Is perception of the whole based on perception of its parts? There is psychological and physiological evidence for parts-based representations in the brain, and certain computational theories of object recognition rely on such representations. But little is known about how brains or computers might learn the parts of objects. Here we demonstrate an algorithm for non-negative matrix factorization that is able to learn parts of faces and semantic features of text. This is in contrast to other methods, such as principal components analysis and vector quantization, that learn holistic, not parts-based, representations. Non-negative matrix factorization is distinguished from the other methods by its use of non-negativity constraints. These constraints lead to a parts-based representation because they allow only additive, not subtractive, combinations. When non-negative matrix factorization is implemented as a neural network, parts-based representations emerge by virtue of two properties: the firing rates of neurons are never negative and synaptic strengths do not change sign.

Learning parts of objects by non-negative matrix factorization

Changes in oceanic primary production, linked to changes in the network of global biogeochemical cycles, have profoundly influenced the geochemistry of Earth for over 3 billion years. In the contemporary ocean, photosynthetic carbon fixation by marine phytoplankton leads to formation of approximately 45 gigatons of organic carbon per annum, of which 16 gigatons are exported to the ocean interior. Changes in the magnitude of total and export production can strongly influence atmospheric CO2 levels (and hence climate) on geological time scales, as well as set upper bounds for sustainable fisheries harvest. The two fluxes are critically dependent on geophysical processes that determine mixed-layer depth, nutrient fluxes to and within the ocean, and food-web structure. Because the average turnover time of phytoplankton carbon in the ocean is on the order of a week or less, total and export production are extremely sensitive to external forcing and consequently are seldom in steady state. Elucidating the biogeochemical controls and feedbacks on primary production is essential to understanding how oceanic biota responded to and affected natural climatic variability in the geological past, and will respond to anthropogenically influenced changes in coming decades. One of the most crucial feedbacks results from changes in radiative forcing on the hydrological cycle, which influences the aeolian iron flux and, in turn, affects nitrogen fixation and primary production in the oceans.

/pdf/biogeochemical-controls-and-feedbacks-on-ocean-primary-4zutq6flns.pdf

Biogeochemical controls and feedbacks on ocean primary production

Diatoms are unicellular algae with plastids acquired by secondary endosymbiosis. They are responsible for approximately 20% of global carbon fixation. We report the 34 million-base pair draft nuclear genome of the marine diatom Thalassiosira pseudonana and its 129 thousand-base pair plastid and 44 thousand-base pair mitochondrial genomes. Sequence and optical restriction mapping revealed 24 diploid nuclear chromosomes. We identified novel genes for silicic acid transport and formation of silica-based cell walls, high-affinity iron uptake, biosynthetic enzymes for several types of polyunsaturated fatty acids, use of a range of nitrogenous compounds, and a complete urea cycle, all attributes that allow diatoms to prosper in aquatic environments.

/pdf/the-genome-of-the-diatom-thalassiosira-pseudonana-ecology-2hwz6jrxaw.pdf

The Genome of the Diatom Thalassiosira Pseudonana: Ecology, Evolution, and Metabolism

Biogeochemistry of the Human Environment.- The Biosphere.- Soils.- Waters.- Air.- Plants.- Humans.- Biogeochemistry of Trace Elements.- Trace Elements of Group 1 (Previously Group Ia).- Trace Elements of Group 2 (Previously Group IIa).- Trace Elements of Group 3 (Previously Group IIIb).- Trace Elements of Group 4 (Previously Group IVb).- Trace Elements of Group 5 (Previously Group Vb).- Trace Elements of Group 6 (Previously Group VIb).- Trace Elements of Group 7 (Previously Group VIIb).- Trace Elements of Group 8 (Previously Part of Group VIII).- Trace Elements of Group 9 (Previously Part of Group VIII).- Trace Elements of Group 10 (Previously Part of Group VIII).- Trace Elements of Group 11 (Previously Group Ib).- Trace Elements of Group 12 (Previously Group IIb).- Trace Elements of Group 13 (Previously Group IIIa).- Trace Elements of Group 14 (Previously Group IVa).- Trace Elements of Group 15 (Previously Group Va).- Trace Elements of Group 16 (Previously Group VIa).- Trace Elements of Group 17 (Previously Group VIIa).

/pdf/trace-elements-from-soil-to-human-gyujylwhl8.pdf

Trace Elements From Soil to Human

The net inputs of silicic acid (dissolved silica) to the world ocean have been revised to 6.1 +/- 2.0 teramoles of silicon per year (1 teramole = 10(12) moles). The major contribution (about 80 percent) comes from rivers, whose world average silicic acid concentration is 150 micromolar. These inputs are reasonably balanced by the net ouputs of biogenic silica of 7.1 +/- 1.8 teramoles of silicon per year in modern marine sediments. The gross production of biogenic silica (the transformation of dissolved silicate to particulate skeletal material) in surface waters was estimated to be 240 +/- 40 teramoles of silicon per year, and the preservation ratio (opal accumulation in sediment/gross production in surface waters) averages 3 percent. In the world ocean the residence time of silicon, relative to total biological uptake in surface waters, is about 400 years.

https://science.sciencemag.org/content/sci/268/5209/375.full.pdf

The silica balance in the world ocean: a reestimate.

During the European Polarstern Study (EPOS 1988/1989) in the Weddell and Scotia Seas, five series of metal enrichment experiments were carried out with natural plankton communities under ultraclean conditions. Despite a clear stimulation of growth by the addition of Fe, control bottles (no additions) also showed rapid buildup of Chl a and complete utilization of a major nutrient within 2 weeks, indicating nonlimiting ambient Fe levels. Effects of Mn additions were less pronounced or absent, whereas extra additions of Zn and Cu in one experiment showed little or no effect. The species composition of the plankton community, monitored by HPLC pigment analysis and microscopic observations, changed in favor of diatoms when Fe was added. The addition of Fe also caused an increase in microzooplankton densities and concentrations of pigment breakdown products. However, metal-mediated shifts in the plankton community were minor compared to major changes resulting from incubation. Changes were most pronounced in experiments where microzooplankton was strongly developed, presumably as a result of excluding mesozooplankton from the bottles. Fe had an impact on plankton growth and species composition, but other factors seem to be responsible for keeping phytoplankton productivity far from its potential in these Antarctic waters.

Metal enrichment experiments in the Weddell-Scotia Seas

/pdf/metal-enrichment-experiments-in-the-weddell-scotia-seas-44jhlld8xl.pdf

Metal enrichment experiments in the Weddell-Scotia Seas: Effects of iron and manganese on various plankton communities

-During leg 2 (November 1988-Jan- uary 1989) of the European Polarstern study (EPOS) in the Weddell-Scotia Confluence zone, we implemented a new, simple, rapid, and sen- sitive method based on 32Si (a P-emitter in equi- librium with 32P) to determine biogenic silica (BSi) production. This method provides simultaneous determination of the production of organic P. Rates of BSi production ranged from 0.09 to 0.95 mmol Si m-2 h-', and the BSi content of the photic layer ranged from 7 to 220 mmol m-2. The Si:C production molar ratios varied be- tween 0.08 (small cell density of diatoms in com- munities dominated by cryptophyceans and dinophyceans) and 0.8 1 (diatom-dominated populations), while the C: P production ratios varied between 47 and 151 with an average of 105, in agreement with the Redfield ratio. We give here the first direct evidence that in the well- mixed waters of the Scotia Sea highly silicified diatom blooms may occur and that unlike earlier descriptions, because of the intense grazing pres- sure of krill in the marginal ice zone (MIZ) of the Weddell Sea, the remaining siliceous phytoplank- ton can grow with only moderate Si production rates. We conclude that the contribution of the MIZ to the total BSi production of the Southern Ocean must not be overstated.

https://aslopubs.onlinelibrary.wiley.com/doi/pdf/10.4319/lo.1991.36.6.1217

Aleido J. van Bennekom

Papers

The silica balance in the world ocean: a reestimate.

Metal enrichment experiments in the Weddell-Scotia Seas

Metal enrichment experiments in the Weddell-Scotia Seas: Effects of iron and manganese on various plankton communities

Production of Biogenic Silica in the Weddell-Scotia Seas Measured with 32Si