Vegetation type determines heterotrophic respiration in subalpine Australian ecosystems



    1. School of Biological, Earth and Environmental Sciences, University of New South Wales, Sydney, NSW 2052, Australia,
    2. Bushfire CRC, 340 Albert St, East Melbourne, Vic 3004, Australia,
    Search for more papers by this author

    1. Bushfire CRC, 340 Albert St, East Melbourne, Vic 3004, Australia,
    2. Faculty of Agriculture, Food and Natural Resources, University of Sydney, Sydney, NSW 2006, Australia
    Search for more papers by this author

Mark A. Adams, Faculty of Agriculture Food and Natural Resources, University of Sydney, Sydney, NSW 2006, Australia, tel. +61 2 9351 2935, fax +61 2 9351 2945, e-mail:


Soils are the largest store of carbon in the biosphere and cool-cold climate ecosystems are notable for their carbon-rich soils. Characterizing effects of future climates on soil-stored C is critical to elucidating feedbacks to changes in the atmospheric pool of CO2. Subalpine vegetation in south-eastern Australia is characterized by changes over short distances (scales of tens to hundreds of metres) in community phenotype (woodland, shrubland, grassland) and in species composition. Despite common geology and only slight changes in landscape position, we measured striking differences in a range of soil properties and rates of respiration among three of the most common vegetation communities in subalpine Australian ecosystems. Rates of heterotrophic respiration in bulk soil were fastest in the woodland community with a shrub understorey, slowest in the grassland, and intermediate in woodland with grass understorey. Respiration rates in surface soils were 2.3 times those at depth in soils from woodland with shrub understorey. Surface soil respiration in woodlands with grass understorey and in grasslands was about 3.5 times that at greater depth. Both Arrhenius and simple exponential models fitted the data well. Temperature sensitivity (Q10) varied and depended on the model used as well as community type and soil depth – highlighting difficulties associated with calculating and interpreting Q10. Distributions of communities in these subalpine areas are dynamic and respond over relatively short time-frames (decades) to changes in fire regime and, possibly, to changes in climate. Shifts in boundaries among communities and possible changes in species composition as a result of both direct and indirect (e.g. via fire regime) climatic effects will significantly alter rates of respiration through plant-mediated changes in soil chemistry. Models of future carbon cycles need to take into account changes in soil chemistry and rates of respiration driven by changes in vegetation as well as those that are temperature- and moisture-driven.


Rising temperatures are likely to significantly increase rates of soil respiration, with the consequent effect of a significant and positive feedback to atmospheric carbon dioxide (CO2). CO2 efflux via soil respiration (Davidson & Janssens, 2006) is increasingly recognized as being strongly linked to plant metabolism and to recent contributions of plant carbon to soil via both aboveground and belowground inputs (Ryan & Law, 2005; Hartley et al., 2006).

Equally, there are major and well-known differences in the quantity and quality of organic matter inputs to soil that result from variation in species composition within community types (e.g. relative abundance of N-fixing species) or from wholesale changes in community phenotype (e.g. grassland, shrubland, woodland, forest) that may result from either management or variation in edaphic conditions at a local scale (e.g. Binkley & Menyailo, 2005; Hart et al., 2005). Despite knowledge that carbon inputs via litter decomposition and root exudates vary with species/community type, only recently has attention been focused on assessing the influence of species or vegetation type on soil respiration (e.g. Michelsen et al., 2004; Fang et al., 2005; Irvine et al., 2007). Indeed, Neff & Hooper (2002) concluded that differences in vegetation type were likely to be better predictors of soil organic matter (SOM) lability than indirect effects of climate on SOM quality and that landscape scale patterns of vegetation distribution will be an important component of modelling regional CO2 emissions.

Edaphic gradients that produce major changes in vegetation or community type have a long history of utilization in ecological studies (e.g. Giesler et al., 1998; Nordin et al., 2001; Toljander et al., 2006). The Betsele transect in Northern Sweden is a clear example of how significant variation in species composition, and/or vegetation type (dwarf shrub, low herb, tall herb) can be produced by short-range (90 m) variation in edaphic conditions (Giesler et al., 1998; Nordin et al., 2001). In the Australian high country there are similarly large variations in plant community types (grassland, shrubland, woodland) over short (100 m) distances (Costin, 1957) and these types form a mosaic across many of the gently undulating landscapes that dominate moderately large areas above 1200 m. In contrast to the Betsele transect where ground water discharge proved the dominant effect on vegetation pattern, the mosaic of vegetation types in the Australian high country reflects a combination of environmental conditions (Costin et al., 2000), but especially the pooling of cold air at night in depressions and valleys (Williams & Ashton, 1987). These patterns in vegetation must also be viewed through the ‘prism’ of fire (Gill, 1981) since this area has burnt many times, most recently in 2003 and 2006/2007. The role of fire in changing patterns of dominance – from grasses to woody shrubs/trees in the absence of fire – is increasingly well recognized in Australia and worldwide (Moore & Williams, 1976; Leigh et al., 1991; Bond et al., 2005).

We hypothesized that the mosaic of vegetation types in the Australian high country would carry with it significant variation in rates of respiration. The importance of this hypothesis is emphasized by the nature of current vegetation patterns. Much of the high country, especially that below the current snowline (∼1450 m), may well be routinely without snow within 20–30 years if current rates of increase in mean annual temperature (MAT) (e.g. Hennessy et al., 2003) are maintained. The absence of snow cover will increase diurnal temperature ranges in winter, and the growth and carbon balance of many plant species is particularly sensitive to night-time temperatures. Likewise, current models predict rainfall is likely to decrease in coming decades in much of the high country. The combined increase in MAT and reduction in rainfall is widely predicted to increase fire frequency and severity (e.g. Lucas et al., 2007). Clearly, both direct changes in climate (e.g. temperature, rainfall) and atmospheric CO2, as well as indirect effects (e.g. fire frequency) are likely to have profound influences on the vegetation mosaic of the high country with concomitant changes in primary production, carbon (C) allocation, rooting depth and nutrient and C cycling (Jackson et al., 1996, 2002).

Changes to vegetation patterns and C cycling in cold climate areas are already being reported from the northern hemisphere. For example, encroachment of shrubs onto areas formerly dominated by tundra (Sturm et al., 2001, 2005; Tape et al., 2006) is associated with changes to albedo, fire regimes, and biogeochemical cycles. The Australian alpine and subalpine regions have been highlighted as potential indicators of climatic change (Hughes, 2003) owing to their tenuous snow-cover. Understanding current relationships between vegetation types and soil respiration are crucial to our ability to predict effects of future changes in climate on global C cycles, especially potentially damaging positive feedbacks.

The temperature sensitivity of soil respiration is well known (Lloyd & Taylor, 1994) and Q10 approaches have been widely used to describe that sensitivity. Difficulties arise when comparing Q10 among studies because of a lack of general consensus in the approaches used (e.g. temperature range) in its calculation. More recently, attention has shifted to re-examination of the models used to describe temperature sensitivity (Lloyd & Taylor, 1994; Hanson et al., 2000; Fang & Moncrieff, 2001). Exponential and Arrhenius models are common (Davidson et al., 2006), especially with data derived from laboratory incubations (e.g. Dalias et al., 2001; Fang & Moncrieff, 2001, 2005; Fang et al., 2005; Rey & Jarvis, 2006).

We applied exponential and Arrhenius models to study variation in soil respiration across the vegetation mosaic found in the high country in south-eastern Australia. Despite their omission from many studies, we included analysis of soils sampled at greater depths since we wished to characterize the depth gradient in respiration (Fang & Moncrieff, 2005). We used a novel gas exchange system to characterize the temperature response of soil respiration across vegetation types and soil depths.

Materials and methods

Field sites

Study sites were located within the Snowy Plains region of the Kosciusko National Park, in NSW, Australia (36°11′S, 148°51′E) at altitudes ranging from 1434 to 1653 m above sea level. The mean annual average minimum and maximum temperature are 4.0 °C and 11.5 °C, respectively, and the average annual precipitation is 1697 mm. Snow cover lasts for approximately 3 months each year, from July to September. The soils developed on Silurian Mowomba granodiorite in all instances (approximately 433±1.5 million years). Soils are best described as Alpine Humus (Costin et al., 1952), and have no distinct boundaries between horizons.

We used an approach similar to that of Högberg and colleagues (Giesler et al., 1998; Nordin et al., 2001; Hoffland et al., 2003; Högberg et al., 2003) in using as the basis of our study, clear and obvious changes in species composition and community type (grassland, woodland with grass understorey, woodland with shrub understorey) over short horizontal and vertical distances (∼100 m; Fig. 1). These short-range gradients allow direct examination of the influence of vegetation on soil properties. Our assignment of community types followed the classification of Costin (1957):

Figure 1.

 Vegetation profile for the Snowy Plains, NSW. Five transects, each approximately 100 m in length were sampled. Starting (lowest) altitudes ranged from 1434 m above sea level, while the final (highest) elevation sampled was 1653 m.

  • 1Sod tussock grassland (G): dominated by Poa spp.
  • 2Subalpine woodland (SGG): Eucalyptus pauciflora (snow gum) woodland typically with a herbaceous understorey dominated by grasses, mainly Poa spp.
  • 3Subalpine woodland (SGS): E. pauciflora (snow gum) woodland typically with an understorey dominated by shrubs (especially the N-fixing Bossiaea foliosa, as well as Leucopogon spp. and Tasmannia xerophila).

A characteristic of this region, and common to subalpine Australia, is the occurrence of an inverted tree line (Fig. 1). In brief, above around 1200 m, Poa spp. dominated grasslands are extensive on gentler slopes in valley bottoms (Williams, 1987), while E. pauciflora (snow gum) woodlands tend to dominate upper slopes and ridges until the tree line is reached at around 1700 m.

Soil sampling

During late February and early March in 2007, soils were sampled in the three vegetation types; SGS, SGG and G. Soils were collected at three soil depths; 0–10 cm with 50 mm soil cores and 10–30 cm and 30–50 cm with a 50 mm soil auger. Each vegetation type was replicated via five separate transects. We established a single 100 m2 sampling plot for each vegetation type along each transect and collected at least three (and up to 10) samples for each soil depth in each plot. Samples were bulked by plot, sieved to 2 mm and roots removed in the field. Soils were then stored at <5 °C until analysis.

Soil chemical analysis

Gravimetric water content of the soils was determined (oven-dried for 24 h at 105 °C). Subsamples of each soil were used for the determination of pH (1 : 2, soil : water) and total nitrogen and carbon (using combustion analysis, LECO CHN2000, St Joseph, MI, USA). Oxidizable organic carbon was determined using a dichromate digestion procedure (Walkley & Black, 1934). Indices of available phosphorus (Bray #1, Bray #2 and Olsen P) were measured as described by Kuo (1996). Microbial biomass C was determined using the fumigation-extraction method (Horwath & Paul, 1994; Ohlinger, 1995; Michelsen et al., 2004). Labile C was determined by K2SO4 extraction of unfumigated soil samples.

Samples were fumigated for 48 h, and carbon content was determined using Mn (III) assay (Bartlett & Ross, 1988) at an absorbance of 495 nm. Cmic was then determined assuming a kec of 0.35.

Soil respiration measurements

A subsample of approximately 100 g fresh weight soil was used for respiration measurements using a novel, flow-through, gas monitoring system (Fig. 2.). This involved specifically designed, polycarbonate incubation cylinders (approximately 200 mm length, 44.6 mm ID), fitted with foam inserts to hold soil in place (polyester-urethane reticulated filter foam, S28/60R and S28/45R, 19 mm width, 37 mm diameter), and capped at each end with Teflon bungs containing luer-lock ports. Once cylinders were packed with soils, up to 48 could be incubated simultaneously and monitored sequentially for CO2 (and O2) using an infra-red gas analyser (IRGA) and differential oxygen analyser (DOX, Qubit Systems Inc., Kingston, Ont, Canada), coupled with gas multiplexing units. Soil-filled incubation cylinders were connected to the multiplexed gas exchange system using PVC tubing (ID 3 mm). A source of compressed air was divided into two: one reference and one sample stream. Both reference and sample streams were then monitored continuously using the IRGA and DOX. Magnesium perchlorate columns were attached to both reference and sample gas streams before entering the DOX and IRGA to ensure water vapour did not interfere with measurements. Before respiration measurements, packed incubation cylinders were equilibrated overnight at 20 °C, to reduce effects of disturbance during preparation. Respiration measurements were made at a range of temperatures over a 10-day period with the incubation cylinders being returned to 20 °C overnight between measurement temperatures. The sequence of temperatures used for measurements was: 20–25–30–35–40–30–20–10–5–20 °C. During respiration measurements, temperature was controlled using a water bath and soils were allowed to equilibrate to water bath temperature (30–60 min) before respiration measurements began. Soil chambers were flushed with air before respiration measurement using at least five cylinder volumes of air (250 mL min−1). Respiration rates were determined on the basis of averaged CO2 concentrations over 60 s, after the concentration had remained steady for 3 m. Soils moisture was maintained by humidifiers fitted in-line with each cylinder. We monitored soil moisture content gravimetrically for the duration of the experiment to ensure field moisture content remained constant (±10%).

Figure 2.

 Schematic of flow-through gas exchange monitoring system used for soil heterotrophic respiration measurements, comprised of infrared gas analyser (IRGA) for CO2 efflux and differential oxygen analyser (DOX) for O2 uptake.

Statistical analysis

Statistical analyses were performed using minitab 12.1 and curve fitting by sigmaplot 9.0. Soil chemical data were analysed by one-way analysis of variance (anova) by soil depth with Tukey's HSD pair-wise comparison (family error rate at 0.05) and Pearson's correlation were conducted for all soils, for soil chemistry and respiration fluxes at 20 °C.

Regression analysis was conducted using linearized data for both Arrhenius (soil respiration vs. 1/T) and exponential functions (ln soil respiration vs. T); equations are shown below for soil respiration measures over seven temperature points.



where y is the actual respiration rate (μg CO2-C g−1 dwt soil h−1), A and Ea are constants related to the temperature sensitivity of the respiration rate, R is the universal gas constant (8.314 J K−1 mol−1) and T is temperature (K).



where a and b are rate constants and T is temperature in kelvin (K).

The sensitivity of temperature dependence (Q10) was calculated from derived parameter from both Arrhenius and Exponential function accordingly:






Soil chemistry

Across all vegetation types and depths, there were strong and consistent patterns in concentrations of C, N and P. Soils from the snow gum with a shrub understorey (SGS) community were consistently richer in C and N and available P, at all depths, than soils from either the snow gum with grass understorey (SGG) or grassland (G), (Table 1). Resultant C : N ratios were also greater in the SGS soils, with the exception of the surface 10 cm. Concentrations of available P differed strongly in the surface soils with vegetation type, while concentrations were reduced in the lower depths of the soil profile and no significant difference in the available P indices were observed. In the 0–10 cm soil depth concentrations of Bray #1 and Bray #2 P in SGG soils were around twice those in G soils and in SGS soils were twice again those in SGG soils. Labile carbon (K2SO4 extractable) concentrations in soils from the two snow gum vegetation types were similar, and around twice those in G soils. As might be expected, microbial biomass C followed a strongly biophilic (meaning similar to the distribution of biological activity; Attiwill & Leeper, 1987) distribution with depth in all vegetation types. Concentrations were notably greater in the surface horizons under the G and SGG, although no significant difference in microbial biomass at each depth was detected.

Table 1.   General soil characteristics for three vegetation types at each soil depth, mean (n=5) with standard errors indicated in parentheses
depth (cm)
pH%C%NC : NBray# 1 P
(μg P g−1 soil)
Bray# 2 P
(μg P g−1 soil)
Microbial biomass
CHCl3 extractable
(μg C g−1 soil)
Labile carbon
K2SO4 extractable
C (μg C g−1 soil)
  1. For each soil depth, vegetation type means followed by the same letter are not significantly different (Tukey's HSD family error rate 0.05).

Snow gum+shrub understorey (SGS)
 0–105.3a8.24 (0.43)a0.80 (0.05)a12.614.8 (1.6)a43.9 (1.7)a195.1 (57.6)a119.4 (12.9)a
 10–305.2a6.69 (0.04)a0.55 (0.02)a17.02.5 (0.3)a22.8 (3.8)a177.2 (48.5)a127.7 (4.2)a
 30–505.2a5.18 (0.26)a0.37 (0.01)a18.32.1 (0.3)a20.2 (3.6)a104.7 (24.1)a109.2 (6.5)a
Snow gum+grass understorey (SGG)
 0–105.2a6.76 (0.28)b0.70 (0.03)a13.68.0 (2.7)ab27.0 (7.0)b287.4(45.6)a99.5 (7.2)a
 10–305.2a5.21 (0.37)b0.46 (0.03)a14.41.4 (0.2)a10.4 (3.5)a228.4 (36.2)a90.7 (5.7)b
 30–505.2a3.51 (0.32)b0.30 (0.03)a14.81.1 (0.2)a7.5 (1.9)a82.68 (17.4)a74.4 (4.0)a
Grassland (G)
 0–105.5a4.37 (0.08)c0.42 (0.02)b13.94.3 (0.5)b10.9 (2.5)b273.9(32.1)a57.1 (5.4)b
 10–305.4a3.49 (0.12)c0.29 (0.03)b15.31.8 (0.5)a9.9 (5.0)a87.2 (26.2)a58.4 (4.4)c
 30–505.4b2.16 (0.14)c0.18 (0.01)b13.32.7 (1.5)a16.9 (7.9)a67.8 (16.4)a44.9 (7.5)a


Rates of respiration increased with temperature across all vegetation types (Fig. 3), from 5 °C (278 K) to 40 °C (313 K). At 40 °C, soils from SGS produced CO2 at rates roughly similar to those recorded for soils from SGG but at more than twice the rate recorded for soils from G sites; a clear example of the effect of vegetation type on soil respiration. Relationships between soil respiration and temperature were closely approximated by both Arrhenius and exponential functions (Table 2). All fitted regressions were highly significant (P<0.01) and the r2 values indicate the strong fit of both functions (Arrhenius or exponential). Rates of respiration were also closely related to soil depth. Rates were greatest for samples from the top 10 cm of the soil profile (Fig. 4) and declined strongly with increasing depth. For example, rates of respiration in surface soils from SGG and G communities were 3.5 times greater in the top 10 cm than at 30–50 cm soil depth. In the SGS community, the equivalent ratio was 2.3. There were strong relationships between respiration and temperature for all soil depths, all r2 values were >0.94 (P<0.001; Table 2).

Figure 3.

 Relationship between soil respiration and temperature for the three subalpine vegetation types (0–10 cm soil depth). Error bars are standard error of the mean (n=5). Arrhenius curves are fitted to the data (parameters are shown in Table 2) as indicated by the dashed lines.•, snow gum with shrub understorey (SGS), ▽, snow gum with grass understorey (SGG) and ▪, grassland (G).

Table 2.   Fitted parameters describing Arrhenius [Eqn (1)] and two-parameter exponential functions [Eqn (2)] for soil respiration measured over seven temperature points
  1. Regression analysis was conducted using linearised data for both Arrhenius (soil respiration vs. 1/T) and exponential functions (ln soil respiration vs. T) for all reported data. All regressions were significant at P<0.001.

Snow gum+shrub understorey (SGS)
 0–10e26.96.11 × 1040.951.15 × 10−100.080.94
 10–30e27.96.46 × 1041.002.63 × 10−110.091.00
 30–50e28.86.78 × 1040.992.69 × 10−120.090.99
Snow gum+grass understorey (SGG)
 0–10e28.96.63 × 1040.991.07 × 10−110.090.98
 10–30e29.16.83 × 1040.992.45 × 10−120.100.99
 30–50e34.58.34 × 1040.952.95 × 10−150.110.94
Grassland (G)
 0–10e26.46.14 × 1041.004.71 × 10−110.081.00
 10–30e35.88.70 × 1040.966.25 × 10−160.120.95
 30–50e26.66.55 × 1040.962.52 × 10−120.090.95
Figure 4.

 Arrhenius functions fitted (as dashed lines) to actual soil respiration and temperature for the Vegetation types, snow gum with a shrub understorey (SGS), snow gum with a grass understorey (SGG) and grassland (G), for the three soils depths: •, 0–10 cm; ▽, 10–30 cm and ▪, 30–50 cm. Error bars are standard error of the mean (n=5), parameters for the Arrhenius functions are shown in Table 2.

Across all communities and soil depths, rates of respiration were also closely related to total nitrogen (r2=0.904, P<0.001) and total carbon (r2=0.868, P<0.001; Table 3). Notably, there was a strong correlation between respiration at 20 °C and indices of plant available P (Bray #1, r2=0.795, P<0.001; Bray #2, r2=0.683, P<0.001; Table 3). Temperature sensitivities (Q10) of respiration (Table 4), as derived from both Arrhenius and exponential equations, were in the range of 2.2–3.5. Irrespective of the model used, Q10 increased with soil depth across all sites, albeit peaking in the grassland soils at intermediate depth. Q10 derived from Arrhenius models show a slight decrease with increasing temperature [calculated using Eqn (3) at either 20 °C or 30 °C].

Table 3.   Pearson correlation matrix for all soils
 Respiration at 20°CTotal COrganic CTotal NBray #1 PBray #2 POlsen PMicrobial
  • *


  • **


  • ***


Total C0.868***       
Organic C0.817***0.957***      
Total N0.904***0.929***0.917***     
Bray#1 P0.795***0.641***0.565***0.732***    
Bray#2 P0.683***0.646***0.587***0.655***0.681***   
Olsen P0.779***0.770***0.737***0.807***0.784***0.860***  
Microbial biomass0.350*0.388**0.449**0.399**0.1400.1600.264 
Labile C0.592***0.795***0.811***0.713***0.379*0.444**0.542***0.110
Table 4.   Temperature dependence of soil respiration rates across vegetation types and depth
depth (cm)
Arrhenius equationTwo-parameter
Q10 at 20°CQ10 at 30°CQ10
  1. Q10 calculated according to Eqn (3) for Arrhenius, and Eqn (4) for two-parameter exponential model. Arrhenius Q10 was calculated at two temperature points, 20°C and 30°C. Two-parameter exponential Q10 calculated using b [Eqn (1), Table 2].

Snow gum+shrub understorey (SGS)
Snow gum+grass understorey (SGG)
Grassland (G)


Despite their almost identical geographic and climatic locations, the different vegetation types at the Snowy Plains had a significant influence on soil chemistry. In addition to total C, concentrations of labile C, total N and both indices of available P, were all significantly greater in the woodlands (SGG and SGS). That plant species affect soil chemistry is well recognized worldwide (e.g. Binkley & Menyailo, 2005) but seldom considered by Australian ecologists who instead focus on the effects of soils on plant communities.

Worldwide, soil C generally increases as MAT decreases (Post et al., 1982) and cool/cold climate regions are characterized by their carbon-rich soils. For example, the cool/cold tundra ecosystems of the northern hemisphere contain a large proportion of the world's soil C (Hobbie et al., 2000). In large part, that soil-stored C reflects the balance between inputs from net primary production (NPP) and outputs due mostly to heterotrophic respiration (Pendall et al., 2004; Trumbore, 2006). The Australian high country should also reflect these general patterns – greater concentrations of soil C in the coldest parts of the landscape. However, the data reported here show the reverse pattern – soil C concentrations were greater in the warmer, upland snow gum woodlands and less in the valley bottom grasslands that are putatively maintained by cold air drainage (Williams & Ashton, 1987). There seem two most likely explanations. First, the balance of NPP and heterotrophic respiration could lie strongly with the former – lesser concentrations of C in grassland soils in the coolest parts of the landscape might be explained by reduced NPP relative to the woodlands at higher elevation, rather than reduced heterotrophic activity. Secondly, fire regime is one of the major influences on species and community distribution in the Australian high country and elsewhere (e.g. Bond et al., 2005) – grasslands and other herbaceous species seldom persist in fire's absence – and much of the present day distribution of plant communities in the Australian high country is undoubtedly dictated by past fire regimes. Rates of respiration in woodland soils (irrespective of understorey, SGS and SGG) were more than twice those in nearby grassland soils (G). Michelsen et al. (2004) also found major differences in rates of respiration between neighbouring plant communities (woodland and forest vs. grassland). They considered that losses of carbon during dry season fires in grasslands contributed to these differences and we conclude similarly, that maintenance of grasslands via a more frequent fire regime is a significant contributing factor to lower concentrations of carbon and slower rates of respiration.

Differences among vegetation types in physiology as well as phenology may play major roles in determining carbon turnover. For example, carbon-to-nitrogen ratios (C : N) are almost invariably and inversely related to rates of carbon turnover. In eucalypt forests and woodlands, the C : N ratio of litter is generally significantly greater than that of green foliage owing to resorption of N during leaf senescence (Attiwill & Leeper 1987). However in this study, the contribution of the nitrogen fixing shrub B. foliosa in the SGS vegetation type is clearly associated with a C : N of litter similar to that of green eucalypt leaves (Table 5). Total C and N concentrations in soils were strongly correlated with soil respiration at 20 °C, in keeping with known relations among these factors (Melillo et al., 1982). The greater carbon concentration in the soil beneath the SGS vegetation was thus better related to total biomass and productivity, and thus the amount of litter, than to litter C : N alone. The improved quality of litter for decomposition, undoubtedly contributes to carbon and nitrogen turnover in SGS soils. The hardy evergreen shrubs dominating the SGS vegetation type are deep-rooted (Wimbush & Costin, 1979), unlike the shallow (<30 cm) and fibrous rooted Poa spp. that dominate the grassland. Many native Australian legumes show considerable capacity to produce specialized roots and mobilize a range of forms of P (e.g. Adams et al., 2002), in keeping with the recent synthesis and framework for N-fixation and P availability proposed by Houlton et al. (2008). The N-fixing shrub B. foliosa may thus simultaneously improve both the N and the P status of the soils it occupies, and thereby serve to increase NPP and rates of carbon turnover, including rates of heterotrophic respiration.

Table 5.   Mean C : N for plant samples collected from the snow gum woodland (SGS; SGG) and grassland vegetation types (G), each value is a mean of three plots and samples for each plot were a composite of at least nine individual plants
 C : N
  1. Included are the dominant species: E. pauciflora, B. foliosa, Lecopogon spp. and Poa spp. After Jenkins (2009).

Bossiaea foliosa leaves29.5
B. foliosa stems43.8
Eucalyptus pauciflora leaves (young)55.0
E. pauciflora leaves (old)48.4
Leucopogon spp. leaves46.7
Leucopogon spp. stems82.0
Litter (woodland)55.0
Litter (grassland)28.0

With the exception of landscape-scale fires, such as those seen in 1939, 2003 and 2006/2007, we lack the detailed knowledge of fire regime required to definitively attribute the causes of the differences in soil C and rates of respiration among vegetation types. Significant circumstantial evidence – lower field temperatures, slower rates of respiration, valley bottom locations – suggests that soil C should be greater in grasslands than in the woodlands, if fire were not involved. Most likely, fire regime and the differential effects of temperature on NPP and heterotrophic respiration, are both involved. Knowledge that fire has direct effects such as volatilization of soil organic C, as well as its indirect effects on distribution of species with contrasting phenotypes (grasses vs. woody shrubs and trees), makes extremely difficult any prediction of future pools of soil C and rates of CO2 efflux to the atmosphere. The need for further research in the area of climate and fire interactions i.e. longer, hotter drier summers increasing bushfire incidence, and their impact on ecosystem C fluxes, was recently noted by Attiwill & Adams (2008).

In general, our measure of the pool of labile C correlated well with that of total carbon and both pools reflected differences in rates of respiration among community types. On the other hand, the carbon content of the microbial biomass was generally greater in soils from the G and SGG sites. Our findings support those of Lipson (2007) who found that soils from alpine dry meadows and adjacent subalpine forests had strikingly different responses to temperature. Lipson's data suggested that microbial communities in alpine meadows were cold-adapted throughout the year, whereas their counterparts in subalpine forest soils changed their temperature responses over the course of a year. In our example from the Australian high country, the consistently colder nocturnal temperatures in grassland communities at lower landscape positions (e.g. Williams & Ashton, 1987) would again be expected to produce a thermally adapted microbial flora.

Inputs of C via leaf litter and root exudates is concentrated in upper soil layers and it is not surprising that heterotrophic respiration was fastest in the upper soil layers (0–10 cm). Respiration rates were well correlated with carbon fractions (including total carbon, organic carbon, and labile carbon) and followed a strongly biophilic pattern in all vegetation types. Our data support those of Fang & Moncrieff (2005) that showed decreasing soil respiration and carbon with depth in the soil profile, and of Michelsen et al. (2004) that showed decreasing carbon concentrations with depth in woodland and forest vegetation but little change with depth in grassland (that had relatively slow rates of soil respiration). Our study showed decreasing concentrations of C with depth in all vegetation types, albeit that the depth gradient was less in the grassland soils. The use of laboratory based systems for respiration measurements allows for direct and controlled comparisons of soils from a range of vegetation types and depths, but further field studies are required to validate likely responses of soils to changed climatic conditions.

Davidson & Janssens (2006) demonstrated the importance of substrate availability and complexity in interpreting Q10. Both Arrhenius and exponential models fitted well with our respiration data. While Arrhenius-derived Q10 should allow for more flexibility in calculation and analysis of low and high temperature responses of soil respiration by comparison with the more static Q10 produced by exponential functions, we observed little variation among the variably derived Q10 from the present study. Our Q10 are similar to those derived for other studies using laboratory incubations of soils (Fang & Moncrieff, 2001; Reichstein et al., 2005; Rey & Jarvis, 2006). Similarly, our data support previous findings of decreasing respiration and carbon concentrations with increasing soil depth (e.g. Nadelhoffer et al., 1991; Fang & Moncrieff, 2005), but increasing Q10. Hence, while heterotrophic respiration shows a strongly biophilic pattern, there remains the clear possibility of a differential response to temperature among different soil horizons, as suggested by Davidson & Janssens (2006).


Alpine and subalpine areas of Australia have been emphasized as among those most likely to be affected by changing climates (Hughes, 2003). Increased summer temperatures will in all likelihood increase bushfire risk and aid expansion of deep-rooted and drought tolerant woody vegetation into areas currently dominated by shallow-rooted herbaceous species. Our data show striking differences in rates of respiration among soils collected from a range vegetation types that were geologically identical and varied only slightly in climate and topography. The temperature sensitivity of respiration varied little among community types but more strongly with soil depth. Q10 varied according to its method of calculation and the function fitted (i.e. Arrhenius or exponential; Table 2). Heterotrophic respiration showed a typically biophilic pattern. Past fire regimes and their effect on vegetation patterns are, in all likelihood, at least as important as current temperature regimes and their effect on both NPP and heterotrophic activity. Much further work is needed to elucidate the influence of carbon substrates, available phosphorus and microbial community composition on soil respiration in the Australian high country if reliable models of the effects of changing climates are to be prepared.


This work was funded by the Bushfire CRC. The authors would like to thank M. Webb, S. Roxburgh and M. Taranto for field and laboratory assistance and Department of Environment and Climate Change, NSW for permission to work within Kosciusko National Park.