Grasstree stem analysis reveals insufficient data for inference of fire history

Abstract

Grinding back dead leaf bases on the stems of arborescent grasstrees (Xanthorrhoea spp.) reveals a pattern of horizontal bands that has been interpreted as a record of the fire history experienced by the plant. The validity of this fire history record has previously been assessed through comparison of 100 grasstree stems sampled from shrubland near Eneabba in Western Australia against a 30 year fire history determined from satellite imagery. This analysis showed that the two records matched more than would be expected by chance, but concluded that the interpretation of the grasstree record as a fire history was not warranted as most of the grasstree fire records did not match satellite fire records. A second analysis of the same two sets of records, published in this journal, also showed that the records matched more than by chance, but concluded that the interpretation of grasstree banding as fire history was valid, though it failed to quantify the strength of this agreement. Here we examine' the approaches and interpretations of the two previously published studies, and provide new analyses to refine estimates of the amount of fire-related data present in the grasstree record. We show that only ~20% of grasstree 'fire' records may be attributable to fire. With eight out of ten of records not attributable to fire, we confirm that the grasstree record in its current form cannot be interpreted as fire history, and therefore claims of the grasstree technique to support management actions are untenable.