Can mud (silt and clay) concentration be used to predict soil organic carbon content within seagrass ecosystems?

The emerging field of blue carbon science is seeking cost-effective ways to estimate the organic carbon content of soils that are bound by coastal vegetated ecosystems. Organic carbon (Corg) content in terrestrial soils and marine sediments has been correlated with mud content (i.e. silt and clay), however, empirical tests of this theory are lacking for coastal vegetated ecosystems. Here, we compiled data (n = 1345) on the relationship between Corg and mud (i.e. silt and clay, particle sizes <63 μm) contents in seagrass ecosystems (79 cores) and adjacent bare sediments (21 cores) to address whether mud can be used to predict soil Corg content. We also combined these data with the δ13C signatures of the soil Corg to understand the sources of Corg stores. The results showed that mud is positively correlated with soil Corg content only when the contribution of seagrass-derived Corg to the sedimentary Corg pool is relatively low, such as in small and fast growing meadows of the genera Zostera, Halodule and Halophila, and in bare sediments adjacent to seagrass ecosystems. In large and long-living seagrass meadows of the genera Posidonia and Amphibolis there was a lack of, or poor relationship between mud and soil Corg content, related to a higher contribution of seagrass-derived Corg to the sedimentary Corg pool in these meadows. The relative high soil Corg contents with relatively low mud contents (i.e. mud-Corg saturation) together with significant allochthonous inputs of terrestrial organic matter could overall disrupt the correlation expected between soil Corg and mud contents. This study shows that mud (i.e. silt and clay content) is not a universal proxy for blue carbon content in seagrass ecosystems, and therefore should not be applied generally across all seagrass habitats. Mud content can only be used as a proxy to estimate soil Corg content for scaling up purposes when opportunistic and/or low biomass seagrass species (i.e. Zostera, Halodule and Halophila) are present (explaining 34 to 91% of variability), and in bare sediments (explaining 78% of the variability).

Thurstonian Boltzmann machines: learning from multiple inequalities

We introduce Thurstonian Boltzmann Machines (TBM), a unified architecture that can naturally incorporate a wide range of data inputs at the same time. Our motivation rests in the Thurstonian view that many discrete data types can be considered as being generated from a subset of underlying latent continuous variables, and in the observation that each realisation of a discrete type imposes certain inequalities on those variables. Thus learning and inference in TBM reduce to making sense of a set of inequalities. Our proposed TBM naturally supports the following types: Gaussian, intervals, censored, binary, categorical, muticategorical, ordinal, (in)-complete rank with and without ties. We demonstrate the versatility and capacity of the proposed model on three applications of very different natures; namely handwritten digit recognition, collaborative filtering and complex social survey analysis. Copyright 2013 by the author(s).