Variegated desert vegetation: Covariation of edaphic and fire variables provides a framework for understanding mulga‐spinifex coexistence

Abstract

Abstract Mulga (Acacia aneura Mimosaceae) and spinifex (Triodia spp. Poaceae) habitats together characterize a large part of arid central Australia. Often very abrupt boundaries form between these two habitats, giving rise to a mosaic pattern of contrasting shrub‐grass alterations across the landscape. Reasons for such patterning remain poorly understood though current niche‐based views relate species' distributions to spatial resource gradients or to fire effects. Field survey work was conducted on central Australian mountain ranges to further quantify floristic, regeneration traits, and structural patterning across mulga‐spinifex transitions and to test resource‐ and disturbance‐models that explain these patterns. Compositional analysis demonstrated variability in transition type – in certain cases boundaries denoted true floristic discontinuity and in others, somewhat more of a structural shift. Moreover, it was shown that minimal between‐habitat floristic overlap coincided with the occurrence of distinct edaphic changes, while greater compositional commonality occurred when soil gradients were more diffuse. This indicated that floristic patterning cannot be ascribed to any one single process. In the case of strong soil gradients, between‐habitat segregation most likely resulted from resource‐based niche differentiation; for weaker gradients, fire‐frequency assumed greatest importance. Disturbance theory most readily accounted for the distribution of woody species' post‐fire regeneration traits across habitat boundaries. The results also suggested that biotic factors –viz competition, facilitation and animal‐mediated dispersal – may be of additional consequence for mulga‐spinifex coexistence. Overall, the study served to emphasize the importance of multi‐factor explanation for within‐ and between‐habitat patterning in these mosaics. It also highlighted the need for experimentation to facilitate distinction between cause and correlation.