Premium
Predictive modelling of ecological patterns along linear‐feature networks
Author(s) -
Ladle Andrew,
Avgar Tal,
Wheatley Matthew,
Boyce Mark S.
Publication year - 2017
Publication title -
methods in ecology and evolution
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 3.425
H-Index - 105
ISSN - 2041-210X
DOI - 10.1111/2041-210x.12660
Subject(s) - feature (linguistics) , feature selection , computer science , metric (unit) , foothills , kriging , linear model , set (abstract data type) , variable (mathematics) , geostatistics , data mining , ecology , cartography , geography , machine learning , spatial variability , statistics , mathematics , mathematical analysis , philosophy , linguistics , operations management , economics , biology , programming language
Summary Ecological patterns and processes often take place within linear‐feature networks, and this has implications when analysing the spatial configuration of such patterns or processes across a landscape. One such pattern is the use of landscapes by human recreationists: an important variable in animal habitat selection and behaviour. Due to the difficulty in obtaining data, ecologists tend to use coarse metrics such as linear‐feature density, while the extent and timing of human activity are often ignored. Remote detector equipment and its increasing use in ecological studies allow for large volumes of data on human activity to be collected. However, the analysis of these data still can be challenging. Using a combination of generalised linear mixed‐effects models and network‐based ordinary kriging, we developed a method for estimating spatial and temporal variations in motorised and non‐motorised activities across a complex linear‐feature network. Trail cameras were set up between 2012 and 2014 and monitored motorised and non‐motorised activities at 238 different trail sites across a 2824 km 2 region of the eastern slopes and foothills of central Alberta's Rocky Mountains. We evaluate the predictive capacity of this approach, demonstrate its application and discuss its merits and limitations. This method offers a straightforward analysis that can be applied to remotely acquired data to give a useful metric for assessing wildlife responses to human activity, and has potential application beyond the highlighted example.