of the session in the middle of the hour, so data for given hour may be incomplete. In any case, trimming/selecting procedure may be crucial for the results. Not doing so may bias the results, or mask the examined phenomena (due to noise in the data/confounding variables). Obviously, and if relevant, parameters may be calculated for each hour/period of the day separately, and then hour/period of the day may be included in the model.


Let’s first look at overall temporal distribution of all nest visits of the focal parents

setwd("C:/Users/User/Dropbox/Working files/Projects/RFID_Starlings/RFID_Starlings")

# Libs
library(tidyverse)

# Data
nests_df <- readRDS("01_output_data.rds")





CONCLUSION: There are hours of the day that seem to be distinct/specific. For instance, look at: 16:00-17:00 (panel/facet: 4) and then 18:00 - 19:00 (panel/facet: 6), for the nest #93


Let’s now have a look at overall number of feeding performed during the day, and separately let’s look at the male contribution, that may also change over the day (e.g. he is busy with other stuff early at the morning/later during the day).


#  data prep 
prop_nest_visit_df <- nests_df %>% 
  group_by(session_id, sex, hr) %>% 
  summarise(n = n()) %>% 
  pivot_wider(names_from = sex, values_from = n) %>% 
  mutate(fm_sum = female + male,
         prop = male/fm_sum)





CONCLUSION: Some hours are of particularly low feeding frequency, including male contribution. Depending on the question being asked, different time window could/should be selected?


For now, for simplicity and further playing with the data, I will simply focus on three early morning hours (7-9), so the time of relatively high feeding frequency and male’s contribution (more data points = better).