“…Notably, CoDA methods overcome constant sum constraints that are characteristic of many analyses ( i.e., the assumption that all samples sum to a constant, such as 1 or 100%), enabling us to analyze DEMs of different total areas, in which compositional components could vary independently from one another ( Aitchison, 2003a ). Whereas the use of traditional statistical techniques on compositional data yields multicollinearity and unreliable estimates ( Kucera & Malmgren, 1998 ; Aitchison, 2003b ; Graham, 2003 ; Dormann et al, 2013 ; Douma & Weedon, 2019 ), advancements in CoDA methods provide valuable and broadly applicable techniques to assess compositional data, which continue to gain recognition across disciplines ( Reyment, 1989 ; Baker, 2008 ; Campbell et al, 2009 ; Pierotti & Martín-Ferńandez, 2011 ; del Pozo Cruz et al, 2020 ). We performed all CoDA analyses in R, version 3.6.1 ( R Core Team, 2013 ), using the compositions , robCompositions , and zCompositions packages ( Palarea-Albaladejo & Martín-Fernández, 2020 ; Templ et al, 2020 ; van den Boogaart, Tolosana-delgado & Bren, 2020 ).…”