In ecology, a common form of statistical analysis relates a biological variable to variables that delineate the physical environment, typically by fitting a regression model or one of its extensions. Unfortunately, the biological data and the physical data are frequently obtained from eparate sources of data. In such cases there is no guarantee that the biological and physical data are co-located and the regression model cannot be used. A common and pragmatic solution is to predict the physical variables at the locations of the biological variables and then to use the predictions as if they were observations.We show that this procedure can cause potentially misleading inferences and we use generalized linear models as an example. We propose a Berkson error model which overcomes the limitations. The differences between using predicted covariates and the Berkson error model are illustrated by using data from the marine environment, and a simulation study based on these data.

The CARS2006 database is derived from all available historical subsurface ocean property measurements (Ridgway et al, 2002). The measurements have been collected primarily using research vessel instrument profiles and autonomous profiling buoys. The observations have been collected over approximately 50 years and have been used to provide an estimate at every depth and every location in the world's oceans for each day of the year, but not for any individual year. CARS2006 spans the southern 2/3 of the world's oceans, from 70o S to 26o N, except in the Atlantic where is reaches only to10o N. The six water properties mapped in are temperature (deg C), salinity (PSU), oxygen (ml/litre), nitrate (micromole/litre), silicate (micromole/litre), phosphate (micromole/litre). It comprises historic mean fields and average seasonal cycles, derived from all available historical subsurface ocean property measurements (primarily research vessel instrument casts and autonomous profiling buoys). There are 12 grids in the dataset. Two for each of the six water properties: mean and standard deviation. Please see the metadata for more detailed information.

Geoscience Australia carried out marine surveys in Jervis Bay (NSW) in 2007, 2008 and 2009 (GA303, GA305, GA309, GA312) to map seabed bathymetry and characterise benthic environments through colocated sampling of surface sediments (for textural and biogeochemical analysis) and infauna, observation of benthic habitats using underwater towed video and stills photography, and measurement of ocean tides and wavegenerated currents. Data and samples were acquired using the Defence Science and Technology Organisation (DSTO) Research Vessel Kimbla. Bathymetric mapping, sampling and tide/wave measurement were concentrated in a 3x5 km survey grid (named Darling Road Grid, DRG) within the southern part of the Jervis Bay, incorporating the bay entrance. Additional sampling and stills photography plus bathymetric mapping along transits was undertaken at representative habitat types outside the DRG. This 81 sample dataset comprises acidextractable concentrations of trace elements (Fe, Mn, Co, Ni, Cu, Zn, Ge, As, Cd and Pb) in surface seabed sediments (~0 to 2 cm) from Jervis Bay.

The ArcINFO grid is one of the outputs of a finescale hydrodynamic model, the Simulating WAves Nearshore (SWAN) model (Booij et al., 1999; Ris et al., 1999).The grid describes the modelled maximum orbital velocity (m/s) which can be used as estimation of seabed exposure on the Carnarvon Shelf.

Geoscience Australia carried out marine surveys in southeast Tasmania in 2008 and 2009 (GA0315) to map seabed bathymetry and characterise benthic environments through observation of habitats using underwater towed video. Data was acquired using the Tasmania Aquaculture and Fisheries Institute (TAFI) Research Vessel Challenger. Bathymetric mapping was undertaken in seven survey areas, including: Freycinet Pensinula (83 sq km, east coast and shelf); Tasman Peninsula (117 sq km, east coast and shelf); Port Arthur and adjacent open coast (17 sq km); The Friars (41 sq km, south of Bruny Island); lower Huon River estuary (39 sq km); D Entrecastreaux Channel (7 sq km, at Tinderbox north of Bruny Island), and; Maria Island (3 sq km, western side). Video characterisations of the seabed concentrated on areas of bedrock reef and adjacent seabed in all mapped areas, except for D Entrecastreaux Channel and Maria Island. The dataset contains 8 bathymetry grids produced from the processed EM3002 bathymetry data using the CARIS HIPS and SIPS software. Please see the metadata informaiton for detailed information.

Geoscience Australia carried out marine surveys in Jervis Bay (NSW) in 2007, 2008 and 2009 (GA303, GA305, GA309, GA312) to map seabed bathymetry and characterise benthic environments through co-located sampling of surface sediments (for textural and biogeochemical analysis) and infauna, observation of benthic habitats using underwater towed video and stills photography, and measurement of ocean tides and wave-generated currents. Data and samples were acquired using the Defence Science and Technology Organisation (DSTO) Research Vessel Kimbla. Bathymetric mapping, sampling and tide/wave measurement were concentrated in a 3x5 km survey grid (named Darling Road Grid, DRG) within the southern part of the Jervis Bay, incorporating the bay entrance. Additional sampling and stills photography plus bathymetric mapping along transits was undertaken at representative habitat types outside the DRG. This 42 sample data set comprises the mineraology of surface seabed sediment (~0-2 cm) in Jervis Bay. .

Geoscience Australia carried out a marine survey on Lord Howe Island shelf (NSW) in 2008 (SS06-2008) to map seabed bathymetry and characterise benthic environments through co-located sampling of surface sediments and infauna, rock coring, observation of benthic habitats using underwater towed video, and measurement of ocean tides and wave-generated currents. Sub-bottom profile data was also collected to map sediment thickness and shelf stratigraphy. Data and samples were acquired using the National Facility Research Vessel Southern Surveyor. Bathymetric data from this survey was merged with other pre-existing bathymetric data (including LADS) to generate a grid covering 1034 sq km. As part of a separate Geoscience Australia survey in 2007 (TAN0713), an oceanographic mooring was deployed on the northern edge of Lord Howe Island shelf. The mooring was recovered during the 2008 survey following a 6 month deployment. Seabed sediment samples were collected by a Smith McIntyre grab at a total of 36 locations (see Brooke et al. 2010). The full sample set represents 16 sampling stations, with multiple (2 or 3) samples taken at 10 of these stations.

This chapter presents a broad synthesis and overview based on the 57 case studies included in Part 2 of this book, and on questionnaires completed by the authors. The case studies covered areas of seafloor ranging from 0.15 to over 1,000,000 km2 (average of 26,600 km2) and a broad range of geomorphic feature types. The mean depths of the study areas ranged from 8 to 2,375 m, with about half of the studies on the shelf (depth <120 m) and half on the slope and at greater depths. Mapping resolution ranged from 0.1 to 170 m (mean of 13 m). There is a relatively equal distribution of studies among the four naturalness categories: near-pristine (n=17), largely unmodified (n = 16), modified (n=13) and extensively modified (n=10). In terms of threats to habitats, most authors identified fishing (n=46) as the most significant threat, followed by pollution (n=12), oil and gas development (n=7) and aggregate mining (n=7). Anthropogenic climate change was viewed as an immediate threat to benthic habitats by only three authors (n=3). Water depth was found to be the most useful surrogate for benthic communities in the most studies (n=17), followed by substrate/sediment type (n=14), acoustic backscatter (n=12), wave-current exposure (n=10), grain size (n=10), seabed rugosity (n=9) and BPI/TPI (n=8). Water properties (temperature, salinity) and seabed slope are less useful surrogates. A range of analytical methods were used to identify surrogates, with ARC GIS being by far the most popular method (23 out of 44 studies that specified a methodology).

Ecoregions are defined in terms of community structure in function of abiotic or even anthropogenic forcing. They are mesoscale structures defined on the potential habitat of species or predicted communities geographic extent. We assume that they can be more easily defined for long-lived species such as benthos or neritic fish in the marine environment. Uncertainties exist for the pelagic realm because of its higher variability, plus little is known about the meso- and bathypelagic zones. A changing environment and modifications of habitats will probably drive new communities from plankton to fish or top predators. We need based-line studies such as those of CAML, databases like SCAR-MarBIN and tools for integrating all of these observations. Our objective is to understand the biodiversity patterns in the Southern Ocean and how these might change.

Mapping of benthic habitats seldom considers biogeochemical variables or changes across time. We aimed to: (i) develop winter and summer benthic habitat maps for a sandy embayment; and (ii) compare the effectiveness of various maps for differentiating infauna. Patch-types (internally homogeneous areas of seafloor) were constructed using combinations of abiotic parameters, and are presented in sediment-based, biogeochemistry-based and combined sediment/biogeochemistry-based habitat maps. August and February surveys were undertaken in Jervis Bay, Australia, to collect samples for physical (%mud, sorting, %carbonate), biogeochemical (chlorophyll a, sulfur, sediment metabolism, bio-available elements) and infaunal analyses. Boosted Decision Tree and cokriging models generated spatially continuous data-layers. Habitat maps were made from classified layers using GIS overlays, and were interpreted from a biophysical-process perspective. Biogeochemistry and %mud varied spatially and temporally, even in visually homogeneous sediments. Species turnover across patch-types was important for diversity, and the utility of habitat maps for differentiating biological communities varied across months. Diversity patterns were broadly related to reactive carbon and redox which varied temporally. Inclusion of biogeochemical factors and time in habitat maps provides a better framework for differentiating species and interpreting biodiversity patterns than once-off studies based solely on sedimentology or video-analysis.