from development (Brownlie et al., 2013; Rajvanshi et al., 2011). Bio-diversity offsets, or alternatively‘biodiversity compensation’,‘netconservation benefit’,‘ecological compensation’,‘environmental com-pensation’,‘compensatory habitats’,‘net conservation benefits’,amongst other terms (Bull et al., 2013;Middle and Middle, 2010;Quintero and Mathur, 2011), can take many forms. These include directcompensation in the form of restoration, rehabilitation and re-estab-lishment of eco-systems, acquiring land for conservation, physicalprotection, or the removal of threats (such as vermin) to biodiversity, aswell as indirect compensation, for example in the form of education,research or contributions to biodiversity funds (EPA, 2006; Madsenet al., 2010). Having continued to grow in popularity (Gibbons andLindenmayer, 2007;Maron et al., 2012), offsets particularly gainedtraction when the Business and Biodiversity Offsets Programme (BBOP)was established in 2004 as an international collaboration of companies,government institutions and other organisations, and published thenow widely read and cited paperBiodiversity Offsets: Views, Experience,and the Business Case(Ten Kate et al., 2004) (260 Google Scholar cita-tions at June 2018).Increasingly, the process of planning offsets in order to compensatefor the negative impacts of development is being integrated within theprocess of environmental impact assessment (EIA), where offsets re-present a‘last resort’when options at the earlier stages in the mitigationhierarchy of avoidance, minimisation and restoration have been ex-hausted (BBOP, 2009;Rundcrantz and Skärbäck, 2003). For the pur-poses of this paper, we define EIA to include the following stages:screening, scoping, baseline studies, impact prediction, mitigation,preparation of report, follow-up and monitoring, i.e. including activitiesoccurring post approval decision (Morrison-Saunders, 2018; BBOP,2009).The wide uptake of offsets as part of EIA can be attributed to thebelief that they serve to reduce the number of difficult trade-offdeci-sions between development and environment that competent autho-rities face during EIAs and supporting the achievement of sustainabledevelopment outcomes (e.g.BBOP, 2009; Bedward et al., 2009; Brownet al., 2014;Cowell, 1997;Kiesecker et al., 2010).Designing and implementing effective offsets is a challenging task(May et al., 2017). It requires good data, good planning, andfinancialsupport and considerable time; asBrownlie et al. (2017, p. 252), note:“The time required to design and implement an offset is often underestimatedby the competent authorities and developers. The planning steps andagreements required prior to implementation can be complex: securing pro-tected area status for an offset site can take years”. The integration ofoffsets planning within the EIA process may help to ensure that thecosts of implementing the offset are appropriately identified and in-corporated into the development budget (BBOP, 2009). There is,however, little consensus on the appropriate timing for the formal in-troduction of offsets into the EIA process, as a comparison of interna-tional biodiversity offsets guidance documents demonstrates. For ex-ample the regulator's guidance in Western Australia at the time of theresearch byHayes and Morrison-Saunders (2007)specified that“Ifoffsets are proposed, these should be described in the referral documenta-tion1”(EPA, 2007, p18), i.e. at the very beginning of the EIA process.This guidance and its implications were a key cause of the concernsraised by practitioners in theHayes and Morrison-Saunders (2007)re-search. Interestingly, the offsets guidance at the Federal level in Aus-tralia that was applicable at a similar time reflects the opposite ap-proach, stating that the Federal legislation“does not allow for offsets tobe considered at the referral stage”(SEWPaC, 2012, p11). It goes on toexplain that this is because it is necessary to fully understand thebaseline environment; the potential impacts of the proposal; and theresidual impacts after options to avoid, minimise and rectify have beenapplied before offsets can be considered. The current Western Aus-tralian guidelines are different again, stating:Offsets are thefinal step in the mitigation hierarchy and are only appliedas a last resort after other avoidance and mitigation measures have beenconsidered. Therefore discussion of the mitigation hierarchy would beginat project scoping stage (Government of Western Australia, 2014, p. 25).It therefore seems that there is little consensus on the most appro-priate point of formal introduction of offsets into the EIA process.Regardless of the wording of applicable guidelines, proponents are ofcourse able to begin considering the need for biodiversity offsets asearly as they like in the EIA process, or even before formal EIA com-mences, that is, in the‘pre-application stage’. Consideration of offsetsmay range from a desktop exercise drawing on available biodiversitydata to identify whether or not offsets are likely to be required, throughto more substantial planning of offset proposals. It seems reasonable toassume that the timing of the introduction of offsets into the EIA pro-cess, and the level of planning undertaken at different stages of theprocess will have implications for the successful delivery of the offset,and potentially for the EIA process itself. We therefore ask:What are the implications of how and when offsets are introducedinto EIA with respect to the principles of best practice biodiversityoffsets?This paper explores this question in the context of EIA in SouthAfrica. South Africa is a developing country where socio-economicdevelopment is paramount, but which is also blessed with extra-or-dinary biodiversity (Cadman et al., 2010). While offsets have beenapplied within the EIA systems in South Africa for some years, untilrecently guidance has been limited and inconsistently applied(Brownlie et al., 2017). This has resulted in both a relatively low uptakeof offsets in relation to the number of EIAs conducted (Lukey et al.,2017)and considerable variety of practice with respect to the point atwhich offsets are introduced to the EIA process. This situation providesthe opportunity to explore how the timing of offset decision-making hasaffected the outcomes in a diverse selection of these case studies.In March 2017, however, after this research was conducted, theSouth African Department of Environmental Affairs (DEA) released aDraft National Biodiversity Offset Policy(DEA, 2017).The objective ofthis policy is to standardise the use of biodiversity offsets in the countryand to ensure that offsets are used in a more structured and consistentmanner across different provincial environmental authorities. Thepolicy also encourages provincial authorities to draft more refined andcontext specific provincial biodiversity offset guidelines to supplementthe national policy document in terms of providing more detailedguidance on issues specifically related to that province.The release of the draft Policy therefore also provided an opportu-nity to ask a second question, namely:To what extent does the South African Draft National BiodiversityOffset Policy address issues associated with timing?Section 2provides a brief overview of the South African EIA systemsthat is necessary for understanding the case studies and the researchmethodology.Section 3then outlines our research methodology, in-cluding an overview of the selected case studies. InSection 4we drawon the literaturefirstly to review and summarise the principles of bestpractice biodiversity offsets which provide the structure for our ana-lysis, and then to review existing literature related to the timing ofbiodiversity offsets in the context of these principles. We present ourresults and discussion with respect to our two research questions inSection 5before drawing conclusions inSection 5.2.1.2. South African contextEIA has been conducted in South Africa since the 1970s, albeit in-itially on a voluntary basis (Sowman et al., 1995). EIA wasfirst legallymandated in 1989 through the Environment Conservation Act (ECA)(Act 73 of 1989). However, thefirst EIA Regulations introduced under1The referral process in Western Australia is part of screening i.e. the pro-ponent refers information about the proposed development to the regulator forthe purpose of determining whether or not formal EIA is required.M. de Witt et al.Environmental Impact Assessment Review 75 (2019) 1–122