Mapping species distributions with social media geo-tagged images: case studies of bees and flowering plants in Australia

Moataz Medhat ElQadi, Alan Dorin, Adrian Dyer, Martin Burd, Zoë Bukovac, Mani Shrestha

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Data sources on species distribution and range are typically expensive and time consuming to build, and traditional survey techniques often have spatial, temporal, or scale-related gaps. Social network sites, on the other hand, can provide massive amounts of cost effective data that may potentially yield information of direct benefit to supplement and understand ecological phenomena. Previous research explored using social network site content to enhance information collected by experts or professional surveys in domains including species distribution and land cover. However, the data quality and general suitability of social network sites data for answering questions related to species distribution and range is highly variable and this aspect of its value to science remains underexplored. In this research we investigate some causes of social network site data unreliability and explore how to mitigate them. We filter data points based on our estimates of reliability and relevance. We then use the filtered data to infer species ranges and distributions in concert with Global Biodiversity Information Facility (GBIF) data. Our proposed methodology was applied to four Australian case studies including two insect pollinators, and two flowering plants. The case studies were chosen from Australia because of its unique geographical features, large landmass, sparse population, and the many tourists and residents who travel across it taking photos and sharing them through social media. We show that, despite some barriers, there are instances where the social network site data clearly complement the existing source, making our technique a valuable means of making repeatable, efficient additions to traditional species distribution data.

Original languageEnglish
Pages (from-to)23-31
Number of pages9
JournalEcological Informatics
Volume39
DOIs
Publication statusPublished - May 2017

Keywords

  • Bio-diversity data
  • Geo-tagged images
  • Social network sites
  • Species distribution mapping

Cite this

@article{a2e736773771441ea2aa9b6ff8a76fd8,
title = "Mapping species distributions with social media geo-tagged images: case studies of bees and flowering plants in Australia",
abstract = "Data sources on species distribution and range are typically expensive and time consuming to build, and traditional survey techniques often have spatial, temporal, or scale-related gaps. Social network sites, on the other hand, can provide massive amounts of cost effective data that may potentially yield information of direct benefit to supplement and understand ecological phenomena. Previous research explored using social network site content to enhance information collected by experts or professional surveys in domains including species distribution and land cover. However, the data quality and general suitability of social network sites data for answering questions related to species distribution and range is highly variable and this aspect of its value to science remains underexplored. In this research we investigate some causes of social network site data unreliability and explore how to mitigate them. We filter data points based on our estimates of reliability and relevance. We then use the filtered data to infer species ranges and distributions in concert with Global Biodiversity Information Facility (GBIF) data. Our proposed methodology was applied to four Australian case studies including two insect pollinators, and two flowering plants. The case studies were chosen from Australia because of its unique geographical features, large landmass, sparse population, and the many tourists and residents who travel across it taking photos and sharing them through social media. We show that, despite some barriers, there are instances where the social network site data clearly complement the existing source, making our technique a valuable means of making repeatable, efficient additions to traditional species distribution data.",
keywords = "Bio-diversity data, Geo-tagged images, Social network sites, Species distribution mapping",
author = "ElQadi, {Moataz Medhat} and Alan Dorin and Adrian Dyer and Martin Burd and Zo{\"e} Bukovac and Mani Shrestha",
year = "2017",
month = "5",
doi = "10.1016/j.ecoinf.2017.02.006",
language = "English",
volume = "39",
pages = "23--31",
journal = "Ecological Informatics",
issn = "1574-9541",
publisher = "Elsevier",

}

Mapping species distributions with social media geo-tagged images : case studies of bees and flowering plants in Australia. / ElQadi, Moataz Medhat; Dorin, Alan; Dyer, Adrian; Burd, Martin; Bukovac, Zoë; Shrestha, Mani.

In: Ecological Informatics, Vol. 39, 05.2017, p. 23-31.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Mapping species distributions with social media geo-tagged images

T2 - case studies of bees and flowering plants in Australia

AU - ElQadi, Moataz Medhat

AU - Dorin, Alan

AU - Dyer, Adrian

AU - Burd, Martin

AU - Bukovac, Zoë

AU - Shrestha, Mani

PY - 2017/5

Y1 - 2017/5

N2 - Data sources on species distribution and range are typically expensive and time consuming to build, and traditional survey techniques often have spatial, temporal, or scale-related gaps. Social network sites, on the other hand, can provide massive amounts of cost effective data that may potentially yield information of direct benefit to supplement and understand ecological phenomena. Previous research explored using social network site content to enhance information collected by experts or professional surveys in domains including species distribution and land cover. However, the data quality and general suitability of social network sites data for answering questions related to species distribution and range is highly variable and this aspect of its value to science remains underexplored. In this research we investigate some causes of social network site data unreliability and explore how to mitigate them. We filter data points based on our estimates of reliability and relevance. We then use the filtered data to infer species ranges and distributions in concert with Global Biodiversity Information Facility (GBIF) data. Our proposed methodology was applied to four Australian case studies including two insect pollinators, and two flowering plants. The case studies were chosen from Australia because of its unique geographical features, large landmass, sparse population, and the many tourists and residents who travel across it taking photos and sharing them through social media. We show that, despite some barriers, there are instances where the social network site data clearly complement the existing source, making our technique a valuable means of making repeatable, efficient additions to traditional species distribution data.

AB - Data sources on species distribution and range are typically expensive and time consuming to build, and traditional survey techniques often have spatial, temporal, or scale-related gaps. Social network sites, on the other hand, can provide massive amounts of cost effective data that may potentially yield information of direct benefit to supplement and understand ecological phenomena. Previous research explored using social network site content to enhance information collected by experts or professional surveys in domains including species distribution and land cover. However, the data quality and general suitability of social network sites data for answering questions related to species distribution and range is highly variable and this aspect of its value to science remains underexplored. In this research we investigate some causes of social network site data unreliability and explore how to mitigate them. We filter data points based on our estimates of reliability and relevance. We then use the filtered data to infer species ranges and distributions in concert with Global Biodiversity Information Facility (GBIF) data. Our proposed methodology was applied to four Australian case studies including two insect pollinators, and two flowering plants. The case studies were chosen from Australia because of its unique geographical features, large landmass, sparse population, and the many tourists and residents who travel across it taking photos and sharing them through social media. We show that, despite some barriers, there are instances where the social network site data clearly complement the existing source, making our technique a valuable means of making repeatable, efficient additions to traditional species distribution data.

KW - Bio-diversity data

KW - Geo-tagged images

KW - Social network sites

KW - Species distribution mapping

UR - http://www.scopus.com/inward/record.url?scp=85016240531&partnerID=8YFLogxK

U2 - 10.1016/j.ecoinf.2017.02.006

DO - 10.1016/j.ecoinf.2017.02.006

M3 - Article

VL - 39

SP - 23

EP - 31

JO - Ecological Informatics

JF - Ecological Informatics

SN - 1574-9541

ER -