The ghosts of forests past: mapping the ghost taxon Beilschmiedia tawa through species distribution modelling and co-occurring species

Abstract

Beilschmiedia tawa (tawa; Lauraceae) is a dominant canopy species in lowland forests of the North Island of New Zealand, yet it is severely under-represented in fossil and modern pollen records. Low pollen production, poor pollen preservation, and low taxonomic resolution all contribute to its extremely weak pollen signal; the three defining characteristics of a ‘ghost taxon’. This lack of paleoecological records translates into a knowledge gap surrounding tawa’s pre-human biogeography. Macphail (1980) suggested that the presence of Beilschmiedia species could be inferred from pollen records through an indicator species, if there are strong associations between Beilschmiedia species and taxa well-represented in pollen records. Using species distribution modelling and species-co-occurrence analyses, my thesis aimed to identify reliable indicator (proxy) species for tawa. I used species distribution modelling to estimate the current distribution of tawa and identify the environmental drivers of tawa’s distribution and tawa’s ecological niche. Using a co-occurrence analysis spanning indigenous forests across tawa’s latitudinal range, I identified strong positive associations between tawa and other forest species. Of those species, Knightia excelsa (rewarewa) was the best potential proxy for tawa, on the basis of high positive co-occurrence, low mutual exclusivity, high range overlap, similar species prevalences, and similar distributions of observations in the modelling dataset. Comparing species distribution models between tawa and rewarewa suggested that this species is an excellent proxy for tawa occurrence, with exciting implications for interpreting tawa’s past distributions from palaeoecological records.