Effects of initial microbial biomass abundance on respiration during pine litter decomposition

Microbial biomass is increasingly used to predict respiration in soil organic carbon (SOC) models. Its increased use combined with the difficulty of accurately measuring this variable points a need to directly assess the importance of microbial biomass abundance for carbon (C) cycling. To test the hypothesis that the initial microbial biomass abundance (i.e. biomass abundance on new plant litter) is a strong driver of plant litter C cycling, we manipulated biomass abundance by 10 and 100-fold dilution and composition using 12 source communities on sterile pine litter and measured respiration in microcosms for 30 days. In the first two days of microbial growth on fresh litter, a 100-fold difference in initial biomass abundance caused an average difference in respiration of nearly 300%, but the effect rapidly declined to less than 30% in 10 days and to 14% in 30 days. Parallel simulations with a soil carbon model, SOMIC 1.0, also predicted a 14% difference over 30 days, consistent with the experimental results. Model simulations predicted convergence of cumulative CO2 to within 10% in three months and within 4% in three years. Rapid microbial growth, evidenced by appearance of visible microbial mats on the litter during the first week of incubation, likely attenuates the effects of large initial differences in biomass abundance. In contrast, the persistence of source community as an explanatory factor in driving differences in respiration across microcosms supports the importance of microbial composition in C cycling. Overall, the results suggest that the initial abundance of microbial biomass on litter is a weak driver of C flux from litter decomposition over long timescales (months to years) when litter communities have equal nutrient availability. By extension, slight variation in the timing of microbial dispersal to fresh litter is likely to be a minor factor in long-term C flux. Importance Microbial biomass is one of the most common microbial parameters used in land carbon (C) cycle models, however, it is notoriously difficult to measure accurately. To understand the consequences of mismeasurement, as well as the broader importance of microbial biomass abundance as a direct driver of ecological phenomena, greater quantitative understanding of the role of microbial biomass abundance in environmental processes is needed. Using microcosms, we manipulated the initial biomass of numerous microbial communities across a 100-fold range and measured effects on CO2 production during plant litter decomposition. We found that the effects of initial biomass abundance on CO2 production was largely attenuated within a week, while the effects of community type remained significant over the course of the experiment. Overall, our results suggest that initial microbial biomass abundance in litter decomposition within an ecosystem is a weak driver of long-term C cycling dynamics.

Reviewer #1: This is a nicely designed, executed and described study showing how initial microbial biomass affects sterilized pine litter degradation in a microcosm experiment. I particularly like the combination of modelling and experimental approaches -it is very informative and shows that the SOMIC model works well also for litter. The paper is very well written, all the parts are of adequate length and convey necessary information (with minor points in Materials and Methods). I don't have any concerns of methodological nature nor pertaining to data presentation.
We thank the reviewer for their positive feedback.
I have only a couple of questions/suggestions of relatively minor importance: 1. The sequencing system could be better described: is it based on custom primers or native Illumina ones? Bioinformatics analysis description suggests the latter, but it is not clear.
We thank the reviewer for pointing out this missing information. We have added text regarding the custom primers in the methods (Lines 222-225).
2. Why have the Authors chosen Pearson's correlation? It assumes linear relation of variables, which might be true or not.
We wanted to test if initial estimates of biomass were directly correlated with respiration and we expected that if correlated, the correlation would be linear. We also visually inspected the graphs and we did not observe a non-linear trend or in fact any correlative trend.
3. The Authors erroneously quote ANOVA instead PERMANOVA when stating that initial biomass affected bacterial community composition and did not affect fungal community (l. 315-316).
We than the reviewer for catching this error. 4. It is not stated how the percent of variance explained was calculated in PERMANOVA analysis. As % of MS or SS? Please specify in the M&Ms.
Cumulative respiration was not significantly correlated with any measurement We have added this information in the M&Ms (Lines 275-276). Percent of variance was calculated as a % of MS. 5. It is not stated in the text which soils were used in the Experiment 2 -it follows from figures, but please put this information in the text.
We have added this information to the M&Ms text (Line 208).
I would like to point out that the paper lacks the accession number for the SRA record containing reads generated in the study, but as the Authors state that the record is being prepared, I think it's fine.
Data is now published in the NCBI SRA under Bioproject PRJNA601499 Reviewer #2: In this manuscript, the authors use experimentation and modelling to assess the quantitative contribution of initial biomass as a driver of litter decomposition. These data are rare and as such the manuscript makes a good contribution to the field. It is generally a well written manuscript but the structuring could be improved to streamline the discussion of the results. I have some specific points below mostly surrounding improving the presentations of hypothesis and the results as well some questions on data analysis. Otherwise the data and the analysis in manuscript are sound and I endorse its publication in PLOS One after those concerns are met.
We thank the reviewer for their positive assessment of the study.
The authors highlight that the rationale of their hypothesis surrounding use of initial biomass abundance is the fact that microbial biomass is increasingly being used in soil C models. But as far as I understand, models often use biomass as a dynamic pool. So more than assessing the implications of biomass incorporation into models, I would say that the study is aimed at investigating the role of initial biomass in litter decomposition.
We agree with the reviewer and we state these points in Lines 85-105 in the introduction. The important distinction that we make on Lines 97-105 is about the relative importance of initial versus equilibrium biomass. If both these biomass pools are equally important then experimentalists need to measure both in field studies, but if initial biomass abundance is only important over very short timescales then this is useful for experimentalists to understand to prioritize resource allocations. We have attempted to clarify further on line 103.
Order of results and figures is random. For eg. figure 3 shows respired CO2 over time and comes after cumulative CO2 was presented in figure 1 followed by analysis of variance. Figure 4 is presenting modelled data. I suggest structuring the results to present the actual data and patterns first followed by statistics and modelling outputs.
We agree with the reviewer that in many instances the reviewer's proposed results structure is most logical, but we do not feel that this is the best approach for this paper. In this paper, the results/figures are not random, they are arranged by concept so that readers do not need to jump between key points. We also have 2 experiments which creates another logical split. The ordered concepts include: 1. Importance of variables tested (Initial biomass vs Source community) (Expt 1) 2. Observed vs Modeled Results (CO2 attenuation) (Expt 1) 3. Community composition (Expt 2).
The authors do not show biomass shifts over the period of the incubation. Did the biomass change over time? They modelled it over time (Fig 4) and use those results to imply that biomass rapidly increases to the environment's carrying capacity.
We have added text (Line 42-43) to indicate that microbial growth was apparent in microcosms from visual inspection. This statement is also qualified in the text by a 'likely' thus we acknowledge that this is not measured.
I wonder if the predicted biomass numbers were verified by observed data. If it's not available, that would be major limitation of the study.
We are glad that the reviewer found the predicted biomass an interesting aspect of the paper. The key conclusions in the paper do not depend on the temporal biomass dynamics, as the focus of the study is on the impacts of initial biomass abundance. Modeling the biomass dynamics is a valuable conceptual addition, but not crucial quantitative data upon which the main conclusions were based.
It looks like the reduction in respiration with diluted initial biomass differed for the various communities used. Was the reduction higher in particular type of community?
The reviewer brings up an interesting point. Looking at this aspect of impact of community type would be interesting but this is beyond the scope of this study. One would need a means to group the 10 communities into 2 or 3 different types in order to do an analysis. We do not have a sensible way of defining community types with this data. We chose to use bar charts because presenting DNA abundance in this way is common in published literature. There are no error bars because the data represent the original soil and multiple DNA extractions were not performed. The only discrepancy that we can find is a red line that was dashed (Fig 3) and solid (Fig 4). We have revised Figure 3 to make the red line solid.
It would also be ideal to merge these two figures to show 6 plots for the observed and modelled changes in respired CO2 over time.
We thank the reviewer for this suggestion. We have added a panel so that  We have added results from a posthoc Tukey HSD test to show significance in Figure 5.