Sensitivity of fine‐scale species distribution models to locational uncertainty in occurrence data across multiple sample sizes

Abstract

Summary1. To generate realistic predictions, species distribution models require the accurate coregistration of occurrence data with environmental variables. There is a common assumption that species occurrence data are accurately georeferenced; however, this is often not the case. This study investigates whether locational uncertainty and sample size affect the performance and interpretation of fine-scale species distribution models. 2. This study evaluated the effects of locational uncertainty across multiple sample sizes by subsampling and spatially degrading occurrence data. Distribution models were constructed for kelp (Ecklonia radiata), across a large study site (680 km 2 ) off the coast of southeastern Australia. Generalized additive models were used to predict distributions based on fine-resolution (2Á5 m cell size) seafloor variables, generated from multibeam echosounder data sets, and occurrence data from underwater towed video. The effects of different levels of locational uncertainty in combination with sample size were evaluated by comparing model performance and predicted distributions. 3. While locational uncertainty was observed to influence some measures of model performance, in general this was small and varied based on the accuracy metric used. However, simulated locational uncertainty caused changes in variable importance and predicted distributions at fine scales, potentially influencing model interpretation. This was most evident with small sample sizes. 4. Results suggested that seemingly high-performing, fine-scale models can be generated from data containing locational uncertainty, although interpreting their predictions can be misleading if the predictions are interpreted at scales similar to the spatial errors. This study demonstrated the need to consider predictions across geographic space rather than performance alone. The findings are important for conservation managers as they highlight the inherent variation in predictions between equally performing distribution models, and the subsequent restrictions on ecological interpretations.

Outstanding Challenges in the Transferability of Ecological Models

“…Model transfers can be further hampered by imperfect detectability, spatial and temporal biases in data collection, insufficient sample sizes, the omission of known drivers, or the use of proxy variables [27]. Additionally, species' characteristics such as range size can impact positional accuracy, leading to erroneous predictions if analyses are conducted at scales corresponding with those of the original locational errors [28]. The magnitude of these effects is ultimately unclear, and data quality therefore represents a substantial source of uncertainty [29].…”

Explicitly integrating a third dimension in marine species distribution modelling

“…MaxEnt; Phillips et al 2006. Use of misrepresentative environmental data, such as that resulting from positional error (Osborne & Leitão 2009, Mitchell et al 2017 or temporal mismatches be tween species and environmental sampling dates (Roubicek et al 2010), is known to reduce SDM performance. Osborne & Leitão (2009) hypothesised that, with positional errors, spatial autocorrelation of environmental data may mitigate substantial mismatch errors between occurrence and environment.…”

“…Ideally, environmental data associated with a species occurrence should be collected at the time and place of sampling to avoid spatial (Osborne & Leitão 2009, Mitchell et al 2017 or temporal (Roubicek et al 2010) mismatches between a confirmed presence and associated environmental data. Doing so is not, however, always practical.…”

The effect of positional error on fine scale species distribution models increases for specialist species

“…inaccurate location of species occurrences) is minimal or mainly associated with relatively older datasets that are often georeferenced from textual descriptions of their locations (which may cause errors of up to hundreds of meters, Wieczorek et al 2004). Additionally, where the marine environment is concerned, species data are often acquired using underwater cameras, in which case the positional error can be affected for example by the camera depth; the deeper the camera is, the greater is the positional error (Rattray et al 2014, Mitchell et al 2017. The positional error of GNSS data may be caused by the use of outdated technology, by poor satellite signal reception (e.g.…”