Environmental and human influences on organic carbon fractions down the soil profile

Highlights

• Soil organic carbon fractions studied at 100 sites at depths >30 cm in Australia.

• Depth strongly influences all three SOC fractions.

• Association between POC and total SOC suggests that these soils are input driven.

• HOC and ROC are additionally influenced by physico-chemical soil characteristics.

• Land-use was not important to the SOC fractions down the whole profile.

Abstract

We investigated the stability of whole profile soil organic carbon (SOC) based upon three mid-infrared predicted fractions – particulate organic carbon (POC), humus organic carbon (HOC) and resistant organic carbon (ROC) – at 100 sites across eastern Australia. Our aim was to identify the controls on SOC stability down the whole soil profile, in particular relating to climate, site and human influences. To do this we used three data-mining algorithms (randomForests, gradient boosting machines and multiplicative adaptive regression splines) to identify and assess the controls on the relative proportions of the three fractions down the soil profile. Depth was the key influence on all three fractions, with the proportion of POC decreasing, and the proportion of HOC carbon increasing with increasing depth. SOC was strongly linked with POC, suggesting that the soils in the region are input driven. HOC and ROC were controlled additionally by climate and soil physico-chemical properties (e.g. clay content, pH), with SOC being less important to these fractions. Human influences (land-use and management) were not important to the proportion of the fractions, implying that the controls humans can exert on SOC stability in these environments may be limited.

Introduction

Soil organic carbon (SOC) represents a large component of the global carbon cycle, with around 1500 Pg of organic carbon stored in the top meter of soil (Jobbagy and Jackson, 2000), and between 2344 (Jobbagy and Jackson, 2000) and 2400 Pg (Batjes, 1996) stored in the top 2–3 m of soils worldwide. Due to its origins in plant photosynthesis and loss via heterogenic respiration (predominately via soil microbial activity) it is a highly dynamic part of the global carbon cycle (Batjes, 1996).

Conceptually, SOC is separated into numerous pools with varying turnover rates ranging from hours to millennia (Trumbore, 1997), which are linked with the bioavailability of SOC and therefore its degradability (Baldock and Skjemstad, 2000). Mechanisms such as mineral-association (Oades, 1988), aggregation (and associated occlusion of SOC in aggregate interiors) (Tisdall and Oades, 1982) and physico-chemical transformation (e.g. via pyrolysis, Skjemstad et al., 1996) can reduce the biological accessibility of organic matter, converting it into more stable forms which are resistant to degradation processes and remain in the soil environment for long time periods.

Such pools of organic matter have been successfully isolated using physico-chemical fractionation procedures (e.g. Christensen, 1992, Golchin et al., 1994) which can be used to initialize soil carbon turnover models (Skjemstad et al., 2004), providing a link between measurable SOC fractions and conceptualized SOC pools. Recent advancements have provided the ability to detect different carbon pools without physico-chemical fractionation of soil samples, instead relying on mid-infrared (MIR) spectra. Baldock et al. (2013a) applied partial least square regression (PLSR) analyses to MIR spectra to derive predictive algorithms to estimate three SOC fractions of conceptually differing bioavailability. Particulate organic carbon (POC) was linked with fresh, more easily decomposable organic material; humus organic carbon (HOC) represented a fraction of SOC which has been altered from its initial state via decomposition, mineral-association and aggregation processes. Resistant organic carbon (ROC) represented a biologically resistant poly-aryl carbon fraction with long residence time in soils (Baldock et al., 2013b).

These three SOC fractions are therefore conceptually associated with different processes and mechanisms driving SOC dynamics, namely production (POC, via photosynthesis, although turnover will reduce POC), turnover and retention (HOC, via microbiological activity and diagenesis with subsequent sorption on mineral surfaces and/or occlusion in aggregates) and abiotic transformation (ROC, i.e. alteration in fire). Thus POC represents a more labile pool of SOC, whereas HOC and ROC represent SOC pools of greater stability which will remain in soils for longer time periods. Identifying what drives not only the total amount of SOC in soils, but also the relative proportions of these fractions is fundamental to our understanding of SOC dynamics and the processes controlling it. This understanding will not only enable us to manage our soil carbon reserves for the future, but also to interpret SOC response to past environmental changes.

Recent work investigating the factors affecting the stock of these three fractions (POC, HOC, ROC) in eastern Australian soils showed that environmental and site properties exert the greatest influence, with human activities contributing a lesser, but still substantial amount to the variance in the fractions (Rabbi et al., 2014). This is consistent with investigations into the drivers of total SOC storage in eastern Australia, which have been identified as climate, soil texture and bulk density related, with the effects of land-use on SOC being negligible at the surface (Rabbi et al., 2015) but becoming more important with increasing soil depth (Hobley et al., 2015).

However, like many studies into SOC dynamics in Australia, these investigations focused on processes within the top 30 cm of soil. Subsoil organic carbon comprises a large (∼70% of SOC is located below the top 20 cm, Jobbagy and Jackson, 2000) yet still poorly understood (Rumpel and Kögel-Knabner, 2011) component of the global carbon cycle. With increasing depth from the surface, soil organic matter generally increases in age (e.g. Hobley et al., 2013, Kaiser et al., 2002), implying enhanced stability, and its utilization as a carbon source by microorganisms is altered (Kramer and Gleixner, 2008) implying different carbon dynamics at depth than at the surface. Understanding soil organic carbon dynamics and its role in the global carbon cycle therefore requires investigation of both topsoils and subsoils.

In this study we investigated controls on SOC dynamics and stability down the soil profile to depths of up to 1 m. Several hypotheses were tested relating to the drivers of the SOC fractions:

• That the relative contribution of POC would be driven by similar factors to the drivers of surface SOC, predominately climate variables related to water availability (Hobley et al., 2015), as this fraction is related to plant productivity. Land-use will negatively affect the relative contribution of POC to SOC via lowered input in agricultural systems (most notably in cropped systems) compared with native vegetation systems (Hobley et al., 2015, Hobley and Wilson, 2016, Rabbi et al., 2014);

• In contrast, we hypothesized that the relative contribution of HOC to SOC will be driven more by soil physico-chemical properties such as texture and pH, as this fraction is associated with SOC turnover and retention on soil minerals, which are influenced by the soil physico-chemical environment. As a driver of SOC turnover, temperature will also be important to this fraction.

• ROC will likely be affected by both climate and anthropogenic factors, as temperature, vegetation type and human fire management will dictate the (unknown) fire histories of the sites.

• Lastly, we hypothesized that, associated with the generally greater age (and by implication stability) of subsoil OC than surface OC, the relative contribution of the labile POC fraction will decrease with depth whereas relative contributions of the more stable ROC and HOC fractions will increase with depth.

To test these hypotheses we quantified the concentrations of SOC fractions as predicted from MIR spectra acquired from different soil depths of 100 soil profiles sampled at 100 sites from across New South Wales, eastern Australia and used several data-mining approaches to model and investigate the factors accounting for variations in SOC composition.