Camera traps for density estimation: Calibration of camera traps using density estimates of voles from capture recapture methods
This study will investigate the ability of camera traps to estimate density of a broad set of small rodent species, including gray-sided vole (Myodes rufocanus), tundra vole (Microtus oeconomus), northern red-backed vole (Myodes rutilus), field vole (Microtus agrestis) and bank vole (Myodes glareolus) For such a calibration an accurate density estimate is crucial and live trapping data analyzed with capture-recapture methods is today seen as most accurate. This method requires animal to be trapped in live traps filled with food and water and checked at frequent intervals. When new animals are trapped they will be pit-tagged so they can be recognized in later encounters (recaptures).
If this study succeed, non-invasive camera traps can replace invasive live traps and snap traps (that kill the animals) that today are commonly used in small mammal research in Scandinavia, and elsewhere. Hence, this study can both replace, reduce and refine the use of live animals in future small rodent population monitoring. Moreover, researchers will be able to monitor small mammals with greater temporal and spatial resolution with lower costs then they have today and even with seasons when standard trapping is very difficult. We have done some pilot studies with the camera trap and we really think this method may change field studies of small rodents from being invasive to completely non-invasive. However, to convince the community of mammalogist that do research on small rodent population ecology in the field we need a robust calibration study with fairly big samples and trials on a set of different species. And that is what we intend to by this proposal.
Most small rodent species in Scandinavia have 3-5-year population cycles and we need to cover all years/phases of these cycles for all the species in two seasons, spring and fall, since movements patterns and trappability are both known to be phase-, season - and species-dependent. We need 10 spatial replicates of local populations of all the species for statistical purposes. With 10 spatial population replicates X 4-years X 2 seasons per species and assuming a mean capture per trap per trapping session of 5 animals (based on previous experience), we conclude that we will need to catch a maximum 400 individuals per species. With 5 species this yields a maximum of 2000 individuals over 4 years. We will fit calibration regressions with subsets of data as the study goes along and we get more information about measurement error. If we find that precise calibration is possible with less replicates or fewer trapped animals we will reduce number of replicates to lower the trapping effort which in turn will reduce number of animals trapped. However, if the effort is too low there is a great risk for the precision of the calibration to be too poor.
If this study succeed, non-invasive camera traps can replace invasive live traps and snap traps (that kill the animals) that today are commonly used in small mammal research in Scandinavia, and elsewhere. Hence, this study can both replace, reduce and refine the use of live animals in future small rodent population monitoring. Moreover, researchers will be able to monitor small mammals with greater temporal and spatial resolution with lower costs then they have today and even with seasons when standard trapping is very difficult. We have done some pilot studies with the camera trap and we really think this method may change field studies of small rodents from being invasive to completely non-invasive. However, to convince the community of mammalogist that do research on small rodent population ecology in the field we need a robust calibration study with fairly big samples and trials on a set of different species. And that is what we intend to by this proposal.
Most small rodent species in Scandinavia have 3-5-year population cycles and we need to cover all years/phases of these cycles for all the species in two seasons, spring and fall, since movements patterns and trappability are both known to be phase-, season - and species-dependent. We need 10 spatial replicates of local populations of all the species for statistical purposes. With 10 spatial population replicates X 4-years X 2 seasons per species and assuming a mean capture per trap per trapping session of 5 animals (based on previous experience), we conclude that we will need to catch a maximum 400 individuals per species. With 5 species this yields a maximum of 2000 individuals over 4 years. We will fit calibration regressions with subsets of data as the study goes along and we get more information about measurement error. If we find that precise calibration is possible with less replicates or fewer trapped animals we will reduce number of replicates to lower the trapping effort which in turn will reduce number of animals trapped. However, if the effort is too low there is a great risk for the precision of the calibration to be too poor.