Beyond data gaps: how prior knowledge in the Gulf of California drives spatial biases in whale survey effort
Abstract
We analysed whale survey routes conducted by multiple research groups to evaluate whether sampling-limited or knowledge-limited inventory processes better explain observed gaps in whale records. Using data from the Gulf of California, survey effort was quantified as the number of navigation routes crossing grid cells at five spatial resolutions. Its spatial distribution was examined in relation to environmental and geographic variables using Generalised Linear Models. Results indicate that survey effort was highly concentrated across all spatial resolutions, with fewer than 10% of grid cells accounting for the highest levels of sampling. Environmental predictors explained only a limited proportion of this spatial variation. However, shallow bathymetry, colder waters, and higher productivity were consistently associated with greater survey effort. Analyses of residuals and spatial autocorrelation revealed systematic underestimation of effort in a small number of heavily surveyed areas, indicating sampling intensities far exceeding those expected based solely on environmental conditions. These over-surveyed areas coincided with independently identified hotspots of whale occurrence, suggesting that survey effort has been preferentially directed towards locations already known to support high whale densities. Biases affecting biodiversity databases are commonly interpreted as resulting from insufficient sampling; however, they may also reflect the influence of prior knowledge guiding where data are collected. Our results therefore support a knowledge-limited inventory scenario, in which accumulated experience and expectations shape survey design and reinforce existing spatial biases.