Soil microbial activity in relation to dissolved organic matter properties under different tree species
Background and aims
The total concentration of dissolved organic carbon (DOC) has often been observed to correlate positively with soil microbial respiration. The aim was to explain the correlation with the properties of dissolved organic matter (DOM).
A dataset from previously published papers was gathered together and subjected to multivariate analyses. Samples were collected from five tree species experiments in Finland. The degradability of DOM was assessed by measuring bacterial and fungal growth in DOM. The chemical properties of DOM were assessed by XAD resin fractionation and molecular weight. Soil microbial activity was assessed as C and N mineralization and microbial biomass.
Both low and high molecular weight compounds, as well as hydrophilic neutral compounds, seemed to be relatively easily degradable. In contrast to our presupposition, easily degradable DOM seemed to be less abundant in soil where variables describing microbial activity were higher. Birch soil with higher microbial biomass N seemed to contain less easily degradable DOM than spruce soil.
We suggest that DOM collected and characterized at a certain point reflects more the accumulation of refractory compounds following high microbial activity than the easily degradable compounds that microbes would be using when measured.
KeywordsDOC Properties of DOM XAD Ultrafiltration Soil Litter Microbial activity
Dissolved organic matter (DOM) in soil is produced during litter and soil organic matter (SOM) decomposition and solubilization. In addition, DOM in root exudates has an important role in soil because of its remarkable quantity and possible priming effect on organic matter decomposition (Neff and Asner 2001; Kuzyakov 2010). Soil microbial activity could be expected to exist in relation to the quantity and properties of DOM, because the dissolved form of elements has, in general, been regarded as a prerequisite for microbial uptake. Indeed, it has been observed several times that the total concentration of dissolved organic carbon (DOC) derived from soil correlate positively with variables describing soil microbial activity (e.g. Michel and Matzner 1999; Sjöberg et al. 2003; Zhao et al. 2008; Smolander and Kitunen 2011; Peterson et al. 2013). The existence of the correlation, as well as its explanation is still, however, under debate and seems to depend on site characteristics such as redox conditions (Zhao et al. 2008; Wang et al. 2013; Hanke et al. 2013). Although it has been assumed that DOM represents a labile part of SOM and that total DOC concentration and especially its easily degradable part resembles soil microbial activity, it has been suggested that a great part of DOM in soil represents a relatively stable by-product of microbial activity (Neff and Asner 2001; Hagedorn et al. 2002; Kalbitz et al. 2003; Wickland et al. 2007; Sanderman et al. 2008; Zhao et al. 2008; Wang et al. 2013).
The correlation between soil microbial activity and the concentration of DOC might be explained by the properties of DOM. The complex chemical structure and the proportion of aromatic compounds in DOM have been used to describe the refractory nature of DOM. The complexity and aromaticity of DOM have been suggested to decrease soil CO2 production (Zhao et al. 2008; Wieder et al. 2008; Bu et al. 2010) whereas the high turnover of low molecular weight (LMW) compounds has been suggested to increase soil CO2 production (reviewed by van Hees et al. 2005). However, large datasets combining soil microbial activity and DOM properties are lacking.
Tree species experiments offer an unique opportunity to study the correlation between soil microbial activity and DOC concentration, because tree species are known to affect soil microbial activity. Microbial activity related to C and N cycling has often been suggested to be higher in soil under deciduous silver birch (Betula pendula Roth.) than coniferous Norway spruce (Picea abies (L.) Karst.) as reviewed by Smolander and Kitunen (2011). They also observed positive correlations between DOC concentration extracted from soil and the rate of C and net N mineralization and amount of C and N in microbial biomass, which were used to assess soil microbial activity. In order to explain the correlation we have gathered together a large dataset by combining our previously published datasets of variables describing microbial activity and the properties of DOC and DON with multivariate analysis. Samples had been taken from five tree species experiments of varying site characteristics in Finland.
First, we relate the degradability of DOM to the chemical fractions, according to molecular weight and adsorption properties (XAD resin fractionation), of DOC and DON in order to discover the easily degradable fractions. Several papers have reported that when hydrophilic or LMW compounds are abundant, the degradability of DOM is high and when hydrophobic compounds are abundant, the degradability of DOM is lower (Marschner and Kalbitz 2003; Qualls 2005; Kiikkilä et al. 2013). The hydrophobic acid fraction contains mainly altered lignin degradation products of relatively high molecular weight while the hydrophilic neutral fraction contains breakdown products of carbohydrates strongly related to soil microbial activity (Guggenberger et al. 1994). DOM degradability has often been measured as the loss of DOC during incubation, the result of which is sensitive to the flocculation of DOM, and small differences in the incubation time and conditions (McDowell et al. 2006). We have, in addition, assessed the degradability of DOM by measuring relative bacterial or fungal growth in DOM, and try to increase the comparability of the results across different studies and sources of DOM. Second, we relate the fractions of soil derived DOM to variables that describe microbial activity in soil.
The aim was to see 1) whether bacterial growth in solution could be used as a measure of the degradability of DOM, 2) whether a relationship exists between soil microbial activity and the properties of DOM and can it explain the previously observed correlation between DOC concentration and soil microbial activity variables, and 3) whether DOM properties separate soil under birch from that of spruce and offer any explanation for the previously observed higher soil microbial activity under birch than under spruce. We hypothesize that easily degradable fractions are more abundant in soil under birch with higher soil microbial activity than in soil under spruce.
Characteristics of the sample sites
Forest site typea
Age of the stand, years
Soil C/N ratiob
Methods have been described in our previous papers listed above in detail. Shortly, soil solution had been collected either with water extraction or centrifugation-drainage technique from fresh samples. DOM had been derived by filtering the solution through 0.45 μm membrane and the concentration of DOC and DON had been measured. The degradability of DOM had been assessed by measuring the relative bacterial or fungal growth in DOM with 3H-thymidine incorporation or 14C-acetate- in-ergosterol technique, introduced to soil studies by Bååth et al. (2001) and Bååth (2001). The degradability of DOM was measured in three papers of Kiikkilä et al. (2006; 2011; 2013). From the same samples also the loss of DOC during short-term (15–20 days) incubations had been measured. The composition of DOM had been characterized in each of our studies by determining DOC and DON in different fractions according to their adsorption properties (XAD-8 resin) (Guggenberger et al. 1994). The technique divides DOC and DON proportionally to hydrophobic and hydrophilic acid, neutral and base fractions. DOM had been also characterized according to molecular weight by ultrafiltration, which divides DOC and DON proportionally to >100 kDa (high molecular weight, HMW), 10−100 kDa, 1−10 kDa and <1 kDa (low molecular weight, LMW) fractions. Soil microbial activity had been described by the rate of C mineralization (CO2 production) and net N mineralization in soil and amounts of C and N in soil microbial biomass (fumigation-extraction) (Smolander and Kitunen 2002, 2011; Smolander et al. 2005; Kanerva and Smolander 2007).
Pearson correlation coefficients were calculated between different measures of the degradability of DOM (from separate datasets and the combined dataset) in order to assess which of the measures are comparable over different studies. Two canonical correlation analyses (CCA), a technique for analysing the relationship between two set of variables, were performed (SAS 9.3). CCA finds a linear combination of variables from each set, called a canonical variable, such that the correlation between the two canonical variables is maximized. Scatter plot diagrams with samples and the most important variables (as vectors) constructing the canonical variables are used to present the result. CCA was first used in order to assess the relation of the chemical fractions of DOM to the degradability of DOM where the measurements for the bacterial and fungal growth and the amount of each chemical fraction of DOC and DON were calculated per amount of DOC in solution. CCA of bacterial growth alone and fungal growth alone against chemical fractions were chosen for the presentation. The second CCA was used in order to assess the relation of the chemical fractions of DOM to the variables describing soil microbial activity. Variables describing soil microbial activity were calculated per soil organic matter. DOC and DON were calculated as the proportional distribution (as %) in order to make DOM derived by centrifugation and water extraction comparable. The combination of soil microbial variables and microbial biomass N alone against chemical fractions were chosen for the presentation.
Degradability of DOM
The loss of DOC in solution during a short-term incubation correlated positively with the bacterial growth in solution in separate datasets. The correlation coefficients were as follows: r = 0.90, n = 27 (Kiikkilä et al. 2006); r = 0.94, n = 14 (Kiikkilä et al. 2011); r = 0.56, n = 27 (Kiikkilä et al. 2013). When the datasets were combined, a very low correlation coefficient (r = 0.3) was observed, indicating no relation. Neither the canonical correlation analysis (CCA) between the loss of DOC and the chemical fractions of DOM was successful. Fungal growth correlated with neither the loss of DOC nor bacterial growth.
Variable loadings of CCA describe the most important variables that construct the canonical variable, i.e. variables that are responsible for the separation of the samples. Loadings are presented as vectors in the figures. Positive vectors along Chem (Fig. 1a) indicated that the concentration of DOC in hydrophilic neutral, HMW (>100 kDa) and LMW (<1 kDa) fraction as well as DON in weak hydrophobic acid fraction were higher when bacterial growth was higher. DON in hydrophilic acid and DOC in 1–10 kDa fraction correlated negatively with bacterial growth according to CCA.
Fungal growth and Chem correlated positively according to CCA (r = 0.84, n = 100) (Fig. 1b). The vectors indicated that the concentration of DOC in the HMW fraction was higher when fungal growth was higher. Several fractions of DOC correlated negatively with fungal growth. CCA sample plot diagram separated neither the degradation stages of DOM or SOM nor the tree species.
Soil microbial activity
Tree species were separated when microbial biomass N (Nmic) was chosen alone to CCA (r = −0.68, n = 66) (Fig. 2b). The vectors indicate that when Nmic was higher the proportion of DON in 10–100 kDa, HMW and hydrophobic acid fraction were higher. DOC and DON in LMW fraction as well as DON in hydrophilic neutral and base fraction were higher when Nmic was lower.
Degradability of DOM
Measuring the loss of DOC in solution during a short-term incubation seemed to be highly dependent on the small differences in the incubation time and conditions and the used inoculum. Instead, the relative bacterial growth in solution, using ultrapure water + soil as a control, seemed to be comparable across different studies and was therefore used here to assess the degradability of DOM.
The compounds supporting highest bacterial growth and thus, with highest degradability seemed to belong to hydrophilic neutral, HMW or LMW fractions. They are regarded here as relatively easily degradable fractions. This easily degradable DOM seemed to decrease fairly consistently during the degradation of organic matter as observed previously (Cleveland et al. 2004; Don and Kalbitz 2005). Easily degradable DOM decreased also during the degradation of DOM in laboratory incubation. LMW or hydrophilic neutral compounds have been regarded mostly easily degradable whereas the HMW fraction has often been regarded relatively refractory and connected to lignin derived compounds (Guggenberger et al. 1994; Marschner and Kalbitz 2003; Qualls 2005). Here it was interpreted to contain also easily degradable compounds. The HMW fraction was in general higher in litter than in humus samples. Thus, in our samples the easily degradable part of the HMW fraction seems to be connected to the litter samples i.e. to the freshness of the source material. This has been suggested earlier concerning surface water samples (Amon and Benner 1996) and is not contradictory to previous observations of lignin derived DOM mentioned above. In contrast, recently Malik et al. (2013) showed that recent plant carbon was less abundant in HMW compounds in DOM derived from soil and microbial biomass. Thus, the picture is not clear.
Leaching of relatively refractory compounds have been suggested to increase at the later stages of litter decomposition (Cleveland et al. 2004; Kalbitz et al. 2006; Kiikkilä et al. 2012). Our analysis showed no general indication of the possible increase in hydrophobic acids during organic matter decomposition. Hydrophobic acid compounds have usually been reported relatively refractory (Marschner and Kalbitz 2003; Qualls 2005; Kiikkilä et al. 2011) but they did not correlate with the degradability of DOM when DOM originating from litter and humus were combined. This can be explained by the high proportion (>60 %) of hydrophobic compounds also in the most degradable litter DOM. The black-and-white division of chemical fractions between easily degradable and refractory may often be too simplistic. None of the fractions according to molecular weight or adsorption properties contain either easily degradable or refractory compounds alone; rather, the exact composition of a fraction depends largely on the origin and decomposition stage of DOM.
The analysis may indicate some differences between fungi and bacteria in utilizing DOM. The degradation stage of DOM seemed to control bacterial growth but not fungal growth (Fig. 1). Fungal growth correlated positively only with the abundance of HMW compounds meaning that HMW compounds supported fungal growth. Several fractions, like the LMW fraction, had, however, negative loadings. This indicates that they correlated negatively with the fungal growth. The negative loadings are difficult to interpret here and fungal growth measured in solution phase may include artifacts and deserves further research. The contribution and dominance of fungi to DOM degradation may be a lot stronger in natural systems than in DOM degradation assays.
Soil microbial activity
The interpretation of the canonical correlation between variables describing microbial activity and DOM fractions was not straightforward. Easily degradable compounds became either more (hydrophilic neutrals) or less (LMW) abundant when variables describing soil microbial activity increased. The high loading for hydrophilic neutrals was interpreted to result from the litter layer samples where the microbial activity variables were highest and where also hydrophilic neutral compounds were relatively abundant. Litter DOM is known to be rich in hydrophilic neutral compounds compared to humus (Uselman et al. 2012; Kiikkilä et al. 2011). The relation between hydrophilic fraction and microbial activity variables seemed not to be strong in the humus samples alone. Therefore the high abundance of hydrophilic neutrals could not explain high microbial activity in the humus layer.
In contrast to our hypothesis, variables describing soil microbial activity or biomass seemed to be negatively related to the degradability of DOM. This was deduced from both canonical correlation analyses (Fig. 2a, b). In analysis with Nmic, the interpretation seemed to be obvious. When Nmic was higher the compounds regarded often easily degradable, i.e. LMW, hydrophilic neutrals or hydrophilic bases, were lower. In contrast, compounds regarded often refractory, i.e. HMW and hydrophobic acid, were higher when Nmic was higher. It is therefore suggested that in organic forest soil with a relatively high microbial activity, and thus, active SOM decomposition, high amounts of DOM are produced but easily degradable compounds are degraded. At the same time, refractory compounds accumulate in the solution. The relative enrichment of refractory compounds during DOM degradation has been suggested previously (Wickland et al. 2007; Kalbitz et al. 2003).
The earlier often observed correlation between soil microbial activity and total DOC concentration in soil probably reflects the accumulation of refractory compounds after high microbial activity. It does not tell about high amount of easily degradable compounds that microbes would be using at the point of measurement. We support the view of Bengtson and Bengtsson (2007) that soil respiration can be explained rather by the turnover of DOC than the concentration of DOC. The turnover of DOC is, however, difficult to assess from the measurements at a certain point, because the turnover is a dynamic process. It is also known that the most easily degradable compounds degrade during the processing of DOM in the laboratory (Rousk and Jones 2010). When DOM has been derived from live foliage or from dried and ground samples using extraction combined with suppression of microbial growth, DOM seems to have a relatively high degradability as well as high amount of easily degradable compounds (Cleveland et al. 2004; Bu et al. 2010; Kiikkilä et al. 2011; Uselman et al. 2012). These techniques may describe better the easily degradable DOM that potentially could release during the decomposition of organic matter, but not the total turnover of DOM.
DOM properties seem to correlate with variables describing soil microbial activity and thus, SOM decomposition, as assumed. However, opposite to general assumption, the correlation was interpreted as indicating the accumulation of refractory compounds after easily degradable compounds have been degraded due to high soil microbial activity. In our soils, signs of higher microbial activities under birch than under spruce had been reported, but, in contrast to our hypotheses, birch soil appeared to have less easily degradable DOM than spruce soil. It has been shown previously that DOM collected from lysimeters is mostly old and stable. We suggest that also DOM that we collect by water extraction or centrifugation-drainage technique is more a leftover product than the active part of SOM decomposition.
Measuring the total concentration of DOC, the amount of DOC that potentially could release or even the chemical properties or degradability of DOM at a certain moment does not tell us about the dynamic processes of DOM. In our soils, easily degradable DOM did not reflect microbial activity when measured at the certain moment. However, we suggest that the total flux of easily degradable DOM is higher under birch than under spruce (Kiikkilä et al. 2012). Few efforts to quantitatively assess the fluxes of various DOM fractions from the forest floor (Qualls et al. 1991) or the importance of refractory DOC to C storage in mineral soil (Kalbitz and Kaiser 2008; Klotzbücher et al. 2012) have been published. A more comprehensive view, by modeling the dynamic processes of DOM quantitatively, taking into account DOM properties, would increase our understanding of the origin and fate of DOM in soil and water ecosystems.
We thank A. Siika for the figures and M. Waller for checking the English. The research was supported by the Academy of Finland.
- Cajander AK (1949) Forest types and their significance. Acta For Fenn 56:1–71Google Scholar
- Kiikkilä O, Smolander V, Kitunen V (2013) Degradability, molecular weight and adsorption properties of dissolved organic carbon and nitrogen leached from different types of decomposing litter. Plant Soil 373:787–798Google Scholar
- Malik A, Blagodatskaya E, Gleixner G (2013) Soil microbial carbon turnover decreases with increasing molecular size. Soil Biol Biochem 62:115–118Google Scholar