SEARCH

SEARCH BY CITATION

Keywords:

  • Australia;
  • correlated evolution;
  • extinction risk;
  • flora;
  • fruit types;
  • genus size;
  • life-history characters;
  • phylogenetic analyses;
  • sex systems

Abstract

Aim  To determine whether life-history characters that affect population persistence (e.g. habit and life span) and those that influence reproductive success (e.g. sexual system and fruit type) are non-randomly correlated with extinction risk (i.e. threat category) in the Australian flora (c. 19,000 species, of which c. 14% is threatened). To identify patterns that present useful conservation directions. To understand patterns of extinction risk in the Australian flora at a broad scale.

Location  Continental Australia.

Methods  A country-wide exploration of four life-history characters in the Australian flora (n = 18,822 species) was undertaken using reference texts, expert opinion, herbarium records and field work. For each character and threat-category combination, a G-test (using a log-linear model) was performed to test the null hypothesis that the two factors were independent in their effects on count. A generalized linear model (GLM) with a logit link and binomial error distribution was constructed with the proportion of taxa in each extinction risk category as the response variable and the habit, sex and fruit-type characters as explanatory terms. In a separate approach, we investigated patterns across the threat categories of non-endangered extant, endangered, and extinct using a multinomial model. We examined whether or not species-poor genera were more likely to contain threatened or extinct species than species-rich genera. A GLM with a binomial error distribution and logit link function was constructed to obtain a weighted regression on the proportion of species listed as extinct or endangered within a genus versus the log of the size of the genus. We also used a supertree analysis and character tracing to investigate the role of phylogeny on extinction risk.

Results  We found that the Australian flora is primarily composed of bisexual shrubs with dry-dehiscent fruits. Dioecious breeding systems (separate female and male flowers on separate plants) in many floras are the predominant unisexual system, but in Australia there are unexpectedly high levels of monoecy (separate female and male flowers on the same plant). Within the extinct data set of 31 species we detected a significant departure from that expected for habit but not for life span, sexual system or fruit type. There are significantly fewer trees on the extinct list than expected. This may reflect the greater resilience of trees than of other growth habits to extinction processes as well as the observation time-frame. Within the endangered data set of 450 species we found significant differences in the representation of the observed characters from that expected within sex systems and fruit types. We show that, depending on the life form, unisexual breeding systems can be significantly and positively associated with endangered species compared with non-threatened species. For example, there are more monoecious species than expected by chance among the tree species listed as endangered but fewer among the herbaceous life forms. Threat category was found to be non-randomly clustered in some clades.

Main conclusions  Life-history characters in certain combinations are predictive of extinction risk. Phylogeny is also an important component of extinction risk. We suggest that specific life-history characters could be used for conservation planning and as an early warning sign for detecting vulnerability in lists of species.