Demographic patterns of a widespread long-lived tree are associated with rainfall and disturbances along rainfall gradients in SE Australia

Abstract

Predicting species distributions with changing climate has often relied on climatic variables, but increasingly there is recognition that disturbance regimes should also be included in distribution models. We examined how changes in rainfall and disturbances along climatic gradients determined demographic patterns in a widespread and long-lived tree species, Callitris glaucophylla in SE Australia. We examined recruitment since 1950 in relation to annual (200-600 mm) and seasonal (summer, uniform, winter) rainfall gradients, edaphic factors (topography), and disturbance regimes (vertebrate grazing [tenure and species], fire). A switch from recruitment success to failure occurred at 405 mm mean annual rainfall, coincident with a change in grazing regime. Recruitment was lowest on farms with rabbits below 405 mm rainfall (mean = 0-0.89 cohorts) and highest on less-disturbed tenures with no rabbits above 405 mm rainfall (mean = 3.25 cohorts). Moderate levels of recruitment occurred where farms had no rabbits or less disturbed tenures had rabbits above and below 405 mm rainfall (mean = 1.71-1.77 cohorts). These results show that low annual rainfall and high levels of introduced grazing has led to aging, contracting populations, while higher annual rainfall with low levels of grazing has led to younger, expanding populations. This study demonstrates how demographic patterns vary with rainfall and spatial variations in disturbances, which are linked in complex ways to climatic gradients. Predicting changes in tree distribution with climate change requires knowledge of how rainfall and key disturbances (tenure, vertebrate grazing) will shift along climatic gradients.

Keywords: Callitris; climatic gradients; demography; disturbances; grazing; tree distribution.

Figure 1

A Callitris glaucophylla tree (center), which established in the 1890s surrounded by more recent recruits in north eastern New South Wales, Australia.

Figure 2

The location of transects used to survey Callitris glaucophylla recruitment in New South Wales, Australia. Transects follow gradients of declining mean annual rainfall (200–600 mm) from east to west and are located in seasonal rainfall zones from summer in the north, uniform in the middle to winter in the south. Sites where wood samples were taken for radiocarbon dating are marked from S1 to S6. Isohyets are labeled with the mean annual rainfall.

Figure 3

Recruitment periods of Callitris glaucophylla specimens, derived from radiocarbon dating versus mean annual rainfall in each seasonal rainfall zone (summer 1 = duplicate 1, summer 2 = duplicate 2, uniform, winter). Cohorts identified in the field are indicated by A, B, C, or D and represented by vertical lines. Cohort A is likely to be older than the dating indicates, as disks were taken at least 20 cm above the ground. Periods of La Niña are also shown.

Figure 4

Results from regression tree analyses showing the factors associated with the average number of cohorts since 1950, when mean annual rainfall was (A) included and (B) excluded. In (B) the number of sites less than (<) and greater than or equal to (≥) 405 mm mean annual rainfall is given below each branch. Seasonally dominant rainfall is in summer (S), uniform (U), and winter (W).

Figure 5

Curves of best fit (based on AIC) for the number of cohorts since 1950 versus mean annual rainfall in each seasonal rainfall zone.

Figure 6

Linear trends in an index of potential opportunities for Callitris glaucophylla regeneration in summer from 1950 to 2008 versus mean annual rainfall in each seasonal rainfall zone (summer, uniform, winter).

Figure 7

Proportion of sample sites with farms, grazing species, and fire versus mean annual rainfall (< and ≥405 mm) in each seasonal rainfall zone (summer, winter, uniform). Asterisks and pluses indicate significant differences between < 405 and ≥405 mm mean annual rainfall and seasons, respectively. As few sites were burned, fire was not statistically analyzed.