Information content in time series of litter decomposition studies and the transit time of litter in arid lands

SARQUIS, AGUSTÍN; SIERRA, CARLOS A.

BIOGEOSCIENCES

COPERNICUS PUBLICATIONS

Lugar: Gottingen; Año: 2023 vol. 20 p. 1759 - 1771

1726-4170

Plant litter decomposition stands at the intersec- tion between carbon (C) loss and sequestration in terrestrial ecosystems. During this process organic matter experiences chemical and physical transformations that affect decompo- sition rates of distinct components with different transforma- tion fates. However, most decomposition studies only fit one- pool models that consider organic matter in litter as a sin- gle homogenous pool and do not incorporate the dynamics of litter transformations and transfers into their framework. As an alternative, compartmental dynamical systems are sets of differential equations that serve to represent both the het- erogeneity in decomposition rates of organic matter and the transformations it can undergo. This is achieved by including parameters for the initial proportion of mass in each compart- ment, their respective decomposition rates, and mass trans- fer coefficients between compartments. The number of com- partments as well as their interactions, in turn, determine the model structure. For instance, a one-pool model can be con- sidered a compartmental model with only one compartment. Models with two or more parameters, in turn, can have differ- ent structures, such as a parallel one if each compartment de- composes independently or in a series if there is mass transfer from one compartment to another. However because of these differences in model parameters, comparisons in model per- formance can be complicated. In this context we introduce the concept of transit time, a random variable defined as the age distribution of particles when they are released from a system, which can be used to compare models with differ- ent structures. In this study, we first asked what model struc- tures are more appropriate to represent decomposition from a publicly available database of decomposition studies in aridlands: aridec. For this purpose, we fit one- and two-pool de- composition models with parallel and series structures, com- pared their performance using the bias-corrected Akaike in- formation criterion (AICc) and used model averaging as a multi-model inference approach. We then asked what the po- tential ranges of the median transit times of litter mass in arid lands are and what their relationships with environmen- tal variables are. Hence, we calculated a median transit time for those models and explored patterns in the data with re- spect to mean annual temperature and precipitation, solar ra- diation, and the global aridity index. The median transit time was 1.9 years for the one- and two-pool models with a paral- lel structure and 5 years for the two-pool series model. The information in our datasets supported all three models in a relatively similar way and thus our decision to use a multi- model inference approach. After model averaging, the me- dian transit time had values of around 3 years for all datasets. Exploring patterns of transit time in relation to environmen- tal variables yielded weak correlation coefficients, except for mean annual temperature, which was moderate and negative. Overall, our analysis suggests that current and historical lit- ter decomposition studies often do not contain information on how litter quality changes over time or do not last long enough for litter to entirely decompose. This makes fitting accurate mechanistic models very difficult. Nevertheless, the multi-model inference framework proposed here can help to reconcile theoretical expectations with the information con- tent from field studies and can further help to design field experiments that better represent the complexity of the litter decomposition process.