Search

Aim Rainfall reliability has been neglected as a determinant of plant trait convergence and divergence in mediterranean-climate ecosystems. This paper reports on patterns of rainfall reliability — quantified as interannual variation in monthly and seasonal rainfall, and as the frequency of individual events in terms of their size, duration and intensity — for four fire-prone mediterranean-climate ecosystems. Location The four mediterranean-climate regions of the world with fire-prone ecosystems, namely SW Cape (South Africa), SW Australia, California and the Mediterranean Basin (Andalusia, Spain). Methods Using long-term monthly rainfall data from stations dispersed across the four regions, we computed monthly means and interannual variation for each month of the year — the latter quantified as the coefficient of variation (CV) — and divided these into winter and summer seasons. We also computed the mean number of rainfall events, the mean frequency in various categories of event duration (days), the amount of rainfall per event (mm) and the rainfall intensity per event (mm/day) per year for winter and summer seasons for a subset of the rainfall stations. Results The fraction of rain falling in summer was lowest in California (5%) and similarly low ( c. 25%) in the other three regions. The hierarchy of values of coefficient of variation (CV) of monthly rainfall during the winter period was as follows: California > Andalusia >> SW Cape > SW Australia; results for summer were: California > > Andalusia >> SW Australia ∼ SW Cape. SW Australian sites experienced the greatest frequency of short, small and low-intensity rainfall events in both seasons; patterns in the SW Cape were intermediate between Australia and the two northern hemisphere sites which both received fewer, larger and more intense events. Overall, the two southern hemisphere regions (SW Australia and the SW Cape) had significantly more reliable regimes than the two northern hemisphere ones (Mediterranean Basin and California). Main conclusions These differences in rainfall reliability regimes may provide a novel perspective on the distribution of certain plant life-history traits in mediterranean-climate ecosystems. Less reliable regimes would select for germination and seedling survival traits that enable persistence of genets in the face of uncertain moisture conditions during the winter and spring establishment phase. Study systems that accommodate for phylogenetic constraints, namely invasive species derived from mediterranean-climate ecosystems, as well as shared lineages, provide good opportunities to develop and test hypotheses on the implication of different rainfall reliability regimes. One of the novel implications of this study is that the distinctive trait of assemblages in the southern hemisphere regions may be a consequence not so much of their shared nutrient-poor soils as of their similarly reliable rainfall regimes. [ABSTRACT FROM AUTHOR]

ABSTRACT Aim We examined the century-long post-fire responses of reptiles to (1) determine the time-scales over which fauna - fire relationships occur, (2) assess the capacity of a conceptual model to predict faunal response to fire, and (3) investigate the degree to which models of fauna - fire relationships can predict species occurrence and are transferable across space. Location A 104,000 km2 area in the semi-arid Murray Mallee region of south-eastern Australia. Methods We surveyed reptiles at 280 sites across a century-long post-fire chronosequence. We developed generalized additive mixed models (GAMMs) of the relationship between time since fire and the occurrence of 17 species in two subregions, and compared modelled responses with predictions derived from the conceptual model. The predictive capacity of GAMMs was then assessed (1) within the subregion the model was developed and (2) when transferred into a novel subregion. Results Eleven species displayed a significant relationship with time-since-fire, with changes in species probability of occurrence continuing up to 100 years post-fire. Predictions of the timing of species post-fire peak in occurrence were accurate for 9 of 13 species models for which a significant fire response was detected, but little success was achieved in predicting the shape of a species' response. GAMMs predicted species occurrence more accurately when applied within the subregion in which they were developed than when transferred into a novel subregion, primarily due to some species responding to fire more strongly in one part of their geographic range. Main conclusions Fire influences the occurrence of reptiles in semi-arid ecosystems over century-long time frames. Habitat-use conceptual models have value in predicting the peak occurrence of species following fire, particularly for species with distributions strongly shaped by fire. Species relationships with fire can differ across their geographic range, probably associated with variation in climatic influences on post-fire succession and the consequent provision of habitat resources. [ABSTRACT FROM AUTHOR]