The nature and trends of agricultural research development in Africa : an informetric study

The study recognizes agriculture as the mainstay activity of most economies in Africa and analyses research nature and trends in the discipline by using descriptive informetrics and focusing on seven indicators, by using the AGRICOLA and ISIE databases from 1991 to 2005. We observed that research output in the discipline is much higher in South Africa and Kenya, and research collaboration is greater than non-collaborative research output and collaboration is less among African countries. The most popular research domains were found to exist in environmental science, soil science, plant/ crop production and [agricultural] economics. Helpful conclusions and recommendations for an agricultural policy, capacity and research orientation have been made.


Introduction
Statistically, the African continent occupies about 11 million square miles of the planet's land mass, and is inhabited by approximately 690 million people (see http://www.maryknollafrica.org/Facts.htm).Most (70%) of the African population live in the rural areas where agriculture is the predominant economic activity.It employs 70% of the continen'ts economically active population and accounts for about 25% of the gross domestic product and 60% of the export income ( Kiplang'at, 2004:328 & 348;Diouf 2003).Whereas there are many ways of improving agricultural production in Africa, research is considered to play a significant role.A widely held view is that research would solve problems existing in the sector, integrate and interrogate its existing theories and hypotheses, create new knowledge, provide new knowledge and information for decisions informed by empirical evidence and rational thinking, and strengthen capacity building for agricultural research whilst developing the sector.A number of methods are used to determine research capacity that also applies to agriculture.Among them are the determination of the number of research workers, research institutions, research output (e.g research publication), research policy, research structures and knowledge, and information systems and services.Arunachalam & Umarani (2001:905) observe that the future of a country's agriculture is dependent on research, public policy and the farming community's cooperative action.Related studies have been conducted in Mexico (De Arenas, Sandoval & Arenas 2003), Saskatchewan (Phillips 2001) and India (Arunachalam & Umarani 2001) but none, to the best of our knowledge, has been conducted in Africa.The aim of this study is to determine the nature of and trends in agricultural research development in Africa between 1991 and 2005 through an informetric/bibliometric analysis on the AGRICOLA (on-line) and Thompson Scientific Science Citation Index Expanded databases.AGRICOLA is considered to be the largest subject based agricultural research database in the world, whilst the Thompson Scientific Science Citation Index Expanded database represents the largest collection of high impact research output in the domain.The study determines the research indicators relating to geographic region or country of publication, national, institutional and individual levels of collaboration, subject, nature of records and research trends over 15 years.With a view of triangulation, the study also makes comparisons in coverage between AGRICOLA and the Thompson Scientifics' Science Citation Index Expanded database.
A total of 2368 documents and 1254 papers were extracted from AGRICOLA and the SCI-E respectively.Six hundred and twenty seven (627) records found in the AGRICOLA database were excluded from the analysis according to the year of publication.The dates of these records were unclear, i.e. were in the following format: 'between 1996'between and 1998'between ', and '1996 or 1997<1992-1995>' OR contained question marks (?) or 1997<1992-1995>' OR contained question marks (?) Where there were two copyright dates of publication, only the most recent date was considered for analysis.This analysis yielded 1741 AGRICOLA records, whilst the SCI-E generated 1254 records (see Table One).Table 1 show that there has been an incremental growth in the number of records indexed in the SCI and a negative growth in the case of AGRICOLA.The number of records fell from 157 in 1991 to 13 in 2005 in AGRICOLA's case, whilst SCI-E's papers rose from 36 to 149 during the same period.Nevertheless, columns six and seven show that cumulatively, the total number of records increased to the current 2368 for AGRICOLA, and 1254 for SCI-E.

Distribution of documents by document type
Agricultural information on Africa and her 53 countries was published in several different document types, amounting to 14 different categories.Ranked first were journal articles, which had 1175 and 1180 postings, followed by books (986), book chapters (151) and book reviews (56).Others, in descending order, were serials (47), editorials (9), meeting abstracts (5), audio-visual materials (4), manuscript collections (2), notes (2), machine-readable data files (2), maps (1), biographical items (1), and news items (1).Worth noting is the large representation of books and book chapters.

Distribution of documents by subject category
The grouping of documents by subject category was useful for establishing the size, scope and nature of subject coverage by the database on Agricultural research, and the various disciplines that are utilising agricultural information.It is important to note that the SCI-E categorises its records according to broader disciplines (subject categories) than AGRICOLA.The dominant subject areas/categories include Environmental sciences (274), Economics (254), Plant production (234), Soil cultivation (209), Agriculture, Multidisciplinary (179), Soil Fertility (170), Agronomy (143) and Agriculture [general] (119).Others are Soil science (116), Water resources (116) and Farm organization and management (114).We have observed that Animal production does not feature among the top categories.

Distribution of documents by geographic region
The distribution of documents by geographic region or country reflects the countries as subjects of research, i.e. the geographic areas of research focus.The analysis sought to determine the most researched country or geographic region in Africa.Table 3 indicates that as far as country output is concerned, South Africa was first and posted 153 (6.46%) records in AGRICOLA, and 147 (11.72%) in the SCI.Other regions/countries that ranked highly in both databases, included Kenya, Nigeria, Angola, Ethiopia, Sub-Saharan Africa, Southern Africa, Zimbabwe and Tanzania.The distribution of documents by country of publication and the author's country of affiliation was based on the analysis of 1006 AGRICOLA and 1254 SCI records.Focus was placed on the country in which the document was published and the author's country of affiliation (i.e.where the author resides or works).Whereas the former provided information on where research in agriculture regarding Africa is published, the latter analysis provided information on the country conducting research on agriculture in Africa (i.e. the country from which the document originated).In other words, the former can be treated as the publisher (publishing country), whilst the latter refers to the country as the author/producer of the record.Leading in both cases (i.e.publisher and author/producer) is the USA, which published 222 records and authored 297 papers.The second ranked country-publisher is France, which posted 100 records, followed by England (85, 8.45%), Egypt (77, 7.65%), South Africa (66, 6.56%), Italy (45, 4.47%), and Ethiopia (41, 4.08%).Besides the USA, the most productive country-authors' were England (160, 12.80%), South Africa (132, 10.50%), Nigeria (109, 8.70%), Kenya (73, 5.80%), Netherlands (68, 5.40%), Germany (65, 5.20%), France (61, 4.90%), Australia (55, 4.40%) and Egypt (48, 3.80%).We observed that most African countries that appeared as authors do not appear as major publishers, suggesting that documents associated with them are not necessarily published within the country.

Discussions and conclusions
We observed, first, that the distribution of documents by year of publication over the duration of 15 years, whilst exhibiting a cumulative growth of records in 1999, does also show largely negative growth, as is the case with AGRICOLA in 2005, reflected in Table 1.This negative growth in turn suggests that either research output is indexed in databases other than AGRICOLA or ISI, or, those published are indexed in non subject gateway databases (e.g those on the web), or indexed in national/regional databases.There could of course be a general decline in the amount of agricultural research output and support in Africa.Unfortunately, besides the South African based SABINET databases, recognisable databases in other parts of Africa are not known, and this in turn affects the visibility of Africa's research.The limited growth between 2002 and 2005 could also be attributed to an indexing time lag factor [the period between the publication of a paper in the public domain and the date it is captured in abstracting and indexing journals] (see Diodato 1994).Second, most indexing and abstracting databases largely index journal articles.AGRICOLA, however, makes use of a unique approach involving the inclusive indexing of a variety of document types, particularly books.Among the 14 document types identified, articles (SCI and AGRICOLA) and books (only AGRICOLA) are dominant.We observe that there is a positive, albeit complex, attempt at incorporating inclusive indexing, which involves the inclusion of books and book chapters as reflected in the AGRICOLA database, that should augur well in diffusing criticism on the over-reliance of journal articles for measuring research output.The small number of documents captured from the other 12 document types is not unusual.Third, a subject category that was created and sequenced/ordered by the frequency of occurrence of records in the particular subject area is illustrated in Table 2.Although a total of 98 subject areas were identified, when some of these domains were grouped into broader concepts or logically within a larger subject field, the dominant subject categories emerging were soil science; plant[crop] production; environmental science; [agricultural] economics; general agriculture; farm administration and management; research and animal production.One of the initial research assumptions was that crop/plant production and animal production would be dominant in the research domain.
Regrettably, research in animal production is insignificant.This perhaps shows that less research is conducted in animal production, even though we believe that it is one of the largest agricultural sectors in Africa.A bibliometric assessment of South African research publications as indexed in the Thomson ISI databases by Pouris (2006), however, reveals that plant sciences was the most published discipline with a total of 2182 publications between 2000 and 2004, followed by animal sciences (2108), and environment ecology (1187).It can be noted that Pouris' findings greatly differ from the results reported by this study in that the latter produced a relatively large volume of publications on the subject area of plant sciences and animal science among others.The pattern reported in our study can be attributed to our limiting the search to only agricultural publications.But this could also mean that research output is not indexed in popular databases.
Although there are other speculative reasons why Agricultural research would be higher in some countries in Africa (such as South Africa and Kenya), stronger research output is observed in less popular agricultural areas such as Angola, Ethiopia and Nigeria, suggesting growing interest in increasing agricultural production in the formerly marginalised but potential agricultural areas.Fifth, a list of journals that publish agricultural research on Africa in AGRICOLA records and 97.69% of SCI papers were published in English, whilst French and German were ranked second and third as illustrated in Table 5.Other languages used in agricultural publications on Africa are Spanish, Afrikaans, Russian, Arabic, Portuguese, Italian, Ukranian, Hungarian, Swahili and Dutch.In some instances, documents were published in two languages, thus catering for a larger audience.This may also have been necessitated by collaboration amongst researchers from different countries, hence requiring the incorporation of multiple languages.
Seventh, this study also focused on collaborative research output.There are several benefits of research collaboration as outlined by Katz & Martin (1997).Among them according to the two authors are: the fact that it enables researchers to share skills and techniques and is one way of transferring knowledge (especially tacit knowledge); through clashing views it may bring about the cross-fertilisation of ideas, which may in turn generate new insights or perspectives that individuals, working on their own, would not have grasped; collaboration provides intellectual companionship (i.e.within a practising community); collaboration plugs the researcher into a wider contact network in the scientific community; and it enhances the potential visibility of the work.

Table 1
Distribution of documents by year of publication

Table 2
Distribution of documents by Subject Category

Table 3
Distribution of documents by Geographic region/territory

Table 4
Sources publishing agricultural research onAfrica, 1991Africa,  -2005 A total of 151 AGRICOLA and 442 SCI sources published agricultural research on Africa and her countries.The top ranking 39 sources that produced 4 or more records (in each database) are presented in Table6.The top 10 most productive sources include Agriculture, ecosystems and environment(197, 42), Experimental agriculture (143, 15), Tropical Agriculture (115, 9), Journal of Sustainable Agriculture (69, 41), Outlook on Agriculture (81, 20), Journal of the Science of Food and Agriculture (79, 1), Agricultural Economics (32, 40), Agricultural Systems (20, 32), FoodPolicy (24, 19)and Agriculture and Human Values (37, 0)3.7 Language of publicationEnglish is dominant.The language was used to publish 2161 (91.26%) papers in AGRICOLA while SCI indexed 1225 (97.69%) agricultural records published in English.Other languages that were used in publishing agricultural research included French, German, Spanish, Afrikaans, Russian, Arabic, Portuguese, Italian, Ukranian, Hungarian, Swahili and Dutch.It was noted that AGRICOLA indexes documents in a variety of languages, while the SCI-E largely prefers Englishbased papers.The distribution of the documents by the language of publication is provided in Table5

Table 5
Distribution of records by language of publication

Table 6
Institutional co-authorships Table4captured some 39 sources out of an identified 595.It illustrated that most journals indexed by AGRICOLA are also indexed by ISI.However, the number of indexed records from the same source is fairly varied (e.g SCI indexes less records).This could be attributed to the large variety of document types indexed by AGRICOLA.Sixth, as is widely known, Africa consists of 53 independent countries whose major non-African languages, according to individual histories, besides the local languages, are 26 English speaking[Anglophone], 29 French speaking [Francophone], 5 Portuguese speaking [Lucophone], 7 Arabic speaking and 2 Spanish speaking.In some countries, however, one or more of the above languages is or are spoken as official and national languages alongside popular African local languages, and ultimately research output is published in these non-African languages.The dominant language of publication with regard to agricultural publications on Africa is English, although publication does occur in other languages as well.English was leading in both databases.91.26% of Thus, collaboration helps speed up problem solving, stimulates creativity and enables inter-disciplinary boundary crossing, which in turn enriches knowledge development and transfer.Co-authored records totaled 1335 producing a collaboration degree (calculated as percentage proportion of co-authored records to singleauthored records) of 58.84 and a collaboration coefficient (the ratio of the number of collaborative records to the total number of records published in a domain) of 0.564 or[56.4].The SCI produced 308 one-author records, 351 two-author records, 251 three-author records and 154 four author-records.The collaboration degree for the SCI was 75.42, with the collaboration coefficient slightly lower, i.e. 0.7536 (75.36).Generally, universities recorded the highest pattern of collaboration, either among the universities or between universities and non-academic institutions (e.g.agricultural research institutions/centers).We noted that research collaboration amongst African based researchers and individuals is very minimal.Finally, we recommend that the research agenda in Africa focuses on how to: support, balance and increase agricultural research; increase research partnership or collaboration in general and among African countries in particular; create and support national indexing databases that would strongly reflect national or regional research output; and audit and map research outputs outside the public domain caused by non-existent indexing services.Further, AGRICOLA and SCI based results be compared with national database indexing records (e.g SABINET based databases such as Current and Completed Research [C&CR] and the Union of Completed Theses and Dissertations [UCTD]) in order to show how national agricultural research trends (reflected in SABINET for South Africa) can be utilised alongside internationally popular databases such as SCI.