A critical scientific review on South African governance of genetically modified organisms ( GMOs )

Credible governance of genetically modified organisms (GMOs) is essential because of public concerns in South Africa (SA) and internationally. In this preliminary study, the opinions of a number of scientists with experience and/or interest in GMO governance were determined by means of two questionnaires to determine their perceptions on the credibility of risk governance of GMOs in SA. The respondents felt ‘some improvement’ was required in criteria related to good governance. Excellence (quality of risk assessment) and effectiveness, such as protracted regulatory processes needed ‘some to much improvement’. The responses were evaluated against an analysis of the South African GMO Act, regulations, policy guidelines and available information. The Act provides a pro-active basis for good governance comparable to internationally described risk governance models, but implementation seemed to follow the less advanced technocratic model. A number of reasons were identified such as unclear roles of decision makers. Some of the causes for protracted decision-making identified by respondents were: a) excessive precaution in decision making, and b) different mandates resulting in no unanimity among government departments. Proposals for improvement in credibility included communication as a critical component of risk governance and continued training of reviewers and decision makers.


INTRODUCTION
A recent study, confirming the need for further investigation of the South Africa's (SA) risk governance of genetically modified organisms (GMOs) (Jaffe, 2008), the risk assessment, to avoid confusion over the functions to be performed by risk assessor and risk managers and to reduce conflict of interest' (CAC, 2011).
The co-evolutionary or democratic model (transparent model according to Renn (2008), includes additional structured reciprocal links between science (risk assessment) and policy (risk assessment framing, management and decision making) as well as an evaluation step to evaluate risks versus benefits (Morris, 2011).This requires stakeholder participation (Millstone, 2007;Dreyer et al., 2009).Presently, more research is being conducted on the concept of participation.The role of democratic decision making depends on 'opening up the governance process' (Stirling et al., 2009).Consultation can occur through advisory committees, ad hoc consultations public hearings and regulatory instruments.The Commission of the European Communities (CEC) identified the following criteria of good governance: independency in risk assessment, ethical conduct, transparency, openness, participation, accountability, excellence in conduct and effectiveness (CEC, 2000(CEC, , 2001)).Models of good governance were used to compare development of SA governance of GMOs with international examples of GMO governance.
The SA GMO Act (South Africa, 1997) legislates all GMOs including GM seed and grain, GM microorganisms and GM animals.All activities are included from importation, contained use, trials, general release, monitoring and exportation.The SA area planted according to 2011 statistics (James, 2011) are a combined 2.3 million hectares of GM maize, GM soybean and GM cotton (maize 72% or 1.873 million hectares of total maize planted; soybean 85 % or 1,873million hectares of total; cotton 100 % or 15000hectares).No microorganisms or GM animals have been approved for general release.GM vaccines are only in clinical trials.
In SA, the GMO Act (Act 15 of 1997) is often held as an example for future food safety governance models (Chanda et al., 2010).This model partly resembles the framework for risk governance in Europe.With the exception of the labelling of GMO foods, governance in the SA of GMOs is legislated in terms of the GMO Act as amended (South Africa, 1997) and administered by the Department of Agriculture, Forestry and Fisheries (DAFF).The Act establishes a decision making body, the GMO Executive Council (EC), constituted in practice of one representative from each of six government departments plus the chairperson of the GMO Advisory Committee (AC).The government departments have mandates in terms of their own legislation except for Department of Science and Technology (DST) whose mandate also derives from the National Biotechnology Strategy for South Africa (South Africa, 2001).DOH reactively regulates the safety of food (South Africa, 1972).The Department of Environmental Affairs (DEA) has additional responsibility for environmental safety of GMOs in terms of the National Environmental Management: Biodiversity Act (South Africa, 2004).The AC consists of independent scientists from academia, research institutes and the private sector, and submits recommendations on permit applications to the EC.
The risk assessment step in risk governance is of necessity and partly non-transparent because of the confidentiality of certain company information.Government policy may reduce transparency.This was exemplified by the fact that the EC discouraged government GMO regulatory scientists from responding to the survey described in this paper, by stipulating that these scientists must first obtain written ministerial approval.
Therefore, determination of credibility could not be based on first-hand information but only on perce-ptions by those who viewed the system from outside.These perceptions do however, indicate problem areas.This study should be regarded as preliminary because policy makers and scientists directly involved in the regulatory process were not allowed to participate.

MATERIALS AND METHODS
The study includes an analysis of applicable legislation, analysis of available official guidelines, interviews with the SA government officials and two questionnaires distributed to the SA scientists with knowledge of GMO legislation and risk assessments.The first questionnaire, targeting a limited number of scientists from industry, public research and academic institutes, was designed to: a) qualitatively describe a credibility profile of the SA governance of GMOs from responses to criteria and sub-criteria described in Table 1a, b) probe perspectives in general on criteria of good governance based on the statements in Table 1b.These statements and results were grouped under three categories of good governance, namely policies/procedures (statements 10.1 to 10.15), excellence scientific (statements 11.1 to 11.8) and transparency (statements 12.1 to 12.9).
For the first questionnaire, 24 responses (response rate of 10.2%including possible responses from regulatory authorities and members of the AC) were obtained and considered as a fair number in this field, considering the constraints encountered.The low response rate could be ascribed to: i) Some potential respondents were unfamiliar with the subject; ii) Government officials as well as advisors to government (a possible 54 responses) did not participate and neither did anti-GM lobby groups; iii) in a number of cases, a single response was received per biotechnology seed company or a research institute, instead of responses from individual persons and iv) The internet approach for questionnaires presented several technical problems.
The participants could be considered as a homogenous group, representing applicants or potential applicants for permits and having in common marketing (or general release) of GMO products or an academic interest.
Statements in the three categories of good governance of the questionnaire indicate a reasonable general understanding of 'good governance' among respondents.Although the response numbers were relatively low, valuable information was obtained that should lead to a more in-depth future study of risk governance.The current study focused on assessment of priority needs to improve legislation, policy and implementation.
In a second questionnaire, a few key scientists from the agricultural biotech seed industry in the SA and scientists responsi-ble for preparation of permit applications were questioned regarding their views on reasons for delays in issuing of permits and proposed remedial actions (Figure 2a, b).Six responses were obtained.The participants were existing or potential permit-holders who had submitted new applications in the last five years.Importers of GM grain were not included as they were not directly involved in new submissions.They were also involved in a legal dispute with government at the time of the study.

RESULTS AND DISCUSSION
The majority view of the scientists was that some improvements were required to ensure credible governance.In particular, results to responses 'transparency', 'openness', legislation' and 'participation' needed 'some improvement'.'Effectiveness', 'excellence' 'scientific' and 'accountability'needed 'some to much improvement', whereas, 'ethical'conduct and 'independency' in risk assessments needed 'no to some improvement' (Figure 1).The results are discussed in more detail in terms of three categories that underpin credible risk governance: i) functional separation between risk assessment and risk management (policies and procedures); ii) excellence in performance (risk assessors and reviewers, review procedures); and iii) transparency in governance (communication, participation).

Functional separation between risk assessment and risk management
The legal framework for governance of GMOs in the SA provides for functional separation between risk assessment and risk management (South Africa, 1997).This brings the SA in line with Codex guidelines (CAC, 2011) but also with the most developed examples of independency in risk governance such as the EFSA (EFSA, http://www.efsa.europa.eu).
Almost all responding SA scientists agreed on the importance of separate roles for the EC (Table 1b, 10.5) and AC (Table 1b,10.3),but felt that some improvement was needed in clarifying those roles (Figure 1, Table 1a, 3).The statement 'EC members do not have a role as reviewers of risk assessment data/information' (Table 1b, 10.4), created some disagreement (21% disagree, 67% agree, 12% unsure).Some felt that government by virtue of its own legislation has a responsibility to conduct independent reviews.This is indeed the case with proactive SA environmental legislation (South Africa, 2004) and is now also included in the amendment to the GMO Act (South Africa, 1997).One respondent felt that 'the legislation needs some work; the DEA calls for 'coordinated regulations of GMOs, but undertakes its own risk assessments'.The respondent suggested: 'the AC …will undertake a complete risk assessment of all safety issues.The EC members must address non-safety issues, such as cultural impact, loss of traditional knowledge, impact on trade and labour etc.' Independency as a criterion for credibility was regarded as important.The majority of respondents (92%) agreed that risk assessment should be conducted independently from risk management (Table 1b, 10.2).Respondents also agreed that risk assessors were independent and not subject to pressure from policy makers or stakeholders.The members of the AC were acceptable to most respondents (Table 1a, 4).One area of future concern, though not identified by respondents to the questionnaire, is that the representation on the AC in terms of the amended GMO Act may include two officials representing the 'public sector' --undefined (South Africa, 1997).Presumably, they could be from semi-state institutes or government departments.The concern is that government officials are bound by political policy and may not be regarded as independent.
A specific characteristic of 'independency' is that the risk assessment should be 'a purely scientific activity' (CAC, 2011) and the majority of respondents agreed that this was the case (Table 1b, 10.1).This is the requirement in Section 3.3.a of the GMO Regulations which states that applicants must submit a 'scientificallybased risk assessment', again repeated in the heading of Section 4 (South Africa, 1997).Some sociologists contest this view as not completely possible, as risk assessors (or reviewers) would to some extent approach the assessment subjectively (Meghani, 2009).This argument is held in favour of more transparency in assessment of risks that need to be considered in new governance models.
Ethical conduct relates to the criterion 'independency from pressure of applicants' and includes 'confidential clearance' and lack of 'conflict of interest'.Some respondents were uncertain of the degree of lack of  conflict, probably because names of the sub-committee and most of the AC members were not publically available (Figure 1; Table 1a).One respondent remarked that it was impossible for scientists not to have vested interests.
The AC has a general advisory function but also speci-fic functions prescribed by the GMO AC (South Africa, 1997).Respondents gave strong support for specific functions in addition to reviewing of dossiers.These included advising on uncertainties and disagreements at EC level (Table 1b, 10.6, 100%), being pro-active by pro- proposing specific activities and projects (Table 1b, 10.7, 91%), and initiating new policies pertaining to assessments such as guidelines (Table 1b, 10.9, 92%).Such pro-activity exists in the self-tasking functions of EFSA, (http://www.efsa.europa.eu).Availability of competent scientists could be a limiting factor for members of the AC as they are mostly employed full-time by universities or research institutes and though appointment to the AC is a prestigious position, their available time is limited.There seems to be a need for more risk assessors (reviewers) (Figure 2a).
Respondents highlighted the need for improved effectiveness in the regulatory system.Only a few procedures were touched on in this study e.g.managerial responsibility such as planning (Table 1a, 2) and decision making (Table 1b,10).Effectiveness was reflected by the time taken to reach a decision on the issuing of a permit; approvals for commodity clearances were delayed from 2005 until 2011.A number of reasons for delays were given (Table 2a).There seemed to be inconsistency in requests for more information from different panels or similar information repeatedly requested, and requests for additional information not requested before.Excessive precaution in decision making, shortage of risk assessors, infrequent EC meetings, no unanimity between government departments on national policy; and inadequate guidelines were cited as main causes.This study did not consider for example the need for improvement of quality of dossiers and more training of applicants as often experienced in regulatory situations by one of us as a former regulator.An example of poor quality and incompleteness of a dossier was the case of a banana application (ACB, 2011).Effectiveness of the regulatory system needs addressing in much detail.

Scientific excellence
The quality of risk assessments appeared to require substantial improvement (Table 1a, 5).Codex guidelines were generally followed and the opinion was that international guidelines were being adhered to (Table 1a, 5.7, 5.9).It is possible to conclude that a reviewer with less experience tends to request much more additional information than an experienced reviewer.Reviewers undertake their tasks independently and therefore may form a variety of opinions.There is little opportunity for reviewers to discuss amongst themselves or for inexperienced reviewers to learn from their more experienced colleagues.In these circumstances, Codex guidelines may also be interpreted differently by different reviewers.It is also unclear what approaches to risk assessments are followed, the comparative risk analysis approach (CAC, 2009;Kuiper et al., 2001) or a toxicological risk assessment (Millstone et al., 1999).Excessive caution in decision making was identified as the main reason for delays in approvals (Figure 2a).Training of risk assessors (reviewers) and decision makers was considered as an important remedial action (Figure 2b).All respondents agreed on the importance of peer review of risk assessment reports (Table 1b, 11.1) and that a single multi-disciplinarian could not replace a team of specialists (Table 1b,11.2).The current procedure (South Africa, 2008) is to appoint, on a case-by-case basis, a panel of three reviewers from the AC subcommittee with a member of the AC as chairperson.Each reviewer's report and recommendations are included in a final report to the EC.Additional information may be requested from the applicant by the reviewers.It is not clear from the Guidelines (South Africa, 2008; DAFF, http://www.daff.gov.za)whether the full AC meeting would discuss or review the final report.The procedure seems to emphasize 'reporting' rather than 'reviewing'.The EC would consider the AC reports, public inputs, as well as reviews from the different government departments.This transferred final reviewing to the EC rather than the AC level.The current procedure reduces time taken for finalizing initial reviews but could also lead to inconsistent recommendations and requests for more information, causing delays, as identified in the second questionnaire (Table 2a).Adequacy of expertise within small panels to cover a number of disciplines such as both food and environmental issues in general release permits, may be a constraint.These aspects may be too broad for the small panel to cover although additional expertise could be co-opted.An option would be to conduct peer reviewing at the level of the full AC meeting where a greater diversity of expertise is present.Another option would be to conduct reviewing by focused subpanels.Against this approach, one respondent remarked that 'there is also a need for overall multi-disciplinary understanding as opposed to over-specialized'.Peer reviewing needs to be critically investigated for optimization of this process.
Respondents agreed that a PhD-degree with at least three years of experience or an MSc-degree with at least five years of experience in relevant disciplines seemed adequate for eligibility for the AC.There was no consensus on any particular number of reviews that should have been handled before being nominated (Table 1b,11.6 to 11.8).Countries such as Brazil require a PhD-degree as a minimum qualification.That may not be a practical suggestion because of lack of capacity at this stage in SA.
The respondents identified the lack of sufficient adequately trained risk assessors as a problem (Table 1b,5.1;Figure. 2a).One reason could be the poor remuneration which is considerably lower than that stipulated in the guidelines of the South African Council for Natural Scientific Professions (South Africa, 2003).Reasons could also include exposure to criticism or lack of interest or inadequate time allocated.Respondents recommended that a detailed roster of potential risk assessors should be kept (Table 1b,11.5).Such a roster of expertise is a requirement of the Biosafety Clearing House (BCH,) established under the Cartagena Protocol on Biosafety to the Convention on Biological Diversity (CPB, http://www.cbd.int/biosafety/protocol.shtml).At present, no experts from SA are listed on the BCH website, despite many names having been submitted, making it difficult to source expertise.In addition, the latest Protocol records show that the Ministry of Environmental Affairs is the national focal point with only one communication record received, both this department and the Ministry of Agriculture are listed as competent authorities, and four records and 13 decisions (permit approvals) received from DAFF, though more than 350 permits are granted annually (DAFF, http://www.daff.gov.za).DAFF's clearing house link seems to be inoperative at times.
As a remedial option to address delays, respondents agreed on the need for AC members to have continued exposure to new information, to attend international meetings and conferences, and to be involved in research and new developments in relevant disciplines including risk assessment approaches (Table 1b,11.3 and 11.4).Training of EC and AC members was regarded as essential to help reduce delays in the regulatory process (Figure 2b).
One respondent felt that assessors (reviewers) also needed 'contact and input from industry experts and practices'.Industry, especially the multi-national industry, has the means to employ world class specialists to conduct research in specialized areas'.Knowledge and experience of scientists from the industry may be considered in specific cases according to the Food and Agricultural Organization and World Health Organization's expert group (FAO/WHO) (2006).
International outreach programmes often focus only on regulatory authorities unless the need of risk assessors (reviewers) is brought to their attention.Where skills and experience are scarce, capturing of institutional memory collectively is important.This has not been well exploited in the SA risk analysis framework.Coordination in this respect is important.The majority of respondents felt that scientists who were members of the AC should be included in government delegations to international meetings/conferences.There seemed to be a general awareness of the importance of access to international scientific exposure and the need for improvement in this regard (Table 2a, 5.3 to 5.5).In a strategic review of Food Standards Australia New Zealand (FSANZ, 2003(FSANZ, -2004)), it was stated that credibility 'was founded on the quality of its science and its scientific risk assessments'.One of the recommendations was that the FSANZ continued to forge strong linkages with international experts and other regulatory agencies.FAO/ WHO (2006) confirmed continuous exposure to international developments was important.

Transparency
The majority of respondents rated transparency as needing some improvement.The questionnaire tested a few pertinent issues, namely clear procedures; communication of uncertainty; and identification of the risk assessors (Table 1a, 9.1-3).The need was identified for some improvement in all three areas.Procedures, guidelines, application forms, and permits are adequately published on the website of the DAFF, but there appears to be a lack of transparency in EC decision making.The minutes of the EC meetings are too cryptic, suffer delayed listing and risk assessment recommendations are not available.The Promotion of Access to Information Act (South Africa, 2000a) that aims to 'foster a culture of transparency and accountability', provides at a nominal fee a legal opportunity to demand additional information.Compared with the extent of accessibility of reports in Australia (FSANZ, http://www.foodstandards.gov.au;http://www.ogtr.gov.au) and EFSA (EFSA 2006(EFSA , 2009)), the SA website information (DAFF, http://www.daff.gov.za) has limitations.

Participation
The majority of respondents agreed that stakeholder participation regarding the evaluation of the potential risk in the context of socio-economic impact was invaluable (Table 1b,12.4).Participatory procedures are not new to the SA decision making but represent a new concept in the risk analysis process.In general, respondents indicated that significant improvement was needed for participation in the SA system, including participation in policy development and risk assessment (Table1a, 8).All respondents agreed that risk analysis policies should be developed in collaboration with stakeholders (Table 1b, 12.1) and that communication was important (Table 1b, 12.2).Opportunity for public input is included in the GMO Act (South Africa, 1997) and should be exploited in a structured way.
In line with Article 26 of the Cartagena Protocol on Biosafety (Secretariat of the Convention on Biological Diversity, 2000), the GMO Act includes provision for socio-economic assessment as part of the risk analysis.This may include cost-benefit and risk-benefit comparisons but Section 5.1 in the regulations is extremely vague on required, but undefined consideration of the socio-economic impact on biodiversity, access to natural resources, cultural traditions, knowledge, and practices.The majority (75%) of respondents agreed that socioeconomic analysis should take place prior to decision making (Table 1b,10.14).At this point, 'opening up the governance process through public participation' (CEC, 2000) would be important.Stakeholders may contribute to 'democratic' decisions of the risk managers at some stage in the process.One respondent stated that, 'nonsafety issues should be considered by the EC (mandate to review) and not the AC (focus on safety and science).Non-safety issues are relevant only to general release, not to contained or confined activities, which are experimental and short term'.Another respondent commented that 'socio-economic effects should have a minor influence on risk decisions'.Although socioeconomic issues may be addressed as a requirement of the GMO Act (South Africa, 1997), little information regarding the procedures and requirements for socioeconomic analysis is available.

Limitations of the study
(1) Statistical analysis of the survey, althoughthe ideal means of analysis, is not possible with small target groups.This is unfortunately the case with a very specialized field such as risk governance of GMOs in a country with limited resources.In an almost similar study by Wentholt et al., (2009) on the risk analysis of GMOs, only 33 out of 106 invited European respondents participated and 19 out of 60 international non-EU participants.They were from a range of professions and occupations.Government officials with intimate knowledge of the system would most likely not be inclined to participate.Their opinion, should they be officially requested, would most likely not be spontaneous.However, their experience in the process of risk assessment is of great importance to identify needs of applicants.
(2) One deficiency identified in the study, is that we did not obtain the opinion of regulators on the quality of applications and provision of required information.
(3) The responses should largely be considered as 'perceptions'.Experience of the respondents are real, perceived, or from hearsay.Perceptions are valid observations that can be changed by improving the system and need be taken into consideration in partici-Jansen van Rijssen et al. 5019 pative policy development, particularly on sociological matters.
(4) This study illustrates the opinions of some stakeholders at the time of the study.This may change with time for the same group.A group (different segment of the population) with priorities other than marketing of their products may have a different perception of credibility.Consumers, in general, are not familiar with the technical side of regulatory matters and, therefore, a more simplified questionnaire will have to be designed for them.
(5) The study should be considered as preliminary as a more balanced opinion would include the experience of those intimately involved in SA risk governance of GMOs according to the mentioned criteria.They should include regulatory authorities, advisors to government as well as members of the AC.

Conclusion
This study tried to determine: How credible is governance of GMOs in SA as perceived by scientists with knowledge of the system?"In general, respondents felt that 'some to much' improvement is necessary to ensure credibility.An analysis of the GMO Act, policy guidelines and delays in issuing of permits, as well as available information on the implementation of the Act, confirmed the perceptions.Based on these results, an indication is given where improvements are most needed to increase credibility of the system.
Although functional separation of risk assessment (AC) and decision making (EC) has been established in the SA legislation, the scope of decision makers' responsibilities needs be clearly defined as it seems as if the EC still functions according to intentions of the 'technocratic' model that may over-ride recommendations by the advisory body.
This could cause increase in the workload and may be outside the expertise and mandate of some EC members, resulting in further delays in decision making.
A clearly identified deficiency was excellence in performance.This implied the need for improved review processes, and elimination of delays.The degree to which potential poor preparation of application forms contributed to delays could not be assessed due to lack of participation by regulators.Continued constructive exposure to new information, research and/or development as well as training of both decision makers and risk assessors were identified as priorities.Improved guidelines that address new challenges such as assessment of stacked traits need urgent attention.
Many aspects of transparency really need some improvement.'Participation', in particular, as a democratic principle, has not yet been clearly developed in the SA governance of GMOs.This includes the contribution of stakeholders and scientists to framing of the assessments.Framing includes scientific inputs, policy development on various matters such as scientific approaches to risk assessments and socio-economic impacts and cultural considerations.New research on how to address participation by stakeholders at different stages of risk analysis needs attention.Government has to consider development of integrated strategies and policies for good governance to achieve greater credibility of risk assessment of GMOs, not only of the risk assessment component itself, but also for all other components of governance.These should include: (1) Harmonization between government departments in the approach to risk analysis, taking into account the National Strategy for Biotechnology for SA (South Africa, 2001) and including policy issues such as the approach to precaution.
(2) Policy on training, recruitment and remuneration of risk assessors (South Africa. 2003).
(3) An additional step could be consideredin the iterative process of risk analysis as described by Codex Alimentarius Commission for development of risk assessment policy that could give direction according to national strategy and good governance policies.(4) The majority of the criteria of good governance could be grouped as communication related, transparency, participation, and openness.Therefore, improved communication is critical.
Although the credibility of the SA governance of GMOs could be improved, it has certain strengths that could be applied to other risk analysis systems.Functional separation of the risk assessment and decision making as contained in GMO legislation could be followed as an example to ensure scientific integrity; however, implementation of new legislation with respect to risk assessment such as for pesticides, should be critically considered with the necessary awareness of good risk governance to ensure credibility of decisions.The availability of experienced risk assessors is a limiting factor in credible governance.The use of regional or subregional independent experts could be considered while in the meantime, educating and training of local candidate risk assessors should be a priority.
In general, the robustness of the governance of GMOs is reflected in the track record of safety since 1990, as no significant impact from accidents or adverse effects on humans or animals have been recorded.Trust and confidence will depend largely on the introduction of more democratic governance for both GMOs and other food safety matters.

Figure 2a .
Figure 2a.Responses from the agricultural biotechnology industry / applicants on the reasons for delays in approval of GMO permits.(Weighed responses, ranking groups: 1 = unimportant to 5 = extremely important).

Figure 2b .
Figure 2b.Respondents from agricultural biotechnology industries/applicants on proposed remedial actions for delays in GMO permit approvals.(Weighed responses from ranking groups: 1 = unimportant to 5 = extremely important).

Table 1a .
South African GMO risk governance: Scores for credibility criteria and sub-criteria.
NI = No improvement; SI = some improvement; MI = much improvement; DNK = do not know

Table 2 .
Contd.Responses to status of GMO risk governance sorted on 'some improvement needed'.