Habitat suitability models for the Siberian jay (Perisoreus infaustus) from Citizen Science and systematic monitoring data: incorporating information about the reporting process

Date:
2018/06/12

Time:
15:30

Room:
K301 Felix

Habitat suitability models for the Siberian jay (Perisoreus infaustus) from Citizen Science and systematic monitoring data: incorporating information about the reporting process

(Oral)

Ute Bradter
ute.bradter@slu.se

Swedish Species Information Centre, Swedish University of Agricultural Sciences, Uppsala, Sweden
,
Louise Mair
Swedish Species Information Centre, Swedish University of Agricultural Sciences, Uppsala, Sweden
,
Mari Jönsson
Swedish Species Information Centre, Swedish University of Agricultural Sciences, Uppsala, Sweden
,
Jonas Knape
Department of Ecology, Swedish University of Agricultural Sciences, Uppsala, Sweden
,
Tord Snäll
Swedish Species Information Centre, Swedish University of Agricultural Sciences, Uppsala, Sweden

DOI: 10.17011/conference/eccb2018/107740

SEE PEER REVIEW

Opportunistically collected presence-only data contributed by volunteer reporters, so called Citizen Science data, are increasingly available for species and regions that lack systematic surveys. However, it is unclear if or how much the biases in opportunistically collected data influence different habitat suitability modelling methods and hence if they can be used with confidence to address different conservation questions.

We evaluated habitat suitability models with opportunistically collected observations against models with systematically collected observations for a forest bird species, the Siberian jay (Perisoreus infaustus) in Sweden. Citizen Science data were obtained from the Swedish Species Observation System and systematically collected data from the Swedish Bird Survey.

The opportunistically collected presence-only data were enhanced by adding high-quality inferred species absences. We obtained these by combining information on reporting behaviour and species identification skills of a few, very active reporters with their observations. We evaluated logistic regression with inferred absences, two versions of MaxEnt, a model combining presence-absence with presence-only observations and a Bayesian site-occupancy-detection model.

All modelling methods produced nationwide habitat suitability maps of Siberian jay that agreed well with results from systematically collected data. At finer geographic scales there were differences among methods. Logistic regression with inferred absences produced results most similar to those from the systematic survey.

Adding high-quality inferred absences to opportunistically collected data is likely possible for many less common species. They may facilitate the modelling and prediction of species distributions in areas or for species for which systematically collected data are not available.

SEE PEER REVIEW

INTRO:
The authors describe a study where they inferred absences for opportunistically collected occurrence data to cope with biases in these kind of data. When comparing preditctions of models incorporating these absences with various other distribution models using only the opportunistic presence, they find that including inferred absences produces more relieble results at a finer spatial scale.
MERITS:
I think the abstract describes an interesting and innovative way for adding absence information to presence-only data and thereby for improving the quality and use of sporadic distribution data. It also seems that they compare their method to an adequate variety of other methods.
CRITIQUE:
I could not think of anything specific that the abstract lacks. It is also well written and does not require any editing.
DISCUSSION:
I think this abstract could be a valuable input at the conference that would probably interest many participants. Developing ways for dealing with biases in sporadical occurrence data is timely as there are large quantities of such data becoming available and an urgent need to enable their use for various purposes, including species conservation.

- - -
INTRO:
There has been earlier attempts to address the problems of utilizing opportunistically sampled observational data. In that sense the research idea is not as unique as the authors are arguing. However, the study setting (comparing two methodologies) of this research is interesting and supports well the research in this field.
MERITS:
It is an interesting approach to promote the presence-only data to presence-absence by defining an observer specific detection probability. This could have a good applicability with other species as well.
CRITIQUE:
The use of Maxent software is not supported by previous studies, which have mostly utilized a Poisson point process. That would probably be a more solid modelling methodology for comparing the two data sets. Furthermore, the robustness of estimated covariate effects would be an interesting point to look at when comparing the models.
DISCUSSION:
The research is solid and would have a clear continuing by comparing different species (rare vs. common) with the methodology.

Habitat suitability models for the Siberian jay (Perisoreus infaustus) from Citizen Science and systematic monitoring data: incorporating information about the reporting process

Date:2018/06/12

Time:15:30

Room:K301 Felix

Habitat suitability models for the Siberian jay (Perisoreus infaustus) from Citizen Science and systematic monitoring data: incorporating information about the reporting process

(Oral)

SEE PEER REVIEW

SEE PEER REVIEW

Date:
2018/06/12

Time:
15:30

Room:
K301 Felix