Ecological opportunity and the evolution of habitat preferences in an arid-zone bird: implications for speciation in a climate-modified landscape

Abstract

Bioclimatic models are widely used to investigate the impacts of climate change on species distributions. Range shifts are expected to occur as species track their current climate niche yet the potential for exploitation of new ecological opportunities that may arise as ecosystems and communities remodel is rarely considered. Here we show that grasswrens of the Amytornis textilis-modestus complex responded to new ecological opportunities in Australia's arid biome through shifts in habitat preference following the development of chenopod shrublands during the late Plio-Pleistocene. We find evidence of spatially explicit responses to climatically driven landscape changes including changes in niche width and patterns of population growth. Conservation of structural and functional aspects of the ancestral niche appear to have facilitated recent habitat shifts, while demographic responses to late Pleistocene climate change provide evidence for the greater resilience of populations inhabiting the recently evolved chenopod shrubland communities. Similar responses could occur under future climate change in species exposed to novel ecological conditions, or those already occupying spatially heterogeneous landscapes. Mechanistic models that consider structural and functional aspects of the niche along with regional hydro-dynamics may be better predictors of future climate responses in Australia's arid biome than bioclimatic models alone.

Figure 1. Geographical distribution and habitat associations in the ATM complex.

The western subspecies of the Western Grasswren (Amytornis textilis textilis, ) occurred in a wide range of shrubland habitats, most commonly acacia or eucalypt shrublands communities, but is now restricted to the far north-west of its range near Shark Bay. The eastern subspecies (A. t. myall, ) is most commonly associated with dense chenopod shrubland (CS) in the southernmost sector of the Lake Eyre Basin. The Thick-billed Grasswren (A. modestus) inhabits dense CS of central Australia with the subspecies modestus () occupying drainages to the west of Lake Eyre, and inexpectatus () drainages to the east. Sample locations for the ATM complex ND2 sequences adapted from. Vegetation map prepared by JN using Adobe Illustrator CS6 software and adapted from Journal of Arid Environments 75, S. R. Morton et al., A fresh framework for the ecology of arid Australia, 313–329, 2011, with permissions from Elsevier.

Figure 2. Alternative models for the evolution of habitat preferences in the ATM complex.

Simplified phylogenetic framework based on analysis of mt ND2 sequences. The most parsimonious model (a) predicts that chenopod shrubland (CS) habitat preference (green) is the ancestral state with a recent transition to acacia-eucalypt shrubland (AES) habitat preference (red) in textilis. This model implies an increase in niche width (ecological release) in response to new ecological opportunity. The alternative model (b) predicts that AES habitat preference is the ancestral state with multiple transitions to CS under a model of constant migration; an early transition in Amytornis modestus with subsequent divergence of the subspecies modestus and inexpectatus, and a more recent transition in myall. This model implies a decrease in niche width and increased habitat specialisation in response to new ecological opportunity.

Figure 3. Temporally calibrated Bayesian maximum clade credibility trees based on ITS sequences for the Australian Camphorosmeae.

The node corresponding to the onset of diversification of the Australian clade (blue branches) is shown () and commenced at 6.1 Ma before present. Bayesian posterior probability for this node was 0.95. Recent speciation events are clustered during the period 0.2–2 Ma as indicated by the dashed line and indicate the likely time period for the formation of contemporary CS communities. * denotes the species Maireana pyramidata a dominant species in the CS communities occupied by the ATM complex.

Figure 4. Temporally calibrated Bayesian maximum clade credibility trees based on ITS sequences for Atriplex.

Australian clades are shown in blue. The nodes corresponding to the onset of diversification of the Australian clades are shown () and are dated to 4.55 (Clade 1) and 2.52 Ma (Clade 2) before present. Bayesian posterior probabilities for these nodes were 0.95 and 0.98 respectively. * denotes the species Atriplex nummularia a dominant species in the CS communities occupied by the ATM complex.

Figure 5. Time-calibrated phylogeny for the ATM complex and formation of chenopod shrubland habitats in central Australia.

The tree was generated using a Bayesian coalescent analysis of 41 unique mt ND2 sequences detailed in the Supplementary Data. The common ancestor of extant lineages in the ATM complex was estimated to have occurred during the interval 8.6–1.7 Ma with 95% confidence intervals (grey bars) estimated using an avian ND2 rate calibration. Chenopods diversified and increased in abundance during the Pliocene (light green shading) becoming widespread and abundant across inland Australia. The formation of contemporary CS habitat in central Australia (dark green shading) most likely occurred during the Pleistocene, and is associated with a shift in habitat preferences from AES to CS in modestus, inexpectatus and myall. The ancestral AES habitat preference is retained in textilis from western Australia. Outgroups occur in Triodia grasslands and sandhill canegrass habitats.

Figure 6. Correlations between paleoclimate change, the evolution of chenopod shrubland habitat, and the timing of shifts in habitat preference in the ATM complex.

Climate profile for Australia adapted from. The timing of events associated with the origin and diversification of the chenopods Atriplex and Camphorosmeae (a–e) were determined from Bayesian analysis of ITS sequences, and the fossil record of the Chenopodiaceae-Amaranthaceae alliance in Australia. Four phases in the evolution of chenopod shrubland (CS) habitats were identified: an initial phase marked by the origin of the Camphorosmeae (a) followed by a period of evolutionary stasis as climate returned to the warm-wet conditions typical of the Miocene; a second phase associated with the onset of diversification in Camphorosmeae (b) and colonisation by Atriplex clade 1 (c) as Pliocene aridification commenced; a third phase in which a dramatic increase in the abundance of chenopod fossils and the diversification of Atriplex clade 1(d) occurred, followed by the origin (e) and diversification (f) of Atriplex clade 2; a final phase associated with recent speciation events in Camphorosmeae (g) and Atriplex (h), indicating the formation of contemporary CS habitats, occurred during the Pleistocene as glacial climate oscillations intensified. The timing of diversification events in the ATM complex (i-iii) were determined from Bayesian analysis of mt ND2 sequences. The ancestor of the ATM complex arose during the Miocene (i) prior to the evolution of CS habitats. A shift from ancestral acacia-eucalypt shrubland (AES) habitat preference to CS is associated with the divergence of A. textilis and A. modestus (ii) during the early Pleistocene, with the divergence of subspecies (iii) occurring in the late Pleistocene. Figure prepared by JN using Adobe Illustrator CS6 software with the climate profile adapted from Molecular Ecology 17, M. Byrne et al., Birth of a biome: insights into the assembly and maintenance of the Australian arid zone biota, 4398-4417, 2008, with permission from John Wiley and Sons.

Figure 7. Bayesian skyline plots of late Pleistocene demographic responses in subspecies of the ATM complex.

Median estimate of log effective population size with 95% confidence intervals (shaded) are shown for inexpectatus (a), modestus (b), textilis (c) and myall (d). A late Pleistocene population decline is evident in textilis from AES commencing at the last glacial maximum approximately 25,000 ka. In subspecies inhabiting CS modest population growth (myall and modestus) or population expansion (inexpectatus) are indicated.