Soil organic matter turnover in British deciduous woodlands, quantified with radiocarbon

Abstract

Soil samples, collected to a depth of 15 cm from 24 plots at six UK deciduous woodland sites in 1971 and 2002, were analysed for ¹⁴C, and total soil carbon pools (gC m^− 2) were estimated. The results, together with data from a further woodland site, were interpreted using steady-state models, driven by the ¹⁴C content of atmospheric CO₂. The average soil pool is 6700 (± 1100) gC m^− 2, where the error term is the standard deviation. Application of a single soil pool model yielded a mean residence time (MRT) of 87 (± 82) years, with an input of 113 (± 52) gC m^− 2 a^− 1. This implies that the remainder of the total litter production, about 300 gC m^− 2 a^− 1, passes through a fast pool with a short MRT (i.e. rapid mineralisation occurs). Slightly better fits, with less bias, were obtained using a model that has two soil C pools with fixed MRTs of 15 (slow pool) and 350 (passive pool), based on the findings of DD Harkness, AF Harrison and PJ Bacon (Radiocarbon 28, 328–337, 1986). From the 21 soils that gave similar results with the two C pool model, it is calculated that, on average, the soil contains 3300 gC m^− 2 in the slow pool (input 215 gC m^− 2 a^− 1) and 3400 gC m^− 2 in the passive pool (10 gC m^− 2 a^− 1); about 200 gC m^− 2 a^− 1 of the litter input is rapidly mineralised. The slow pool is approximately constant for all the soils, variations in total soil C being due to the range of passive pools (1800 to 5900 gC m^− 2). Equally good fits of the bulk soil ¹⁴C data were obtained with different assumptions about the MRTs of the slow and passive pools; therefore additional information about soil fractions is needed for more exact characterisation of soil organic matter cycling on timescales of decades to millennia.

Introduction

The turnover rate of soil organic matter (SOM) is a key factor in the terrestrial carbon cycle with respect to carbon exchange with the atmosphere, responses of SOM to global warming and plant fertilisation by C and N, and the management of soils to increase carbon storage (Schlesinger, 1997, Amundson, 2001, Kimble et al., 2003, Janzen, 2004, Johnston et al., 2004). Furthermore, future changes in SOM pools may affect soil functions such as water retention, nutrient status, especially N cycling, and cation exchange.

Natural-abundance radiocarbon data have been used extensively to obtain insight into soil carbon cycling and age, originally by making use just of radioactive decay since C fixation (Tamm and Östlund, 1960), but mainly by exploiting “bomb carbon” produced by the atmospheric enrichment of ¹⁴CO₂ due to weapons testing in the 1950s and 1960s (O'Brien and Stout, 1978, Harkness et al., 1986, Jenkinson et al., 1992, Hsieh, 1993, Scharpenseel, 1993, Harrison, 1996, Bol et al., 1999, Gaudinski et al., 2000, Harrison et al., 2000, Hahn and Buchmann, 2004). The radioactive decay of ¹⁴C gives information on processes occurring over hundreds or thousands of years, whereas the bomb carbon peak works on a decadal time scale. In most studies, soil has been analysed by horizon, regular depth interval or extracted pools. Turnover rates have been derived either by simple box modelling or by the application of more sophisticated models such as RothC (Jenkinson, 1990). The results have yielded a range of mean residence times (MRT) of carbon in soil (see e.g. Trumbore, 1997, Trumbore, 2000), and it is clear that, in general, SOM contains components with MRTs ranging from a few years (material closely related to plant litter) to millennia (recalcitrant or mineral-protected material). If the available data are sparse, as is often the case with radiocarbon measurements, derived turnover times are likely to be model-dependent and should be regarded as characteristic or relative rather than absolute.

The two most detailed bomb ¹⁴C studies of soils were conducted within 20 years of the initial bomb enrichment. O'Brien and Stout (1978) measured the ¹⁴C content of New Zealand pasture soil samples that had been collected from different depths on 9 different occasions over the period 1959–1973. Harkness et al. (1986) monitored ¹⁴C in the soil of a British deciduous woodland (Meathop Wood) at different depths over the period 1961 to 1984. Whereas O'Brien and Stout (1978) interpreted their results using a model that represented the downward mixing of soil by pseudo-diffusion, Harkness et al. (1986) considered turnover rates of different pools, the approach that has been adopted most often since. They showed that soil organic C comprises several pools differing in the rate at which they incorporated ¹⁴C. They suggested that near-surface SOM needs to be conceptualised with at least two pools, one with an MRT of less than 20 years, the other which a longer time span of several hundred years.

These two early studies benefited from both depth-based sampling and also the substantial temporal variability associated with the rapidly-changing atmospheric signal. Later studies of different individual soils were restricted by the lesser temporal variability, and Trumbore (2000) pointed out the limitations associated with single value bulk soil ¹⁴C values. Harrison (1996) overcame this problem to an extent by amalgamating data from many studies on different soils in order to generate a data set of topsoil ¹⁴C values covering the period 1957 to 1991. He represented SOM in general using a passive pool (turnover ∼ 6500 years) and an active pool (25 years). Contemporary investigations at new sites are restricted, unless archived soil samples can be analysed (e.g. Jenkinson et al., 1992, Torn et al., 2002). The scarcity of suitable archived samples means that we should take every opportunity to exploit them for ¹⁴C studies.

In the present work, we carried out radiocarbon analyses of 24 surface soils (15 cm deep) sampled in 2002 as part of the Resurvey of British Woodlands (Kirby et al., 2005), and of archived samples from the same sites taken in 1971. These plots, together with Meathop Wood (see above), provided a reasonable representation of deciduous woodlands and their soils in Great Britain. We analysed the data using steady-state models, deriving MRT values and combining them with estimated soil C pools to estimate carbon fluxes. Thus the study considered variations in soil ¹⁴C from both temporal and spatial perspectives, with the aim of quantifying soil carbon turnover in this ecosystem type, which is widespread in the UK and the northern and southern hemisphere temperate zones.