This analysis is trying to get a handle on how weather (particularly temperature) is related to corticosterone in tree swallows. Samples are from adults and nestlings measured in Ithaca from 2013-2020. Samples collected after direct manipulations are excluded, but some samples after manipulations that we don’t think had any impact on corticosterone are still included (e.g., color manipulations). Samples that are collected after injections other than post-dex are also excluded. There is probably a bit more cleaning and checking that could be done here for a final analysis, but this should be pretty good as a general overview given how large the sample sizes are. Weather data is from the Ithaca Airport or Game Farm Road station and is summarized in various ways as described below.
For each combination of weather variable and cort measurement, I’m fitting essentially the same models that are as simple as possible. Basically, they include just latency and (usually) random effects for year and individual identity. In some cases the random effects are estimated at 0 and are removed to get around errors (they explain no variance). For females, I also have a categorical predictor for capture number, wich roughly equals early incubation (intercept), late incubation (capture 2), and provisioning (capture 3). Males and nestlings don’t have that since they are almost exclusively samples from ~day 4-8 (males) and day 12 (nestlings) after hatching. The plots are not controlling for the other variables that are in the models, they are just straight scatter plots for illustration. The fit lines are simple loess smoothed regressions so also aren’t controlling for anything. More could of course be done for any of these that we want to follow up on in more detail.
In some cases, I’m also modeling and plotting log values because a few points skew so high that it’s hard to see them. I’ve also trimmed the y-axis in some cases so that a few of the very highest value points are not visible in plots (e.g., in some basecort plots there are a handful of points >50 ng/ul but expending the y axis that high makes it impossible to see anything in the normal range).
One important thing to note is that the x axes differ between figures because nestlings and males are only ever sampled later in the season and because of that we have much less data (or none) at some of the very lowest temperatures that females are sampled at. There are obviously a lot of different plots here and more different ways of summarizing weather that we could think about, and the exact relationships or strength of patterns can differ between some of these choices, but I think the main summary points are:
This is using the average temperature for the three hours immediately before capture. The exact time of day included differs depending on the time of capture and is much later for nestlings. I did just the morning for females because not all entries in the database had times carried over and they’re all captured at similar times. Males and nestlings are more variable so I used the three hours prior even though that excluded some records.
|Adult Female Baseline||Adult Male Baseline||Nestling Baseline|
|Intercept||3.70||0.31 – 7.09||0.032||5.01||1.69 – 8.33||0.003||18.14||14.72 – 21.55||<0.001|
|Temperature 6-8am (C)||-0.20||-0.37 – -0.03||0.018|
|Capture 2||1.91||0.36 – 3.47||0.016|
|Capture 3||-0.12||-2.34 – 2.10||0.917|
|Latency (min)||1.79||0.43 – 3.15||0.010||2.39||1.27 – 3.51||<0.001||0.97||-0.00 – 1.94||0.050|
|Temperature 3 hrs prior (C)||-0.30||-0.46 – -0.15||<0.001||-0.61||-0.76 – -0.47||<0.001|
|τ00||0.96 band||0.77 band|
|N||494 band||199 band|
|Marginal R2 / Conditional R2||0.018 / 0.026||0.120 / 0.153||0.075 / 0.073|