Methods
E. superba
Pygoscelis sp.
Individual penguin counts were acquired from the dataset located at Mapping Application for Penguin Populations and Projected Dynamics (MAPPPD) \cite{Humphries_2017}. Using only stations within the West Antarctic Peninsula (CCAMLR region 48.1), nest counts were extracted for Pygoscelis sp.. In order to use the most relevant and representative counts, only stations with at least 10 counts per species were considered (n = 22 unique sites). The MAPPPD included counts of nests, adults, and chicks, however, to lower variability, only nest counts were used in our analysis. Observations were carried out mostly by ground counts, but aerial surveys and video recordings also contributed. The selected measurements ranged from 1979 to 2018 with a variable number of count surveys per year (Adelie: n = 213; chinstrap: n = 324; gentoo: n = 630). All counts were averaged per year for each species (n = 40).
Statistical analyses
Statistical analyses were carried out using Microsoft Office 16®️ Excel (2016) and Minitab 17®️ Statistical Software (Minitab Inc. 2017). Mean counts and hauls per year were automatically calculated after importing to Minitab®️. All species were first analyzed for observable population trends over time. Significant differences between mean populations in late-19th and early-20th centuries were tested for using ANOVA and paired t-test models. Correlation (Pearson's correlation coefficient) and regression analyses were run between E. superba and each Pygoscelis sp.