- Top of page
- Materials and methods
- Supporting Information
Global changes such as elevated atmospheric CO2 concentrations and nitrogen (N) deposition rates are altering the input rates of carbon (C) and N to soils (van Groenigen et al. 2006; Hyvönen et al. 2007). There is experimental evidence that the resulting changes in C and N availabilities in soils alter the magnitude of the imbalance between soil organic carbon (SOC) decomposition and formation processes, thereby affecting total SOC contents (Neff et al. 2002; Fontaine et al. 2004; van Groenigen et al. 2006). An improved understanding of how altered C and N input rates to soils will affect SOC decomposition and formation processes may enable more accurate prediction of how SOC stocks will respond to global change, and feedback to atmospheric CO2 concentrations and soil fertility (Lal 2004).
Mechanistic investigations typically treat SOC as consisting of a continuum of C-containing fractions whose cycling rates vary from < 1 year to millennia (Olk & Gregorich 2006; von Lützow et al. 2007). It is often defined as material containing, at least partially, humified matter (Paul, Morris & Böhm 2001). Following this definition then, new C-inputs to soils (e.g. plant litter) do not increase SOC stocks; instead these inputs must first be processed by decomposers before ‘formation’ of SOC. The size of SOC fractions is determined by the balance of SOC ‘formation’ and ‘decomposition’. The latter is often used as a catch-all term (e.g. Cardon et al. 2001) to include loss processes such as mineralization of SOC to CO2, as well as leaching of C solubilized from SOC as it is further humified (Fig. 1).
Figure 1. Conceptual representation of the fractions and processes that determine the size of the total SOC pool (broken-line box). This representation is an operational definition for the purposes of the current study; SOC fractions and processes may be conceptualized a number of different ways. Here, total SOC is separated into two fractions (solid-line boxes: particulate organic matter and mineral-associated C), which have different turnover times and hence C sink capacities. SOC formation occurs when C inputs to the soil are humified (e.g. through microbial activity); they may enter either SOC fraction. It is also likely that SOC in the more rapidly cycling fraction (i.e. the particulate matter) moves into the mineral-associated fraction across the experiment but this arrow is not shown. Carbon that is added to the soils during the experiment, and that enters either SOC fraction, is classified as ‘new’ SOC. In contrast, C present prior to the experiment is classified as ‘old’ SOC; the net decrease in the size of this pool of old C occurs through processes referred to collectively as ‘decomposition’. The operational definitions used in the current study neither permit movement of old C between fractions to be estimated nor the ‘decomposition’ of new SOC that occurred during the study. This is because, by definition, no new SOC was present in the soils at the start of the experiment. Decomposition of old SOC therefore represents treatment effects on the net decrease in old SOC within a fraction; formation represents the net increase in new SOC within a fraction. See text for further details.
Download figure to PowerPoint
That we cannot reliably explain varying responses of SOC stocks in different environments to increased inputs rates of C and N to soils, is perhaps the best evidence that we do not have the required mechanistic understanding of SOC formation and decomposition processes. For example, the positive relationship between soil N fertilization rate and SOC stocks under elevated CO2 (van Groenigen et al. 2006) mirrors the situation in cropping systems where increases in residue inputs under N fertilization result in higher SOC contents (Gregorich et al. 1996; McLauchlan 2006). However, effects of N fertilization on SOC stocks in non-agricultural grasslands and forests have not necessarily translated to greater SOC stocks, despite increases in plant inputs to soils. Indeed, positive (Waldrop et al. 2004), negative (Mack et al. 2004; Waldrop et al. 2004) and negligible (Neff et al. 2002) effects of N fertilization on SOC stocks have been observed.
One reason why positive, above-ground plant productivity responses to N fertilization may not necessarily translate to greater SOC stocks (Neff et al. 2002) is that the majority of plant-C in non-cropped systems enters mineral soils via roots (van Groenigen et al. 2006). There is now substantial evidence that of the root-C entering these soils, the dominant input is in the form of labile-C compounds (i.e. rhizodeposition) (van Hees et al. 2005; Högberg & Read 2006; Boddy et al. 2007). Enhanced rhizodeposition rates under elevated atmospheric CO2 are likely to increase supply rates of labile plant-C to mineral soils (Pendall, Mosier & Morgan 2004). Undoubtedly, some of this rhizodeposited C will be incorporated into SOC as microbially derived compounds (Lundberg, Ekblad & Nilsson 2001). Yet despite evidence for elevated rhizodeposition, Heath et al. (2005) observed decreasing SOC formation of root-derived C under increasing atmospheric CO2 concentrations. If this effect is observed in other systems, then the negligible or positive effects of elevated atmospheric CO2 on SOC stocks (Jastrow et al. 2005; van Groenigen et al. 2006) suggest that enhanced root inputs under elevated CO2 must also reduce rates of SOC decomposition. That is, for decreased SOC formation to be associated with positive or negligible effects on total SOC stocks requires that SOC decomposition must simultaneously decrease.
The possibility that enhanced rhizodeposition might reduce rates of SOC decomposition needs to be explored. Indeed, some experimental observations suggest that labile C inputs to soils result in no SOC decomposition (Dalenberg & Jager 1981; Wu, Brookes & Jenkinson 1993; Fontaine & Barot 2005). The apparent absence of a ‘priming effect’ (Cheng 1999; Kuzyakov, Friedel & Stahr 2000) on SOC in such instances has been attributed to the utilization of labile C inputs by microbes which do not mineralize SOC and out-compete those that do (Fontaine et al. 2004; Fontaine & Barot 2005). This ‘preferential substrate utilization’ theory (Dalenberg & Jager 1981; Wu et al. 1993; Fontaine & Barot 2005) contrasts with priming effect theory because the latter proposes that the increase in microbial activity with elevated labile C inputs will concomitantly stimulate SOC decomposition. Notably, Heath et al. (2005) observed greater SOC formation of root-derived C with N fertilization, mitigating decreases in SOC formation under elevated CO2. This effect, combined with the possibility that exogenous N addition inhibits SOC decomposition (Franklin et al. 2003; Hyvönen et al. 2007), may explain why SOC contents increase when systems are exposed simultaneously to N and CO2 fertilization (van Groenigen et al. 2006), but not necessarily N or CO2 fertilization alone (Neff et al. 2002; van Groenigen et al. 2006). To understand the mechanisms underlying these responses requires studies that quantify both SOC formation and decomposition, and not just the net balance of these processes, under experimentally controlled, labile C and N input rates to soils.
Here, we amended intact soil monoliths with C, N and phosphorus (P) across 1 year. We manipulated P, as well as N, because much of our understanding of how N fertilization affects SOC comes from studies where P is added simultaneously with N (e.g. Cardon et al. 2001; Mack et al. 2004; Heath et al. 2005); also P may both affect SOC dynamics itself and modify the impacts of N alone (Cleveland & Townsend 2006). Furthermore, as with C and N, human activities are enhancing P availabilities in soils (Wassen et al. 2005). We amended soils chronically (twice-weekly) across a range of input levels to investigate the potential for input rate dependent effects of C and nutrients (e.g. as seen in van Groenigen et al. 2006). We utilized 13C-techniques to partition the effects of C, N and P additions into those on SOC formation and decomposition (Fig. 1). We resolved these dynamics for the total SOC pool and for individual SOC fractions which differ in their sink potentials. Following from our current understanding of labile C and N effects on SOC processes, we hypothesized that (i) there is no net effect of labile C input rate on SOC stocks because decreased SOC decomposition is offset by decreased SOC formation. Conversely, we hypothesized that (ii) with simultaneous nutrient addition total SOC stocks will increase because N will inhibit SOC decomposition, and also mitigate decreases in SOC formation observed under elevated labile-C inputs alone. We hypothesized that (iii) this increase in SOC stocks will be a product of increased SOC formation and decreased decomposition of slower-cycling, mineral-associated SOC fractions; whereas less stable, particulate SOC fractions will simply turnover faster (i.e. greater formation balanced by greater decomposition).
- Top of page
- Materials and methods
- Supporting Information
Based on the theory of preferential substrate utilization and previous experimental observations (Dalenberg & Jager 1981; Wu et al. 1993; Fontaine & Barot 2005; Heath et al. 2005; van Groenigen et al. 2006), we hypothesized that increasing labile C input rates would slow SOC decomposition and formation rates. This hypothesis was supported to some degree considering that the decomposition of SOC was attenuated at higher C amendment rates (Fig. 2). This observation supports the hypothesis that certain soil microbes utilize only fresh plant-C inputs and that these microorganisms can out-compete SOC decomposing microbes, thereby slowing SOC decomposition (Fontaine & Barot 2005). However, when no C was added the amount of SOC decomposition was similar to that under the high C amendment rate, and low rates of C amendment stimulated SOC decomposition (Fig. 2). These observations suggest that the extent to which SOC decomposers are out-competed, and hence the magnitude of the priming effect, will be dependent on the C input rate.
We predicted that SOC formation would decline with increasing C amendment rate. We made this prediction based on the fact that although elevated CO2 concentrations increase below-ground C inputs, they have been observed to reduce SOC formation from these inputs (Heath et al. 2005) and/or have negligible effects on total SOC stocks (van Groenigen et al. 2006). In contrast, we observed greater SOC formation with high C amendment rates (Figs 2–4), which we would expect if much of this new SOC is in the form of microbially derived products (Lundberg et al. 2001). Indeed, microbial biomass and activity were greatest at the highest C amendment rate (Table 2). SOC stocks declined at the lower C amendment rate because SOC decomposition was greater than formation, yet increased at the higher C amendment rate because formation was greater than decomposition (Fig. 2). These results emphasize that to predict accurately how changes in below-ground C supply to soils will impact SOC stocks will require quantification of C input rates via rhizodeposition.
Our second hypothesis was that nutrient (i.e. NP) amendment, in addition to C amendment, would increase SOC content because it would both inhibit SOC decomposition (Waldrop et al. 2004; Knorr, Frey & Curtis 2005a; Hyvönen et al. 2007) and enhance SOC formation. In support of the first part of our second hypothesis we observed lower SOC decomposition with N amendment (Fig. 5). Notably, however, P amendment enhanced SOC decomposition compared to soils which did not receive additional P (Fig. 5). It seems likely that the enhanced SOC decomposition with combined N and P amendment in the Factorial Experiment (Fig. 2) was caused by the added P. The NP Regression Experiment indicated that at high rates of NP amendment the N may ameliorate some of the P effects on decomposition (Fig. 5), which may explain why the NP effect on decomposition in the Factorial Experiment was slightly more pronounced at the low amendment rate (Fig. 2). Thus the first part of our second hypothesis, that nutrient amendment combined with C amendment would further inhibit SOC decomposition compared to C amendment alone, was not supported. We did, however, observe greater SOC formation with nutrient addition. The absence of an interaction between C and NP amendment suggests that reported interactions between elevated CO2 and N deposition which increase SOC stocks (van Groenigen et al. 2006) likely are additive (as opposed to statistically non-additive).
The expectation that simultaneous NP and C addition would increase SOC content (Hypothesis 2) was based on a differential response by POM and mineral-associated C fractions. Specifically, our expectation was that the POM C fraction would turnover faster (i.e. enhanced decomposition and formation of SOC) but that decomposition of the mineral-associated C fraction would decline and its formation increase (i.e. Hypothesis 3). A number of studies (Ågren & Bosatta 2002; Kirschbaum 2004; Eliasson et al. 2005; Knorr et al. 2005b) have highlighted the importance of considering SOC as a conglomeration of multiple fractions of C cycling at different rates. Our results support this perspective: as hypothesized we found that the different C fractions did not respond uniformly to C, N or P amendment. However, the responses of each fraction were not necessarily as we hypothesized. For example, in the Factorial Experiment decomposition of POM C at the low C amendment rates was greater with NP amendment, but unaffected by NP amendment at the high C amendment rate (Fig. 2). In contrast, decomposition of mineral-associated C was significantly accelerated by NP amendment but not C amendment (Table 1). Differences in SOC fraction responses to C and NP amendment were also observed for SOC formation. That is, formation of POM C was greater under C and NP amendment but these amendments did not interact to determine the response (Table 1). In contrast, formation of mineral-associated C was dependent on interactions between C and NP amendment (Table 1). Interestingly, Cardon et al. (2001) examined impacts of elevated CO2 and differing soil N availability on decomposition and formation responses of SOC fractions but found that the CO2 and N factors did not interact. Given our results and those of Cardon et al. (2001) it is not clear if interactive (i.e. non-additive) effects of C and N need be considered if we want to predict how altered resource availability will affect decomposition and formation of SOC in fractions of differing sink potential. Further research in this area is required.
Despite positive plant biomass responses, N fertilization of non-agricultural systems has had negative, positive and negligible effects on SOC stocks (Neff et al. 2002; Mack et al. 2004; Waldrop et al. 2004). Neff et al. (2002) observed more rapid turnover of faster-cycling SOC fractions under N fertilization but slower turnover of more stable SOC fractions, the net result being no change in total SOC content. In contrast, NP amendment in our Factorial Experiment increased decomposition of mineral-associated C and also, generally, accelerated formation of new SOC in this fraction (Fig. 3). These changes would be expected to be associated with more rapid turnover of this more stable SOC pool. Notably, Neff et al. (2002) did not amend with P as well as N and when both were applied in a field experiment SOC stocks were decreased (Mack et al. 2004). Our results, and those of Mack et al. (2004), highlight the need for P to be explicitly considered when interpreting SOC responses to altered resource availability. As to the mechanism underlying the mineral-associated C response to P we are at a loss. It may suggest that decomposition of this more recalcitrant SOC fraction requires enzymes that are P-rich, or a high ATP supply rate to provide the energy required for decomposition. Given P eutrophication of soils (Newman et al. 1995; Anderson & Downing 2006), elucidation of the mechanism appears essential to predict accurately SOC dynamics. What may also be essential, given those interactions between C and NP amendment on SOC fraction dynamics that we observed, is quantification of how below-ground C input rates are affected by altered N and P availability.
We chronically amended our soils with C in an attempt to simulate conditions that might recreate, at least in part, those of mineral soils receiving rhizodeposition. This process is a dominant pathway through which low-molecular weight, C compounds enter soils but its ecological significance is understudied (van Hees et al. 2005; Boddy et al. 2007; Pollierer et al. 2007). Given that the effects of more recalcitrant C inputs to soils on total SOC contents differed to those we observed, in that they were dependent on nutrient availability (Fontaine et al. 2004), we may need to consider two sets of dynamics relating to how SOC will respond to altered resource availability. The first should focus on dynamics associated with discrete litter inputs and the second on more continuous inputs of labile C compounds. The expectation that these inputs are processed differently within soils is already explicit in much of the theory relating soil decomposer food web structure to function (e.g. Hendrix et al. 1986). It does not, however, seem to have been coupled with priming effect theory.
Although disaccharides are a dominant constituent of rhizodeposition, there are certainly many more C-compounds found in root exudates, of which sucrose is but one (van Hees et al. 2005). An outstanding question, therefore, from our study is whether sucrose is a useful approximate for the general effects of rhizodeposition. Also outstanding is whether the high rates of sucrose amendment (e.g. 800 g C m−2 year−1) we used are realistic; it is likely for example that many of the compounds in fast-cycling C pools (e.g. Gu et al. 2004) are significantly more recalcitrant than sucrose. Given that global changes are likely to influence not only the amount but also composition of rhizodeposition, the effects of different rhizodeposits alone and in combination on SOC dynamics would likely be a useful research theme for understanding how the soil C sink may respond to future global change.
As in other studies (e.g. Fontaine et al. 2004) that have examined the impacts of C and nutrient amendments on SOC dynamics, our study did not include plants and hence may have missed a critical feedback in determining SOC dynamics (Fontaine & Barot 2005; Luo, Field & Jackson 2006). For example, increasing C amendment was negatively correlated with extractable, inorganic N levels (Table 2), which might be expected to alter plant-C inputs below-ground and hence SOC dynamics (Fontaine & Barot 2005; Luo et al. 2006). Indeed, N addition may itself feedback to decrease C allocation below-ground to roots and rhizodeposition (Bowden et al. 2004), although direct effects of N on microbial activity are also common (Bowden et al. 2004; Burton et al. 2004). The presence of plants might also have served to stimulate competition for N, and would also have introduced other, more recalcitrant C compounds (e.g. cellulose and lignin) to the soil. All of these plant effects might have altered how the microbes utilized SOC and the sucrose inputs. Further, mycorrhizae may play a central role in the turnover of SOC fractions (Manning et al. 2006; Osler & Sommerkorn 2007) and in the absence of plants their functional symbiosis was omitted from our experiment. Our work was also restricted to a single soil type; further research is required to evaluate whether our results can be extrapolated to other soils and in the presence of plants. A potentially informative starting point would be to select soils of differing C : N ratios, given the expectation that this ratio will influence SOC responses to C amendments (Osler & Sommerkorn 2007). Even then, caution should be extended in extrapolating the results to the field because periodic amendment of soils with limiting resources may not accurately represent the manner in which C, N and P are made available in soils.
Our results do, however, highlight that our current understanding of SOC dynamics is insufficient for accurate forecasting of how the SOC sink will respond to altered resource availability under global change. They also provide a mechanistic basis for understanding apparently contradictory effects on SOC dynamics of nutrient amendment in the field (e.g. Neff et al. 2002 and Mack et al. 2004). Based on our findings, we highlight four areas for future investigations to focus on to improve our understanding of SOC responses to altered C, N and P availabilities. First, the response of the total SOC, in terms of both decomposition and formation processes, to variation in C and nutrient amendment rates did not predict that of SOC fractions of differing sink strengths. Thus, there is a need to focus on SOC fractions and not just total SOC contents if we are to determine the soil sink potential for C under global change. Second, we demonstrated that chronic amendment with labile C has the potential to reduce total SOC contents (e.g. Fig. 2), a hitherto unexpected result for low molecular weight C compounds such as mono- and disaccharides (Dalenberg & Jager 1981; Wu et al. 1993; Fontaine et al. 2004). This finding, in addition to the potential for labile C to constitute the dominant form of below-ground C input (van Hees et al. 2005) and the magnitude of its input to vary under global change (Pendall et al. 2004), necessitates expansion of priming effect theory to consider the impacts of chronic, labile C amendment on SOC dynamics. Third, in our study the effects of C and nutrient amendment on SOC dynamics, and interactions between these elements, were typically dependent on resource input rates. These data emphasize the need for studies which assess multi-factor and multi-level resource manipulations and hence the recognition that responses to altered resource availability may be non-additive. Fourth, our results and those of others (Mack et al. 2004; Cleveland & Townsend 2006) suggest that P availability may play a critical role in determining SOC sink strengths. Notably, P amendment in our study affected decomposition of the SOC fraction of greater not lesser sink strength (i.e. it decreased the mineral-associated SOC fraction). This highlights the pressing need for naturally or anthropogenically varying P availabilities to be explicitly considered when SOC responses to altered C and N availabilities are investigated.