The contents and scope of environmental impact statements: how do they evolve over time?

Abstract

It can be expected that, over a long time span, both the contents and the scope of environmental impact statements (EISs) will evolve to incorporate new issues and new analytical approaches. A diachronic review was conducted of a sample of EISs produced by one consultancy in Brazil, specializing in mining and quarrying. Nine EISs prepared in the period 1987–2010 were selected, covering different minerals and locations. Content analysis showed that the scope of the EISs has widened and the content has deepened over time, particularly in terms of level of detail and issues dealt with in baseline studies, while impact analysis has evolved to incorporate some quantitative predictions and mitigation measures have become structured along a consistent framework, as compared with a simple list of loose commitments that appeared in the first EISs. Interviews conducted with both regulators and consultants suggest that the main driver of change has been new legislation, followed by better regulation and administrative control exerted by the environmental agencies and improved consultants' capacity, while proponents' policies had little influence on the observed evolution. If EISs are to evolve further in order to incorporate new analytical approaches and become less descriptive, the initiative should come mainly from regulators.

Keywords:

• environmental impact assessment
• EIS quality
• content analysis
• mining
• Brazil

Introduction

The quality of environmental impact assessment (EIA) documents is often pointed to as one determinant of their effectiveness. A number of studies have discussed the quality of such documents, in particular of the assessment report often named an environmental impact statement or study (EIS).1 A summary of previous studies on EIS quality is shown in Table 1. An acceptable, adequate or satisfactory EIS is usually considered to be one that suits the needs of a particular decision-making context, which precludes any universal characterization of a ‘good’ EIS. However, several attempts have been made to make sense of what constitutes a quality EIS.

Table 1 A summary of previous research on EIS quality.

Author	Sample	Country	Period	Methods
Lee and Brown (1992)	83 EISs/different sectors	UK	1988–1991	Lee and Colley package
Lee and Dancey (1993)	40 EISs/different sectors	Ireland	1988–1992	Lee and Colley package
Glasson et al. (1997)	50 EISs/different sectors	UK	1988–1996	Pairs of EISs – matching type of project, location and other characteristics – in two periods, to make a comparative review
Bojórquez-Tapia and García (1998)	33 EISs/highway projects	Mexico	1989–1994	A set of criteria developed for purpose resulting in a score for each EIS
Wende (2002)	145 EISs/different sectors	Germany	1990–1997	Statistical analysis of 11 variables
Cashmore et al. (2002)	72 EISs/different sectors	Greece	1990–1999	Lee and Colley package
Gray and Edwards-Jones (2003)	89 EISs/forest sector	UK	1988–1998	Lee and Colley package
Canelas et al. (2005)	46 EISs/different sectors	Spain and Portugal	1998–2003	European Commission EIS review guidance
Carrasco et al. (2006)	43 EISs/highway projects	United States	1980–1999	Measuring a ‘completion index‘ to quantify the amount of information included in an EIS
Gontier et al. (2006)	38 EISs/different sectors	Sweden, France, UK and Ireland	1999–2003	Analysis of contents pertaining to biodiversity
Pinho et al. (2007)	13 EISs/small hydro projects	Portugal	1990–2003	A set of 12 review criteria and 43 subcriteria, resulting in a score for each document
Sandham and Pretorius (2008)	28 EISs/different sectors	South Africa	Not specified	A modified version of the Lee and Colley package
Jalava et al. (2010)	15 EISs/incineration plants	Finland	Not specified	Comparing ratings assigned by consultants and public officials, according to the European Commission EIS review guidance
Kabir and Momtaz (2012)	30 EISs/different sectors	Bangladesh	Not specified	Lee and Colley package
Sandham et al. (2013)		South Africa	before and after 2006	A modified version of the Lee and Colley package

Most papers report on studies aiming to determine the quality of samples of EISs in a particular jurisdiction. Research methods vary. Several authors (Table 1) used the well-known Lee and Colley package (see Lee et al. 1999 for an updated version), while Sandham and Pretorius (2008) and Sandham et al. (2013) adapted this approach, developed in the UK, to the South African context. Canelas et al. (2005) and Jalava et al. (2010) used the guidance document prepared by the European Commission (EC 2001) to help competent authorities to review EISs. Bojórquez-Tapia and García (1998), Wende (2002), Carrasco et al. (2006) and Pinho et al. (2007) developed their own criteria to review the documents. Gontier et al. (2006) looked for biodiversity-related contents in their sample of EISs from four countries. A number of authors graded the EISs, using the above-mentioned review package, defining a scale of completeness (Carrasco et al. 2006) or endeavouring to describe how elaborate an EIS is (Wende 2002).

Sample size in these studies varied from 13 to 145 EISs. Not all authors claim representativeness of their samples. The biggest sample is from Germany, where Wende (2002) performed a cross-sector evaluation of 145 EISs prepared between 1990 and 1997. Out of his conclusions, the author points that EIA procedures did have an influence on decisions related to project approval and that a variable that statistically explains this influence is the degree of ‘elaboration’ of impact analysis and prediction which, for the purposes of the present paper, can be interpreted as a proxy for EIS quality. Having graded the EISs in three levels, Wende also concluded that their quality had improved over the period. A secondary finding of the author, although not statistically tested, is that EISs became more ‘extensive’ over time.

Cashmore et al. (2002) found that, in the Thessalonica region of Greece, better studies were prepared for larger, more complex and more controversial projects, as well as for types of projects for which more accumulated experience existed, especially road schemes.

Out of the 89 EISs analysed by Gray and Edwards-Jones (2003), 69 were graded as ‘very poor’ and 19 as ‘poor’. They highlighted ‘unfocused baseline data collection, inadequate identification of impacts, and inadequate determination of impact significance’.

In Mexico, Bojórquez-Tapia and García (1998) found that the EISs ‘are descriptive, inconsistent, and unsystematic’ (p. 235). In South Africa, Sandham and Pretorius (2008) found that the most satisfactory parts of the documents were those ‘descriptive and presentational’ (like project description) in contrast with analytical components.

Although review criteria and methods vary across these studies, they all identify deficiencies in EIA documents. A feature of those studies that covered a time span is that they captured improvements over time in EISs prepared in different countries and for different kinds of projects. Glasson et al. (1997) compared and graded pairs of EISs – matching type of project, location and other characteristics – prepared in two periods, finding an improvement in quality, although many in the recent group were considered to be ‘still unsatisfactory’ (p. 457). Canelas et al. (2005) found ‘a clear improvement’ (p. 222) in EIS quality over time, but expressed concern about the significant percentage of poorly ranked EISs. Pinho et al. (2007) obtained higher scores for the reports prepared in the period 2000–2003 than scores for those prepared in the period 1990–2000. They speculate that a modification of the Portuguese legislation, which came into force in 2000, may have influenced the quality of the EISs.

Using a different approach, Tzoumis and Finegold (2000)2 reviewed ratings assigned by the United States Environmental Protection Agency to 9405 EISs filed from 1970 to 1997. As the EPA rates the adequacy of information presented in the statements on a scale of three levels, the authors expected to find an improvement over time, considering that ‘increased knowledge and learning by agencies preparing these documents should have resulted in better information quality’ (p. 560). However, their conclusions did not support this hypothesis, as the supposed improvement was not observed.

Other researchers didn't seek to capture changes over time. Carrasco et al. (2006) reviewed 43 EISs prepared over a 20-year time span (1980–1999). They developed a ‘completion index’ to quantify the amount of information included in an EIS and link the quality of an EIS to the accuracy and completeness of environmental data. In their sample, they found a statistical correlation between this index and both environmental and economic characteristics of a project, such as, respectively, number of endangered species and cost of construction. The approach of Jalava et al. (2010) was to compare ratings assigned to EISs by two different groups of consultants and public officials. The authors found that the latter group grade EISs slightly and consistently lower than the former. They did not seek to detect changes over time. In contrast, Sandham et al. (2013) aimed at detecting possible improvements in EIS quality after a legal reform in the South African EIA regime, but found that overall quality did not improve, implying that regulatory change did not drive practice change.

The accuracy and the reproductibility of ratings is a concern for Peterson (2010) and Põder and Lukki (2011). The former reported on an experiment in which 24 professionals (officers within EIA authorities and consultants) individually reviewed one EIS using the European guidance (EC 2001). Subsequently, this EIS was reviewed by groups of these professionals, who assigned lower grades to the main sections of the EIS than those assigned by themselves acting as individual reviewers. The latter authors take issue with the use of checklist approaches for EIS evaluation, arguing that results are barely reproducible. They submitted an EIS of an oil refinery in Estonia to 41 students who had just completed an EIA course, asking them to use the EC (2001) guidance, finding significant variation in results and interpreting such interpersonal variation as a fundamental flaw of any checklist-based EIS review.

Departing from the literature that inquires into the quality of EISs, this paper focuses on a diachronic examination of the contents of EISs. The research aimed at detecting and documenting a possible evolution of EISs, looking at the work of a particular type of practitioner, the consultant team. Analysis of actual consultant EIA practice is scarce in the literature. Recent studies were performed by Morrison-Saunders and Bailey (2009), who interviewed and surveyed two groups of practitioners in Western Australia – consultants and government officials – to gauge their perceptions of collaboration and tensions between them and by Morgan et al. (2012), investigating practice in New Zealand and finding different perceptions of impact assessment ‘adequacy’ between professional groups involved in the preparation of EISs, as well as between reviewers in government.

One aspect insufficiently discussed in the literature on EISs quality is the identification of reasons or drivers of change. Pinho et al. (2007) did mention that new legislation in Portugal could be credited with the improvement in quality they observed. In South Africa, however, Sandham et al. (2013), aiming to detect this possible relationship, found no influence of new regulations on the quality of the documents. Glasson et al. (1997) interpreted their findings about quality improvement in the UK as a process of ‘learning from experience’, speculating that amelioration of government guidance and training may also explain gains in EIS quality. On the other hand, in the United States, the extensive review of quality information ratings performed by Tzoumis and Finegold (2000) found no sign of learning among practitioners. Wende (2002), who grounded his conclusions of improvement over time on statistically tested evidence, did not advance possible reasons for the observed change.

All cited literature is context-specific, that is, it discusses practice in a particular jurisdiction. Also, previous research is based either on reviewing documents or on interviewing practitioners. This paper is no exception, as the research was conducted in Brazil by reviewing EISs and interviewing practitioners.

In summary, the research objectives are twofold: (1) identifying and documenting a possible temporal evolution in the contents of EISs; and (2) exploring the possible drivers of change. It is expected that more recent EISs are more detailed and broader in scope than those prepared in the past. It is also expected that (1) more stringent legislation and (2) consultant experiential learning could explain the anticipated evolution.

Methods

A consultancy firm established one year before EIA regulations came into force in Brazil (1986) was selected to provide a sample of EISs produced over a long time period. Few of the specialized consultancies that pioneered EIA practice are still active, and although many were established afterwards, a forerunner was needed to allow for the maximum possible time frame. The authors are aware of only two firms that fulfill this requirement, both specializing in mining. One of those firms was approached and agreed to participate in the research. The fact that the firm specializes in mining and quarrying projects is advantageous for the research, as this characteristic limits the range of variations that could stem from a consultancy assessing multiple types of projects.

Once the firm had agreed to participate, the first step was setting a database and compiling information from the consultancy's files. Select data include: (1) year of completion of EIS; (2) project location; (3) minerals involved; (4) project proponent; and (5) whether the EIS was prepared for a new, greenfield project, or for the expansion of an existing mine.

Cases for review were intentionally selected based on the following criteria: (1) preference for greenfield projects, especially for the first decade of implementation, since, in this initial period, regulatory authorities often required an EIA for existing mines; (2) projects suggested by the consultancy team because of characteristics such as complexity, innovation or extensive public involvement; (3) preference for an array of minerals, ranging from relatively simple assessments of sand pits to more complex mining and industrial complexes in the fertilizers sector; (4) projects located in different states, as regulations often differ and the capacity of environmental agencies are also different; and (5) assessments distributed along the whole time span of the consultancy's activities.

A checklist was prepared to guide the review of the EISs. Guidance for structuring the review was obtained from the content analysis literature. Content analysis is a tool extensively used in research about communication that has also been employed in research pertaining to problems in the field of environmental planning and management, such as analysing corporate sustainability reports (Pérez & Sánchez 2009; Monteiro & Guzmán 2010; Sen et al. 2011), reviewing corporate environmental strategies (Park & Ahn 2012), identifying which sustainability-related concepts are most frequently used by companies to describe their sustainability programmes (Jones et al. 2010) and at least one case of EIS review (Gontier et al. 2006).

For the present research, content analysis techniques were used to check a text (in the occurrence, EISs) against research hypothesis about its contents. By reducing the complexity of a text to a number of categories of analysis, this technique allows for replicable inferences to be made (Krippendorf 2004). Bauer (2000) mentions six research approaches to content analysis, one being longitudinal, where comparisons over a long period, but restricted to one particular context, allow for detecting fluctuations in contents and inferring context changes. This longitudinal research aimed at encompassing as long a period as possible of EIA practice.

To review the contents, semantic units of analysis were selected. These are broadly understood as those ‘designed to describe and explicate a communication or set of communications in a systematic, objective and quantitative manner’ (Crano & Brewer 2002, p. 245). The semantic units of analysis are the major chapters or subchapters of an EIS. From the structure usually seen in Brazil, the following major units were chosen: project description, alternatives, baseline, impact analysis and mitigation. These units are partially coincident with the ‘review sections’ of the European guidance (EC 2001), which asks reviewers to consider seven items, including non-technical summary and quality of presentation, which have no interest for this research. Baseline was divided up in 13 subunits relevant to the type of project considered and drawn from the table of contents of the most recent EISs reviewed. For other types of projects, a different structure for reviewing the baseline could be adopted. Data collected from each document was tabled and compared to search for regularities, temporal changes or innovations.

The number of pages was used as a descriptor of the contents of each unit of analysis. Ideally, the number of words should be counted as a more accurate descriptor of the level of detail in project description, but although this can be easily done for a digital EIS, older studies exist only in paper support and word counting would be extremely time-consuming. As a project is not only explained in words, but in technical drawings, plans or artistic renderings, the number of such figures was also computed.

Finally, hypotheses to explain observed temporal changes were formulated and tested against results from an interview applied to the current and two former heads of the mining branch of São Paulo State Department of Environmental Impact Assessment (a division of the Environment Secretariat). This state was selected because the consultancy is based there and a significant portion of its work, especially in the early years, was performed in São Paulo. Further enquiry was conducted with staff at the consulting firm, based on semi-structured interviews to cross-check responses provided by the public officials.

Hypothesis about possible drivers of change were drawn from the reviewed literature on EIS quality that speculates on reasons for observed improvement, from papers on controls on the EIA process (Ortolano 1993) and from the authors' experience with the Brazilian regulatory context. The selected possible drivers are: (a) more capacity built into consultancies; (b) standards adopted by project proponents; (c) advances in legislation; (d) terms of reference for EISs more carefully designed; and (e) mandatory provision of supplementary information promoting learning within consultancies.

Possible limitations of this study include the portfolio of the particular consultancy firm selected, whose projects concentrate on aggregates and industrial minerals. However, as previous studies suggest that variations in EIS quality do appear to relate to the type of project (Barker & Wood 1999; Wende 2002), it is expected that variations in scope and content may follow a similar pattern.

Results and discussion

Cases selected

Company files displayed 63 EISs prepared in the period 1987–2010. Applying the criteria described in the previous section, nine were selected for review (Table 2). The company's portfolio features several small projects (mostly sandpits), very few metal mining projects and a concentration in industrial minerals, including limestone for cement manufacturing and phosphate rock for fertilizer production. Thus, the sample includes one sand pit (sandstone), one granite quarry, a set of three limestone mines, one metallic project (bauxite) and three phosphate rock cases, one being doubled, that is, two EISs prepared for the same project at a 17 year interval. This project was selected, among other reasons, for its complexity and the significant impacts it may cause. The proponent gave up implementing the project in the early 1990s owing to changing economic conditions and resumed it in 2005–2006. The sample features cases located in four states. Although the consultancy's portfolio includes EIA projects in eight other Brazilian states, in reviewing the project briefings, they were discarded for the sake of fitting into the selection criteria and to limit the workload in reviewing the documents. Project proponents include a range of companies, as depicted in Table 2.

Table 2 Cases selected for review.

Serial number	Year of completion	Short description of the project's main features and location	Profile of project proponent
1	1987	A granite quarry in São Paulo metropolitan area producing aggregates	A Brazilian company from the construction sector
2	1989	A phosphate rock mine, an ore concentration plant, a tailings dam and a chemical fertilizer plant in a rural area in Santa Catarina State	A Brazilian company from the industrial sector
3	1990	A limestone mine in Minas Gerais; three new open pits near an existing lime plant	The biggest cement manufacturer in Brazil
4	1993	Expansion of a small quartzite mine in an environmental protection zone in Minas Gerais State	A multinational company from the glass manufacturing sector
5	1998	A sandstone mine and treatment plant for glassmaking in São Paulo State	Another multinational company from the glass manufacturing sector
6	2002	A set of small bauxite mines and stockyard in a rural area in São Paulo State	One of the biggest world aluminium producers
7	2006	A second EIS for project 3, which had not been implemented and was proposed again with modifications	A joint venture of two multinational corporations
8	2007	A new phosphate rock pit adjacent to an existing mine in Minas Gerais State	A multinational corporation
9	2010	A phosphate rock mine, an ore concentration plant, two tailings dams, a water dam and ancillary facilities in Tocantins State	A junior mining company incorporated in Canada and operating only in Brazil

Project description

There are many kinds of mining projects submitted to EIA, from large undertakings featuring outputs of several million tons per year to small open pits. It is expected that smaller and simpler projects will deserve a short description in an EIS, while a large and complex project will require several pages, tables, figures and maps. In order to make sense of the projects' level of complexity, a scale was established: level I – the project comprises only open pits, waste dumps and ancillary facilities; level II – besides the precedent components, the project comprises a treatment plant; level III – the project comprises also a tailings dam; level IV – the project comprises also further industrial processing, such as a cement or a fertilizer plant. Table 3 shows this classification alongside the annual output (run-of-mine), which is a descriptor of project size. It can be noticed that, compared with large metal or coal mines, all cases could be referred to as small to medium-sized. Projected outputs from the selected cases are not always presented in the same way in the EIS; they can be featured as ore actually mined and transferred to treatment (e.g. the bauxite mine) or as output of concentrate from the plant (the three phosphate cases which comprise concentration). As the actual amount of ore excavated and transported is a better descriptor of project size, estimates were made from data presented in the studies. Yet, data about waste rock is not included. Another possible descriptor could be the project's footprint, but this data is not consistently reported in the documents and could not be collected for comparative purposes.

Table 3 Projects’ complexity and their description in the EISs.

Serial number	Level of complexity	Annual mine output (tons/year)	No. of pages of project description	No. of plans and drawings
1	II	320,000	15	5
2	IV	∼5,000,000	82	24
3	I	300,000	31	12
4	I	∼10,000	19	20
5	III	1,600,000	39	11
6	I	135,000	20	2
7	IV	∼5,500,000	93	50
8	I	3,625,000	86	23
9	IV	∼3,000,000	123	103

Table 3 also shows the number of pages used for project description in the EIS, as well as the number of maps or plants pertaining to this topic. The average number of pages for level I projects is 23, the only level II project has 15 pages for this topic, while level III projects has 39 and the average for level IV is 99. Thus, there is an apparent relationship between these variables. Table 3 also shows the number of drawings in each case, the three more complex cases featuring more drawings.

Alternatives

Alternatives start being mentioned in case 2. Space dedicated to alternatives is, in all cases, limited to two or three pages. Technical alternatives are described only in cases 2 and 9. Location alternatives are mentioned in all other EISs, for project components such as waste rock piles and transportation routes. In case 5, the regulator requested the EIS to be supplemented by a study comparing location alternatives for the industrial plant and the tailings dam. In case 6, the project proponent voluntarily restrained from excavating in fragments of native vegetation.

In all cases, consideration of alternatives remained essentially descriptive. However, the scope of alternatives is inherently limited in a mining project, as the pit itself can only be located over the ore body.

Baseline

The scope of baseline studies and the space dedicated in each EIS were considered in the review. Table 4 shows a list of themes featured in the baseline description. Thirteen themes were chosen. As the structure of baseline description changed over time, the featured topics are organized along the three more recent reports and do not necessarily appear as such in the earlier reports. The number of pages reported includes, where appropriate, any appendix containing additional or detailed information (e.g. lists of species appear included in the main text in some EISs and as an appendix in others), but excludes any appendices featuring raw data, such as chemical analysis of water samples. Likewise, pages containing bibliographical references were not considered. For the category ‘maps’, the table includes any cartographic representation in format A4 or larger prepared by the consultancy. Many EISs, especially the most recent, feature several small-scale maps, often from secondary sources, and these were not considered as ‘maps’ for the purposes of this research, as they all are small-scale maps presented for contextualization. Table 4 also shows the number of professionals in the team for each report.

Table 4 Selected contents of baseline studies.

Theme^a	No. of pages/no. of maps
	1	2	3	4	5	6	7	8	9
1 – Geology, geomorphology, soils	9/1	18/6	10/2	8/2	11/2	12/5	23/7	17/5	103/6
2 – Speleology	–	–	8/6	–	–	–	–	–	10/5
3 – Climate	25/1	1	3	–	3	3	7/1	4	18/1
4 – Hydrology, hydrogeology	6/1	22	–	–	12/2	4	13/5	8/1	12/1
5 – Air quality	–	–	–	–	3	5	1	7/1	9
6 – Water quality	–	5	10/1	13/1	34	10	5/2	6/1	55/2
7 – Noise and vibrations	–	–	–	–	1	5	11/2	15/2	6
8 – Terrestrial ecosystems – flora	4	28/1	9	23/1	12/1	16/2	31/1	23/2	72/5
9 – Terrestrial ecosystems – fauna	4	30/1	15	12	22/1	61	98	89	109/1
10 – Cave fauna		2	6	–	–		–	–	62
11 – Aquatic ecosystems^b	–	30	22	–	17	–	–	–	35
12 – Social environment	80	50/1	14	8/1	20/2	14/1	43/2	25	54/1
13 – Archaeology^c	–	–/1	5	–	–	–	–	–	139
Total no. of pages	128	184	102	64	135	130	232	194	553
Total no. of maps	3	9	9	5	8	8	20	12	22
No. of fauna survey surveys	n.a.	1	2	2	2	5	2	1–4^d	2
No. of days of fauna survey effort	n.a.	6	10	12	8	n.a.	12	1–6^e	20
Professionals in the technical team	9	15	10	10	18	15	31	18	23
Notes:
^a EISs do not have the same structure of baseline description; adaptations were made in order to fit baseline chapters or items into the categories of the table.
^b Aquatic ecosystems include descriptions of fish, benthic and plankton communities.
^c Archaeology often featured as part of the human environment description.
^d Different number of surveys for different faunal groups.
^e Field survey effort varies according to faunal group.

A number of findings of interest to this study can be obtained from Table 4. They are discussed in terms of (1) size, (2) cartography, (3) scope and (4) level of detail. In terms of size, no clear trend suggesting that baseline description became more extensive appears at a first glance, but the three most recent EISs have longer descriptions of the existing environment, especially case (9). However, examining the printed copies of the earliest and the latest EISs, the amount of empty space in each page is clearer bigger in the former, suggesting fewer words per page than more recent reports. Hence, it can be asserted that baseline descriptions have become more extensive over time.

The use of maps is also larger in the last three cases. Overall, the topic that makes more extensive use of cartography is geology, geomorphology and soils. The scope of studies did not show any clear evidence of enlargement, that is, more topics being treated in the EISs, and appears to be more linked to the environmental setting of the project. In terms of level of detail of information presented, if size does matter, more recent reports feature more text and more maps than older reports. In seven out of nine cases, the biggest part of the EIS is fauna description. Noise starts being dealt with as a stand-alone topic in 1998 (case 5) and its importance has increased ever since; noise maps were added in 2006.

The use of primary data also increases over time, but appears to be related to characteristics of the local environment. If case 1 is almost entirely based on secondary data, in cases 2 and 3, the level of detail of baseline is clearly higher and based on field studies, while cases 4 and 5 again largely rely on secondary data. Both are small projects, such as case 6, but here field surveys are prominent again. Cases 7–9 feature more extensive collections of primary data. These findings from the review of each EIS are, to some extent, confirmed by the number of fauna surveys and the number of days of field effort, both shown in Table 4. However, baseline studies remain largely descriptive and do not adopt new analytical approaches such as landscape ecology, ecosystems services or community vulnerability.

Impact analysis

Methods for impact identification and assessment changed in the period. Four styles can be identified. In the first EIS (1987), impacts are presented in a list unrelated to their causes. In the second (1989), a list of impacts is arranged in a matrix relating them to project actions. In cases 3–5 (1990–1998), actions are related to effects (defined as modifications or natural processes) which, in turn, are related to impacts (modifications in the quality of environmental resources). In cases 6–9 (2002–2010), a matrix features two fields: the first relates actions to environmental aspects (in the sense of ISO 14001 standard), which are then related to corresponding impacts in a second field, following a procedure proposed by Sánchez and Hacking (2002). Thus, except for the first EIS, an attempt is made to establish cause–effect relationships for impact identification (Perdicoúlis & Glasson 2009).

Impact prediction is dealt with in a systematic way from case 6 onwards, when indicators of magnitude of environmental aspects (in the meaning of ISO 14001 standard) or impacts are presented; a table summarizing such indicators is featured in this and subsequent EISs. However, analysis evolved to incorporate only a few quantitative predictions and does not make extensive use of modelling.

The assessment of importance of impacts also changed over time, from an unspecified and unstructured approach used in the first EIS to an argumentative style in cases 2–5 to a simple scaled-weighted approach used in cases 6–9, which integrates only the perspective of the assessing team and fits the ‘technical approach’ described by Lawrence (2007). In no case does any attempt to integrate the perspective of stakeholders or affected communities appear, nor is it required by regulations.

Mitigation

In all reports, mitigation is presented as a list of actions to be implemented by the project's proponent. Brazilian regulations call for an environmental management plan (EMP) to be submitted in a later stage of the licensing process, but every EIS must at least briefly describe the intended mitigation. An EMP will present detailed description of the programmes proposed in the EIS and must incorporate any additional programme as determined by the licensing authority. EMPs were not reviewed for this research, as they are not necessarily prepared by the same consultancy, the proponent being free to hire any.

In all reviewed cases, except for the first, mitigation programmes were related to the identified impacts, usually featuring a synoptic correlating chart. Table 5 synthesizes the information on mitigation proposed in each EIS. All EISs include provisions for monitoring, as they are required to under the national regulations. Proposal for offsets appear consistently since 2002 (case 6), in response to new legislation. The table indicates that more mitigation measures are proposed in the studies, suggesting the measures are more focused and less aggregated than those proposed in earlier EIS. This may be a signal that EISs are more directed towards project and impact management and not simply focused on impact identification or prediction (Marshall 2002). An integration with environmental management systems has been recommended since case 6 (2002), whose EIS was prepared for a company whose operations were ISO 14001 certified. Case 7 inaugurates a set a programmes – repeated in the following EISs – towards building capacity in the construction and operation teams, in the sense of transmitting key findings from the assessment phase and explaining the reason for the other proposed programmes, thus aiming to raise awareness with both management and construction and operation crews. This is justified by the fact that, once an EIS is approved and the project implemented, operations management tends to ignore assessment documents (Prado Filho & Souza 2004).

Table 5 Selected features of mitigation proposed in the EISs.

Serial number	Level of complexity	No. of mitigation programmes	Includes offsets?	Includes capacity building?
1	II	n.a.	No	No
2	IV		No	No
3	I	11	No	No
4	I	11	No	No
5	III	16	No	No
6	I	12	Yes	No
7	IV	41	Yes	Yes
8	I	38	Yes	Yes
9	IV	37	Yes	Yes

Case 3 is particularly interesting, because the proposed mitigation was a project modification to (1) protect the aquifer and not disrupt the underground flow; and (2) protect the caves identified in the area intended for limestone mining. Thus, avoidance of impacts, one of the main aims of EIA, was a result of this study. Avoided impacts and project modifications are not clearly accounted for in most EISs in the sample. One exception is case 6, where the proponent had a policy of systematically avoiding any clearing of natural vegetation, unambiguously mentioned in the EIS and demonstrated by maps.

Key findings from document review

The main findings from the review of contents of the nine EISs are that over time:

1.	EISs are longer, feature more maps and improved graphic presentation;
2.	project description tends to be more comprehensive and detailed for both more recent and more complex schemes;
3.	presentation and comparison of alternatives did not significantly evolve over time – EISs describe only the proponent's preferred alternative and seek to justify the choice;
4.	baseline studies tend to feature broader scope (i.e. describe more themes), to be more detailed and to present more primary data;
5.	impact identification evolved to establish clearer cause–effect relationships – impact prediction rarely employs models and assessment of significance is based on a simple scaled–weighted approach;
6.	mitigation and environmental management plans evolved to be more disaggregate and to seek linkages with management systems.

The evolution of EISs as seen by the regulators and the consultants

Seeking to confirm the key findings above and aiming to gauge reasons that could explain the observed changes, perception of both the consultancy's staff and a third party was sought. There are not many persons or institutions who could provide an independent and comprehensive view about the evolution of EIA, except officials within government agencies in charge of reviewing EISs. Although the projects were distributed in several states, the consultancy is headquartered in São Paulo, where it started business and is still active. Hence, it was decided to conduct interviews with the current and two former heads of the mining division within the State Department of Environmental Impact Assessment. Together, they cover the period 1991 to the present.

The respondents at this Department supervised a staff that reviews EISs filed by several consultant firms. They were asked about a possible evolution of these documents in terms of scope and content. Scope was defined as the array of themes or subjects dealt with in the EIS and content as the detail presented in the main chapters of an EIS (namely project description, baseline, impact analysis and mitigation). Eight statements reflecting the expected changes over time were presented and the respondents asked to rank their agreement using a Likert scale. The responses are presented in Table 6, in the lines labelled ‘regulator’.

Table 6 Perception of government staff and consultants about the evolution of EISs – number of responses.

Compared with past practice, more recent EISs prepared for mining projects feature		TA	PA	N	PD	TD
1. More detailed baseline	Regulator	2	1
	Consultant	2	1
2. More topics or themes in the baseline	Regulator	3
	Consultant	2	1
3. Alternative analysis and comparison	Regulator		2			1
	Consultant		2	1
4. More consistent impact analysis methods and approaches	Regulator		2	1
	Consultant	3
5. More detailed mitigation and management programmes	Regulator		2			1
	Consultant	3
6. Broader mitigation and management programmes (dealing with more issues)	Regulator	1	1		1
	Consultant	2	1
7. A demonstration of effectiveness of the proposed mitigation	Regulator		1			2
	Consultant		3
8. Impacts of mine decommissioning and closure	Regulator				1	2
	Consultant		2	1
TA, Totally agree; PA, partially agree; N, sees no change or does not know; PD, partially disagree; TD, totally disagree.

In order to contrast the views of the regulators, three consultancy employees were interviewed and asked the same questions. The most experienced staff member worked in the period 1992–1993 and then continuously from 2000 onwards. The two others had been working for the consultancy respectively since 1998 and 2000. Their responses are featured in Table 6, in the lines labelled ‘consultants’.

Results show ample agreement that baseline features in more recent EISs are more detailed and cover more topics. As the opinions of the regulators also reflect EISs prepared by other consultancies and coincide with results of content analysis, they provide strong evidence for this finding. Another undisputed opinion from the regulators is that the EISs do not consider the impacts of mine closure, although consultants see some advancement over time. For other topics, respondents were split. Thus, it could be useful to explore the reasons advanced by the respondents for their ratings.

Relative to analysis of alternatives, the two government respondents whose experience encompassed the period 1991–2002 justified their partial agreement by stating that the first EISs they reviewed simply failed to mention any alternative, but those filed in the early 2000s discussed issues such as where best to locate a waste rock pile. On the other hand, in the more recent period, no further improvement was perceived by the third respondent. A similar pattern was found in the regulators' ratings for statement 5 (recent EISs feature more detailed mitigation and management programmes), with the respondent whose experience is more recent disagreeing. However, the same respondent sees that mitigation and management are now broader than they were in the past, including actions to engage the community and to provide better communication. This perception in coherent with the results of EISs analysis about evolution of mitigation proposals.

Respondents were also asked to list any other improvements they may have detected. Two major groups were mentioned. The first includes a more extensive use of modelling techniques, especially in hydrogeology (modelling the effects of groundwater pumping in opencast mines), but none of the selected cases used this tool. The second group of improvements is related to the more extensive use of information technologies, especially geographic information systems. They also noticed significant changes in the use of maps and graphics for communicating with users or readers of EISs. Both regulators and consultants agreed about this change.

Possible drivers for change and innovation

The interviews with regulators and consultants included questions about what could explain the observed and perceived changes. Five possible factors were advanced to explain the evolution of contents: (1) more capacity built into consultancies; (2) policies or standards adopted by project proponents; (3) advances in legislation; (4) terms of reference for EISs more carefully designed; and (5) mandatory provision of supplementary information promoting learning within consultancies. Respondents were asked if they agreed that these factors contributed to a possible evolution and were asked to rank them in order of importance. They all agreed that these factors actually concur to a perceived evolution, with the exception of one respondent in the government, who felt unable to judge factor 2.

However, respondents in government largely disagreed about the most important factors. For the respondents who had witnessed the 1990s period, the most important was factor 4. Such perception probably stems from the fact that, until 1994, scoping was not part of the State EIA procedure, that is, the only available guidance was provided by the EIA regulations. On the other hand, for the respondent who had witnessed the more recent period, changes in legislation played a more significant role in influencing the EISs. When this respondent was asked about which laws were more influential, two were highlighted: a Federal law protecting the Atlantic rainforest, which is perceived as leading to more accurate and detailed flora and fauna surveys for projects in this region, and Federal regulations about the protection of caves, which call for a cave inventory to be prepared and the relevance of all caves to be assessed against a set of prescribed criteria. Atlantic rainforest is a biodiversity hotspot (Carnaval et al. 2009) and formerly covered about 80% of the state's area, as well as significant portions of other states where the consultancy is active. The acknowledgement of the environmental and cultural value of caves (Trajano 2000) led to regulations being adopted nationally.

The consultants confirmed this view, emphasized that capacity had grown within the consultancy, dismissed any possible influence coming from standards or codes subscribed to by clients and added that technology contributes positively to the evolution of EISs. Moreover, they pointed that a major criteria used by project proponents to select an environmental consultancy being price, they cannot push too far towards larger or more detailed EISs.

Evolution and innovation

The main drivers of change being legislation and regulation, it appears that the central concern of those preparing EISs is meeting regulatory demands, in accordance with the findings of Morgan et al. (2012) about EIA practice in New Zealand as a ‘pragmatic, procedural approach’ (p. 14) to meet legal requirements and the results obtained by Morrison-Saunders et al. (2001), who studied the perception of practitioners about which pressures on proponents lead to ‘good’ EISs and identified regulators as the most important.

The hypothesis that policies adopted by project proponents could be a driver for evolution or innovation in EIA did not meet with any evidence. In five out of nine cases, clients were subsidiaries of multinational corporations operating in multiple sites, thus potentially in need of demonstrating adherence to internal procedures, but in no case did clients provide the consultant with specific guidance on the EIA, including for the most recent EISs reviewed.

The administrative control of the EISs (Ortolano 1993; Sánchez 1993), that is, their review by a government EIA authority, systematically accompanied by requests for supplementary information in the case of São Paulo and frequent in other states, was also identified as one reason for the consultancy preparing documents with enlarged scope and more detail. Strengthening administrative control is a suggestion that also arises from studies about EIS quality, as clearly stated by Cashmore et al. (2002), while Carrasco et al. (2006), considering the treatment of biodiversity issues in their sample of US EISs, also conclude that ‘real improvements’ (p. 74) will come only as a response to legislation or the administrative process. These perspectives place a much heavier burden on regulators than on consultants as drivers of change and improvement.

Notwithstanding, factors internal to the consultancy may also have played a role in the observed trend. In the period, the firm became bigger and more professionally managed. From a technical staff of seven in 1987 and high reliance on external consultants, the firm grew to a dedicated technical staff of 14 in 2011, using external consultants only for specific items, such as archaeology and for field studies of some zoological groups (Costanzo & Sánchez 2012). That EIA consultants feature a learning curve is suggested by Glasson et al. (1997), who found that in the UK the quality of EISs was related to the experience of consultants, those with a larger portfolio producing more EISs ranked as satisfactory in these authors' quality assessment. They observed a positive relationship between the length of an EIS and its quality, suggesting that the length can be a proxy for proponent commitment and consequent resources allocated to EIA, which supposedly result in better EISs.

Cashmore et al. (2002) also found a statistical positive correlation between length and quality of an EIS. As longer EISs tend to be prepared for more complex or controversial projects, these authors also consider this variable to reflect resources allocated by project proponents to EIA. Large documents can, of course, contain large amounts of irrelevant information or lack focus, a major reason for scoping to be considered as an essential component of EIA good practice.

Beyond length, however, this research found improvements over time in this sample of EISs, converging with conclusions of other diachronic examinations of EISs prepared in other countries for different types of projects (Glasson et al. 1997; Wende 2002; Canelas et al. 2005; Pinho et al. 2007).

Conclusions

Comparisons of EIS contents over a long period detected fluctuations in contents which in turn, allow for context changes to be inferred. Out of the main chapters of such documents, (1) project description tends to be more comprehensive and detailed for both more recent and more complex schemes; (2) baseline studies tend to feature broader scope (i.e. describe more themes) and to be more detailed; and (3) mitigation measures are structured along a consistent framework, as compared with a simple list of loose commitments that appeared in the first EISs.

In terms of form, EISs have become larger, and make more extensive use of maps and graphic editing resources. Geographic information is used both as an analytical tools and for display purposes.

The main drivers of contextual change that may explain the observed evolution were identified as (1) legislation, which became more stringent, (2) regulation, which grew to provide more guidance to EIS preparation, and (3) administrative control exerted by environmental agencies, resulting in requests for supplementary surveys, project modifications or additional mitigation. More capacity built into the consultancy is credited as positively influencing change. No evidence was found concerning a possible influence on the observed changes deriving from policies, standards or codes of conduct subscribed by proponents.

In spite of the observed improvements, baseline studies remain largely descriptive and do not adopt new analytical approaches such as landscape ecology, ecosystems services or community vulnerability. Similarly, consideration of alternatives remained essentially descriptive over the period. As clients were found to have very little or no influence on the evolution of contents and scope of EIS, one can conclude that, if EISs are to evolve further in order to incorporate new analytical approaches and to become less descriptive, the initiative should come mainly from regulators.

Acknowledgements

The authors are grateful to the Brazilian National Council for Scientific and Technological Research for providing a junior research scholarship for the first author and to the São Paulo State Research Support Foundation for providing a research grant for the second author. We also are thankful to the consultancy's founder and owner, for agreeing to participate and facilitating access to files and documents, as well as to senior staff who participated in the interviews and to the three current and former officials with the State Department of Environmental Impact Assessment, for the interviews and for sharing their views.

Notes

^1. These documents receive different names in different countries. For the sake of clarity, they will be referred to as EISs in this text, irrespective of the denomination they may have in the cited literature.

^2. This paper is not shown in Table 1 because it is a meta-study that reviewed ratings attributed to EISs by government officers.