As we move into the second century of publishing Journal of Ecology, it’s a good time to see how the Journal fared in its 100th year. With this in mind, we compared uptake of the 150 papers published in the Centenary volume 100 in 2012. To make it interesting, we look at citations as they are the grist of our obsessive compulsive Impact Factor chase, and a number of the upstart altmetrics – the new kid on the block. Citation counts were obtained from Web of Science while altmetrics were obtained from Altmetric.com using the rAltmetric package in R. A visual comparison and summary of these metrics is provided in the Heat Map below showing uptake of Journal of Ecology papers from 2012 according to citations and altmetrics. Each row corresponds to a paper. Papers are rank-ordered vertically according to Altmetric score (Alt). M=Mendeley readers, Tw=Twitter mentions, WoS=Web of Science citations, FB=Facebook wall posts, CiteUL=CiteULike readers, g+ = Google+ posts, Feeds = Blog posts. Scores were log10 transformed to improve visual interpretation.
Citations take a while to accumulate, so it is perhaps no surprise that the papers with the most citations so far (6 each) include two from issue one. These are Bullock et al.’s paper on dispersal under climate change, and Malhi’s paper on carbon cycling in tropical forests. Both of these were part of our Centenary Symposium Special Feature. However, Baraloto et al.’s paper on functional traits and phylogeny of tropical tree communities has also already attracted 6 citations, and it came out in issue 3.
The altmetrics tell a different story, but that story depends on which altmetric you look at. The Altmetric score is a general measure of the attention that an article has received online and for ecologists might be thought of like a diversity measure such as Shannon’s H’. Latz et al.’s paper on the interaction between plant diversity, soil pathogens and bacterial communities had the largest Altmetric score (17.6), followed by Crawford & Rudgers’s paper on the importance of genetic diversity of a dominant species on plant community biomass (14.5).
Overall, there were tweets about 119 papers, the most popular being the controversial Forum paper by Rees et al. on measuring competition and stress (9 tweets), followed by García-Palacios et al.’s review on spatial heterogeneity and climate change (8 tweets) and Crawford & Rudgers’s paper (7 tweets). The twitterverse seems to notice Journal of Ecology papers!
Ninety-eight papers were picked up by Mendeley users. Of these, Baralato et al., was again a big hit accumulating 55 readers, followed by Spasojevic & Suding’s study of community assembly and functional traits (54 readers). Eighteen other papers accumulated 20 or more readers.
The number of posts about Journal of Ecology papers from 2012 on Google+ (4 posts), Blog feeds (7 posts), CiteULike (13 readers), and Facebook (26 wall posts) were limited in comparison to uptake by Twitter and Mendeley. There were no Reddit submissions, no posts in internet forums (e.g. SEQanswers), and no bookmarked articles on Delicious.
Statistically, how are the altmetrics related? The altmetric score was positively correlated with several altmetrics, i.e., Twitter counts (Spearman’s rank r=0.39, P<0.0001, n=128), Blog posts (r=0.39, P<0.0001), and Google+ posts (r=0.18, P=0.04). However, there were limited statistical correlations among the other altmetrics. Twitter counts and Mendeley uptake were negatively correlated (r = – 0.21, P=0.01), whereas Mendeley uptake was positively correlated with Web of Science citations (r=0.30, P=0.0009) and CiteULike shares (r=0.26, P=0.028). Correlations among other metrics were not significant reinforcing the view that citation counts and the various altmetrics focus on different aspects of article use by readers.
What does all this tell us? The analysis presented here scratches the surface as it is preliminary and based upon a limited sample of 150 papers from a single year. However, it would seem, as perhaps suspected, that with only a few exceptions, each metric of journal article performance reflects a different use by readers. Does this mean that we can ignore some metrics because we are not interested in that audience? Or, do we have to flog ourselves to death chasing all metrics? Do one or two metrics best reflect the mission of our journal? These are questions that journal editors have to think about, and at Journal of Ecology we are only just starting to grapple with these issues. Of course we want our article citations to be high as they reflect use of our publications by the academic community. But professional journals have a broader mission than that and we want to serve the research community in many ways. It is clear that altmetrics are a new way of helping us understand how we are doing.
UPDATE: Thanks to @altmetric for suggesting that we look at article downloads in our survey. This was so obvious a metric to consider, that we, um, forgot about it. So, now we do…
During 2012 there were 156,331 downloads of abstracts for Journal of Ecology articles published in 2012, and 84,677 full-text downloads. Of these, abstracts of 10 papers were downloaded over 2,000 times, with Zhang et al.’s meta-analysis of forest productivity and species and trait diversity topping the list with 2,891 downloads. There were five articles with over 2,000 full text downloads, with Moles’ et al.’s article on exotics species invasion and disturbance from the Centenary Symposium Special Feature having the largest number of downloads at 2,331. There did not appear to be much correspondence between highly downloaded articles and those with either the highest number of citations (so far) or high altmetric scores except perhaps Baraloto et al.’s study of forest tree phylogeny which is being highly cited and was also one of the papers with the highest number of Mendeley readers.
Abstract downloads were positively correlated with several altmetrics including twitter mentions (Spearman’s Rank Correlation r=0.17, P=0.049, n=127), blog posts (r=0.26, P=0.003), Mendeley readers (r=0.30, P=0.0006), and most strongly with altmetric score (r=0.45, P<0.0001). Full-text downloads were correlated with the similar altmetrics including blog posts (r=0.26, P=0.003), Mendeley (r=0.59, P<0.0001), CiteULike readers (r=0.23, P=0.008) and altmetric score (r=0.19, P=0.03) (not mentioned correlations were all nonsignificant). While these relationships again indicate the variety of ways in which Journal of Ecology articles are used, the relationship between full-text downloads and Mendeley and CiteULike readers is consistent with “traditional” use by academics as providing source material for their own work.
David Gibson, Andrea Baier and Scott Chamberlain