Introduction

Accurate data on growth rates of individual species are critical for: informing ecological restoration; estimating changes in tree biomass and timber production potential; facilitating species selection for seed source augmentation and ecosystem rehabilitation; and also for determining commercial cutting cycles, recovery time frames and silvicultural prescriptions (Korning and Balslev 1994; Silva et al. 2002; Bowman et al. 2013; Stephenson et al. 2014). Furthermore, in the last decade, tree planting for ecological restoration has been used as a greenhouse gas emissions offset (carbon farming) option by individuals, governments, and corporations in efforts to mitigate climate change (Bekessy and Wintle 2008; United Nations 2008; Booth and Williams 2012). Additionally, acquisition, protection and management of native regrowth forests complemented with the enrichment planting of native species has also been used as an offset option intended to mitigate biodiversity losses associated with mining and infrastructure development (Commonwealth of Australia 2012).

There are at least 27 forestry-based carbon offset providers in the Australian market (Low Carbon Australia and the University of Queensland 2011) and similar initiatives can be found worldwide with over 200 forestry-based carbon offset providers in the marketplace (The Ecosystem Marketplace 2011). Most carbon offset plantings in Australia use propagules of locally occurring native plant species with the aim of delivering multiple benefits that include carbon storage and restoration of biodiversity values of degraded ecosystems (Atyeo and Thackway 2009; Booth and Williams 2012; Booth et al. 2012).

Currently, estimation of the benefits and risks of all these activities is hampered by a dearth of knowledge on rates of growth of native tree species. There is a vital place for models of forest dynamics (Botkin 1993; Ngugi et al. 2000; Ngugi and Botkin 2011; Ngugi et al. 2011) in the prediction of tree and forest growth in different environments and under different management regimes. Forest growth models depend on models of individual tree development, and these in turn depend on a reliable knowledge of the rates of stem diameter growth and the associated changes in height, stem volume and tree biomass (Stephenson et al. 2014). Therefore, stem diameter increment data are fundamental to sustainable forest management and carbon farming initiatives.

Although we present analyses for Australian forests of native tree species, these findings have broad application because Australian native tree species are cultivated and studied globally. For example, there are over 20 million hectares of Eucalyptus plantings in many countries including Brazil, India, China, Uruguay, Chile, South Africa, Portugal, Spain, Vietnam, Sudan and Thailand (GFI Forestry Consulting 2009; Ferraz Filho et al. 2014; Wu et al. 2014). The successes of these initiatives continue to be limited by a dearth of information on growth rates and growth patterns of numerous Australian native tree species that have potential but are of limited natural occurrence and cultivation (Ngugi et al. 2003, 2004; Dale and Dieters 2007).

In this study, we investigated mean annual diameter at breast height (DBH) increments of tree species in native forests receiving about 500–2000 mm mean annual rainfall in subtropical Queensland, Australia. We used diameter measurements made over periods up to 75 years (1936–2011) on more than 86,400 trees in 641 permanent forest plots. Our specific objective was to determine absolute estimates and the pattern of variation in annual stem diameter increment of selected species across vegetation types, diameter classes and among rainfall zones.

Materials and methods

Study area and location of plots

Our study area covered about 2.6 million hectares of State owned uneven-aged mixed species native forests in southern Queensland, between latitudes 21° and 29°S and longitudes 146° and 154°E. The forests ranged in elevation from 14 m.a.s.l on Fraser Island to 900 m.a.s.l at Blackdown Tableland National Park in central Queensland (Fig. 1). The regional climate is subtropical to semi-arid, characterised by hot humid summers (December–February) and cool, dry winters (June–August). Mean annual rainfall within the study area varies from 500 to 2000 mm. These subtropical forests in Queensland range from tall moist forests along the eastern coast (e.g. Mapleton and Ringtail State Forests, and Bauple National Park) to dry woodlands of central Queensland (Pluto and Umbercollie State Forests) (Tables 1, 4). Spatial data for rainfall zones 500–600, 600–800, 800–1000, 1000–1200, 1200–1600 and 1600–2000 mm were provided by Australian Bureau of Meteorology (BOM) based on data collected between 1961 and 1990 (BOM 2012). Rainfall zones are identified here by the upper value.

Fig. 1
figure 1

Average annual rainfall isohyets and distribution and location of permanent forest plots (white circles) in State forests and National parks (black outlines) in southern and central Queensland, Australia. A legend for forest estate names and plot counts is provided in Table 4

Full size image
Table 1 Brief broad vegetation group (BVG) description of forests represented by permanent plots in the study area, a count of number of plots located in each rainfall zone, and mean diameter increment and 95 % confidence interval (in italics) for the each BVG and within each rainfall zone
Full size table
Table 2 Mean diameter growth rates (cm yr−1) of selected tree species with >200 increment data values (n) across observed rainfall zones
Full size table
Table 3 Regressions of diameter increment (Di, cm yr−1) as a function of tree diameter (D, cm) for species and rainfall zones in which the regression coefficient of determination (r2) was >0.5
Full size table
Table 4 Assigned forest identification numbers (FID), gazetted area and number of native forest permanent plots located in each forest, shown in Fig. 1
Full size table

History of management of native forest permanent plots

In the past 100 years these forests have been used for multiple purposes, including timber production and conservation of biological diversity (Queensland Government 1959) They are now held under various tenures, including Conservation Parks (3506 ha), Forest Reserves (120,711 ha), National Parks (536,387 ha) and State Forests (1,949,200 ha). The management goals for State forests in Queensland, as in all States and Territories in Australia, include perpetual protection of watersheds and biodiversity, and the supply of forest products (Queensland Government 1959).

Silvicultural treatments such as thinning of non-commercial trees and selective logging occurred during the history of these forests. Logging practices have included selective harvesting on 20–30-year cutting cycles for sawn timber, poles, girders, railway sleepers and fencing. Silvicultural guidelines determined the stem form, spacing and animal habitat quality of trees to be retained. In the lower rainfall zones (500–1200 mm), commercial harvesting usually used the single tree selection method, and in the wetter zones (1600 mm and above) group selection was more commonly practiced (Florence 1996). Prescribed burning and cattle grazing have been part of the fire risk management strategy in the eucalypt forests, but fire has generally been purposely excluded from the fire-sensitive Callitris forest communities. Since 1995, these guidelines have been consistent with the principles of the Montreal Process (Montreal Process 2009).

Broad vegetation classification of native forests

The Queensland Herbarium is the agency now primarily responsible for mapping and classification of vegetation in the study area (Neldner et al. 2012). Their classification system is hierarchical, ranging from broad classes (~1:5 million scale) down to regional ecosystems (RE) suitable for mapping at about 1:100,000 scale. These RE are combinations of vegetation types which are consistently associated with particular combinations of geology, landform, soil and climate (Neldner et al. 2012). We used a classification of broad vegetation groups (BVG) with nominal scale of 1:2 million, which divided the sub-tropical forests in this study into 16 broad vegetation groups (BVGs), ranging from tall moist forests along the eastern coast to dry woodlands of central Queensland (Table 1). Each plot was individually assigned to a BVG based on the dominant tree species, basal area stock of dominant species, height and foliage projected cover of the forests recorded at the site (DERM 2012). Consequently, this vegetation classification provided a detailed methodology of aggregating plots on moisture availability, basal area, species combination and growth conditions. In other studies, stand basal area attribute, a component of this classification, has been used in growth modelling as a proxy for inter-tree competition (see Weiskittel et al. 2011; Prior and Bowman 2014).

Native forests permanent sample plot data

The tree growth data used in this study were collected from 641 long-term permanent inventory plots located in 118 forest estates (Fig. 1; Table 4). The earliest plots were established in 1936 at Braemar State Forest and the last batch in the late 1990s. All permanent plots established prior to 1993 were subjectively selected from a systematic grid network of temporary forest inventory plots, to cover the range of stocking, species composition and site quality of major commercial stands within forest types (Beetson 1992). Plots established after 1993 were based on forest stratification data, to cover under-represented species mixtures and geographic distributions (Cant and Mannes 1995). There were 609 rectangular plots, varying in size from 0.2 to 0.5 ha, within which every tree with DBH > 10 cm that was suitable for commercial wood production was individually numbered, tagged and measured, complying with an established measurement manual (Cant and Mannes 1995). In a separate study on Fraser Island (Applegate 1982), 32 circular plots were established, within which trees with DBH > 10 cm were measured in an inner circle of radius of 17.8 m and trees with DBH > 30 cm measured in an outer circle with a radius of 39.9 m.

The measurement protocol was amended in 1993 to include sampling of all tree species irrespective of their potential for timber production. This amendment was aimed at promoting recovery of species diversity in forests used for timber production. In addition, trees <10 cm DBH but >3 m tall and or DBH > 2 cm were sampled using small circular plots (≤5 m radius) located at the centre of each quarter of the rectangular plot. All plots irrespective of establishment year were re-measured every 2–10 years after their establishment, with additional measurements carried out after a logging, wildfire or silvicultural treatment. The total number of times a plot was monitored ranged from two to 18, with an average of five times per plot.

Data analyses

Data analysis was undertaken using R programming language (R Development Core Team 2011). Annual diameter increment was calculated as the change in individual tree diameter between two consecutive measures as a function of time period in years (time period ≥ 1 year). Filtering was done to remove gross typing and recording errors, and errors associated with shifts in height at which diameter was measured due to recorded development of deformities at DBH. Trees with a defect at 1.3 m (DBH measure point) were excluded during diameter increment calculations. The growth rates were calculated for each 10 cm diameter class and expressed as mean, and 95 % confidence interval (CI) about the mean. Relationships between tree diameter class (D) and diameter increment (Di) for individual species within a rainfall zone were explored graphically. Preliminary model fitting used different model forms (exponential, logarithmic, power function, polynomial and linear) and aggregations of data from several rainfall zones in an attempt to fit species-specific generalised models. Fitted models were compared using the coefficient of determination (r2). Models that best fitted the data and had r2 > 0.5 were retained.

Results

Growth rates of broad vegetation groups and species representation

The ranges of mean diameter increment for forest types among rainfall zones (0.18–0.27 cm yr−1) was less than that among BVGs (0.11–0.29 cm yr−1) (Table 1). The highest mean increment among BVGs (0.29 ± 0.18 cm yr−1) was observed in BVG 8b, moist open-forests to tall open-forests dominated by Eucalyptus pilularis on coastal sands, sub-coastal sandstones and basalt ranges. For all BVGs, the 95 % confidence interval was about 57 % of the mean increment (data not shown).

Mean diameter growth rates and 95 % confidence intervals of 23 of the most abundant tree species are presented in Table 2. The most abundant data were available for C. glaucophylla (84,517 increment values), C. citriodora subsp. variegata (18,505 increment values), and E. crebra (16,409 increment values). Mean stem density for trees >10 cm DBH ranged from 287 to 399 stems ha−1 but without any consistent relationship to rainfall. Mean stand basal areas for each rainfall zone, calculated across the entire sampling period, ranged from approximately 7.5 m2 ha−1 (600 mm rainfall zone) to 29 m2 ha−1 (1600 mm rainfall zone) (Fig. 2). The relatively low stand basal area (15 m2 ha−1) in the 1200 mm rainfall zone reflects the extent of commercial utilisation in this forest zone.

Fig. 2
figure 2

Mean stand basal area for rainfall zones calculated from all measurement occasions on all plots. Rainfall zones (range of mean annual total): 600 mm (500–600 mm); 800 mm (600–800 mm); 1000 mm, (800–1000 mm); 1200 mm, (1000–1200 mm); 1600 mm, (1200–1600 mm); 2000 mm, (1600–2000 mm)

Full size image

Species growth rates across diameter size classes within rainfall zones

Comparisons of growth rates across diameter classes for species with the highest number of annual increment determinations within each rainfall zone are presented in Fig. 3. Within the 600 mm zone, growth rates of E. populnea, E. crebra and C. glaucophylla remained relatively constant (between 0.15 and 0.3 cm yr−1) across the observed diameter range while E. melanophloia showed a maximum growth rate (0.34 cm yr−1) in the 30–40 cm diameter class (Fig. 3a). In the 800 mm rainfall zone, the growth rates of all frequent species were relatively consistent in stems up to 50 cm DBH but were more variable (0.04–0.49 cm yr−1 in larger stems (Fig. 3b). Growth rates for the frequent species within the 1000 mm zone ranged from 0.13 to 0.41 cm yr−1 (Fig. 3c). Within species, there was no consistent pattern of variation in diameter increment with DBH but E. moluccana showed generally higher increments than E. siderophloia and E. propinqua.

Fig. 3
figure 3

Comparative species diameter increments within each rainfall zone as defined in Fig. 2

Full size image

Diameter growth rates for trees in the 1200 mm zone ranged between 0.2 and 0.52 cm yr−1 for trees in the 10–70 cm DBH range, with generally consistent and higher rates in E. microcorys and in E. sphaerocarpa, a species confined to Blackdown Tableland National Park (Fig. 3d). Highest growth rates in other species were generally observed in the 30–50 cm diameter range. There was far greater variation in growth rate among species in the 1600 mm rainfall zone (Fig. 3e). E. grandis and E. pilularis showed maximum growth rates between about 0.8 and 0.9 cm yr−1) in contrast to about 0.4 cm yr−1 in other frequent species in the zone. Highest growth rates were observed in the 30–70 cm DBH range. Great variability in diameter growth rates was also observed in the 2000 mm zone, where E. microcorys, E. racemosa and E. pilularis showed growth rates between 0.4 and 0.7 cm yr−1 within the 30–70 cm DBH range (Fig. 3f).

Diameter growth rates across rainfall zones

Over 45 species in our dataset occurred across at least three rainfall zones (Tables 2, 5). The growth rates among species and rainfall zones ranged from 0.01 to 0.96 cm yr−1, with an overall mean of 0.25 ± 0.13 cm yr−1. A sample comparison of diameter growth rates of 10 species that occurred across 800 mm to 2000 mm rainfall zones is presented in Fig. 4. Relatively constant growth rates across rainfall zones were observed for Corymbia trachyphloia subs. trachyphloia (0.12–0.16 cm yr−1) but for the other nine species, two patterns were generally observed. Eucalyptus siderophloia, E. propinqua, E. tereticornis, E. acmenoides and Lophostemon confertus demonstrated increased diameter growth rates (up to 0.27–0.35 cm yr−1), with increasing rainfall (Fig. 4a). Corymbia citriodora subsp. variegata, E. biturbinata, Corymbia intermedia and Angophora leiocarpa, demonstrated peak growth rates (0.21–0.31 cm yr−1) in the 1200 or 1600 rainfall zones (Fig. 4b).

Table 5 Mean diameter growth rates (cm yr−1) of species across rainfall zones
Full size table
Fig. 4
figure 4

Variation in diameter growth rate of individual species with diameter class across 800–2000 mm rainfall zones with a showing species that demonstrated highest growth at highest rainfall and b species that demonstrated a peak and a decrease in diameter growth rates

Full size image

Growth rate variation within diameter classes and across rainfall zones

Figure 5 illustrates the different patterns of variation in growth rate with diameter class for 10 common species that were distributed across several rainfall zones. There was little difference in diameter growth rates of Allocasuarina luehmannii, Callitris glaucophylla, Corymbia intermedia or Eucalyptus crebra within across rainfall zones and within diameter classes (Fig. 5). Eucalyptus acmenoides, E. fibrosa subsp. fibrosa, and E. siderophloia showed consistently lower growth rates in the lowest observed rainfall zones while Corymbia citriodora subsp. variegata and, E. fibrosa subsp. fibrosa showed higher growth rates in the highest rainfall zone of their occurrence. The variation in growth rates with diameter class and rainfall zone in Lophostemon confertus was inconsistent (Fig. 5).

Fig. 5
figure 5

Diameter increments across diameter size classes for ten species with distributions across several rainfall zones. Equations for the regression lines are presented in Table 3

Full size image

Regression analyses of diameter increment as a function of tree diameter are presented in Table 3 for the species illustrated in Fig. 5. Although most species were observed across several rainfall zones, only in a few rainfall zones were the coefficients of determination values (r2) > 0.5. Species with the highest r2 were Eucalyptus fibrosa subsp. fibrosa (r2 = 0.91), Corymbia citriodora subsp. variegata (r2 = 0.87) and Corymbia intermedia (r2 = 0.85) growing in the 1200 mm rainfall zone.

Discussion

Although the desirability of increasing forest carbon stocks in addressing climate change is well established (Bekessy and Wintle 2008; United Nations 2008; Booth and Williams 2012) and there are many estimates and methods for estimating carbon stocks (e.g. IPCC 2006), the rates at which these stocks accumulate in native forests are by comparison known imprecisely (Stephenson et al. 2014). Reliable empirical estimates of growth rates and patterns for 74 native tree species growing in natural forests in subtropical Australia were obtained from an extensive individual tree increment dataset (211,189 records). Growth rates and patterns varied markedly among broad vegetation groups and species in the observed rainfall zones and diameter size classes and the patterns were difficult to generalise. In the low rainfall zones (<800 mm) growth rates among species and across diameter classes were generally similar. Much greater fluctuations in growth rates were observed in the >1000 mm zones. Individual growth rates among all the species ranged from 0.01 to 0.96 cm yr−1.

Mean annual diameter growth rates of species within each rainfall zone

The generally similar growth rates among frequently occurring species in the low rainfall areas in contrast to the pronounced differences observed in higher rainfall areas suggests that potential growth differences among species are small under low rainfall conditions as compared to higher rainfall conditions. The highest growth rates for most species in all rainfall zones were observed in trees with DBH between 20 and 60 cm. In these native forests, trees in the <20 cm DBH class are often in the understorey and their growth is often suppressed by larger trees. Moreover, even though small trees may be numerous in a stand, their contribution to stand basal area and carbon stock is small compared with that of larger trees (Stephenson et al. 2014). Compared to other studies, the growth rates observed in our study were considerably lower than those of 1–5 cm yr−1 reported from 4 year-old plantation trial of eucalypts in the 1000 mm rainfall zone in southeast Queensland (Ngugi et al. 2004). However, unlike the natural growth conditions in our study, growth conditions under experimental trials or commercial plantations are modified through intensive ground preparation procedures involving deep ripping, fertilization and weed control (Ngugi et al. 2004) and the absence of competition for light, water and nutrients from established (old) trees.

The observed mean diameter increment for Corymbia citriodora subsp. variegata (0.25 ± 0.10 cm yr−1, n = 18,493) is consistent with results obtained for the species in southeast Queensland and northern New South Wales that ranged from 0.20 ± 0.03 in the lower rainfall areas, up to 0.29 ± 0.02 in the higher rainfall areas (Lewis et al. 2010). The mean individual tree growth rate range for all trees used in this study was 0.01–0.5 cm yr−1 for 94 % of the data, with an overall mean of 0.25 cm yr−1. This range is consistent with that reported for Eucalyptus forests in temperate Australia (Bowman et al. 2014) but the mean value is higher than that for woodland (0.20 cm yr−1) and for open forests (0.16 cm yr−1) reported in north Australia savanna in a two-year study (Prior et al. 2004). Our mean is also within the range of 0.2–0.37 cm yr−1 reported for subtropical forest trees in Puerto Rico (Brandeis 2009) and close to the 0.27 cm yr−1 reported for tropical lowland tree species in the Amazonian Ecuador where the mean annual precipitation was 3244 mm (Korning and Balslev 1994).

Tree plantings for carbon capture often target species with the highest growth prospect within a rainfall area and geographic location. However, because some species have a wide geographical distribution, it is often difficult to assess relative environmental requirements and growth potential (Booth et al. 2012). For example, many species observed in our dataset occur across several rainfall zones (between 600 and 2000 mm). Because there is limited knowledge on where such species are likely to exhibit their highest growth rates, the selection of fast growing species for carbon accumulation for a particular local area is difficult. Using a sample of ten species that occurred between the 800 and 2000 mm rainfall zones, we attempted to identify the zone in which each of the species recorded the highest growth rates. For example, although Corymbia intermedia is widespread, the growth increment recorded in the 1200 mm zone was approximately four times higher than that recorded in the 800 mm zone and about 55 % higher than that recorded in the 1600 and 2000 mm rainfall zones. Contrary to the fluctuating growth increment patterns observed for the other eight species, there was a consistent three-fold increase in growth rates of E. siderophloia and E. acmenoides with increase in rainfall from 800 to 2000 mm.

Growth rates of individual species across diameter classes among rainfall zones

Growth rates of C. glaucophylla were relatively independent of tree diameter for trees greater than 10 cm DBH, with growth rates of trees <10 cm DBH being about half that of the larger trees. A similar pattern was observed for Eucalyptus crebra growing between the 800 and 1000 mm zones. However, in several species, including C. citriodora subsp. variegata, Lophostemon confertus, E. acmenoides, E. siderophloia, Corymbia intermedia and E. fibrosa subsp. fibrosa, there was an increase in growth rate with increase in diameter up to a DBH of 50 cm, generally followed by a decrease in larger trees. The large differences in growth rates across diameter classes suggest that the use of an average increment value for these species should be adopted cautiously and only when more detailed descriptions of increment are not available.

Natural multi-age forests in Australia are characterised by a large number of species and varying growing conditions. This study did not document all possible tree species and growing conditions but instead used a dataset that was initially aimed at collecting data for the purpose of estimating timber production in State Forests in Queensland. As a result, commercial timber species are well represented in the dataset relative to those perceived to have limited timber value. Increment estimates for some non-commercial species are now being obtained as a result of the changes in 1993 to the plot measurement protocol. Some of these non-commercial species indicate growth rates much higher than the over-storey eucalypt species (e.g. 0.96 cm yr−1 for Acacia melanoxylon in the 2000 mm zone, n = 11; and 0.76 cm yr−1 for Angophora floribunda in the 1600 mm zone, n = 6). The accuracy of estimates for understorey and short-lived species is currently limited by small sample sizes but will be enhanced as subsequent re-measurements occur. Estimates of growth rates for large trees (>80 cm) should be used very cautiously because of measurement errors resulting from bole deformities associated with aging of predominantly habitat trees and bias related to their limited representation in the dataset.

Conclusions

Stem diameter increments for native species growing naturally in uneven-aged mixed species forests in subtropical Queensland typically range from 0.01 to 0.50 cm yr−1 (94 % of observed values). While tree growth rate is influenced by a range of genetic, edaphic and climatic factors, this study related increment to growth conditions (geology, landform, soil and rainfall), stem diameter (tree size) and rainfall, which are the most commonly recorded factors in a region where temperature variation is relatively small. In most species, increment is related to both rainfall and stem diameter class, but the relationships vary substantially between species and sometimes between rainfall zones. In some species, particularly from lower rainfall environments, increment does not vary consistently between either rainfall or stem diameter classes. For species that occurred across several rainfall zones, the highest growth rates were not necessarily observed in the highest rainfall zone. Within a rainfall zone, the highest growth rates within species were often observed in trees within the DBH range of 20–60 cm, following a unimodal curve within that range or with the increment either increasing linearly with stem diameter (often in lower rainfall zones and especially in Callitris glaucophylla and Eucalyptus crebra).

The dataset used for this study provided a consistent record of individual tree growth that avoids the errors and costs of repeated de novo assessments. However, many of the plots have not been remeasured in the last decade. With continued monitoring, the network of plots could provide a robust scientific basis for directing sustainable forest management and assist in enhancing our understanding on carbon storage, impacts of rising atmospheric CO2 concentration on tree growth, forest dynamics and improve accuracy of forest growth prediction tools.