**Note from Chao**: John and I investigated the research focus and collaboration of faculties at Odum School of Ecology. I posted the analyses on the diverse research areas a few days ago. John has graciously agreed to analyze the collaboration network and write this guest blog. Here are what he has to say about the collaborative research in Odum School of Ecology.

Chao and I, driving back from Atlanta, began discussing the diversity of research areas in the Odum school. We were amused that the school’s foundations are based in ecosystem ecology (the quote from Odum “*The whole is greater than the sum of its parts” *comes to mind) but now hosts multiple scientists focusing in disease, population, community, freshwater, marine, terrestrial, evolutionary, and conservation ecology.*

The scientific community at Odum represents many branches of ecology and these often intersect resulting in new ideas and findings. But, how often do the intersections happen and are there individuals who are central in creating these intersections? Here we report the results of our investigation and include all data and code for full reproducibility in R.

**What we did…**

When thinking about how to approach this we decided we only wanted to only focus on current, regular faculty. From these individuals, we surveyed each individuals publication list counting the number of publications they share with another faculty member. Data was primarily collected from Google Scholar. However, if the individual did not have have a Google Scholar profile, then we used either ResearchGate or the CV/publication list from their website.

This results in a symmetric matrix (*M*) with the diagonal containing the total number of papers that a faculty member has published. All other elements (*M _{ij}*) are the number of publications that faculty member

*j*co-authored with faculty member

*i*.

The average number of publications that an individual has published is 63.2 (sd=50.7). The number of publications range from 8 to 210 papers. On average, a faculty member has co-authored a total of 10.23 papers with any faculty member. The number of co-authored publications range from 0 to 58 papers.

require(igraph) fac.data = as.matrix(read.csv("OSE_Pub_sharing_R.csv", row.names = 1, header=TRUE)) fac.data.manip = fac.data diag(fac.data.manip) = 0 fac.data.manip = 2*fac.data.manip/max(fac.data.manip) par(mfrow=c(1,2)) boxplot(diag(fac.data), horizontal=TRUE, main="Total Published") points(mean(diag(fac.data)),1, pch=16, col="firebrick") boxplot(rowSums(fac.data.manip.1), horizontal=TRUE, main="Total Co-Authored") points(mean(rowSums(fac.data.manip.1)),1, pch=16, col="firebrick") mtext("Number of Publications", side=1, line=-1, outer=TRUE, cex=2.5)

Figure 1. Boxplot representing the number of publications that each faculty member has published (left panel) and the number of publications co-authored with a fellow faculty member (right panel). The middle bar represents the median number of publications and red point is the mean number of publications. The box represents the interquartile range with the left limit as the first quartile and the right limit as third quartile. The whiskers represent the minimum (left) and the maximum (right) within 1.5 of the interquartile range. The point represents an outlier in the data.

The matrix can be visualized as a undirected, weighted network where each node is a faculty member and an edge represents a co-authorship. The weight of each edge is a linear scaling of the number of co-authored publications. The left side of the figure shows the full network using all faculty members. To clearly show all the faculty connections, the right side shows a zoom in on the connections in the lower/bottom cluster.

fac.graph.2 = graph.adjacency(fac.data.manip, mode="undirected", weighted=TRUE) E(fac.graph.2)$width &lt;- E(fac.graph.2)$weight + min(E(fac.graph.2)$weight) + 1 # offset=1 plot(fac.graph.2, vertex.size = 10, vertex.shape="none", vertex.label.cex=1.0, margin=c(0,0,0,0), frame=TRUE) fac.keep = c("Drake","Rohani","Gittleman","Kramer","Schmidt","Park","Stephens","Ezenwa","Byers","Altizer","Hall","Davis") fac.data.keep = sapply(fac.keep, grep, x=row.names(fac.data.manip), value=FALSE) fac.data.zoom = fac.data.manip[fac.data.keep, fac.data.keep] fac.graph.3 = graph.adjacency(fac.data.zoom, mode="undirected", weighted=TRUE) E(fac.graph.3)$width &lt;- E(fac.graph.3)$weight + min(E(fac.graph.3)$weight) + 1 # offset=1 plot(fac.graph.3, vertex.size = 10, vertex.shape="none", vertex.label.cex=1.0, margin=c(0,0,0,0), frame=TRUE)

Figure 2. The network of publications linkages within the Odum School of Ecology. Nodes of the network represent a current faculty members and edges connecting nodes are representative of co-authored publications. Each edge is weighted with thicker edges representing a higher number of papers co-authored. The left panel is the full network and the right panel zooms into the clustered subnetwork to clearly show all connections.

To measure the centrality of each faculty member, we can calculate the betweenness of each individual. Higher valued members have the potential to “break” the connectedness of the graph when removed.

betweenness(fac.graph.2)

I have only reported top five with non-zero betweenness values:

- Byers: 154.75
- Drake: 102.50
- Rosemond: 66.50
- Pringle/Wenger: 45.00
- Ezenwa: 31.75

*This is by no means a comprehensive list of all the great scientists found in Odum.

Chao, I think you have demonstrated that our “strategic plan” for developing multiple areas of strength is working in the Odum School. And if collaboration with other universities and research centers around the globe were included it would likely show we have an extensive network that is growing fast! Well done! Alan

LikeLiked by 1 person