During my postdoc I’ve had the opportunity to take part in a really interesting piece of research on landscape change. The research was part of the EU project HERCULES, and explored the driving forces of landscape change in Europe. Although literature reviews is the classic starting point of academic research, systematic reviews are increasingly used in all fields of research. We followed the guidelines from Centre for Evidence-Based Conservation: Guidelines for Systematic Review in Environmental Management.
The guidelines divide the systematic review into 8 steps:
Question > Protocol > Search > Article screening > Critical appraisal and data extraction > Data synthesis > Write report > Finalised Review
As Europe underwent transformations in the last two hundred years, its landscape has also dramatically changed. Societal and industrial transformations, as well as large scale events like both World Wars, have shaped the landscape in many different ways; although common history is likely to have produced common changes. Many researchers have investigated these changes at a regional scale; so our research intended on bringing together all these pieces to find trends of landscape change drivers.
We defined objective criteria to select which scientific articles we would be looking at:
- landscape had undergone anthropogenic change
- research had in-depth field investigations,
- and published quantitative change
- the study area considered regional to local scale within Europe (1–10,000 km2)
- discussed the driving forces of landscape change in the study areas
We used four database of scientific articles: ISI Web of Science, GEOBASE (Ovid), CABI: CAB Abstracts (Ovid), and Scopus. Each database had different syntax, so it was a bit of a challenge to perform our complex search in each.
We searched the terms:
“Landscape change” OR “landscape dynamics”,
Europe* OR EU OR Albania OR Andorra OR Armenia OR Austria OR Azerbaijan OR Belarus OR Belgium OR “Bosnia and Herzegovina” OR Bulgaria OR Croatia OR Cyprus OR Czech* OR Denmark OR Estonia OR Finland OR France OR Georgia OR Germany OR Greece OR Hungary OR Iceland OR Ireland OR Italy OR Kazakhstan OR Latvia OR Liechtenstein OR Lithuania OR Luxembourg OR Malta OR Moldova OR Monaco OR Montenegro OR Netherlands OR Norway OR Poland OR Portugal OR Romania OR Russia OR “San Marino” OR Serbia OR Slovak* OR Slovenia OR Spain OR Sweden OR Switzerland OR Macedonia OR Turkey OR Ukraine OR “United Kingdom” OR England OR Wales OR Scotland
Not all databases allowed the same criteria (search in all text, abstract, keywords, and so on) or the same number of criteria (that was a lot of keywords!).
Some of the caveats in this search were:
- landscape change could refer to something else than transformation of landscape, and so we picked up many studies in geology for instance,
- some studies in landscape change do not use landscape change, but land use or land cover change,
- including the list of countries sometimes included articles for which the authors was in one of these countries
- it also included places such as Georgia (US) or New England.
- and probably the biggest caveat; it focused on articles with at least an abstract or keywords in English. Although the search yielded some articles in German, French, Polish, Spanish, among others, if the database was not properly filled then we could not read the article. Since the articles we were after related to local to regional change, it is very likely that some of the research published has been written in a regional language. However, most of the articles not published in English did not fell into our criteria, and most often were either not quantitative or not addressing the drivers of landscape change.
After going through the four databases, we put all the articles together, and obtained 4034 papers. Obviously, there were up to 4 duplicates per article, and after using Endnote, zotero, and Qiqqa to remove most of the duplicates (these reference managers did not find the same duplicates!), we had 2190 papers left.
This part of the research was definitely part of the most boring hours of my working life. Digitising almost sounded appealing at the time! To filter through the papers with our criteria, I read first the titles, then the abstract, and finally the entire article. To ensure that my selection was not biased, Tobias (first author of the paper) also performed the selection on a subset (10%) and we compared and discussed our selection. In case of doubt, I also kept the article in the selection. The number of article decreased as follow:
We then included some articles suggested by experts on landscape change, and which had not been found through our selection process. It gave us 144 articles to start our study.
Not all articles were of similar quality; considering the scientific merit of the study, the quality of English, or the language. Some of them were actually of very low quality, while a few of them were really interesting. On the bright side, I have to say not only the quality was increasing with time, and the latest articles were of much higher quality, but also, in total objectivity, Tobias’ papers (with whom I was working) were really among the best.
Critical appraisal and data extraction
I read the 144 articles left, and synthesised the findings in an excel spreadsheet.
Data synthesis > Write report > Finalised Review
Tobias, first author of the paper and expert in landscape studies, synthesised the findings, and wrote most of the article we just published.
Based on these articles, we found that:
- Land abandonment is the most frequent driver of landscape change in Europe.
- Combinations of underlying drivers determine landscape change, not single drivers.
I also made the maps (in QGIS), and graphs (in Scribus). Land Use Policy, in which the article is published, also allowed us to share the location of each case study area as a KML.