The coral reef research field has grown markedly in terms of both human power and technological capacity in recent years, a fortuitous occurrence given the rapidly diminishing nature of Earth's reefs on account of climate change and other anthropogenic stressors. Unfortunately, most coral biologists lack the statistical background to realize the full analytical potential of "big" datasets emerging from (non-exhaustively) 1) expanding reef survey efforts, 2) satellite and in-water (e.g., photomosaic) coral reef imaging projects, and 3) "next-generation" molecular approaches (i.e., 'OMICs); statistical training has not advanced commensurately with dataset size, a significant short-coming when considering the utility of these data in informing coral reef ecosystem management and conservation. One notably pervasive issue in 'OMICs research in particular is the general omission of multivariate statistical approaches (MSA), which universally outperform the more commonly employed, less statistically conservative univariate alternatives when attempting to A) model experimental results and B) make predictions about reef coral health and fate. Herein I have attempted to make a case for coral biologists to strongly re-evaluate the merit of MSA, as well as explain why relying on univariate approaches alone may actually lead to spurious findings that do not advance our knowledge of corals and coral reefs.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。