Printer Friendly

Carbon sequestration under subtropical perennial pastures I: overall trends.

Introduction

The introduction of perennial grasses, especially in lower and more seasonal rainfall zones, can provide production and environmental benefits (Sanford et al. 2003; Nichols et al. 2012). As such, many farmers have looked, or are looking, to convert some or all of their annual pastures to perennial based systems. In addition to these agronomic benefits, converting to a perennial-based pasture system may increase soil organic carbon (SOC) stocks. With the advent of the Australian government's Carbon Farming Initiative in July 2012, farmers now have a potential added incentive to convert some of their annual pastures over to perennials. However, there has been very little quantitative research exploring the SOC implications of converting to a perennial-based pasture system.

Perennial grasses, compared with annual grasses, generally allocate a greater fraction of productivity to the maintenance of a deeper and more extensive root system (Jackson and Roy 1986; Nie et al. 2008), including associated mycorrhizal fungi (Wilson and Hartnett 1998). Increased below-ground C allocation coupled with an increase in duration of C inputs and decreased surface erosion (Kort et al. 1998) due to greater grass cover through dry months, relative to annual grass pasture, should lead to increased soil C stocks. Respiratory losses of SOC may also be reduced under perennial pastures because off-season rains are effectively and rapidly utilised relative to typical summer fallow of annual grasses (Paydar et al. 2005). However, not all perennial grasses exhibit these growth forms and strategies that are favourable for building SOC stocks (Nie et al. 2008).

Analytical data to support these theoretical assertions remain limited. Young et al. (2009) found significant increases in SOC stocks (0.15-0.35 Mg C [ha.sup.-1] [year.sup.-1]) over 6 years of repeated-measurements when various perennial pastures were established on a formerly cultivated Vertosol in central New South Wales (NSW). In a replicated field trial, Chan et al. (2011) found no difference in sequestration rates to 30cm between annual and perennial [mixture of phalaris (Phalaris aquatica L.), cocksfoot (Dactylis glomerata L.), and lucerne (Medicago sativa L.)] pastures when established on a formerly rotationally cropped, acidic soil in NSW. Those authors attributed the lack of difference to similar levels of biomass production for the two pasture types (Li et al. 2006). Chan et al. (2010) also found no difference in SOC stocks between pairs of commercial annual- and perennial-based pasture systems in southern NSW and northern Victoria. Low replication (n=4) in this latter study may have prevented detection of differences given the natural variability in SOC across the landscape. In a warm temperate region of NSW with non-limiting irrigation, Neal et al. (2013) found that kikuyu (Pennisetum clandestinum Hochst. ex Chiov.) led to a significant increase in SOC after 3 years. Importantly, those authors found that temperate forages, despite matching the above-ground yield of kikuyu, did not change SOC levels.

The present study was designed to investigate whether sowing subtropical perennial grasses into temperate pasture systems leads to increased SOC levels. We have focussed on examining changes in SOC where the [C.sub.4] subtropical perennial grasses kikuyu and a mix of Gatton panic (Panicum maximum Jacq.) and Rhodes grass (Chloris gayana Kunth) have been sown into formerly [C.sub.3]-dominated pastures or cropping systems. By limiting this study to locations where [C.sub.3] to [C.sub.4] transitions have occurred, we can utilise stable carbon isotope measurements to distinguish how much of the SOC found under the perennial grasses is attributable to the new perennial grasses, because [C.sub.3] and [C.sub.4] plants have unique isotopic fingerprints (Cerling et al. 1989). Thus, isotope measurements yield an independent means to assess changes in SOC levels between annual- and perennial-based pastures that is less reliant on the quality of the perennial-annual pairing.

Here, we report on changes in SOC stocks and the accumulation of [C.sub.4]-derived SOC following the introduction of subtropical perennial grasses in several agronomically important regions across southern agricultural Australia. Known and potential sources of variability in the calculation of SOC change are then explored to build a case for the integrity of these findings. Finally, potential reasons for the variation in results between regions and species type are discussed. In the companion paper (Sanderman et al. 2013, this issue), we further explore reasons for regional differences in SOC sequestration rates by tracing the flow of new C into fractions of SOC.

Methodology

We have adopted a paired-site approach in this study. For each identified pasture where subtropical perennial grasses have replaced [C.sub.3] temperate vegetation, we also sampled a neighbouring [C.sub.3]-based system that was representative of the state of the perennial pasture at the time of conversion from temperate to subtropical perennial vegetation. Within each region, field sites were selected to span a range of perennial pasture ages, thus enabling us to adopt a chronosequence approach to comment on the multi-decadal changes that may be occurring in SOC. As the extent of use of subtropical perennial grasses is still fairly limited, we had to be opportunistic in our site selection. In each region we worked with local agronomists and/or farming systems groups to identify appropriate sites. To ensure that the paired sites were as similar as possible, all sites were inspected before sampling to ascertain similarity in soil type and management history.

Field locations

Forty-two locations were included in this project from five agricultural regions (Table 1):

(1) Fleurieu Peninsula, South Australia (SA). In this high-rainfall region, four kikuyu-based pastures were sampled in undulating terrain, with ages ranging from 3 to 45 years. Paired sites typically consisted of a mixed grass pasture dominated by perennial ryegrass (Lolium perenne). Soils were primarily high-fertility loams; however, site FP6 was a degraded sandy duplex soil. Dairy and beef cattle were the dominant land uses in this region.

(2) Kangaroo Island, SA. Five gently sloping pastures sown to kikuyu and ranging in age from 3 to 14 years were sampled, all on high-fertility sandy soils. Sheep grazing for meat and wool production dominates in this region.

(3) Northern Agricultural District of Western Australia (WA). Fourteen panic--Rhodes grass based perennial pastures were sampled, ranging from 3 to 18 years since establishment, on low-fertility sandy soils. Terrain ranged from flat plains to rolling hills. As with the southern district, set-stocking of sheep and cattle and low but variable fertiliser histories dominate in this region.

(4) Southern Agricultural District of WA. Fourteen kikuyu-based pastures on flat to gently sloping terrain were sampled, ranging in age from 3 to 33 years, all on low-fertility sandy soils. Set-stocking of sheep and cattle and low but variable fertiliser histories dominate in this region. Nine of the locations within this zone were sampled by the University of Western Australia as part of a broader study on SOC levels in WA (D. V. Murphy, F. C. Hoyle, A. Wherrett, D. Hall, J. Sanderman, K. W. Holmes, unpubl.). At these locations, only a single composite sample at each depth was available for each pasture.

(5) Namoi Catchment, NSW. Here, formerly cropped paddocks on gently sloping terrain have been sown to kikuyu-based pastures instead of temperate pasture grass species. Composite soil samples from five sets of kikuyu pasture and paired cropped sites were contributed to this project by King and Wilson (2010) as part of the NSW Monitoring, Evaluation and Reporting (MER) program (for collection methodologies see Chapman et al. 2011). Pasture ages ranged from 2 to 18 years since establishment.

At each sampling location, mean climatic data were extracted from the SILO database (www.longpaddock.qld.gov.au/silo/) and various topographic attributes were extracted from the Smoothed Digital Elevation Model of Australia, of which the focal mean slope was used in further analyses in this study (details can be found in Sanderman et al. 2012a).

Sampling strategy

Due to the fact that this project has used a paired sampling approach, a different field sampling strategy than was adopted by other projects in the Soil Carbon Research Program was needed to capture whether or not significant differences existed between each pair of perennial and annual pastures (Sanderman et al. 2012a). Placing a fixed grid 25 m by 25 m at any one location within a large paddock may not give an accurate representation of the mean and variance of SOC content of that paddock. To test which sampling strategy would best capture these properties, we intensively sampled two perennial paddocks--one sown to kikuyu and the other to a combination of panic and Rhodes grass. At five randomly chosen locations within a 4-ha region of each paddock, a grid 3 m by 3 m was laid down, and every 1-m intersection of that grid was sampled in 10-cm increments to 30 cm using hand-driven volumetric cores. Using hand-driven volumetric cores ensured that accurate bulk density values would be collected with each sample. This sampling design resulted in 80 cores being retrieved from the panic--Rhodes grass pasture and 64 cores from the kikuyu pasture (one grid had to be rejected in the kikuyu pasture due to a large contrast in soil type).

From these intensively sampled paddocks, Sanderman et al. (2012b) found that: (1) a random sampling approach was superior to a small, fixed grid; and (2) eight randomly placed cores were more than adequate to capture representations of the mean and variance of SOC stocks for the region of interest.

For the remainder of the sampling in this project except for the samples contributed from different studies, eight randomly located cores over an area of ~4 ha within the pasture of interest were collected using hand-driven volumetric cores (5 cm internal diameter) (AMS Soil Core Samplers, AMS Inc., American Falls, ID, USA) in 10-cm increments. In each perennial pasture, vegetation samples (shoot and root samples) were collected from three random locations to determine the isotope ratio of the soil carbon input material. All samples were collected between June and August to minimise variability due to seasonality of grass productivity.

Analytical methods

All soil samples were air-dried in a 40[degrees]C forced-air oven within a few days of field collection. Samples were then weighed and sieved to 2 mm. Material >2 mm was weighed to determine gravel content, and for samples where it was amenable (primarily in the sandy soils of WA), coarse root material was further separated from this fraction. The fine-earth fraction (<2 mm) was then subsampled using a riffle-box splitter for subsequent analyses. Residual moisture content was determined by drying a 20-g subsample at 105[degrees]C for 24 h. A second subsample was finely ground (MM400 Mixer Mill; RETSCH, Haan, Germany) before further analyses. Details of sample preparatory procedures are given in Baldock et al. (2013a, this issue). Plant samples were dried in a 60[degrees]C oven and then finely ground before isotope analysis.

Total C and nitrogen (N) content was determined by high-temperature oxidative combustion (CNS2000; LECO Corp., St. Joseph, MI, USA). Almost all samples were found to be non-calcareous; thus, the total carbon (TC) measurement equates to total organic C (TOC). The few samples that were found to be calcareous based on a fizz test were pre-treated with sulfurous acid to remove inorganic C (IC) before analysis for TC. Analytical precision determined by repeat TOC analysis of 10% of the samples was better than 0.05 mg C [g.sup.-1] for samples containing >4 mg C [g.sup.-1] and better than 0.2 mg C [g.sup.-1] for samples containing less C.

The isotopic composition of soil and plant samples was determined by coupled continuous isotope ratio mass spectrometry on a 20-20 Isotope Ratio Mass Spectrometer (Sercon Ltd, Crewe, UK) coupled with a Europa Solid/Liquid Elemental Analyser (Sercon Ltd) for sample introduction. Stable carbon isotope data are reported relative to the Pee Dee Belemnite (PDB) standard (Craig 1957) as:

[delta][sup.13]C = ([R.sub.sample] - [R.sub.std])/[R.sub.std] x 1000

where r is [sup.13]C/[sup.12]C of the sample and standard, respectively. Precision of the isotope analysis determined by duplicate analysis of 10% of the samples was typically better than 0.05[per thousand].

At all sampled paddocks, a composite sample was created for each depth by combining aliquots of each individual core in proportion to the bulk density of that core. A subsample from each of these composite samples was finely ground and analysed by mid-infrared (MIR) spectroscopy. Clay content (mg clay [g.sup.-1] <2 mm soil) and pH were then predicted from the MIR spectra using a partial least-squares regression approach (Janik and Skjemstad 1995; Janik 2006; Baldock et al. 2013b, this issue).

Data analyses

Carbon stock data have been reported as the C content (Mg C [ha.sup.-1]) for each soil layer by multiplying the TOC concentration (mg C g [soil.sup.-1]) by the bulk density (Mg [m.sup.-3]) by depth and by a correction for gravel content.

Data have also been reported on an equivalent-mass basis. For each pair of sites, the SOC content was adjusted to the lesser soil mass of the perennial and control treatment. For the paired site with the greater soil mass, SOC was removed from the 20-30 cm layer in proportion to the amount of soil mass that needed to be subtracted to reach the equivalent mass of the lighter pair.

Changes in SOC, whether on a fixed-depth or equivalent-mass basis, were reported as [DELTA]SOC ([C.sub.perennial] - [C.sub.control], Mg C [ha.sup.-1]), and the annual rate of change was defined as [DELTA]SOC divided by the perennial pasture age (Mg C [ha.sup.-1] [year.sup.- 1]).

Isotope data were interpreted using a two-component linear mixing model between the old [C.sub.3] soil ([C.sub.C3]), measured in the paired control site, and new soil carbon from the [C.sub.4] vegetation ([C.sub.C4]):

[C.sub.s] x [delta][sup.13][C.sub.s] = ([C.sub.C3] x [delta][sup.13][C.sub.C3]) + ([C.sub.C4] x [[delta].sup.13] [C.sub.C4]);

where [C.sub.S] is the carbon stock of the entire soil ([C.sub.S] = [C.sub.C3] + [C.sub.C4]). The fraction of total SOC attributable to the new C4 carbon inputs ([f.sub.C4] = [C.sub.C4]/[C.sub.S)] was calculated as:

[f.sub.C4] = [[[delta].sup.13] [C.sub.S] - [[delta].sup.13] [C.sub.C3]]/[[[delta].sup.13] [C.sub.C4] - [[delta].sup.13] [C.sub.C3]]

Uncertainty in end-member values was accounted for by propagating the variance in each term following Phillips and Gregg (2001). The [delta][sup.13]C and [f.sub.C4] values for the 0-30 layer were calculated as SOC mass-weighted mean values of the individual horizons. Finally, the mass of new Ca-derived SOC was calculated by multiplying [f.sub.C4] by the mass of SOC found in that perennial pasture.

Statistical analyses

At the regional level, two tests were performed, a two-way ANOVA (region x pasture type) and a pair-wise comparison such as a one-way, repeated-measures ANOVA (pasture type) for each region, to test for significant increases in SOC and [delta][sup.13]C. Due to the fairly low numbers of replications within each region, the variability between different farms within each region may swamp any potential differences between pasture types using the first of these approaches. The repeated-measures test eliminates some of the site-to-site variability within each region and only tests that the mean pasture type difference is significantly different from zero. These two statistical tests were performed on the 0-10 and 0-30 cm data. The SOC mass was log-transformed to normalise data. A critical assumption here was that the control paddock is representative of the state of the perennial paddock upon conversion to the perennial vegetation.

As we sampled to span a range of pasture ages, a chronosequence approach was also utilised in which the changes in SOC properties are plotted against pasture age. The chronosequence approach relies on the assumption that all other state factors (i.e. soil properties, land form, climate, and vegetation) were constant within a region and only pasture age varies (Jenny 1941). Soil type and land form were broadly similar within each region (Table 1). Vegetation was purposely selected to be similar across sites within each region. There was found to be a significant range of climatic properties within each region. Importantly, with the exception of a weak correlation ([R.sup.2] = 0.33, P=0.03) in the Southern District of WA, climatic properties (rainfall or temperature) were unrelated to pasture age (Fig. 1). The oldest site in this region was found to be driving this correlation, and when removed, there was no residual correlation between rainfall and age ([R.sup.2] =0.08). Given the similarity in soil type, landform, climate, and vegetation, the two regions in SA have been combined for the purposes of the regression analysis. Several different regression models were tested on the chronosequence data. All parametric analyses were performed in SigmaPlot, version 12 (Systat Software Inc., Chicago, IL, USA).

Results

Regional mean changes in SOC

Across the project (n=42), subtropical perennial pastures stored 3.3 [+ or -] 8.6 Mg C [ha.sup.-1] (mean [+ or -] s.d.) more SOC than their paired temperate counterparts (P = 0.014) to a depth of 30 cm (Table 2), translating to a mean annualised rate of change of 0.23 [+ or -] 0.85 Mg C [ha.sup.-1] [year.sup.-1] (P=0.098). At a regional level, significant increases in SOC stocks were only found for kikuyu in the Southern District, WA, and the Fleurieu Peninsula, SA. The annualised rate of change was only found to be significantly different from zero (P = 0.001) in the Southern District, with a mean rate of change of 0.61 [+ or -] 0.51 Mg C [ha.sup.-1] [year.sup.-1]. A much more detailed summary of bulk data from all sites is provided in Sanderman et al. (2012b).

In the Northern District of WA, SOC levels under pastures sown to a mixture of panic and Rhodes grass were, on average, no different from their annual counterparts (Table 2), with 24.8 [+ or -] 6.6 and 24.4 [+ or -] 7.4 Mg C [ha.sup.-1] found in the perennial and annual paddocks, respectively. In fact, only one of 14 sites was found to have significantly more SOC under the panic-Rhodes pasture (data not shown).

On Kangaroo Island, the paired annual pasture for sites KI1-KI4 was on a slightly different soil type but it was the best representation of the perennial paddocks that we could find. The annual soil had a higher watertable at the time of sampling due to a shallower depth to the clay pan. As a result, the rates of change calculated from the difference between the perennial and annual pairs (Table 2) should be interpreted with caution. For these sites on Kangaroo Island, the trend over time (i.e. Fig. 2) is likely the more accurate representation of the SOC dynamics.

On the Fleurieu Peninsula in South Australia, there was a large range in SOC content (36.5-83.5 Mg C [ha.sup.-1] in the perennial pastures), but by focusing only on the [DELTA]SOC values, we see that the subtropical perennial pastures had on average 5.8 Mg C [ha.sup.-1] more than the temperate grass based pastures (Table 2).

In the Namoi Catchment of NSW, where kikuyu pastures have been sown into previously cropped paddocks, small but non-significant increases in SOC were found (Table 2), with the SOC stocks under the subtropical perennials averaging 26.2 [+ or -] 7.7 v. 22.9 [+ or -] 9.3 Mg C [ha.sup.-1] in the cropped paddocks. Importantly, only one pasture in this region had been established >5 years ago.

Accumulation of [C.sub.4]-derived SOC

Isotope measurements, as reported as the fraction of [C.sub.4]-derived SOC ([f.sub.C4]) present in the perennial soils, proved to be a more sensitive indicator of change in SOC stocks, with significant differences found in most regions (Table 2) and at greater depths than for SOC stocks alone (Sanderman et al. 2012b). Across all sites, the perennial paddocks contained an average of 9.2% new [C.sub.4]-derived SOC, ranging from a mean of 16.9% under kikuyu in the Southern District of WA to a mean of only 4.8% under panic-Rhodes in the Northern District of WA. In the Namoi Catchment, while SOC stocks had not changed, there was a significant shift in the isotopic composition of the SOC even after only a few years under kikuyu.

Chronosequence results

Several different regression models were fitted to the chronosequence data including linear, logarithmic, exponential rise to maximum, and a sigmodal function, but for all three regions included in the chronosequence analyses, the linear model gave the best fit (Fig. 2). The annual rate of change in [DELTA]SOC was thus calculated as the slope of a linear regression between [DELTA]SOC and perennial pasture age (Table 3). Results from this analysis indicated that the regression-based statistical approach yielded similar patterns between regions but with values ~25% greater than found in the paired ANOVA approach (Table 2). For the entire dataset, the regression-based analysis indicated that the annual rate of change in [DELTA]SOC with pasture age (mean [+ or -] s.e.m.) in the 0-30 cm layer was 0.51 [+ or -] 0.13 Mg C [ha.sup.-1] year l, with an increase of 0.41 [+ or -] 0.10 Mg C [ha.sup.-1] [year.sup.-1] for the 0-10 cm layer only. The greatest rate of increase was found in the Southern District of WA. Importantly, the trend in the Southern District was independent of the oldest sites. If these two sites were removed, the slope of the linear regression only dropped from 0.90 [+ or -] 0.13 to 0.77 [+ or -] 11 Mg C [ha.sup.-1] [year.sup.-1], with [R.sup.2] = 0.58 (P= 0.004).

When the bulk SOC data were combined with the isotope data, we found that the annual rate of change in C4-C across all sites was 0.41 [+ or -] 0.05 and 0.35 [+ or -] 0.04 Mg C [ha.sup.-1] [year.sup.- 1] for the 0-30 and 0-10 cm layers, respectively (Table 3). The ratio of these regression slopes suggested that >80% of the total increase in SOC stocks was attributable to the new perennial vegetation. The Namoi Catchment data were not included in the regression analyses because of low replication and a narrow range of pasture ages.

Discussion

Sources of variability in SOC stock

There are a myriad of reasons why differences in SOC stocks may be found. Before even considering environmental and anthropogenic drivers of SOC change, the calculation of SOC stocks, itself, incorporates three separate measurements, all of which contribute to variation: SOC concentration, bulk density, and the rock fragment content. For the mean 0-10 cm data across all 42 sites, we found that variation in SOC concentration contributed the greatest amount of variance (66%), with variation in rock fragment content contributing 27%, and variability in bulk density contributed the remaining 7% (Fig. 3). On Kangaroo Island, where rock fragment content varied from 14 to 33% in the top 10 cm, nearly 40% of the variance in SOC stocks can be attributed to this single factor, which is independent of land use and management.

Beyond the error inherent in the calculation of SOC stocks, climate, soil properties, and resource quality are known to exert strong direct and indirect controls on SOC cycling (Sanderman and Amundson 2003), and this was the case for total SOC stocks within each region in this study (Fig. 1). However, much less is known about whether these environmental parameters affect the ability to change SOC stocks due to a shift in land management. The parameters [DELTA]SOC and [f.sub.c4] were found to be uncorrelated with the measured covariates in all regions except SA, where there were weakly significant correlations with mean annual precipitation ([DELTA]SOC: r= 0.69, P= 0.038; [f.sub.c4]: r = 0.63, P=0.069), but this trend was primarily driven by the fact that greater SOC changes were found on the Fleurieu Peninsula where rainfall was 200-300 mm higher than on Kangaroo Island (Table 1).

Importantly, the ability to build SOC under subtropical perennial grasses appears to be independent of the amount of SOC already in the soil (Fig. 4). For all data and within each region, there were no correlations between [DELTA]SOC or [f.sub.c4] and the amount of SOC found in the paired annual pasture, despite [C.sub.annual] varying from 13 to 73 Mg C [ha.sup.-1]. This finding is in contrast to the often-stated conventional wisdom (e.g. Bruce et al. 2009) that it is harder to increase SOC levels in soils already high in SOC. It is possible that none of these sites were close to their 'potential maximum' SOC level. Additional modelling work would be needed to further explore this possibility. An alternative explanation is that the new SOC is simply accumulating in an unprotected fraction and that the more stable SOC pool has been saturated (cf. Stewart et al. 2007). This latter possibility appears to be supported by analysis of the isotopic composition of size-fractioned soil samples from the Southern District of WA (see companion paper, Sanderman et al. 2013).

A final source of variability in SOC stocks that we considered in this study was due to differences in soil mass found within our fixed sampling depth. Carbon accounting in Australia occurs to a fixed depth (Richards 2001). However, when trying to ascertain the true impact of a particular management practice on SOC stocks, calculating the stocks to an equivalent mass is preferred (Dalai and Mayer 1986) because differences due to compaction or inflation of the soil are removed. In this study, it was found that the extensive root system under kikuyu significantly reduced the bulk density in the upper 10 cm of soil, leading to a decrease in soil mass compared with the paired control sites (Fig. 5). The panic-Rhodes grass systems examined in the Northern District of WA did not change the soil bulk density. Given that there was less soil mass under kikuyu, the fixed-depth SOC calculation under-estimated the rate of change in SOC by 20% on the Fleurieu Peninsula, 13% on Kangaroo Island, and 7% across the Southern District of WA (Table 2). Due to these underestimations in SOC sequestration rates using a fixed-depth sampling strategy, it is recommended for systems such as kikuyu which build an extensive perennial root system that an equivalent-mass sampling strategy is undertaken.

Interpretation of isotope results

The fact that there were significant differences in the [delta][sup.13]C value of SOC under the [C.sub.4] grasses does not necessarily equate to increases in SOC stocks, because the new [C.sub.4] carbon may simply be replacing older [C.sub.3] carbon that is being respired away by microorganisms. Modelling of the steady-state replacement of the existing [C.sub.3] SOC with new [C.sub.4] inputs would give an indication of whether the rate of change in isotope values was greater than would be expected if there were no changes in total SOC stocks occurring. This modelling exercise is beyond the scope of this study as it requires detailed knowledge of pasture composition and pasture production throughout the history of each of the pastures, and assumes that the model fully captures the dynamics of C in the soil (for elaboration see companion paper, Sanderman et al. 2013).

The isotope results can be combined with the [DELTA]SOC results to build a case that the changes in [delta][sup.13]C were likely greater than would have been expected based on steady-state replacement of the older [C.sub.3]-SOC. The finding that the trends in [DELTA]SOC and [C.sub.4]-SOC were similar (Fig. 2) strongly suggests that it was the new perennial vegetation that contributed to the gains found in SOC in the perennial-based pastures. Additionally, the agreement of findings between the paired-site approach (i.e. [DELTA]SOC) and the isotope approach validates the quality of the pairings and builds a strong case that real changes in SOC can occur with the introduction of kikuyu into pasture systems.

Does conversion to subtropical perennial pastures increase SOC levels?

Regardless of whether an ANOVA-based or regression-based statistical approach was taken, the data presented here suggest that subtropical perennial pastures can increase SOC levels, but it is dependent upon grass species and, to a lesser degree, soil characteristics. In WA and SA, pastures sown to kikuyu had higher SOC levels than paired annual-based systems. Pastures comprising a mix of panic and Rhodes grass had SOC levels similar to those found in the paired annual systems, and the isotope results suggested that the new perennial C is just slowly replacing the older [C.sub.3]-SOC. Recently, Lawes and Robertson (2012) examined differences between 16 pairs of panic-Rhodes grass based perennial pastures and annual crop-pasture rotations and also found no significant differences between the land-use types. Because the panic-Rhodes grass systems were only in one region, it is a difficult to ascribe the lack of difference to one cause. However, the soils and climate in the Northern District are broadly similar to the Southern District of WA (Table 1), and the two pasture systems differed markedly in the amount of belowground biomass. Nearly three times more coarse root (defined as the roots retained on 2-mm sieve) C was found in samples from soils under kikuyu compared with the panic-Rhodes grass pastures and the annual-based pastures in WA (Fig. 6). In a trial of several forage species, Neal et al. (2013) found that despite similar yields for several species, only kikuyu was able to increase SOC levels, suggesting that below-ground C allocation is indeed significantly greater than for many commonly sown temperate [C.sub.3] grasses. Thus, it appears that the spreading growth form of kikuyu combined with substantial below-ground C compared with both annual grasses and panic and Rhodes grass is the most likely explanation for the differences in SOC response between systems. Findings from a radiocarbon pulse-chase experiment conducted by Roper et al. (2013, this issue) support this notion. Those authors found that significantly more fixed [sup.14]C[O.sub.2] was recovered below ground under kikuyu than under panic grasses.

The effectiveness of kikuyu for increasing SOC levels varied between regions, with the largest increases found in the sandy soils of WA, smaller but still significant gains found in SA, and no changes found in the Naomi Catchment of NSW. Cooler winter temperatures in the SA regions relative to the Southern District of WA (11.9 v. 14.2[degrees]C) may confer a greater relative advantage to winter-active grasses, because kikuyu growth is greatly restricted over cold winter periods (Hill et al. 1985; Moore et al. 2006). In addition, the soils in SA are generally more fertile and, based upon farmers' responses, were more actively managed with regard to nutrient status. Thus, it was likely that the traditional, temperate grass based pastures in SA were already producing at high levels, and introduction of kikuyu was only leading to smaller net gains in productivity and C return to the soil. We did not sample along the southern coast of Victoria where kikuyu is also becoming an important part of the grazing landscape. Conditions are generally similar to the Fleurieu Peninsula but with slightly cooler temperatures and different soil types. Future research in this region would be valuable for confirming or refuting the above hypothesis regarding a smaller relative advantage for kikuyu in cooler, more fertile regions.

The results from the Naomi Catchment were particularly striking as that region receives the most summer rainfall and has the greatest annual temperature (Table 1); thus, it might be expected that a summer-active subtropical grass would have a greater advantage in this region over the cooler, Mediterranean regions along Australia's south coast. Additionally, in the Namoi Catchment region, the kikuyu was sown into formerly cropped paddocks. Conversion of cropped land to pasture typically increases SOC levels regardless of pasture type (Guo and Gifford 2002; Sanderman et al. 2010; Chan et al. 2011). Although the SOC stocks have not shifted significantly, there has been a rapid replacement of older SOC with new kikuyu-derived carbon (Table 2). The lack of significant differences in SOC stocks here may have been due to a combination of low replication and a low mean age of the perennial pastures.

The regression-based analysis is likely the more accurate of the two statistical approaches used in this study because the age of the perennial pastures varies substantially, thus potentially confounding results when only the mean values were computed for each region. We specifically sampled sites that spanned a range of pasture ages to be able to comment on the trend over time. At least for the first few decades after conversion to kikuyu, the increase in SOC appeared to be fairly linear (Fig. 2). A logarithmic model did no better job explaining the data for two main kikuyu regions than a simple linear model (data not shown). Our best estimates of the rate of change in SOC when a pasture was sown to kikuyu in the Southern District of WA was 0.90 [+ or -] 0.25 Mg C [ha.sup.-1] [year.sup.-1] and in South Australia was 0.26 [+ or -] 0.13 Mg C [ha.sup.-1] [year.sup.-1] over a period of at least 20 years.

Unfortunately, more specific management history (i.e. fertiliser history, grazing management, etc.) for these pastures was not variable enough within a region, or the data were not of a consistent enough quality to quantitatively test for more specific management drivers of changes in SOC or [delta]sup.13]C values.

Conclusions

The combination of elemental and isotope measurements proved to be a powerful means for assessing changes in SOC stocks where subtropical perennials were sown into formerly [C.sub.3]-based agricultural systems. We found that the ability of subtropical perennial grasses to sequester SOC varied with species and region.

Levels of soil C increased in kikuyu-based systems in the Southern Agricultural District of WA, and in Kangaroo Island and the Fleurieu Peninsula of SA. In these regions, the increase in [C.sub.4]-SOC within the perennial pasture averaged 80% of the overall increase in SOC, indicating that new SOC attributed to the kikuyu was the dominant driver of SOC change. The majority (>70%) of the 0-30 cm change in SOC and [C.sub.4]-SOC occurred in the upper 10 cm of soil. The SOC difference between the kikuyu and [C.sub.3]-based pasture increased linearly with the age of the perennial pasture for 33 years in WA and for 45 years in SA. Mean SOC sequestration rates in the upper 30 cm were 0.90 [+ or -] 0.25 and 0.26 [+ or -] 0.13 Mg C [ha.sup.-1] [year.sup.-1] in WA and SA, respectively.

Given that ~120 000 ha has been sown to kikuyu in the Southern District of WA (Nichols et al. 2012), using results presented in Table 3, these pastures may be sequestering 0.29-0.51 Tg C[O.sub.2]e annually. However, to assess the full greenhouse gas benefit of converting to a kikuyu-based pasture system, any additional methane or nitrous oxide emissions due to running higher livestock numbers would have to be discounted from this value (Thomas et al. 2012).

In the Namoi Catchment, NSW, we found no changes in overall SOC levels when cropped paddocks were converted to kikuyu-based pasture systems, despite rapid accumulation of [C.sub.4]-SOC in these systems. Replication was low and there were only two pastures that had been in the ground for >5 years in this region; thus, confidence in the broader applicability of these findings is limited.

Perennial pastures based on a mixture of panic and Rhodes grass do not appear able to build SOC in the Northern Agricultural District of WA. Soil carbon levels were no different between the perennial and annual pastures and there was little indication of accumulation of new [C.sub.4]-SOC in the perennial pastures (<5%), even after 18 years. The most likely reason for the greater response in the kikuyu-based pastures relative to the panic-Rhodes grass and annual grass based systems is a combination greater below-ground C allocation and more complete pasture coverage due to the spreading growth form of kikuyu.

http://dx.doi.org/10.1071/SR13111

Received 29 November 2012, accepted 31 May 2013, published online 20 December 2013

Acknowledgements

Funding for this study was provided to CSIRO by the Australian Government's Department of Agriculture, Fisheries and Forestry. We gratefully acknowledge all of the farmers who provided access to their land; Paul Sanford, Lyn Dohle, Tim Prance, Rod Dowie, and David Hall for providing expert assistance in locating suitable sampling locations; Elizabeth Sanderman, Mark Farrell, Claire Farrell, and Andrew Wherrett for additional assistance in the field; Bruce Hawke and Cathy Dandy for MIR analyses; and Tom Carter, James Kerr, Kiralee Jury, and Christina Asanopoulos for assistance in the laboratory.

References

Baldock JA, Sanderman J, Macdonald LM, Puccini A, Hawke B, Szarvas S, McGowan J (2013a) Quantifying the allocation of soil organic carbon to biologically significant fractions. Soil Research 51, 561-576.

Baldock JA, Hawke B, Sanderman J, Macdonald LM (2013b) Predicting contents of carbon and its component fractions in Australian soils from diffuse reflectance mid-infrared spectra. Soil Research 51, 577-595.

Bruce S, Sims J, Walcott J, Baldock J, Grace P (2009) 'Soil carbon management and carbon trading.' (Bureau of Rural Sciences: Canberra, ACT)

Cerling TE, Quade J, Wang Y, Bowman JR (1989) Carbon isotopes in soils and paleosols as ecology and paleoecology indicators. Nature 341, 138-139. doi:10.1038/341138a0

Chan KY, Oates A, Li GD, Conyers MK, Prangnell RJ, Poile G, Liu DL, Barchia IM (2010) Soil carbon stocks under different pastures and pasture management in the higher rainfall areas of south-eastern Australia. Australian Journal of Soil Research 48, 7-15. doi:10.1071/SR09092

Chan KY, Conyers MK, Li GD, Helyar KR, Poile G, Oates A, Barchia IM (2011) Soil carbon dynamics under different cropping and pasture management in temperate Australia: Results of three long-term experiments. Soil Research 49, 320-328. doi: 10.1071/SR10185

Chapman G, Gray J, Murphy B, Atkinson G, Leys J, Muller R, Peasley B, Wilson B, Bowman G, McInnes-Clarke S, Tulau M, Morand D, Yang X (2011) Assessing the condition of soils in NSW. Monitoring, Evaluation and Reporting Program, Technical Report Series. Office of Environment and Heritage, NSW Government, Sydney.

Craig H (1957) Isotopic standards for carbon and oxygen and correction factors for mass spectrometric analyses of carbon dioxide. Geochimica et Cosmochimica Acta 12, 133-149. doi:10.1016/0016-7037(57) 90024-8

Dalal RC, Mayer RJ (1986) Long-term trends in fertility of soils under continuous cultivation and cereal cropping in southern Queensland. 1. Overall changes in soil properties and trends in winter cereal yields. Australian Journal of Soil Research 24, 265-279. doi: 10.1071/SR98 60265

Guo LB, Gifford RM (2002) Soil carbon stocks and land use change: a meta analysis. Global Change Biology 8, 345-360. doi: 10.1046/j.1354- 1013.2002.00486.x

Hill MJ, Pearson CJ, Campbell LC (1985) Growth of seedlings of prairie grass and tall rescue in small swards of kikuyu at different temperatures. Australian Journal of Agricultural Research 36, 213-220. doi: 10.1071/AR9850213

Isbell RF (2002) 'The Australian Soil Classification.' Revised edn (CSIRO Publishing: Melbourne)

Jackson LE, Roy J (1986) Growth-patterns of Mediterranean annual and perennial grasses under simulated rainfall regimes of southern France and California. Acta Oecologica Oecologia Plantarum 7, 191-212.

Janik LJ (2006) 'Predict Ver 6.0 MIR Soil Analysis Software.' (CSIRO Land and Water: Adelaide, S. Aust.)

Janik LJ, Skjemstad JO (1995) Characterization and analysis of soils using mid-infrared partial least-squares. 2. Correlations with some laboratory data. Australian Journal of Soil Research 33, 637-650. doi:10.1071/SR9950637

Jenny H (1941) 'Factors of soil formation: A system of quantitative pedology.' (McGraw-Hill: New York)

King D, Wilson B (2010) Soil Carbon Monitoring Project. Project Report to Namoi CMA, NSW Department of Environment and Climate Change, Sydney.

Kort J, Collins M, Ditsch D (1998) A review of soil erosion potential associated with biomass crops. Biomass and Bioenergy 14, 351-359. doi:10.1016/S0961-9534(97)10071-X

Lawes RA, Robertson MJ (2012) Effect of subtropical perennial grass pastures on nutrients and carbon in coarse-textured soils on a Mediterranean climate. Soil Research 50, 551-561. doi:10.1071/SR11 320

Li GD, Helyar KR, Welham SJ, Conyers MK, Castleman LJC, Fisher RP, Evans CM, Cullis BR, Cregan PD (2006) Pasture and sheep responses to lime application in a grazing experiment in a high-rainfall area, southeastern Australia. I. Pasture production. Australian Journal of Agricultural Research 57, 1045-1055. doi: 10.1071/AR05298

Moore G, Albertsen T, van Gool D, Titterington J (2006) Poor persistence of subtropical grasses over winter. In 'Proceedings of the 13th Australian Agronomy Conference'. 10-14 September 2006, Perth, W. Aust. (Eds NC Turner, T Acuna, RC Johnson) (Australian Society of Agronomy/ The Regional Institute Ltd: Gosfurd, NSW) Available at: www.regional.org.au/au/asa/2006/concurrent/environment/4537_moorega.htm

Neal JS, Eldridge SM, Fulkerson WJ, Lawrie R, Barchia IM (2013) Differences in soil carbon sequestration and soil nitrogen among forages used by the dairy industry. Soil Biology & Biochemistry 57, 542-548. doi: 10.1016/j.soilbio.2012.09.019

Nichols PGH, Yates RJ, Loo V, Wintle BJ, Titterington JW, Barrett-Lennard EG, Stevens JC, Dixon KW, Moore GA (2012) Establishment of subtropical perennial grasses in south-western Australia. Future Farm Industries CRC Technical Report No. 9, Perth, W. Aust.

Nie ZN, Miller S, Moore GA, Hackney BF, Boschma SP, Reed KFM, Mitchell M, Albertsen TO, Clark S, Craig AD, Kearney G, Li GD, Dear BS (2008) Field evaluation of perennial grasses and herbs in southern Australia. 2. Persistence, root characteristics and summer activity. Australian Journal of Experimental Agriculture 48, 424-435. doi: 10.10 71/EA07136

Paydar Z, Huth N, Ringrose-Voase A, Young R, Bernardi T, Keating B, Cresswell H (2005) Deep drainage and land use systems. Model verification and systems comparison. Australian Journal of Agricultural Research 56, 995-1007. doi: 10.1071/AR04303

Phillips DL, Gregg JW (2001) Uncertainty in source partitioning using stable isotopes. Oecologia 127, 171-179. doi:10.1007/s004420000578

Richards GP (2001) 'The FullCAM Carbon Accounting Model: Development, calibration and implementation for the National Carbon Accounting System.' (Department of Climate Change: Canberra, ACT)

Roper MM, Fillery IRP, Jongepier R, Macdonald LM, Sanderman J, Baldock JA (2013) Allocation into soil organic matter fractions of [sup.14]C captured via photosynthesis by two perennial grass pastures. Soil Research 51, 748-759.

Sanderman J, Amundson RG (2003) Biogeochemistry of decomposition and detrital processing. In 'Treatise on geochemistry. Vol. 8: Biogeochemistry'. (Eds HD Holland, KK Turekein) (Elsevier Science Ltd: Oxford, UK)

Sanderman J, Farquharson R, Baldock JA (2010) Soil carbon sequestration potential: A review for Australian agriculture. Report to the Australian Department of Climate Change. Technical Report, CSIRO Land and Water, Adelaide, S. Aust. Available at: www.csiro.au/resources/SoilCarbon- Sequestration-Potential-Report.html

Sanderman J, Baldock J, Hawke B, Macdonald L, Massis-Puccini A, Szarvas S (2012a) National Soil Carbon Research Programme: Field and laboratory methodologies. Report to the Australian Department of Agriculture, Fisheries and Forestry.

Sanderman J, Fillery I, Jongepier R, Massalsky A, Roper M, Macdonald L, Maddern T, Murphy D, Baldock J (2012b) Quantification of carbon input to soils under important perennial pasture systems used in Australian agriculture: C3/C4 transitions. Report to the Australian Department of Agriculture, Fisheries and Forestry.

Sanderman J, Fillery I, Jongepier R, Massalsky A, Roper M, Macdonald L, Maddern T, Murphy D, Baldock J (2013) Carbon sequestration under subtropical perennial pastures II: Carbon dynamics. Soil Research 51, 771-780.

Sanford P, Wang X, Greathead KD, Gladman JH, Speijers J (2003) Impact of Tasmanian blue gum belts and groundwater recharge in south-western Australia. Australian Journal of Experimental Agriculture 43, 755-767. doi: 10.1071/EA02226

Stewart CE, Paustian K, Conant RT, Plante AF, Six J (2007) Soil carbon saturation: concept, evidence and evaluation. Biogeochemistry 86, 19-31. doi: 10.1007/s10533-007-9140-0

Thomas DT, Sanderman J, Eady S J, Masters DG, Sanford P (2012) Whole farm net greenhouse gas abatement from establishing kikuyu-based perennial pastures in south-western Australia. Animals 2, 316-330. doi: 10.3390/ani2030316

Wilson GWT, Hartnett DC (1998) Interspecific variation in plant responses to mycorrhizal colonization in tallgrass prairie. American Journal of Botany 85, 1732-1738. doi:10.2307/2446507

Young RR, Wilson B, Harden S, Bernardi A (2009) Accumulation of soil carbon under zero tillage cropping and perennial vegetation on the Liverpool Plains, eastern Australia. Australian Journal of Soil Research 47, 273-285. doi: 10.1071/SR08104

Jonathan Sanderman (A,B,F), I. R. P. Fillery (A,C), R. Jongepier (A,C), A. Massalsky (A,B), M. M. Roper (A,C), L. M. Macdonald (A,B), T. Maddern (A,B), D. V. Murphy (D), B. R. Wilson (E), and J. A. Baldock (A,B)

(A) CSIRO Sustainable Agriculture National Research Flagship.

(B) CSIRO Land and Water, Private Bag 2, Glen Osmond, SA 5064, Australia.

(C) CSIRO Plant Industry, Private Bag 5, Wembley, WA 6913, Australia.

(D) Soil Biology and Molecular Ecology Group, School of Earth and Environment, Institute of Agriculture, University of Western Australia, Crawley, WA 6009, Australia.

(E) University of New England, Armidale, NSW 2351, Australia.

(F) Corresponding author. Email: jonathan.sanderman@csiro.au

Table 1. Selected regional attributes

Soil type presented to Suborder of the Australian Soil
Classification (Isbell 2002), with dominant surface soil texture
given in parentheses. MAT, MAP: Mean annual temperature
and precipitation, respectively; standard deviation of MAT, MAP,
and slope given in parentheses

                                                 Pasture age
Region                   Perennial grass   n    range (years)

Fleurieu Peninsula, SA   Kikuyu             4       3-45
Kangaroo Island, SA      Kikuyu             5       3-14
Northern district, WA    Panic-Rhodes      14       3-18
                         grass
Southern district, WA    Kikuyu            14       3-33
Namoi catchment, NSW     Kikuyu             5       2-18

Region                     Dominant land use

Fleurieu Peninsula, SA   Dairy and beef cattle
Kangaroo Island, SA      Sheep (wool and meat)
Northern district, WA    Cattle and sheep
                           (wool/meat)
Southern district, WA    Sheep (wool/meat)
                           and cattle
Namoi catchment, NSW     Beef cattle

                                                        MAT
Region                      Dominant soil type      ([degrees]C)

Fleurieu Peninsula, SA   Red Kurosol (variable)     14.2 (0.5)
Kangaroo Island, SA      Red Sodosol (sandy loam)   15.1 (0.3)
Northern district, WA    Bleached-Orthic Tenosol    19.4 (0.8)
                         (sand)
Southern district, WA    Yellow-Orthic Tenosol or   16.5 (0.3)
                           Yellow Sodosol (sand)
Namoi catchment, NSW     Red Chromosols (clayey     18.2 (0.1)
                           loam)

                                     Summer/annual    Focal mean
Region                   MAP (mm)    rainfall ratio   slope (%)

Fleurieu Peninsula, SA   804 (104)        0.20        9.3 (1.1)
Kangaroo Island, SA      579 (50)         0.18        1.7 (0.4)
Northern district, WA    515 (60)         0.11        4.3 (2.0)
Southern district, WA    547 (106)        0.25        1.2 (0.8)
Namoi catchment, NSW     618 (16)         0.51        1.6 (0.7)

Table 2. Overall soil organic carbon (SOC) trends: paired
regional means reported on a fixed-depth and an basis
equivalent-mass total and for the 0-30 cm total for the
0-10 cm horizon only

The [f.sub.C4] data are only reported on a fixed-depth basis
because calculation on an equivalent-mass basis for the 0-10 cm
horizon is not possible. P-statistic: one way repeated-measures
ANOVA

                     [DELTA]SOC (Mg C [ha.sup.-1]

                     Fixed depth

Region               Mean [+ or -] s.d.      P

0-30 cm total
Fleurieu Peninsula    5.77 [+ or -] 5.03   0.053
Kangaroo Island      -4.05 [+ or -] 4.34   0.099
Northern District     0.46 [+ or -] 4.20   0.494
Southern District     9.06 [+ or -] 11.7   0.003
Namoi Catchment       3.33 [+ or -] 4.63   0.279
All                   3.33 [+ or -] 8.55   0.014

0-10 cm horizon
Fleurieu Peninsula    3.28 [+ or -] 2.18   0.028
Kangaroo Island      -3.47 [+ or -] 2.02   0.018
Northern District    -0.66 [+ or -] 2.78   0.357
Southern District     6.31 [+ or -] 9.58   0.028
Namoi Catchment       1.34 [+ or -] 3.01   0.438
All                   1.72 [+ or -] 6.69   0.099

                     [DELTA]SOC (Mg C [ha.sup.-1]

                     Equivalent mass

Region                Mean [+ or -] s.d.      P

0-30 cm total
Fleurieu Peninsula    8.15 [+ or -] 5.35    0.071
Kangaroo Island      -3.05 [+ or -] 4.89    0.213
Northern District     0.46 [+ or -] 4.27    0.488
Southern District     9.67 [+ or -] 12.43   0.002
Namoi Catchment       2.98 [+ or -] 4.33    0.262
All                   3.81 [+ or -] 9.05    0.008

0-10 cm horizon
Fleurieu Peninsula    9.60 [+ or -] 5.84    0.046
Kangaroo Island      -3.83 [+ or -] 2.50    0.027
Northern District    -0.50 [+ or -] 2.88    0.495
Southern District     7.87 [+ or -] 10.84   0.018
Namoi Catchment       3.21 [+ or -] 5.55    0.266
All                   3.12 [+ or -] 8.12    0.014

                     Rate of change (Mg C
                     [ha.sup.-1] [year.sup.-1])

                     Fixed depth

Region               Mean [+ or -] s.d.      P

0-30 cm total
Fleurieu Peninsula    0.58 [+ or -] 0.55   0.064
Kangaroo Island      -0.83 [+ or -] 1.01   0.138
Northern District     0.17 [+ or -] 0.94   0.483
Southern District     0.61 [+ or -] 0.51   0.001
Namoi Catchment       1.42 [+ or -] 2.41   0.325
All                   0.23 [+ or -] 0.85   0.098

0-10 cm horizon
Fleurieu Peninsula    0.33 [+ or -] 0.28   0.050
Kangaroo Island      -0.63 [+ or -] 0.69   0.111
Northern District    -0.06 [+ or -] 0.54   0.668
Southern District     0.40 [+ or -] 0.16   0.001
Namoi Catchment       0.80 [+ or -] 1.36   0.328
All                   0.05 [+ or -] 0.56   0.574

                     Rate of change (Mg C
                     [ha.sup.-1] [year.sup.-1])

                     Equivalent mass

Region               Mean [+ or -] s.d.      P

0-30 cm total
Fleurieu Peninsula    0.75 [+ or -] 0.50   0.091
Kangaroo Island      -0.74 [+ or -] 1.04   0.189
Northern District     0.17 [+ or -] 0.96   0.482
Southern District     0.65 [+ or -] 0.61   0.001
Namoi Catchment       1.30 [+ or -] 2.26   0.334
All                   0.26 [+ or -] 0.87   0.060

0-10 cm horizon
Fleurieu Peninsula    0.82 [+ or -] 0.64   0.084
Kangaroo Island      -0.69 [+ or -] 0.70   0.092
Northern District    -0.02 [+ or -] 0.58   0.89
Southern District     0.56 [+ or -] 0.08   0.000
Namoi Catchment       0.43 [+ or -] 0.82   0.387
All                   0.18 [+ or -] 0.69   0.108

                     [f.sub.C4] (%)

                     Fixed depth

Region               Mean [+ or -] s.d.      P

0-30 cm total
Fleurieu Peninsula   12.0 [+ or -] 6.0     0.028
Kangaroo Island       3.5 [+ or -] 3.9     0.111
Northern District     4.8 [+ or -] 4.1     0.000
Southern District    16.9 [+ or -] 9.5     0.000
Namoi Catchment       7.4 [+ or -] 6.3     0.097
All                   9.2 [+ or -] 8.6     0.000

0-10 cm horizon
Fleurieu Peninsula   17.5 [+ or -] 9.6     0.036
Kangaroo Island       4.8 [+ or -] 5.2     0.108
Northern District     4.8 [+ or -] 4.0     0.000
Southern District    20.5 [+ or -] 10      0.000
Namoi Catchment      11.5 [+ or -] 9.3     0.096
All                  11.3 [+ or -] 10      0.000

Table 3. Regression summary for selected SOC measurements for the
Southern District (SD) and Northern District (ND) of Western Australia,
and for South Australian sites Fleurieu Peninsula (FP) and Kangaroo
Island (KI) combined

Slope of linear regression [+ or -] s.e.m. only given when regression
was significant ([alpha] [less than or equal to] 0.10). n.s., Not
significant

                                0-30 cm

Region          Slope          [R.sup.2]       P

Perennial SOC mass (Mg C [ha.sup.-1])

SD        1.84 [+ or -] 0.36     0.68        0.000
ND                               0.01        n.s.
FP + KI                          0.22        n.s.

Perennial [delta][sup.13]C ([per thousand])

SD                               0.15        n.s.
ND        0.04 [+ or -] 0.03     0.19        0.090
FP + KI   0.07 [+ or -] 0.02     0.59        0.019

[DELTA]SOC (Mg C [ha.sup.-1])

SD        0.90 [+ or -] 0.25     0.61        0.002
ND                               0.06        n.s.
FP + KI   0.26 [+ or -] 0.13     0.38        0.070
All       0.51 [+ or -] 0.13     0.27        0.001

[C.sub.4]-C (%)

SD        0.47 [+ or -] 0.24   0.30 (A)    0.047 (A)
ND        0.46 [+ or -] 0.25     0.19        0.099
FP + KI   0.39 [+ or -] 0.11     0.64        0.007
All       0.51 [+ or -] 0.13     0.27        0.001

[C.sub.4]-C (Mg C [h.sup.-1])

SD        0.51 [+ or -] 0.06     0.84        0.000
ND        0.13 [+ or -] 0.06     0.27        0.041
FP + KI   0.29 [+ or -] 0.09     0.60        0.013
All       0.41 [+ or -] 0.05     0.61        0.000

                                0-10 cm

Region          Slope          [R.sup.2]       P

Perennial SOC mass (Mg C [ha.sup.-1])

SD        1.35 [+ or -] 0.24     0.73        0.000
ND                               0.09        n.s.
FP + KI                          0.12        n.s.

Perennial [delta][sup.13]C ([per thousand])

SD                               0.18        n.s.
ND        0.05 [+ or -] 0.02     0.17        0.100
FP + KI   0.11 [+ or -] 0.03     0.52        0.010

[DELTA]SOC (Mg C [ha.sup.-1])

SD        0.77 [+ or -] 0.20     0.66        0.001
ND                               0.08        n.s.
FP + KI   0.15 [+ or -] 0.08     0.35        0.090
All       0.41 [+ or -] 0.10     0.28        0.000

[C.sub.4]-C (%)

SD        0.54 [+ or -] 0.27   0.33 (A)    0.030 (A)
ND        0.44 [+ or -] 0.26     0.18        0.100
FP + KI   0.61 [+ or -] 0.16     0.67        0.007
All       0.71 [+ or -] 0.15     0.38        0.000

C4-C (Mg C [h.sup.-1])

SD        0.42 [+ or -] 0.05     0.86        0.000
ND        0.08 [+ or -] 0.03     0.29        0.033
FP + KI   0.24 [+ or -] 0.08     0.58        0.017
All       0.35 [+ or -] 0.04     0.62        0.000

(A) [R.sup.2] and P given for regression with log(age) as
the log-transformation significantly improved the fit.
COPYRIGHT 2013 CSIRO Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2013 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Sanderman, Jonathan; Fillery, I.R.P.; Jongepier, R.; Massalsky, A.; Roper, M.M.; Macdonald, L.M.; Ma
Publication:Soil Research
Article Type:Report
Geographic Code:8AUST
Date:Oct 1, 2013
Words:9061
Previous Article:Allocation into soil organic matter fractions of [sup.14]C captured via photosynthesis by two perennial grass pastures.
Next Article:Carbon sequestration under subtropical perennial pastures II: carbon dynamics.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters