Evolution of Collaboration within the US Long Term Ecological Research Network

Professional Biologist Evolution of Collaboration within the US Long Term Ecological Research Network The US Long Term Ecological Research (LTER) program began in 1980 with the mission of addressing long-term ecological phenomena through research at individual sites, as well as comparative and synthetic activities among sites. We applied network science measures to assess how the LTER program has achieved its mission using intersite publications as the measure of collaboration. As it grew, the LTER program evolved from (a) a collection of independent sites (1981–1984) to (b) multiple ephemerally connected groupings with a gradual increase in collaboration (1985 to about 1998) to (c) a largely collaborative, densely connected network (from approximately 1999 on). Some sites demonstrated “preferential attachment” by contributing more to the evolution of network cohesion than others. Collaborative efforts of LTER scientists included cross-site measurements and comparisons, information technology transfer, documentation of methodologies, and synthesis of ecological concepts. Network science provides insights that not only document the evolution of research networks but also may be prescriptive of mechanisms to enhance this evolution. Keywords: centrality, homophily, LTER, preferential attachment, social network analysis A growing number of ecological programs around the world have been founded on the recognition that many ecological processes occur at, or are driven by, factors at long temporal and large spatial scales (e.g., the International Long Term Ecological Research Network; http://ilternet.edu). Networks of sites foster evaluation of processes at these large scales and comparison and synthesis of their patterns and responses (Callahan 1984, Hobbie et al. 2003). Formal association in a network is expected to promote standardization of activities and shared opportunities for common research; therefore, networks might begin as a loose aggregation of research sites but should evolve greater network cohesion as more cooperation and interdependence develop. Collaboration among sites within national programs should increase over time, and collaboration among national programs is ultimately expected to increase (Christian et al. 1999). Here we use tools of social network analysis (Borgatti et al. 2009) to assess the evolution of research cooperation among sites within one national program. The US Long Term Ecological Research (LTER) Network, created by the National Science Foundation (NSF) in 1980 to encourage long-term and comparative research on ecological processes (Callahan 1984), has grown from 6 to 26 sites, and has added a central coordinating office (the LTER Network Office). The US LTER Network is the oldest such network (Gosz et al. 2010) and has perhaps had the most cumulative financial support of any national program. The program has produced more than 10,000 publications (Hobbie et al. 2003) that initially emphasized site findings; however, intersite collaboration has gradually become more central to the broader mission of the LTER program (USLTER 2007). Decadal reviews of the program have emphasized the importance of cross-site and networkwide collaboration and synthesis. Analyzing the evolution of the US LTER Network and correlative information has helped us isolate variables important for collaboration and those that may foster a cohesive and productive consortium. Collaboration among scientists may take different forms, some of which are more easily tracked than others. Collaboration across sites may occur through simple communication and sharing of ideas; it may involve exchanges of scientists, students, or even equipment among sites. But collaboration and cooperation can be more easily tracked by the documents that arise from those activities, such as proposals or publications. Proposals, however, often are not tracked as effectively as publications. Moreover, publications are frequently a primary outcome of other forms of interaction, including joint proposals. Therefore, we have used joint publications among sites as the metric of collaboration; our analysis traces the annual patterns of publications by researchers within the LTER Network as grouped by the site. We used social network analysis to assess patterns of cross-site publication. Social network analysis has become a valuable tool in the evaluation of social interactions among people and larger groupings (Borgatti et al. 2009), and it BioScience 60: 931–940. ISSN 0006-3568, electronic ISSN 1525-3244. © 2010 by American Institute of Biological Sciences. All rights reserved. Request permission to photocopy or reproduce article content at the University of California Press’s Rights and Permissions Web site at www.ucpressjournals.com/ reprintinfo.asp. doi:10.1525/bio.2010.60.11.9 www.biosciencemag.org December 2010 / Vol. 60 No. 11 • BioScience 931 Jeffrey C. Johnson, Robert R. Christian, James W. Brunt, Caleb R. Hickman, and Robert B. Waide Professional Biologist can provide insights into how groups of scientists, artists, and members of other vocations cooperate, and how traits of both individuals and groups promote or deter high levels of collaboration, functionality, and performance (Guimera et al. 2005). We applied both visualization and quantitative analysis techniques to describe the evolution of networks of collaboration, and attempted to understand involvement of the network principles of homophily (“birds of a feather flock together”) and preferential attachment (“the rich get richer”; see box 1 for more detailed definitions.). Constructing and analyzing networks We used UCINET (Analytic Technologies), a software program for social network data analysis. We use the vocabulary of network science in this article, and provide a glossary of social network science in box 1 for biologists unfamiliar with the lexicon. We organized the data into a series of annual, two-mode matrices of accession numbers (i.e., database publication number) along rows, and sites along columns. Network relations among LTER sites were produced by calculating the number of shared authors across sites for a given paper in a given year. This process yielded a series of annual one-mode affiliation networks that were visualized using Netdraw in UCINET with spring embedding Box 1. Definitions for some social network terms. Degree centrality: the number of edges a node has, or the number of nodes adjacent or directly attached to any given node. In this case, it represents the number of other Long Term Ecological Research (LTER) sites with coauthored publications to a site. Density: the ratio of the number of observed dyadic (nodal pairs) connections in a network to the total number of possible dyadic connections. The density runs from 0, where no nodes are connected, to 1, where every node in the network is connected to every other node in the network. This is often referred to as connectance in network ecolo (...truncated)