What factors influence the extent of midstorey development in Mountain Ash forests?

The midstorey is a critical component of the structure of many kinds of forest globally. We constructed statistical models of the factors influencing the percentage cover of two dominant Acacia spp. (Montane Wattle [Acacia frigiscens]) and Silver Wattle [Acacia dealbata]) in the midstorey of Mountain Ash (Eucalyptus regnans) forests in mainland south-eastern Australia. We modelled the influence on the percentage cover of two these two species of Acacia of : (1) the age of the overstorey eucalypts (which corresponded to the time elapsed since the last major stand-replacing disturbance), and (2) environmental drivers (slope, aspect, elevation, and topographic wetness).

Stand age was an important factor influencing the percentage cover of both Montane Wattle and Silver Wattle. We found evidence of a non-linear, humped-shaped percentage cover-stand age relationship for the percentage cover of Montane Wattle, with the highest values in stands of Mountain Ash that were 30-60 years old. There were no differences in percentage cover among other age classes. The highest values for the percentage cover of Silver Wattle were for stands regenerating after the 2009 fire with markedly lower levels of cover in other age classes. There were no differences in cover between other age classes. Although our data contained evidence of inter-specific differences between Montane Wattle and Silver Wattle in their response to stand age, both species persisted as a midstorey component in old growth Mountain Ash forest.

 No environmental covariates influenced the percentage cover of Montane Wattle or Silver Wattle. Both tree species occur well beyond our study region and the set of environmental conditions we modelled may therefore not be limiting the occurrence of these tree species. We suggest that disturbance is the key driver of site occurrence of the Montane Wattle and Silver Wattle in the Mountain Ash forests of the Central Highlands of Victoria.

Methods
Vegetation surveys

We completed detailed vegetation surveys at each of our 178 long-term sites. We measured the projective (percentage) foliage cover of each Acacia spp. across six 10 m x 10 m plots located at 20 m increments along a 100 m transect on each of our sites. We completed measurements of percentage cover in summer 2019-2020.

Environmental and other variables

We calculated values for a suite of covariates for each of our 178 field sites for subsequent use in constructing statistical models. We assigned the age of the forest at each site to one of five age classes: 1 = old-growth dominated by trees that germinated before 1900 (13 sites), 2 = 1939 regrowth (dominated by trees that regenerated as a result of the 1939 wildfires) (85 sites), 3 = 1960–1990s regrowth (i.e. trees that regenerated between 1960 and 1990) (15 sites), 4 = sites were regenerated after the 2009 wildfire (31 sites), and 5 = mixed-aged forest (in which there were two or more distinct age cohorts of trees in the stand) (31 sites). Our age class classification was based on the dominant age cohort of living overstorey trees in a stand. However, we note that the vast majority of the mixed-aged stands supported an old-growth component with a number of individual large old living trees.

We interrogated a 20 m resolution Digital Elevation Model to extract data on slope, aspect and elevation for the centroid of each site. We also calculated values for a Topographic Wetness Index (TWI) (Moore and Hutchinson 1991) for each site. TWI which is a measure of relative position in the landscape and thus potential water distribution. Calculation of TWI requires a Digital Elevation Model (DEM) that has hydrological integrity, and we used the ANUDEM algorithm (Hutchinson 2011) to generate a DEM of our study region at a grid resolution of 20 m. For each cell, the size of the catchment that flows to it was divided by its width, adjusted geometrically by the aspect of inflow direction. This ‘specific catchment’ was then divided by the cell’s local slope. Lower values indicate ridges and upper slopes that have little to no contributing catchment, with values increasing for lower slopes, valley bottoms, and drainage lines.

林中层（midstorey）是全球多数森林结构的关键组成部分。本研究构建了统计模型，用以解析影响澳大利亚东南部大陆区域山桉（Mountain Ash，学名*Eucalyptus regnans*）林分林中层内两种优势金合欢属（Acacia）物种——山地金合欢（Montane Wattle，*Acacia frigiscens*）与银叶金合欢（Silver Wattle，*Acacia dealbata*）——投影盖度的调控因子。我们针对这两种金合欢属物种的盖度，建模分析了两类因子的影响：（1）上层桉树的林分年龄（对应末次大规模林分更替型干扰以来的时间跨度）；（2）环境驱动因子，包括坡度、坡向、海拔及地形湿度。

立地年龄是影响山地金合欢与银叶金合欢盖度的重要因子。我们发现山地金合欢的盖度与林分年龄呈非线性的驼峰型关系，在树龄30~60年的山桉林分中盖度达到峰值，其余年龄组的盖度无显著差异。银叶金合欢的盖度峰值出现在2009年野火后更新的林分中，其余年龄组的盖度显著更低，且其余年龄组间盖度无差异。尽管数据显示两种金合欢属物种对林分年龄的响应存在种间差异，但二者均作为林中层组分持续存在于老龄山桉林分中。

未检测到环境协变量对山地金合欢或银叶金合欢的盖度产生显著影响。这两种乔木物种的自然分布范围远超本研究区域，因此本研究建模所涵盖的环境条件可能并未限制其种群发生。我们提出，干扰作用是维多利亚州中部高地山桉林分中山地金合欢与银叶金合欢林分发生的关键驱动因子。

## 研究方法
### 植被调查
我们在全部178个长期样地中开展了详细的植被调查。在每个样地内，沿100m样线以20m间隔设置6个10m×10m的样方，对每个金合欢属物种的投影枝叶盖度进行测定。调查于2019-2020年夏季完成。

### 环境及其他变量
我们为全部178个野外样地计算了一系列协变量，用于后续统计模型的构建。我们将每个样地的森林年龄划分为5个年龄组：1组为1900年前萌发的老龄林（共13个样地）；2组为1939年火烧后形成的再生林（以1939年野火更新的桉树为主，共85个样地）；3组为1960-1990年代更新的再生林（即1960至1990年间萌发的桉树，共15个样地）；4组为2009年野火后更新的样地（共31个样地）；5组为混合年龄林分（林分内存在两个及以上明确的年龄组群，共31个样地）。本研究的年龄组划分基于样地内上层活立木的优势年龄组群，但需说明，绝大多数混合年龄林分均包含老龄林组分，且存在多株大型活立木个体。

我们通过解析20m分辨率的数字高程模型（Digital Elevation Model, DEM），提取每个样地中心点的坡度、坡向与海拔数据。同时，我们为每个样地计算了地形湿度指数（Topographic Wetness Index, TWI，Moore与Hutchinson 1991）。TWI反映了景观中的相对位置，进而指示潜在的水分分布状况。TWI的计算需要具备水文完整性的数字高程模型，本研究采用ANUDEM算法（Hutchinson 2011）生成了研究区域20m网格分辨率的DEM。对于每个栅格单元，将汇水至该单元的集水区面积除以其宽度，并根据入流方向的坡向进行几何校正，得到“比集水面积”；随后将该值除以单元的局部坡度。TWI值较低的区域对应山脊与上坡位，这些区域几乎无汇水区域；而下坡位、河谷底部与排水线区域的TWI值则相对更高。