The time series graph of mean Pygoscelis spp. nest abundance from 1979 to 2018 yielded one observable negative trend: P. adeliae. Both variability and population number seem to drastically decrease around the turn of the 20th century (Fig. ???). This agrees with the significant differences between mean P. adeliae abundances before and after the year 2000 (ANOVA: P-value = 0.000, Paired t-test: P-value = 0.000; α = 0.05). When plotted against one another, this decline in adelie nests and the variability shift of E. superba both appear to occur around this time (Fig. ????). However, neither regression plots nor correlation tests yielded significant relations between E. superba and any of the Pygoscelis spp. (Regression: Adelie R-sq = 1.9%; chinstrap R-sq = 0.5%; gentoo R-sq = 0.000; Correlation: Adelie r = 0.138, P-value = 0.522; chinstrap r = -0.073, P-value = 0.735; gentoo r =-0.014, P-value = 0.946, α =0.05). Notably, though none of the r-values were statistically significant, the correlations between both chinstrap and gentoo were negative. We hypothesize that this could either be a factor of a lesser reliance on krill than adelie, or a relationship influenced by outliers. Outliers seemed to heavily affect trendlines (Fig. ???), however as the reason for these outliers was unknown, we did not exclude them from our analysis.
We attribute the lack of significant results to the selected data. The use of haul numbers as a proxy for E. superba populations may not be representative, especially considering the modern demand for krill products and, thus, more rigorous fishing regimes. While this could cause E. superba declines, it could also cause a bias toward higher numbers of hauls. Furthermore, though the haul dataset is considered a standard for krill population estimates, it is somewhat lacking in both modern and historic numbers; there is a gap between 1939 and 1980 as well as the last count being in 2003. More recent haul estimates would be extremely informative and could change estimates completely. On the other hand, while the MAPPPD is a very impressive and useful resource, the sheer amount and variation in types of data (species, sites, range in years, count methods, etc.) makes estimating accurate and representative numbers difficult. The choice to use only sites with 10 or more measures seemed sufficient, though it probably could have been more conservative. The more well-monitored a site is (e.g., 30-50 counts), the more influential it probably is to the total population \cite{Humphries_2017}.
While we do reject most of our hypotheses based on the results, the time series plots do suggest a lowering population of adelie and a lowering of E. superba population variability. However, as we question the representation of the datasets used, we are tentative to conclude significant declines and accept the "krill surplus" hypothesis. Additionally, the International Union for Conservation of Nature's Red List lists Adelie as "population increasing" (IUCN 2017). This raises the possibility of contrast between subpopulations and the overall global population. A disparity between Pygoscelis spp. reliance on krill could not be confidently concluded, though the r-values for chinstrap and gentoo were negative, suggesting possible krill-independence.
In conclusion, we can say that there may be a small correlation between the two populations. Realistically, it displays the need for more data and better methods of analyses. A more representative way of measuring E. superba rather than by haul numbers would be ideal, however the pelagic and planktonic nature of krill makes exact counts all but impossible \cite{NICOL_2006}; this contrasts with penguins, which can be relatively simple to count from land. Moreover, the addition of other relevant variables could draw further conclusions (e.g., sea ice cover, average temperatures, other top krill predator populations), especially in a rapidly changing climate. Admittedly, our hypotheses could not be explicitly accepted, though our results pose many new and exciting questions.