Collaboration in Africa: Networks or clusters?

Abstract

Recent discussion about the increase in international research collaboration suggests a comprehensive global network centred around a group of core countries and driven by generic socio-economic factors where the global system influences all national and institutional outcomes. In counterpoint, we demonstrate that the collaboration pattern for countries in Africa is far from universal. Instead, it exhibits layers of internal clusters and external links that are explained not by monotypic global influences but by regional geography and, perhaps even more strongly, by history, culture and language. Analysis of these bottom-up, subjective, human factors is required in order to provide the fuller explanation useful for policy and management purposes.

Full text

1
International collaboration clusters in Africa
Jonathan Adams(1), Karen Gurney(1), Daniel Hook(2) and Loet Leydesdorff(3)
jonathanzadams@gmail.com
(1) Evidence Thomson Reuters, 103 Clarendon Road, LEEDS LS2 9DF, UK
(2) Symplectic, 10 Crinan St, LONDON N1 9XW, UK
(3) Amsterdam School of Communication Research (ASCoR), University of Amsterdam, Kloveniersburgwal 48, 1012
CX Amsterdam, The Netherlands
Abstract
Recent discussion about the increase in international research collaboration suggests a
comprehensive global network centred around a group of core countries and driven by generic
socio-economic factors where the global system influences all national and institutional outcomes.
In counterpoint, we demonstrate that the collaboration pattern for countries in Africa is far from
universal. Instead, it exhibits layers of internal clusters and external links that are explained not by
monotypic global influences but by regional geography and, perhaps even more strongly, by history,
culture and language. Analysis of these bottom-up, subjective, human factors is required in order to
provide the fuller explanation useful for policy and management purposes.
Introduction
Georghiou (1998) drew attention to the phenomenon of increasing co-authorship in research
publications and, noting previous studies including Frame & Carpenter (1979), associated this
particularly with major global research facilities and cooperative programmes. He identified a
number of barriers to the potential growth of ‘bottom up’ cooperation but concluded that formal
‘top down’ enabling arrangements were emerging through, for example, the European Framework
Programmes (FPs). Persson et al (2004) noted deepening scientific collaboration and increasing
citation impact in all science areas. In particular, authorship was increasing exponentially while the
number of collaborative papers was increasing linearly so connectedness was intensifying. Greene
(2007) confirmed this trend and King (2012) described the growing frequency of massively multi-
authored papers.
Wagner & Leydesdorff (2005) argued that patterns in international collaboration in science can be
considered as network effects and that only the European FPs noted by Georghiou (op. cit.)
mediated relationships at that level. Their global network shares features with other complex
adaptive systems in which order emerges from interactions between many agents pursuing self-
interested strategies. Adams, Gurney & Marshall (2007) pointed to the intense levels of interactions
between leading research economies. Leydesdorff & Wagner (2008) suggested that the global
network reinforced a core group of (fourteen) cooperative countries with strong national systems.
They argued that peripheral countries could be disadvantaged by increased strength at the core.
Wagner (2008) argues, from complex systems theory, that the self-organizing global system
influences all lower systems (Wagner et al, in prep). Here, we accept the meta-pattern but contest
the network as a sufficient explanatory model and concur with Georghiou, that there are other
agents such as major facilities (e.g. CERN – see King, 2012) and cooperative programmes (e.g. WHO,

2
FAO, climate change) which have been important. In addition, we argue that the effects of history,
culture and language continue to have a profound human influence on collaboration patterns,
mediated through personal preference rather than strategic logic (Adams, 2012).
In this paper we illustrate these effects through an analysis of collaboration patterns in Africa (see
also Adams, King & Hook, 2010). Africa contains more than 50 nations, hundreds of languages, and a
welter of ethnic and cultural diversity. OECD’s African Economic Outlook (OECD et al., 2011) sets out
in stark detail the challenge for the research base in Africa and the extent to which current global
economic problems may make this worse and further compromise the value of the commitment
made in by developed nations 2005 to double official development assistance to Africa by 2010.
More than half the African nations are off-track or regressing on objectives to achieve universal
primary education by 2015. Internet penetration is good only in North Africa, constraining
communication and access to knowledge. It needs international research partners.
Is the uniform, generic pattern perceived elsewhere also found in Africa, or does the continent
exhibit more subtle influences in its patterns of research collaboration? And, picking up the
visualisation methods compared by Leydesdorff et al (in press), how can we best represent what we
see?
Methods
We focussed on research publications with one or more addresses for a country within Africa as
defined by the UN. We sampled data for the period 2000-2012 (data to current indexing, not year-
end). We also collated data on GDP for each country for which publication data were available.
Volume and subject analyses used Thomson Reuters National Science Indicators. Collaboration
analyses were carried out using Research Performance Profiles data in InCites™, a web-based
platform for research evaluation from Thomson Reuters. Database years were used to delineate
years, and only article, note and review document types were considered.
We counted all collaborations between countries represented by co-authorship on the publications
we collated. The counts are by paper not by number of researchers. For example, a paper co-
authored by two researchers from Ghana, three from Nigeria and one from Kenya counts as a single
paper in each country’s total and as one link between each pair of countries.
Analysis was extended using Wolfram Mathematica® 7 to create maps and collaboration diagrams.
We also had access to the data collected about 2011-publications by Leydesdorff et al. (in press), and
extracted the subset of African countries.
Results
Total research output for Africa increased from 13,271 publications indexed on Thomson Reuters
Web of Science in 2000 to over 35,000 publications in 2012 (34,528 catalogued at Dec 15, 2012). For
reference purposes, the total output of Africa is about the same as that of the Netherlands.
The percentage of Africa’s publications that were substantive research papers (that is to say, articles
or reviews) declined from 88% to 82.6%, which reflects an increasing number of proceedings papers
and other contributions authored within Africa. (Figure 1)

3
The number of articles and reviews that have been authored wholly within Africa (i.e. that have no
collaborative co-author from outside the region) has doubled since 2000 from 6,319 to 12,089. This
is a decline as a percentage of total research paper output from 54% to 42%. However, the relative
collaborative output of G8 countries rose much faster over the period: collaboration is increasing
everywhere. Thus, in fact, the autonomous research output of Africa clearly grew in the last decade
and Africa is becoming increasingly self-reliant in this regard.
A breakdown of the figures demonstrates the extent to which each region—and African science as a
whole—is dominated by three nations: Egypt in the north, Nigeria in the middle, and South Africa in
the south. In this millennium, since 2000, Egypt produced nearly 58,000 publications which was
more than twice the total for Tunisia, its next-place and regional neighbour. In west-central Africa,
Nigeria’s total for the same period was over 20,000, compared to roughly 12,800 for Kenya which is
the leading research economy in the east of the continent. South Africa’s dominance, as might be
expected, is even more pronounced: over 95,000 publications since 2000, compared to the southern
region’s next-most-prolific nation, Tanzania, which fielded just over 6,300. (Figure 2a, Table 1)
What happens when we break the publication data down by field of research? In our recent Global
Research Report on Africa we showed a discernible pattern in Africa’s relatively high
representation—as a share of world publications—in fields that are relevant to natural resources.
The highest percentage of any field, for example, is South Africa’s 1.55% share of Plant & Animal
Science. Not far behind is the same country’s 1.29% share of Environment/Ecology. A review of the
more detailed analyses in Thomson Reuters Essential Science Indicators shows that many of South
Africa’s most highly cited papers in this field pertain to climate change and its effects on plant
propagation. Following this theme, South Africa’s 1.13% share of Geosciences is in keeping with the
region’s mineral richness.
In short, Africa is a continent abundant in natural resources. How much does Africa itself benefit
from those resources? Absolute volume of published papers is one indicator of research activity
and—indirectly—of research capacity. It will therefore be obvious that the output of a country
reflects how much money is going in to its research system, and that is likely to be partly dependent
on its general economy. Bigger countries with a larger economy should be producing more papers, if
they invest at the same level as smaller countries. However, land area, population density and
resources vary a great deal. We have compared publications with Gross Domestic Product (GDP) for
each country, reasoning that proportionate investment in the knowledge economy is a good index of
a government’s commitment to maximize the longer term benefit of resource development and
exploitation for the general wealth of its people.
The leading countries by output are South Africa, Egypt, Nigeria, Tunisia, Algeria and Kenya (Figure
2a). Four of these (South Africa, Egypt, Nigeria and Algeria) are also leading countries in terms of
GDP while Kenya and Tunisia fall in a lower GDP tier. Indexing output against GDP (Figure 2b)
provides further interpretation. Zimbabwe is highlighted as relatively the most productive country
in terms of publications per unit GDP but this is anomalous because it retains its legacy research
base despite a collapsing economy and very low current GDP. The real leaders are Tunisia and
Malawi with very different economic bases but strong relative productivity in both cases. South
Africa, Kenya and Egypt all have significant relative productivity, as do a number of other countries in
East Africa (Ethiopia, Uganda, Tanzania) and West Africa. (Cameroon, Ghana).
It is clear, however, that despite Nigeria’s high volume output it is not producing as much research
as would be expected given the size of its economy. The value of its resources is not yet being felt in

4
its knowledge base. In fact, the same research productivity gap between potential and actual
investment applies to several other countries. This is an area where Africa is not yet benefitting
from the best use of its own natural resources.
Africa’s research can be boosted by collaborative international partnerships. The countries
collaborating most frequently with partners in Africa are the USA (39,292 papers between 2000 and
2012), France (31,421), the UK (25,753), Germany (13,879) and Canada (7,604). This looks like a roll-
call of ‘the usual suspects’ among major research producers. It is therefore worth noting that Saudi
Arabia collaborated on 6,285 papers, albeit almost entirely with countries in North Africa of which
Egypt (4,939 joint publications) was the pivotal link. Ethiopia’s research base is distinctive in being
substantial, growing and yet almost entirely domestic. The most substantial links between countries
in Africa and the USA, UK, France, Saudi Arabia are summarised in Table 1.
The research axis between Egypt, Saudi Arabia and the USA is an instructive example of new and
changing collaboration patterns. The numbers of papers co-authored between Egypt and the USA
has grown but has remained around 10% of Egyptian output since 1995. The numbers of papers co-
authored between Egypt and Saudi Arabia has been much smaller historically but reached 100 (4%
of Egypt’s output) in 2002 and exceeded 1000 (15%) in 2011. This is regionally, not globally, driven:
only a small fraction of these papers also have the USA as a co-author. (Table 2)
France also has a niche relationship with Africa. It is unusual in studies of international collaboration
to find it high in any ranking, and here to be 2nd behind the USA, ahead of the UK and with much
more than twice the collaboration links of Germany. Among the 31,421 total co-authorships by
partner then we find that Tunisia (Table 1: 7,400 23.6%) leads with Algeria (19%), Morocco (18.3%)
and then Cameroon (5.5%). France is focussed on a small set of countries just across the
Mediterranean in North Africa.
The USA and the UK, by contrast, collaborate diversely with South Africa, Kenya, Egypt, Nigeria and
others (Table 1). The UK has much greater collaboration with specific countries, such as the Gambia
and Malawi, than any other partner. Clearly, no collaboration pattern in Africa is general or uniform.
For each of six key research economies in Africa we have analyzed collaborative research links by
collating co-authorships with other countries and analyzing collaboration with the USA and the UK as
the most frequent partner for most countries, and three other frequent partners. (Figure 3)
How can we best visualise research co-publication within Africa? First, to create a simplified picture
useful as an indication of major links for policy purposes, we used a threshold to clarify where
relatively strong and persistent collaborations occurred. The threshold was set at a minimum of five
papers per year, or 25 papers in total over the recent five-year period. This meant that some
countries did not appear at all in the analysis because they had too low a level of recent
collaboration. We then used a grouping algorithm to associate the countries around the rim of the
wheel until groups with strongest cross-links were placed close together. (Figure 4)
Second, we created a more complete but necessarily more complex picture of the entire Africa
network 2011 using VOSViewer. (Figure 5)
Discussion
This analysis presents a complex picture of diverse research collaboration links, internationally
(Table 1, Figure 3) and within Africa (Figure 4, Figure 5). It is difficult to argue that these outcomes

5
are a response to a common global network phenomenon rather than local, cultural and historical
factors that play into research opportunities and create the highly individualistic and specific African
outcome. We do not disagree with the concept that international research collaboration is a
common phenomenon but we do believe in the need to determine the bottom-up regional and local
factors that properly explain complex outcomes departing from a notional top-down global
template. Only by understanding this detail will research performance analysis engage with the
theory and practice of research policy and management.
The research output of Africa is growing although remains small compared to established economies
(Figure 1). Africa has enormous natural resources but, while there is a broad relationship with
investment as GDP (Figure 2), some richer countries have yet to commit to substantial investment in
their knowledge economy. It is therefore to be anticipated that further research development will
continue to benefit from extensive external support and collaboration.
External collaborative links vary significantly by country. France is the second most frequent
collaborator with Africa, after the USA, with concentrated links to North-West Africa and to central
West Africa. It is interesting to note that, after normalization for size, Leydesdorff & Wagner (2008:
321) found France as highest on betweenness centrality because of its intermediating function with
the EU networks. The UK is the most frequent collaborator with other African countries, such as
Malawi and Gambia (Table 1). These links are not driven by global phenomena but by local historical
and cultural factors and by targeted international cooperative health and food programmes. Many
links are mediated through cooperative health and agricultural programs. Gambia is the site for
long-term research into tropical diseases for the UK’s Medical Research Council (Adams, Gurney &
Pendlebury, 2012) which also works in Uganda. The Wellcome Foundation has similar, major
research investments in Kenya and Malawi. A significant intellectual benefit is thus secured outside
Africa.
Another exceptionally strong link is that between Egypt and Saudi Arabia, which is not mediated by a
third party such as the USA (Table 2) but through their axis in supporting regional growth in research
capacity in the Arabian Middle East. (Figure 3; Adams et al, 2011)
How can we create a picture of Africa’s research network that would be helpful for policy
engagement? If we apply a threshold on the strength of interaction we find no single network
within Africa. Interface with African countries requires awareness that collaboration is driven partly
by geography but also by shared culture and—very strongly—by language. (Figure 4)
(1) There is a marked interaction between the countries in North Africa which share both language
and culture and are also relatively prolific. Thus, this network is probably the strongest group
overall since it links countries which are individually research active across multiple fields. The
group does little research with the rest of Africa, however, other than through an Egypt-South
Africa link.
(2) A West Africa group (Benin-Togo) pivots around Cameroon, a relatively research productive
country. The common factor within this group is almost certainly their common use of French as
the cross-national business language.
(3) Language also gives us the clue to the large group which includes Kenya and geographical
neighbours in East Africa but also includes Nigeria, Ghana and Gambia. Those countries appear
to have English as a common language or have had a strong Anglophone influence.

6
(4) The Southern African Development Community (SADC) does not emerge as a research network
since it is split between that group linked to Kenya and Nigeria and a second group most closely
linked to South Africa, but which also includes Sudan and Gabon. The overall collaboration
network, to the extent that one exists at all, is dependent on a small number of key players
linking these regional and cultural groupings.
The simplified collaboration cartwheel of Figure 4 is useful for managers and planners. It is
expanded and developed in Figure 5 into a complementary visualisation where completeness adds
complexity requiring additional interpretation. It is therefore of greater value for the analyst. The
map highlights the pivotal role of South Africa: the research hub in every sense. The map shows that
Egypt is not embedded in the separated cluster of North Africa but is an outlier due to its wider
attachments. There is a strong East Africa group, as in Figure 4, but there is also the development of
two distinct groups in West Africa.
In Figure 3, there is a striking difference between the countries pulled out in North Africa (Figure 5)
and those in other regions. Globally, the most frequent collaborative partner is the USA. Often this
is a consequence of researchers who have studied in the USA maintaining links with those research
groups when they return home. The UK and Germany are the other common partners to the
countries featured here and France has a major role. This is the influence of the global network
(pace Wagner op. cit.): between them the USA, UK, Germany and France have authors on half the
world’s research papers every year.
Nigeria sits at a research crossroads between East, West and South Africa. Despite its disappointing
level of research investment, it has an important connecting role. Not only is it a part of the
Anglophone collaborative network but it also has significant—albeit weaker—connections with its
West African neighbours, and it connects strongly to South Africa. South Africa is a similarly strong
node with a spread of links into other groups. These two, with Kenya, create strong cross-continent
links and are key nodes into global networks.
China and Brazil’s rapidly expanding research bases collaborate only weak with Africa. Nigeria’s
global reach is marked by some collaboration with China. It is theoretically well-positioned to
extend its links westwards and partner with the emerging Brazilian research base. It could thus
serve as a key doorway into both the West African and the Anglophone African research base for
some of the exciting research which is now appearing in Asia and Latin America. But it has yet to
realise this opportunity.
References
Adams, J. (2012). Collaborations: the rise of research networks. Nature, 490, 335-336
Adams J, Gurney, K. A. & Pendlebury, D. (2012). Neglected Tropical Diseases, pp 1-16. Global
Research Report, Thomson Reuters, Philadelphia. ISBN 1-904431-31-3
Adams, J., King, C., Pendlebury, D., Hook, D., Wilsdon, J. & Zewail, A. (2011). Exploring the changing
landscape of Arabian, Persian and Turkish research, pp 1-8. Global Research Report, Thomson
Reuters. ISBN 1-904431-27-5
Adams, J., King, C. & Hook, D. (2010). Africa, pp 1-9. Global Research Report, Thomson Reuters.
ISBN 1-904431-25-9

7
Adams, J., Gurney, K. & Marshall, S. (2007). Patterns of international collaboration for the UK and
leading partners. Report commissioned by the Office of Science and Innovation, 27 pp. Department
of Innovation, Universities and Skills, London. http://www.berr.gov.uk/files/file40396.pdf
Frame, J.D. & Carpenter, M.P. (1979). International research collaboration. Social studies of
Science, 9, 481-497.
Georghiou, L. (1998). Global cooperation in research. Research Policy, 27, 611-626.
Greene, M. (2007). The demise of the lone author. Nature, 450, 1165.
King, C. (2012). Multiauthor papers: onward and upward. Science Watch, 23, 1-2
Leydesdorff, L. & Wagner, C.S. (2008). International collaboration in science and the formation of a
core group. Journal of Informetrics, 2(4), 317-325.
Leydesdorff, L., Wagner, C.S., Park, H. W., & Adams, J. (in press). International Collaboration in
Science: the Global Map and the Network. El Profesional de la Información.
OECD, African Development Bank, United Nations Economic Commission for Africa and United
Nations Development Programme. (2011). African Economic Outlook 2011: Africa and its emerging
partners. OECD Publishing, Paris. ISBN 9789264111752
Persson, O., Glanzel, W. & Danell, R. (2004). Inflationary bibliometric values: the role of scientific
collaboration and the need for relative indicators in evaluative studies. Scientometrics, 60, 421-432.
Wagner, C. S. (2008). The New Invisible College. Washington, DC: Brookings Press.
Wagner, C.S. & Leydesdorff, L. (2005). Network structure, self-organization, and the growth of
international collaboration in science. Research Policy, 34, 1608-1618.
Wagner, C.S., Leydesdorff, L. & Adams, J. (in prep.) Policy implications of the global network of
science.

8
Table 1 Research output and collaboration (all publications on Web of Science, 2000-2012) between
the USA, UK, France, Saudi Arabia and their most frequent partners in Africa
USA France UK Saudi
Arabia
Africa - Total 296,351 39,292 31,421 25,753 6,285
South Africa 95,309 14,264 3,801 10,131
Egypt 57,741 5,900 1,019 2,409 4,939
Kenya 12,769 4,260 460 2,791 18
Uganda 6,317 2,318 231 1,402 11
Nigeria 21,909 1,945 243 1,426 54
Tanzania 6,299 1,693 171 1,625 8
Ghana 4,945 1,159 156 1,003 14
Malawi 2,909 990 124 1,087
Morocco 17,518 956 5,738 559 186
Ethiopia 5,579 933 218 576 26
Cameroon 5,915 832 1,730 548
Tunisia 24,724 755 7,400 421 326
Senegal 3,634 573 1,622 275 3
Algeria 14,846 412 5,961 292 367
Gambia 1,294 297 86 857
Gabon 1,188 241 504 205
Table 2 Growth of Egypt’s research output and its collaboration with the USA and with Saudi Arabia
over thirty years from 2000-2012 (part year). Egypt has increased collaboration with Saudi Arabia
and little of this is driven by its prior links with the USA.
Year Egypt total
Egypt +
USA
USA as %
Egypt
Triple co-
authors
Saudi as %
Egypt
Egypt +
Saudi
2000 2,577 286 11 2 4 95
2001 2,707 227 8 3 3 94
2002 2,894 295 10
4 115
2003 3,238 312 10 7 6 181
2004 3,212 318 10 4 5 169
2005 3,338 326 10 3 5 164
2006 3,847 358 9 6 5 190
2007 4,280 424 10 8 5 199
2008 4,710 439 9 15 6 261
2009 5,725 597 10 20 7 416
2010 6,281 708 11 33 10 614
2011 7,416 823 11 55 15 1,093
2012 (part) 4,386 428 10 47 19 832

9
Figure 1 Africa’s output of publications indexed on Thomson Reuters Web of Science databases
between 2000 and 2012
Figure 2. Output per country in 2008 as total volume (Figure 2a) and as volume/GDP (Figure 2b).
South Africa is absolutely the most productive country. Zimbabwe appears to be relatively
productive but this is an anomaly due to very low recent GDP and a strong historical base. Tunisia is
relatively the more productive on current performance.
-
10,000
20,000
30,000
40,000
2000 2004 2008 2012
Publications per year
Total output Articles and reviews

10
Figure 3 Most frequent intercontinental research collaborations for six key African research
economies

11
Figure 4: Collaboration between countries within Africa. This dependency graph uses Wolfram
Mathematica® 7 to provide a new visual interpretation of collaboration, by paper not by number of
researchers, and reveals clusters of countries with strong and persistent partnerships. Links
displayed between each country meet a threshold of five publications per year for a continuous
period of five years.

12
Figure 5: Coauthorship relations between 46 countries within Africa in 2011. VOSViewer was used
for the clustering and mapping. This map highlights the pivotal role of South Africa, shows the
separated cluster of North Africa with Egypt as an outlier due to its wider attachments, and
recognizes not only the East Africa group but also the development of two distinct groups in West
Africa. The map can be web-started at
http://www.vosviewer.com/vosviewer.php?map=http://www.leydesdorff.net/intcoll/afr_map.txt&n
etwork=http://www.leydesdorff.net/intcoll/afr_netw.txt&label_size=1.25&n_lines=1000 for further
exploration.