Data was collected as described in:  Bradley, M., Baker, R., Nagelkerken, I., & Sheaves, M. (2019). Context is more important than habitat type in determining use by juvenile fish. Landscape Ecology, 34(2), 427–442. doi:10.1007/s10980-019-00781-3

Video were collected from a range of locations throughout the Indo-Pacific, including: Australia (Queensland, Northern Territory, Western Australia), Papua New Guinea (New Britain), French Polynesia (Tahiti). Videos are point census surveys of fish assemblages, and are either unbaited or baited. This provided a point census of fish taxa present, as well as an indication of local biological and structural habitat characteristics. Surveys were conducted during daylight hours and periods of relatively low turbidity to ensure conditions appropriate for video sampling. Accordingly, sampling occurred between June and December, outside the Austral monsoon. Sampling was carried out during the full breadth of variation in tidal inundation in a region both in terms of the tidal cycle (high-low) and tidal periods (springs and neaps). Independence of video samples was maintained by leaving more than 20 m between deployments, with cameras usually spaced 50–100 m apart. Efforts were made to sample the entire breadth of habitat variation present in the region. Initially, the study areas were sampled randomly, with cameras placed at random intervals from the shore, to achieve broad spatial replication. After reviewing the metadata, targeted sampling was performed to boost replication in substrate/biota combinations with low representation, though for some rare combinations of biota and substrate, this could not be achieved. 

This data record also includes a series of spreadsheets. This data was prepared as described in: Bradley, M., Nagelkerken, I., Baker, R., Travers, M., & Sheaves, M. (2021). Local environmental context structures animal-habitat associations across biogeographic regions. Ecosystems (in press). Videos were viewed to record the taxa present, and fish identifications were reviewed by relevant experts. Only presence/absence of families are included in these spreadsheets. To ensure that all analyses were robust to the influence of rare taxa, taxonomic assemblage data were treated in three different ways.

In the first more conventional treatment, an arbitrary dataset-wide cut-off was used to determine the inclusion of taxa in the analysis. All taxa present in less than 10% of total sites were excluded, which removed all taxa with low occurrences regardless of location-specific occurrence rates, providing a dataset with minimal influence of rare taxa: IPMangrove_Fish_Family_total_percentage_treatment_1.

In the second, more inclusive treatment, taxa present in at least 10% of sites from any one location were retained in the analyses. This only removed taxa that were rare throughout all locations, and provided a dataset where all non-trivial differences between locations would be maintained: IPMangrove_Fish_Family_site_percentage_treatment_2.csv

This treatment resulted in substantial outliers in subsequent analysis. These outliers were excluded from the dataset, providing a third and final data treatment: IPMangrove_Fish_Family_site_percentage_excl_outliers_treatment_3

Spreadsheets contain the following columns. Variables are described in full detail in the above publication. video.number : unique identifier for each video sample region.site : text location identifier log.tide.amp.max : log of maximum tidal amplitude for the location of the video sample log.distance.reef : log of distance to reef salinity : salinity of each sample (ppt) biogeog : biogeographical region