Temporal and spatial patterns of tree mortality in a montane New Zealand mountain beech forest

Abstract

Tree mortality is a fundamental population process that, in part, controls the structure and dynamics of natural forests. Sources of mortality, such as competition and disturbance, are often size-dependent. It has been suggested that the size-specific mortality function for natural all-aged forests could be U-shaped, because mortality of small and large trees mainly results from different processes; for the smallest trees, mortality rates should be comparatively high because of the effects of size-asymmetric competition, whilst for the largest trees mortality rates should be high as a consequence of size-dependent disturbances, or senescence.

To examine patterns of size-related mortality, I used long-term permanent plot data from mountain beech (Nothofagus solandri var. cliffortioides) forest of the Harper, Avoca and Broken River catchments, on the eastern slopes of the Southern Alps, New Zealand, where the fates of more than 20,000 trees have been followed for >25 years. To determine whether the size-specific mortality function was U-shaped, size-specific mortality rates were calculated, and graphed. This was done at a range of spatial and temporal scales, for example separately for adjacent catchments and separately through periods of forest stability and decline. Patterns of size-specific mortality were then related to known information about the forests disturbance history.

To determine whether different processes were the major cause of mortality in small (<20.0 cm DBH) versus large (≥20.0 cm DBH) trees, individual-tree mortality models were constructed. Mortality probability was modelled as a function of tree size (DBH), using local crowding as a measure of neighbourhood interactions, and local change in basal area as a measure of disturbance.

Size-specific mortality functions were U-shaped when measured over large spatial scales or long (c. 25 y.) time periods, but disturbance caused variations over smaller spatial or temporal scales. Small trees (< 20.0 cm DBH) were more likely to die if they had many larger neighbours, suggesting that size-asymmetric competition (for light) was a major cause of mortality. Trees of all sizes were more likely to die if there was disturbance in their immediate neighbourhood, but this effect was most pronounced for large (≥20.0 cm DBH) trees. A key finding of this study was a shift from competitive neighbourhood interactions for small trees, to positive neighbourhood interactions for large trees. These positive neighbourhood interactions meant that large trees with many neighbours were less likely to die than those with few neighbours.

This study is unique, in that size-specific mortality functions were examined at a range of spatial and temporal scales, in conjunction with an individual based mortality model. The results emphasize the need to consider size-specific mortality rate patterns over large spatial or temp9ral scales. In addition, this study indicates that as well as competition and disturbance, positive neighbourhood interaction can also control the shape of size-specific mortality functions. Accounting for positive interactions between large trees may be essential for developing accurate models of forest dynamics.