Benthic foraminifera of the central Indian Ocean

Late Oligocene to late Pliocene vertical water-mass stratification along depth traverses in the northern Indian Ocean is depicted in this paper by benthic foraminifer index faunas. During most of this time, benthic faunas indicate well-oxygenated, bottom-water conditions at all depths except under the southern Indian upwelling and in the Pliocene in the southern Arabian Sea. Faunas suggest the initiation of lower oxygen conditions at intermediate depths in the northern Indian Ocean beginning in Oligocene Zone P21a. Lower oxygen conditions intensified during primary productivity pulses, possibly related to increased upwelling vigor, in the latest Oligocene and throughout most of the late middle through late Miocene. During times of elevated primary production, there may be more oxygen flux into sedimentary pore waters and the shallow infaunal habitat may become more oxygenated. One criterion for locating the source of "new" water masses is vertical homogeneity of benthic foraminifer indexes for well-oxygenated water masses from intermediate through abyssal depths. In the northern Mascarene Basin, this type of faunal homogeneity with depth corroborates the proposal that the northern Indian Ocean was an area of sinking well-oxygenated waters through most of the Miocene before Zone N17. Oxygenated, possibly "new" intermediate-water masses in the low- to middle-latitude Mascarene and Central Indian basins first developed in the late Oligocene. These well-oxygenated waters were probably more fertile than the Antarctic Intermediate Waters (AAIW) that cover intermediate depths in these areas today. Production of intermediate waters more similar to modern AAIW is indicated by the sparse benthic population of epifaunal rotaloid species in the northern Mascarene Basin during middle Miocene Zone N9 and from early through late Pliocene time. Deep-water characteristics are more difficult to interpret because of the extensive redeposition at the deeper sites. Redeposited intermediate, rather than shallow, water fossils and erosion from north to south in the Mascarene Basin are incompatible with the sluggish circulation from south to north through the western Indian Ocean basins today. Such erosion could result from the vigorous sinking of an intermediate-depth water mass of northern origin. Before late Oligocene Zone P22, benthic faunas indicate a twofold subdivision of the troposphere, with the boundary between upper and lower well-oxygenated water masses located from 2500-3000 mbsl. No characteristic bottom-water fauna developed before the end of late Oligocene Zone P22. Deep and abyssal benthic indexes suggest the development of water masses similar to those of the present day in the latest Miocene. Faunas containing deep-water benthic indexes, including the uvigerinids, suggestive of a water mass similar to modern Indian Deep Water (IDW), appeared during the late Miocene in the northern Mascarene and Central Indian basins. In the early Pliocene, this deep-water fauna was found only in the Central Indian Basin, whereas a fauna typical of modern Antarctic Bottom Water (AABW) spread through deep waters at 2800 mbsl in the Mascarene Basin. By late Pliocene Zone N21, however, deep-water faunas similar to their modern analogs were developed in both the eastern and western basins. Abyssal faunas, studied only in the Mascarene Basin, show more or less similarity to those under modern AABW. Bottom-water faunas containing Nuttallides umbonifera or Epistominella exiguua were first differentiated at the end of Zone P22, then appeared episodically during the early Miocene. These AABW-type faunas reappeared and migrated updepth into deep waters during the glacial episodes at the end of the Miocene and at the beginning of the Pliocene. By late Pliocene Zone N21, however, a bottom-water fauna similar to that under eastern Indian Bottom Water (IBW) developed in the Mascarene Basin. Modern bottom-water characteristics of the Mascarene Basin must have developed after ZoneN21.

Stable oxygen isotope record and Calcium concentrations of sediment core GeoB3910-2

There is increasing evidence that the preceding Holocene climate was as unstable as the last glacial period, although variations occurred at much lower amplitudes. However, low-latitude climate records that confirm this variability are sparse. Here we present a radiocarbon-dated Holocene marine record from the tropical western Atlantic. Aragonite dissolution derived from the degree of preservation of the pteropod Limacina inflata records changes in the corrosiveness of the bottom water at the core site due to the changing influence of northern versus southern water masses. The delta18O difference between the shallow-living planktonic foraminifera Globigerinoides sacculifer and the deep-living Globorotalia tumida is used as proxy for changes in the vertical stratification of the surface water, hence the trade wind strength at this latitude. We compared our data to high-latitude records of the North Atlantic region. A good agreement is found between the aragonite dissolution and the strength in the Island-Scotland Overflow Water, which contributes significantly to the North Atlantic Deep Water. This suggests that large-scale variations in the Atlantic thermohaline circulation occurred throughout the Holocene. Concurrently, the comparison of our Delta delta18O with the GISP2 glaciochemical records points to global Holocene atmospheric reorganizations seen in both the tropics and high northern latitudes.

Geochemistry, lithology, and bulk density of Cricket Flat paleosol samples

The high-resolution marine isotope climate record indicates pronounced global cooling during the Langhian (16-13.8 Ma), beginning with the warm middle Miocene climatic optimum and ending with significant Antarctic ice sheet expansion and the transition to "icehouse" conditions. Terrestrial paleoclimate data from this interval is sparse and sometimes conflicting. In particular, there are gaps in the terrestrial record in the Pacific Northwest during the late Langhian and early Serravallian between about 14.5 and 12.5 Ma. New terrestrial paleoclimate data from this time and region could reconcile these conflicting records. Paleosols are particularly useful for reconstructing paleoenvironment because the rate and style of pedogenesis is primarily a function of surface environmental conditions; however, complete and well-preserved paleosols are uncommon. Most soils form in erosive environments that are not preserved, or in environments such as floodplains that accumulate in small increments; the resulting cumulic soils are usually thin, weakly developed, and subject to diagenetic overprinting from subsequent soils. The paleosol at Cricket Flat in northeastern Oregon is an unusually complete and well-preserved paleosol from a middle Miocene volcanic sequence in the Powder River Volcanic Field. An olivine basalt flow buried the paleosol at approximately 13.8 ± 0.6 Ma, based on three 40Ar/39Ar dates on the basalt. We described the Cricket Flat paleosol and used its physical and chemical profile and micromorphology to assess pedogenesis. The Cricket Flat paleosol is an Ultisol-like paleosol, chemically consistent with a high degree of weathering. Temperature and rainfall proxies suggest that Cricket Flat received 1120 ± 180 mm precipitation y-1 and experienced a mean annual temperature of 14.5 ± 2.1 °C during the formation of the paleosol, significantly warmer and wetter than today. This suggests slower cooling after the middle Miocene climatic optimum than is seen in the existing paleosol record.

Foraminifers from DSDP Leg 87 holes

Leg 87 investigated two sites in the Nankai Trough, off southeastern Japan, and one in the Japan Trench, off northeastern Japan. Several holes at the Nankai Trough sites penetrated mostly Quaternary interbedded sandy turbidites and hemipelagic mud. Foraminifers are common only in certain turbidite sands because both sites are at or just below the carbonate compensation depth. The planktonic assemblages from these sandy layers consist of mixed cool-temperate and warm-water species, and include both solution-resistant and solution-prone species. The benthic assemblages from these same layers are composed of mixtures of shelf to abyssal species. The northward-flowing Kuroshio is important in producing the mixed planktonic faunas, whereas turbidity currents are the primary agents in mixing benthic faunas and in the rapid burial of both planktonic and benthic foraminifers, which protects them from solution. Interbedded hemipelagic muds are barren or contain sparse faunas. Hole 582B penetrated through the trench-fill deposits into hemipelagic sediments that originated in the Shikoku Basin. These muds contain a dissolution facies of solution-resistant planktonic species, partially dissolved tests, and deep bathyal benthic species. Drilling at Site 584, on the landward midslope of the Japan Trench, penetrated a section of dominantly diatomaceous mudstone. This section contains a meager Pliocene calcareous fauna in its upper third and a nearly monospecific assemblage of Martinottiella communis in the lower two-thirds. Diatom biostratigraphy indicates that this change in assemblages occurs near the Miocene/Pliocene boundary. Similar biofacies changes are observed in neighboring sections drilled during Legs 56 and 57. The change from agglutinated to calcareous faunas is probably related to a relative drop in the carbonate compensation depth at the end of the Miocene.

Oxygen and hydrogen isotopes measured in the Northeast Atlantic

Only a few studies have examined the variation of oxygen and hydrogen isotopes of seawater in NE Atlantic water masses, and data are especially sparse for intermediate and deep-water masses. The current study greatly expands this record with 527 d18O values from 47 stations located throughout the mid- to low-latitude NE Atlantic. In addition, dD was analyzed in the 192 samples collected along the GEOTRACES North Atlantic Transect GA03 (GA03_e=KN199-4) and the 115 Iberia-Forams cruise samples from the western and southern Iberian margin. An intercomparison study between the two stable isotope measurement techniques (cavity ring-down laser spectroscopy and magnetic-sector isotope ratio mass spectrometry) used to analyze GA03_e samples reveals relatively good agreement for both hydrogen and oxygen isotope ratios. The surface (0-100 m) and central (100-500 m) water isotope data show the typical, evaporation related trend of increasing values equatorward with the exception for the zonal transect off Cape Blanc, NW Africa. Off Cape Blanc, surface water isotope signatures are modified by the upwelling of fresher Antarctic Intermediate Water (AAIW) that generally has isotopic values of 0.0 to 0.5 per mil for d18O and 0 to 2 per mil for dD. Along the Iberian margin the Mediterranean Outflow Water (MOW) is clearly distinguished by its high d18O (0.5-1.1 per mil) and dD (3-6 per mil) values that can be traced into the open Atlantic. Isotopic values in the NE Atlantic Deep Water (NEADW) are relatively low (d18O: -0.1 to 0.5 per mil; dD: -1 to 4 per mil) and show a broader range than observed previously in the northern and southern convection areas. The NEADW is best observed at GA03_e Stations 5 and 7 in the central NE Atlantic basin. Antarctic Bottom Water isotope values are relatively high indicating modification of the original Antarctic source water along the flow path. The reconstructed d18O-salinity relationship for the complete data set has a slope of 0.51, i.e., slightly steeper than the 0.46 described previously by Pierre et al. (1994, J. Mar. Syst. 5 (2), 159-170.) for the tropical to subtropical Northeast Atlantic. This slope decreases to 0.46 for the subtropical North Atlantic Central Water (NACW) and the MOW and to 0.32 for the surface waters of the upper 50 m. The dD-salinity mixing lines have estimated slopes of 3.01 for the complete data, 1.26 for the MOW, 3.47 for the NACW, and 2.63 for the surface waters. The slopes of the d18O-dD relationship are significantly lower than the one for the Global Meteoric Water Line with 5.6 for the complete data set, 2.30 for the MOW, 4.79 for the NACW, and 3.99 for the surface waters. The lower slopes in all the relationships clearly reflect the impact of the evaporation surplus in the subtropics.

Rock magentic investigations of ODP Site 162-983

We report natural remanent magnetization (NRM) directions and geomagnetic paleointensity proxies for part of the Matuyama Chron (0.9-2.2 Ma interval) from two sites located on sediment drifts in the Iceland Basin. At Ocean Drilling Program Sites 983 and 984, mean sedimentation rates in the late Matuyama Chron are 15.9 and 11.5 cm/kyr, respectively. For the older part of the record (>1.2 Ma), oxygen isotope data are too sparse to provide the sole basis for age model construction. The resemblance of the volume susceptibility record and a reference d18O record led us to match the two records to derive the age models. This match, based on Site 983/984 susceptibility, is consistent with available Site 983/984 benthic d18O data. Paleointensity proxies were derived from the slope of the NRM versus anhysteretic remanent magnetization plot for alternating field demagnetization in the 30-60 mT peak field range. Paleointensity lows correspond to polarity reversals at the limits of the Jaramillo, Olduvai, Cobb Mountain, and Réunion Subchrons and to seven excursions in NRM component directions. Magnetic excursions (defined here by virtual geomagnetic polar latitudes crossing the virtual geomagnetic equator) are observed at 932, 1048, 1115, 1190-1215 (Cobb Mountain Subchron), 1255, 1472-1480, 1567-1575 (Gilsa Subchron), and 1977 ka. The results indicate that geomagnetic directional excursions, associated with paleointensity minima, are a characteristic of the Matuyama Chron and probably of polarity chrons in general.

Distribution of benthic foraminifera in lower Cretaceous sediments of the Cuvuer Abyssal Plain in the Indian Ocean

Analysis of 66 samples from DSDP Site 263 (Cores 263-4R-4 to 263-29R-4) reveals a unique faunal composition with a predominance of agglutinated taxa, many of them previously unrecorded from any other DSDP and ODP Indian Ocean sites. A total of 66 agglutinated and 31 calcareous taxa are documented and five new species are described: Hippocrepina gracilis n.sp., "Textuluriopsis" elegans n.sp., Aaptotoichus challengeri n.sp., "Gaudryinopsis" pseudobettenstaedti n.sp. and "Gaudryina" cuvierensis n.sp. Three assemblages are recognized based on changes in the composition of dominant taxa and occurrences of stratigraphically important species: (1) a high-diversity Valanginian to Barremian Bulbobaculites-Recurvoides Assemblage (Cores 263-29R to - 18R), comprised of numerous elongate agglutinated forms, rare nodosariids, and variable numbers of tubes and ammodiscids; (2) a moderately diverse Aptian to Albian Rhizammina-Ammodiscus-Glomospira Assemblage (Cores 263-18R to -7R) with highly fluctuating numbers of the nominate taxa and Haplophragmoides, Trochammina, Verneuilinoides spp., and Verneuilina howchini; (3) a very low diversity Albian or younger Assemblage (Cores 263-6R to -4R) containing sparse agglutinated foraminifera, rare nodosariids and rotaliids. We interpret the assemblages as shelf to lower slope and consider them to reflect a deepening palaeobathymetry as the Cuvier margin subsided after the initial breakup of East Gondwana during the Valanginian. Our interpretation is in sharp contrast with initial palaeodepth estimates of less than 100 m, as well as with original chronostratigraphic interpretations based on foraminifera and nannofossils which correlated the base of the recovered interval with the Aptian. The absence of many cosmoplitan forms, despite high diversity suggests strong faunal differentiation in the Austral realm or endemisn within the Cuvier Basin during the Early Cretaceous.

Age determination and faunal composition of sediments off northern Greenland

We present a record encompassing marine isotope stages 7-1 from a hitherto unexplored and heavily ice-covered area of the Arctic Ocean, the Lomonosov Ridge off the northern Greenland-Canada continental margin, using nannofossil and benthic foraminifera stratigraphy. Planktic foraminifera assemblages are used as a key paleoceanographic proxy, and a surprisingly large variability is found for an interior Arctic Ocean site. Abundant small (63-125 µm) subpolar Turborotalita quinqueloba occur in two sections, possibly representing substages 5e (last interglacial) and 5a (warm interstadial). However, the present-day circulation pattern and the very distant location of high productive regions cannot explain such high abundances of subpolar specimens in the interior, perennially sea ice-covered Arctic Ocean. Hence our proxy record indicates that last interglacial sea ice concentrations were reduced off some areas of northern Greenland-Canada. Whether this was part of a larger regional pattern or it represents the influence of polynya areas with locally increased productivity remains to be solved. With respect to glacial conditions, increased ice-rafted debris (IRD) deposition in the area appears to be associated with glacial stages 6, 4, and late 3. Stage 2 sediments (including the Last Glacial Maximum) are condensed with a sparse IRD content only.

Silicoflagellates in ODP Leg 108 holes

The occurrence of Cenozoic silicoflagellates at three Ocean Drilling Program (ODP) Holes (660A, 662A, and 667A) was investigated to determine biostratigraphic and relative paleotemperature relations in the tropical Atlantic Ocean. This report presents the data obtained from a study of 37 samples and some preliminary comments on the data. The age of the single sparse assemblage at Hole 660A is late middle Eocene or late Eocene (Dictyocha hexacantha Zone); the sparse to common assemblages of Hole 667A are Oligocene and early Miocene and the common to abundant assemblages of Hole 662A are early Pliocene to Quaternary. Dissolution thinning of silicoflagellates is noted in most samples, even in Hole 662A, which is under the present productive Benguela Current.

Neogene palaeoceanographic and palaeoclimatic events inferred from palynological data: Cape Basin off South Africa, ODP Leg 175

Sites 1085, 1086 and 1087 were drilled off South Africa during Ocean Drilling Program (ODP) Leg 175 to investigate the Benguela Current System. While previous studies have focused on reconstructing the Neogene palaeoceanographic and palaeoclimatic history of these sites, palynology has been largely ignored, except for the Late Pliocene and Quaternary. This study presents palynological data from the upper Middle Miocene to lower Upper Pliocene sediments in Holes 1085A, 1086A and 1087C that provide complementary information about the history of the area. Abundant and diverse marine palynomorphs (mainly dinoflagellate cysts), rare spores and pollen, and dispersed organic matter have been recovered. Multivariate statistical analysis of dispersed organic matter identified three palynofacies assemblages (A, B, C) in the most continuous hole (1085A), and they were defined primarily by amorphous organic matter (AOM), and to a lesser extent black debris, structured phytoclasts, degraded phytoclasts, and marine palynomorphs. Ecostratigraphic interpretation based on dinoflagellate cyst, spore-pollen and palynofacies data allowed us to identify several palaeoceanographic and palaeoclimatic signals. First, the late Middle Miocene was subtropical, and sediments contained the highest percentages of land-derived organic matter, even though they are rich in AOM (palynofacies assemblage A). Second, the Late Miocene was cool-temperate and characterized by periods of intensified upwelling, increase in productivity, abundant and diverse oceanic dinoflagellate cysts, and the highest percentages of AOM (palynofacies assemblage C). Third, the Early to early Late Pliocene was warm-temperate with some dry intervals (increase in grass pollen) and intensified upwelling. Fourth, the Neogene "carbonate crash" identified in other southern oceans was recognized in two palynofacies A samples in Hole 1085A that are nearly barren of dinoflagellate cysts: one Middle Miocene sample (590 mbsf, 13.62 Ma) and one Upper Miocene sample (355 mbsf, 6.5 Ma). Finally, the extremely low percentages of pollen suggest sparse vegetation on the adjacent landmass, and Namib desert conditions were already in existence during the late Middle Miocene.

10Be data for sediments from DSDP Leg 56 and ODP Leg 170

Sediment accretion and subduction at convergent margins play an important role in the nature of hazardous interplate seismicity (the seismogenic zone) and the subduction recycling of volatiles and continentally derived materials to the Earth's mantle. Identifying and quantifying sediment accretion, essential for a complete mass balance across the margin, can be difficult. Seismic images do not define the processes by which a prism was built, and cored sediments may show disturbed magnetostratigraphy and sparse biostratigraphy. This contribution reports the first use of cosmogenic 10Be depth profiles to define the origin and structural evolution of forearc sedimentary prisms. Biostratigraphy and 10Be model ages generally are in good agreement for sediments drilled at Deep Sea Drilling Project Site 434 in the Japan forearc, and support an origin by imbricate thrusting for the upper section. Forearc sediments from Ocean Drilling Program Site 1040 in Costa Rica lack good fossil or paleomagnetic age control above the decollement. Low and homogeneous 10Be concentrations show that the prism sediments are older than 3-4 Ma, and that the prism is either a paleoaccretionary prism or it formed largely from slump deposits of apron sediments. Low 10Be in Costa Rican lavas and the absence of frontal accretion imply deeper sediment underplating or subduction erosion.

Late Quaternary siliciclastic/carbonate sedimentation in the Cabricorn Channel

A model is presented for hemipelagic siliciclastic and carbonate sedimentation during the last glacial-interglacial cycle in the Capricorn Channel, southern Great Barrier Reef (GBR). Stable isotope ratios, grainsize, carbonate content and mineralogy were analysed for seven cores in a depth transect from 166 to 2892 m below sea level (mbsl). Results show variations in the flux of terrigenous, neritic and pelagic sediments to the continental slope over the last sea level cycle. During the glacial lowstand terrigenous sediment influenced all the cores down to 2000 mbsl. The percentages of quartz and feldspar in the cores decreased with water depth, while the percentage of clay increased. X-ray diffraction analysis of the glacial lowstand clay mineralogy suggests that the siliciclastic sediment was primarily sourced from the Fitzroy River, which debouched directly into the northwest sector of the Capricorn Channel at this time. The cores also show a decrease in pelagic calcite and an increase in aragonite and high magnesium calcite (HMC) during the glacial. The influx of HMC and aragonite is most likely from reworking of coral reefs exposed on the continental shelf during the glacial, and also from HMC ooids precipitated at the head of the Capricorn Channel at this time. Mass accumulation rates (MARs) are high (13.5 g/cm**/kyr) during the glacial and peak at ~20 g/cm** 3/kyr in the early transgression (16-14 ka BP). MARs then decline with further sea level rise as the Fitzroy River mouth retreats from the edge of the continental shelf after 13.5 ka BP. MARs remain low (4 g/cm**3/kyr) throughout the Holocene highstand. Data for the Holocene highstand indicate there is a reduction in siliciclastic influx to the Capricorn Channel with little quartz and feldspar below 350 mbsl. However, fine-grained fluvial sediments, presumably from the Fitzroy River, were still accumulating on the mid slope down to 2000 mbsl. The proportion of pelagic calcite in the core tops increases with water depth, while HMC decreases, and is present only in trace amounts in cores below 1500 mbsl. The difference in the percentage of HMC in the deeper cores between the glacial and Holocene may reflect differences in supply or deepening of the HMC lysocline during the glacial. Sediment accumulation rates also vary between cores in the Capricorn Channel and do not show the expected exponential decrease with depth. This may be due to intermediate or deep water currents reworking the sediments. It is also possible that present bathymetry data are too sparse to detect the potential role that submarine channels may play in the distribution and accumulation of sediments. Comparison of the Capricorn Channel MARs with those for other mixed carbonate/siliciclastic provinces from the northeast margin of Australia indicates that peak MARs in the early transgression in the Capricorn Channel precede those from the central GBR and south of Fraser Island. The difference in the timing of the carbonate and siliciclastic MAR peaks along the northeast margin is primarily related to differences in the physiography and climate of the provinces. The only common trend in the MARs from the northeast margin of Australia is the near synchronicity of the carbonate and siliciclastic MAR peaks in individual sediment cores, which supports a coeval sedimentation model.

(Table 1) Calcite, siderite, and stable isotopes d13C and d18O at DSDP Hole 93-603B

Diagenesis of the fine-grained, feldspathic sandstones in the Lower Cretaceous submarine fan complex cored in DSDP Hole 603B can be considered to have occurred in three stages: (1) replacement of matrix and framework grains by pyrite, siderite, phillipsite (?), and particularly by ferroan calcite; (2) dissolution of ferroan calcite and feldspars to produce secondary macroporosity; and (3) development of sparse feldspar and quartz overgrowths, and authigenic modification of remnant matrix. Only ferroan calcite is a volumetrically important diagenetic mineral phase (up to 50 vol.%). Matrix in thin sandstone turbidite deposits has been extensively replaced by ferroan calcite. Carbon stable isotope data suggest that organic diagenesis had only a minor influence on calcite precipitation. Oxygen stable isotope data indicate that the minimum average calcite precipitation temperature was 40° C. Preliminary calculations show that steadystate diffusion of Ca+ + from the dissolution of nannoplankton skeletal material in the interbedded pelagic marls to the associated sandstones is a feasible transport mechanism. A thick sandstone unit from 1234-1263 m sub-bottom is extensively replaced by calcite near the upper and lower contacts. Farther into the sand body away from the contacts, the sandstone has good secondary porosity resulting from the dissolution of ferroan calcite that partially replaced matrix and framework grains. The central portion of the thick sand appears to be a channel with high-energy clean sand. We believe that the channel provided a conduit for focused flow of diagenetic compactional fluids responsible for dissolution. Focused flow may be the result of the earlier lithification of the pelagic limestones and thin-bedded sandstones which, then formed vertical permeability barriers. Calcite dissolution has occurred and may still be occurring at temperatures less than 65°C.

On nonlinearity in neural encoding models applied to the primary visual cortex

Within the regression framework, we show how different levels of nonlinearity influence the instantaneous firing rate prediction of single neurons. Nonlinearity can be achieved in several ways. In particular, we can enrich the predictor set with basis expansions of the input variables (enlarging the number of inputs) or train a simple but different model for each area of the data domain. Spline-based models are popular within the first category. Kernel smoothing methods fall into the second category. Whereas the first choice is useful for globally characterizing complex functions, the second is very handy for temporal data and is able to include inner-state subject variations. Also, interactions among stimuli are considered. We compare state-of-the-art firing rate prediction methods with some more sophisticated spline-based nonlinear methods: multivariate adaptive regression splines and sparse additive models. We also study the impact of kernel smoothing. Finally, we explore the combination of various local models in an incremental learning procedure. Our goal is to demonstrate that appropriate nonlinearity treatment can greatly improve the results. We test our hypothesis on both synthetic data and real neuronal recordings in cat primary visual cortex, giving a plausible explanation of the results from a biological perspective.

Learning an L1-regularized Gaussian Bayesian Network in the Equivalence Class Space

Learning the structure of a graphical model from data is a common task in a wide range of practical applications. In this paper, we focus on Gaussian Bayesian networks, i.e., on continuous data and directed acyclic graphs with a joint probability density of all variables given by a Gaussian. We propose to work in an equivalence class search space, specifically using the k-greedy equivalence search algorithm. This, combined with regularization techniques to guide the structure search, can learn sparse networks close to the one that generated the data. We provide results on some synthetic networks and on modeling the gene network of the two biological pathways regulating the biosynthesis of isoprenoids for the Arabidopsis thaliana plant