BRAZILIAN SCIENTIFIC OUTPUT IN EVOLUTIONARY BIOLOGY : 2000-2012

The international dimensions of contemporary science have significant effects on patterns of production and use of scientific knowledge. This work evaluates the references and citations of Brazilian scientific output in Evolutionary Biology and its contribution to the internationalization of Brazilian science. The analyses are based on data from the Science Citation Index of the Web of Science and involve 1,450 articles published between 2000 and 2012, 60,454 cited references and 18,059 citing documents. The results reveal high indices of international references and citations, averaging 90.5% and 84.0%, respectively. It concludes that Brazilian Evolutionary Biology is an eminently international field, whose standards of internationality extend the audience of research and entail in greater visibility for Brazilian science.


INTRODUCTION
Scientifi c collaboration has increased signifi cantly over the last decades, in different areas and at different levels of aggregation, at a macro (when collaboration occurs and is studied across different countries, continents and areas of knowledge), meso (when it occurs between different institutions, regions or departments in the same university) or micro level (between research groups and individuals) (Glänzel 2003).The interest in the collaboration phenomenon has also thrived among researchers and governments all over the world.At the same time that the scientifi c community is conducting research to understand the structure and dynamics of collaboration, governments are increasingly giving attention to it in order to support the planning of scientifi c policies.
The analysis of the scientifi c output resulting from collaboration may expand the understanding DIRCE M. SANTIN, SAMILE A.S. VANZ and IDA R.C. STUMPF of the behavior patterns of scientific agent's production (Maltrás Barba 2003).Collaboration indicators refl ect the structures of scientifi c fi elds by analyzing data on co-authorship and author affiliation.The analysis of networks, in turn, may indicate the players that establish greater connectivity and exert greater infl uence in certain contexts (Newman 2003).
Bibliometric studies, among which the analysis of collaboration indicators is included, focus on a variety of approaches and dimensions and can be used to evaluate the scientifi c community, and the scientifi c output of a country or institution (Meis and Leta 1996).They allow for the evaluation of the scientific development of players, their participation in international scientifi c cooperation and their contributions to science and economic progress.Furthermore, they can be useful to outline national and institutional strategies, for the planning and evaluation of science policies (Vanz and Stumpf 2010).
This study analyzes collaboration networks present in the Brazilian scientific output in Evolutionary Biology, considering papers published from 2000 to 2012 in journals indexed by Web of Science (WoS).Based on co-authorship data, the study aims to reveal the growth of collaboration during that period, their characteristics and the structure of collaboration between authors, institutions and countries.
Evolutionary Biology is dedicated to the study of changes accumulated by organisms over generations that grant them different physiological and morphological characteristics (National Institutes of Health 2011).Its studies seek to explain the origin and diversification of life, adaptive changes, and extinction of living beings, among others.The fi eld emerged in the nineteenth century, with the studies of Lamarck, Darwin and Mendel, who revolutionized science and led Biological Sciences to a higher level in the global scenario (Salzano 1979).In the mid-twentieth century, the discovery of the DNA structure by Watson and Crick (1953) expanded the possibilities of the fi eld with the establishment of a physical basis for Genetics.The emergence of Molecular Biology in the second half of the twentieth century, integrating knowledge from areas such as Biochemistry and Genetics, has transformed Evolutionary Biology research by granting it new techniques for solving problems, in addition to raising new questions about natural history (Barton et al. 2007).
Currently, Evolutionary Biology is one of the "cornerstones" of science (Barton et al. 2007) and plays a central role in Biological Sciences (Townsend 2010).The structure of the field is complex, gathering both experimental biologists, who work with organism models in laboratory, and naturalists, who observe populations in their natural habitats.Its studies integrate knowledge from disciplines such as Ecology, Biodiversity Conservation, Molecular Biology, Developmental Biology, Genetics, among others.Not only does this confi guration strengthen the area in the context of Biological Sciences and basic research in general, but it also expands the possibilities for using new skills in applied areas such as Health Sciences, Human Genetics, Agriculture, and Biodiversity Conservation (Futuyama 2009).
In Brazil, the fi eld has mostly developed since the 1930's and 1950's as a result of its approximation with Genetics (Salzano 1979).Its association with Genetics is still considered extremely relevant in the structure of Brazilian Evolutionary Biology, both with regards to its integration in university departments, such as the Department of Genetics and Evolutionary Biology at the Universidade de São Paulo (USP) 1 , as well as its graduate programs, such as the Graduate Program in Genetics, Conservation and Evolutionary Biology linked to the Instituto Nacional de Pesquisas da Amazônia (INPA)2 .It should be noted, however, that Evolutionary Biology is a different fi eld from Genetics, Molecular Biology and other disciplines, because it has solid theoretical foundations and its own object of research, although part of its studies can be shared by other fi elds (Barton et al. 2007).
The analysis of collaboration networks in the Brazilian scientific output in Evolutionary Biology is justifi ed by the importance of the fi eld in the national context, by the international impact achieved by publications and by the extremely high rates of international collaboration observed in previous studies (Meneghini and Packer 2006, D'Abadia and Rodrigues 2012, Santin and Silva 2013, Santin et al. 2015).The analysis of collaboration networks in Brazilian co-authored articles on Evolutionary Biology is believed to be able to contribute to the understanding of the structure and the dynamics of output and collaboration in this area and the properties that are inherent to it.The study of the interactions between the main players can also reveal patterns and trends of national and international collaboration in Brazilian science, focusing on the Natural Sciences.

MATERIALS AND METHODS
The paper consists of a bibliometric study at a macro level of aggregation, focusing on collaboration indicators (Glänzel 2003).Social Network Analysis is used as an important complement to the bibliometric indicators (Vanz and Stumpf 2010), since it enables the understanding of the relationship among players in a graphical representation of the characteristics and dynamics of collaboration.
The main source of data was the Science Citation Index (SCI) on WoS, produced by Thomson Reuters and regarded as one of the most important sources for bibliometric studies.Data were collected in December 2013 using the search by country strategy (CU), limited to the subject category Evolutionary Biology (WC), the period from 2000 to 2012 (PY), article as document type and limited to SCI.Combined with a strategy of search by topic (TS), using the terms Evolutionary Biology OR Biological Evolution, by country and in the same period of publication, the search resulted in 1,450 articles that make up the corpus of this research.
The records were exported from the database in full format using the Save to Plain Text option, in groups of 500 according to the maximum number allowed for export by WoS, and subsequently placed into a single fi le.The next procedure was the normalization of the names of authors and institutions listed in the Author (AU) and Address (AD) fi elds in the bibliographic records, seeking to ensure the reliability of the data analyzed.The software used for data analysis was Microsoft Excel, BibExcel and Gephi.The connections between players were defi ned using the Force Atlas algorithm, integrated to Gephi, which defi nes the approximation of the nodes by the strength of their connections and enables groups within the network (clusters) to be identified.The quantification of the articles was of the complete type, assigning an article to each author, institution or country involved (Glänzel 2003).
The methodology consisted of investigating the number of articles and their increase in the period, the levels and dynamics of collaboration, comparing national and international collaboration, and analyzing networks of collaboration between authors, institutions and countries.The study of the increase in scientifi c production and of collaboration structures was based on the full set of data (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012).This set was divided into four periods (2000-2003, 2004-2006, 2007-2009, 2010-2012), from which the data were processed separately in the analysis of the collaboration dynamics.Except for the fi rst time window, the others correspond to the triennial evaluation of Brazilian graduate DIRCE M. SANTIN, SAMILE A.S. VANZ and IDA R.C. STUMPF programs by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES).Although the fi rst period has an additional year in comparison to the subsequent ones, it gathers a considerably smaller number of articles.It is believed, therefore, that this division seems suited to the research and does not imply prejudice to the observations made sequentially over time.

RESULTS AND DISCUSSION
The Brazilian scientific output in Evolutionary Biology published from 2000 to 2012 and indexed on the WoS comprises 1,450 articles, which represent approximately 0.5% of the Brazilian articles published in the period.Figure 1 shows the distribution of articles by year of publication, followed by the annual growth rate.Their growth was observed to be positive in most years, especially 2002, 2004and 2008. Negative growth rates, noted in 2005, 2006, 2009and 2011, do not compromise the increase in the number of articles in the period, because the average growth was 15.7% per year.This increase, therefore, is higher than the average increase of 10.7% in the Brazilian scientifi c output from various fi elds, as pointed out by Almeida and Guimarães (2013) based on the growth seen in the 1980's, 1990's and 2000's.
The increase in the number of articles published by Brazilian researchers and indexed on the WoS follows the trend towards a greater presence of Brazilian science in the international setting and can be associated with several factors, among which are: a. the increase and continuity of public investments intended for training human resources for science in Brazil (Guimarães 2004 It is important to mention that the increase in the number of Brazilian journals indexed in international databases, which started especially in the second half of the 2000s and is highlighted by authors such as Leta (2011Leta ( , 2012)), Vargas and Vanz (2014), among others, does not seem to interfere signifi cantly with the increase in the number of Brazilian Evolutionary Biology articles published in the 2000-2012 period, since only 0.5% of the articles considered in this research were from Brazilian journals.Therefore, the confi guration is shown to be distinct in this fi eld whose articles are published primarily in foreign journals, surpassing even the patterns of Brazilian Biological Sciences, traditionally characterized by their preference to publicize research results in international journals (Meneghini and Fonseca 1990, Melo et al. 2011, Santin et al. 2015).
The evolution of the Brazilian scientifi c output in Evolutionary Biology was also highlighted by Johnson (2009) in a comparative analysis on the increase in the number of publications in South Africa, Switzerland, New Zealand, China and Brazil in the years 1990-2008.According to the author, Brazilian contributions to global scientifi c production are substantial and reached around 7% in 2008, outpacing the increase seen in the other countries included in the analysis.The contributions from Brazil in this area turn out to be much higher than the percentage achieved by other areas (2.7% of the world total in 2011) (Cruz 2013) and may be indicative of the thematic specialization that is taking place in Brazilian Evolutionary Biology, which may be associated with the natural advantages of the country in terms of biodiversity, among other factors (Johnson 2009).

LEVELS AND DYNAMICS OF COLLABORATION
The analysis of article co-authorship reveals the predominance of collaboration, representing a consolidated pattern in Evolutionary Biology.Among the 1,450 articles analyzed, 52 (3.6%) were published by a single author, while 1,398 (96.4%) had multiple authors.The proportion of articles with a single author is similar to the value obtained by Vanz and Stumpf (2012) regarding authorship in the Brazilian scientifi c output published in the years 2004 to 2006, where the percentage was 3.9%.The average number of authors per article was 5.2, and the median of four authors per article, reinforces a value close to the average.However, the standard deviation of 8.7 indicates discrepancy between the number of authors, given that articles with 135 and 149 authors were found.
The pattern of article co-authorship is consistent with the high level of collaboration in Brazilian Biological Sciences research over the last decades, in which approximately 60% of researchers have been networked, including relationships that are internal and external to the discipline itself (Mena-Chalco et al. 2014).The large number of authors identifi ed in some articles indicates the participation of Brazil in the Big Science, which is characterized by complexity and participation of a large number of researchers in studies of global interest (Price DIRCE M. SANTIN, SAMILE A.S. VANZ and IDA R.C. STUMPF 1976).The Evolutionary Biology interfaces favor the integration of the discipline into the model of major collaborations in Molecular Biology and biomedical research, including, for example, contributions to the Genome Projects.
The general trend towards an increase in scientifi c collaboration can be seen in Evolutionary Biology through the time window defi ned for this study.Figure 2 shows the evolution and dynamics of co-authorship in the Brazilian articles, published in each period.As these are collaborative networks, they are of the undirected type and represent only the authors who had at least one connection in each period.
The first network (2000)(2001)(2002)(2003) is made up of several circles which, although connected by strong bonds forming small clusters, are kept separated at some points.The average level of the network, represented by the average number of edges (lines) connected to each node (author) is 3.1.Weak ties, represented by nodes with fewer connections or located at the ends, play an important role in the expansion of co-authorship, because they operate connections between parts of the network not directly connected by strong bonds (Barabási 2009).The central players in the network and the cluster leaders, however, establish greater connectivity and exert more infl uence on the fi eld (Newman 2003).The formation of two clusters that are external to the network and are mostly made up of foreign authors is also observed.
The configuration remains similar in the second period, although there is a trend towards the integration of groups into the large network.Between 2004 and 2006 the connections become more intense and the clusters are larger.Nevertheless, the average level of connections remains stable at 3.1.The groups, which were previously small and isolated, extend their internal and external connections and make the network denser.The stronger nodes take on a prominent place in the clusters and generate greater integration, indicating the formation of so-called "small worlds", where the players in the network are highly interconnected through bonds that establish the co-authorship of publications (Watts 1999).
The third network (2007)(2008)(2009) represents the phase of highest increase in Brazilian co-authored articles on Evolutionary Biology.The size of the network widens greatly and the average level of connections increases to 12.6, indicating a high amount of links between authors.At this stage, the increased collaboration is extraordinary and the network is quite complex, requiring closer examination and the identifi cation of key players in order to understand the collaboration phenomenon in the fi eld.This option is adopted in this study to globally analyze the structure in the networks of authors, institutions and countries, presented in more detail in the following sections.
The fourth and final network (2010-2012) reveals the consolidation of collaboration in the fi eld, reaching an average level of 16.7 connections per author.The picture is totally occupied by nodes and connections, and the network reaches its maximum size in this period, revealing the complexity and intensity of collaborative relationships.By adding the numbers of nodes and edges of the four networks, an average collaboration level of 11.5 is achieved.This rate exceeds the average level of 8.0 identifi ed in the co-authorship of publications on Biological Sciences in the years 1990-2010 by Mena-Chalco et al. (2014), with appropriate reservation for the covered period, indicating higher levels of collaboration in the scientifi c production of Evolutionary Biology when compared with the greater fi eld of Biological Sciences.
Co-authorship relationships represent the socalled "scale-free networks", in which the grouping of authors takes place randomly (Barabási 2009).This characteristic may reveal the potential of a network for absorbing new players and enhancing preexisting collaborative relationships.The results of this research indicate that the expansion of cooperation in Evolutionary Biology is strongly associated with the inclusion of new authors in scientifi c production (Barabási et al. 2002, Newman 2004), because the increase in the number of nodes associated with each network represented in Figure 2 is high.However, the number of edges increased more than the number of authors observed during the four time window, revealing a significant increase in collaboration between authors, whose relationships become more intense over time.
The dynamics of collaboration shown in Figure 2 confi rm that the scientifi c activity in Brazilian Evolutionary Biology has gradually became more collaborative in the early twenty-fi rst century.This situation follows the trend towards the evolution of collaboration over time in world science as DIRCE M. SANTIN, SAMILE A.S. VANZ and IDA R.C. STUMPF

COLLABORATION BETWEEN INSTITUTIONS
The institutions to which the authors are affi liated also stand out regarding the network of coauthorship existing among the most productive institutions, which totaled more than 15 articles in the period and established inter-institutional collaboration relationships.This situation strengthens the central role of the main authors and institutions in Brazilian research and indicates the existence of highly productive groups, which engaged in national and international collaboration relationships.The choice of 15 articles as the cutoff point aims to facilitate viewing the relationships within the network.The institutions represented in Figure 4 correspond to 3.3% of the 1,318 institutions identifi ed in the research corpus.
The network comprises 43 institutions, defi ned on the map for the intensity of the links established between them.The most intense lines and the central positions of the nodes indicate the institutions that have higher connectivity and exert greater infl uence on the fi eld (Newman 2003).The centrality of these institutions reveals their importance in relation to the collaboration they established with other institutions in the network.USP, with 424 articles (29.2%);UNICAMP, with 192 (13.2%);UFRGS, with 143 (9.9%); Universidade Federal do Rio de Janeiro (UFRJ), with 116 (8.0%); and UNESP, with 111 articles (7.6%), stand out.
The geographical and institutional diversity is highlighted in the network, although the central positions are occupied primarily by institutions of southeastern Brazil, following the leadership patterns achieved by the region in the Brazilian scientific production (FAPESP 2010).This characteristic is very positive and can be linked to the biological diversity found in the different regions of the country and the confi gurations of the fi eld with regard to aspects such as structure, interdisciplinarity and institutional affi liation of researchers.
Despite the leadership of the universities, a strong presence of national and international research institutes is observed, such as Embrapa in Brazil, with 66 articles (4.5%), and Centre National de la Recherche Scientifique (CNRS), in France, with 43 articles (2.9%).This situation demonstrates the complex structure of the area, which brings together universities, research institutes and research centers, among other organizations, integrating both experimental and applied studies (Futuyama 2009).The presence of foreign institutions in the network follows this trend and indicates the importance of international collaboration in Brazilian Evolutionary Biology.

INTERNATIONAL COLLABORATION
Collaboration with other countries was found in 753 articles (51.9%) and includes 86 partner countries.The rate is high, especially when compared to the rates of international collaboration noted in the Brazilian scientifi c output from recent years, in which co-authorship with authors from other countries occurred in approximately 30% of the publications (Leta and Chaimovich 2002, Vanz and Stumpf 2012, Leta et al. 2013).This situation highlights a strong trend towards international collaboration in Brazilian Evolutionary Biology research and unveils a specifi c characteristic of this fi eld in Brazil.Figure 5 shows the collaboration network with the main partner countries.
The network is formed by Brazil and 29 other countries, selected based on the number of articles and intensity of connections established in coauthoring.The collaborating countries orbit around Brazil, with greater or shorter distance, according to the intensity of the collaboration established with the central node.The most intense lines reinforce the importance of the main partners in the Brazilian scientific production in the field, especially the United States (USA) with 361 articles (24.9%);United Kingdom, with 161 articles (11.1%);France, with 85 articles (5.9%); Germany, with 73 articles (5.0%); and Argentina, with 72 articles (5.0%).
The relative centrality taken on by the USA in the network reveals how important the participation of that country in Brazilian Evolutionary Biology research is, and confi rms its position as Brazil´s main partner in the production of scientific knowledge (Leta and Chaimovich 2002, Glänzel (2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012).Source: research data.DIRCE M. SANTIN, SAMILE A.S. VANZ and IDA R.C. STUMPF et al. 2006, Packer and Meneghini 2006, Vanz and Stumpf 2012).The strong participation of the USA is also consistent with the leadership exercised by the country in the international scientific community, with emphasis on the Natural Sciences (Mueller and Oliveira 2003).
The presence of European countries follows the existing collaboration trend in Brazilian science (Leta andChaimovich 2002, Packer andMeneghini 2006) and can be associated to the tradition of Europe in biological evolution studies, especially in nations such as the United Kingdom, Germany and France (Barton et al. 2007).The tradition in research from an evolutionary perspective and the use of advanced and expensive molecular techniques make these countries take precedence over Argentina among the main partners, although this country has been pointed out by several studies as the second main partner of Brazil in the scientifi c output of different fi elds (Glänzel et al. 2006, Vanz andStumpf 2012).
The international collaboration noted in the articles can also have its level of cooperation analyzed for the number of countries involved in co-authoring.Among the items produced through international collaboration, 484 (64.3%) consisted of bilateral collaboration, 148 (19.6%) of trilateral collaboration, and 121 (16.1%) of multilateral collaboration.The predominance of bilateral collaboration follows the trend found by Vanz and Stumpf (2012) in the Brazilian scientific output published in the 2004-2006 period, in which most of the articles resulting from international co-authorship had the participation of Brazil and another country.The last level, which brings together articles produced by authors from four or more countries, included seven articles published by authors from ten to fi fteen countries, and eight articles produced in collaboration by sixteen or more countries, including Brazil.The average number of countries per article was 1.9 and the median was 2.0.The mode of the number of countries per work was 1.0, and the standard deviation was 1.8.The stability of the data proves the regularity of international collaboration in Brazilian Evolutionary Biology and reinforces the existing pattern of collaboration in the fi eld.

CONCLUSIONS
Scientific collaboration has achieved major importance over the last decades, becoming a strategic issue for planning and evaluating national and institutional scientific policies.Scientific collaboration networks constitute dynamic phenomena that are constantly changing according to the movements of researchers and the connections between them, as well as external factors.The increase in production and the consolidation of collaboration as a scientifi c practice were highlighted in this study, which underscored two distinct and complementary aspects: the dynamics and the structure of co-authorship relationships in Brazilian articles on Evolutionary Biology published between 2000 and 2012.
The analysis of collaboration dynamics between authors demonstrated the growth of co-authoring relationships in four distinct and sequential periods, allowing the evolvement of the phenomenon over time to be understood.While the period from 2007 to 2009 represents the phase of largest increase in Brazilian co-authored articles on Evolutionary Biology, the period from 2010 to 2012 is marked by the consolidation of collaboration in the area, revealing the complexity of the relationship between researchers.On the other hand, studying the structure of collaboration between authors, institutions and countries enabled us to know the main actors and the relationships they established in the national and international scientifi c community.The analysis of collaboration networks also indicated the central authors and institutions whose infl uence on studies is expressed by the variety and intensity of the relationships established in article co-authorship.
The pattern of international collaboration, found in more than half of the articles is unusual in the Brazilian scientifi c output and may result in broadening the impact of publications in the global scenario.The collaboration with authors from other countries also indicates the internationalization of scientifi c output and the capacity of the area to take part in the science produced within the international scientific community.From this perspective, the results of this study point to new possibilities for understanding the phenomenon of major collaborations in Evolutionary Biology, in addition to enabling an analysis of the possible correlation between the high level of international collaboration and the impact achieved by the publications.

Figure 2 -
Figure 2 -Evolution of collaboration between authors in the Brazilian scientifi c output in Evolutionary Biology (2000-2012).Source: research data.