Insight, Part of a Special Feature on Understanding Adaptive Capacity in Forest Governance Comparing Sustainable Forest Management Certifications Standards: a Meta-analysis

To solve problems caused by conventional forest management, forest certification has emerged as a driver of sustainable forest management. Several sustainable forest management certification systems exist, including the Forest Stewardship Council and those endorsed by the Programme for the Endorsement of Forest Certification, such as the Canadian Standards Association – Sustainable Forestry Management Standard CAN/CSA-Z809 and Sustainable Forestry Initiative. For consumers to use certified products to meet their own sustainability goals, they must have an understanding of the effectiveness of different certification systems. To understand the relative performance of three systems, we determined: (1) the criteria used to compare the Forest Stewardship Council, Canadian Standards Association – Sustainable Forestry Management, and Sustainable Forestry Initiative, (2) if consensus exists regarding their ability to achieve sustainability goals, and (3) what research gaps must be filled to improve our understanding of how forest certification systems affect sustainable forest management. We conducted a qualitative meta-analysis of 26 grey literature references (books, industry and nongovernmental organization publications) and 9 primary literature references (articles in peer-reviewed academic journals) that compared at least two of the aforementioned certification systems. The Forest Stewardship Council was the highest performer for ecological health and social sustainable forest management criteria. The Canadian Standards Association – Sustainable Forestry Management and Sustainable Forestry Initiative performed best under sustainable forest management criteria of forest productivity and economic longevity of a firm. Sixty-two percent of analyses were comparisons of the wording of certification system principles or criteria; 34% were surveys of foresters or consumers. An important caveat to these results is that only one comparison was based on empirically collected field data. We recommend that future studies collect ecological and socioeconomic data from forests so purchasers can select certified forest products based on empirical evidence.


INTRODUCTION
Anthropogenic pressures on forests for products are major contributors to reductions in global forest ecological integrity.The consequences of these pressures differ by geographic location and ecosystem (Gullison 2003, FRA 2005), and include the degradation of forest capacity to protect and clean water, air, and soils; and loss of biodiversity, employment, and other social and ecological services (Kneeshaw et al. 2000).When combined with natural factors, such as climate change, the threats to global forest health and integrity are significant (Easterling and Apps 2005).Forest certification for sustainable forest management ([SFM] i.e., forest management that prevents the negative effects of forestry in the long-term while maintaining the benefit to society) emerged in the early 1990s as a remedy to anthropogenic forest degradation (Auld et al. 2008, Vogel 2008).
Certification systems allow consumers to directly influence forest management by purchasing certified products.As demand for certified products increases, so does the pressure on forest companies to become certified to maintain their market share Ecology and Society 16(1): 3 http://www.ecologyandsociety.org/vol16/iss1/art3/(Auld et al. 2008).Citizens assume that certified forest products come from sustainably managed forests, making certification a de facto "quality assurance" mechanism for the sustainability performance of a forest product.Furthermore, consumers may assume that all certification standards are equivalent, which may not be true.A major problem is that consumers cannot determine which labels signify the most sustainably managed forests.
The sizes of future market shares of different certification systems may have the potential to impact the rate at which SFM occurs.For instance, widespread market acceptance of a certification system may cause more firms to become certified under that label than others.If that certifier meets fewer SFM goals than other certification systems, the total benefit of certification to global forest health will be depleted.Therefore, it is important that the forest industry and consumers have accurate information about each system's strengths and weaknesses.
For consumers, there is an increasing amount of choice in certified forest products, covered by two main certification bodies.The Forest Stewardship Council (FSC) was created in 1993 to stop irresponsible tropical forestry (Auld et al. 2008).FSC now certifies 2.7% of all the world's forests (FPAC 2009, Metafore 2009).As well, the Programme for the Endorsement of Forest Certification (PEFC) is an umbrella group that provides international markets to national and regional certification bodies.The combined certification schemes endorsed by PEFC certify 5% of all the world's forests (Metafore 2009).
In this paper, we review the literature that compares different certification systems to determine which most effectively meets SFM goals.We focus on two PEFC certification systems in Canada: the Canadian Standards Association -Sustainable Forest Management Standard CAN/CSA -Z809 (CSA-SFM), a Canadian government derived system, and the Sustainable Forestry Initiative (SFI), an American industry derived system.However, we also included the FSC system in our review.CSA-SFM and SFI certify 26% and 12% of all Canadian forests, respectively; FSC certifies 9% (FPAC 2009).
We also examined the methods used in these comparisons to find knowledge gaps that are impeding the understanding of how effectively certification systems meet SFM goals.Our research questions were as follows: (1) how have certification systems been compared, (2) is there consensus in their effectiveness at achieving sustainability goals, and (3) what gaps in comparative analyses must be filled to determine which forest certification systems improve SFM the most?Understanding the knowledge gaps will speed the empirical research needed to provide the forestry industry and consumers with the ability to select the certifier with the best performance.

Literature review
We searched for comparisons of FSC, CSA-SFM, and SFI within the grey and primary literature.Books, technical reports, and other industry or nongovernmental organization (NGO) publications not published in academic journals were considered grey literature.Primary literature was defined as articles published in peer-reviewed academic journals.
Database searches were the main technique used to find the appropriate literature.Primary literature searches were conducted with specific search terms in ISI Web of Science (Thompson ISI) in July 2008, March 2009, and November 2009.We used several search terms and combinations of search terms to ensure we found all literature pertaining to FSC, CSA-SFM, and SFI.The search terms were:  2008-August 2008, 15 March 2009-1 April 2009, and November 2009.The Google search term used was "forest certification comparison".
Informal discussions with local SFM experts from academia and industry identified credible websites that were sources of comparisons between different certifiers.Those involved in the project identified the experts based on previous relationships.The experts recommended many of the same websites that were found through Google, but they also provided others.Grey literature was considered credible and was included in our analysis if it was recommended by an expert or was referenced in another publication, and if it compared two or more certification systems.Literature from industry and NGO groups was included regardless of the source.It was not our intention to investigate the existence of bias in NGO, industry, or academic papers.We excluded review papers.Pertinent literature that compares certified forests to non-certified forests is referenced in the discussion.

Studies of individual certification systems
Peer-reviewed articles that were excluded from our analysis were filtered according to whether they examined FSC, CSA-SFM, or SFI individually.We did this only with peer-reviewed literature because we assumed that the publication process eliminated bias that might appear in unreviewed documents published by industry or NGOs.

Determining how comparisons have been made
The themes, criteria, and indicators used to assess certification system effectiveness were identified from the literature that met our search criteria, and for 11 single certifier studies that did not meet our search criteria.Themes were defined as overarching categories pertaining to a functional aspect of certification.Criteria were defined as tools used to judge whether principles of SFM have been fulfilled (Prabhu et al. 2001).Indicators were defined as any measure of the performance of a certification system (Prabhu et al. 2001).Indicators were placed within criteria, and criteria were grouped within appropriate themes.
Comparisons were made in two time periods, before and after 2002, which correspond with revisions to both CSA-SFM and SFI.CSA-SFM revised its standards in 2002 (CSA-SFM 2009), and SFI introduced its 2002-2004 standards at that time (SFI 2009).References are displayed according to these time periods to denote changes after revision (Appendix 1).
Comparative methods varied between studies, and as a result, the data we used for indicators were either quantitative or qualitative.Quantitative variables often came from surveys, and provided a variety of numerical and nominal variables.These were often linked to themes, criteria, or indicators.Qualitative variables were descriptions of certification systems found in the discussions of the reviewed literature.They provided descriptions about how a certification system failed or succeeded in promoting a specific SFM principle, and we linked them to themes, criteria, or indicators accordingly.

Meta-analysis to determine consensus of certifier performance
The principles of meta-analysis were used to determine the level of consensus regarding FSC, CSA-SFM, and SFI performance.Meta-analysis combines data sets that have been collected during multiple studies, and searches for statistical trends that are unobservable in single studies (Gurevitch and Hedges 2001).We could not conduct a statistically robust meta-analysis because the response variables used in each study were often inconsistent, and the number of studies using each indicator was highly variable.We relied on graphs alone for our assessment.http://www.ecologyandsociety.org/vol16/iss1/art3/With meta-analysis we sought to answer which certification system best achieves SFM goals.To do this, we converted the qualitative and quantitative data found for each indicator into a binary value (1 or 0), where values of 1 were awarded to successes and 0 was awarded to failures.
One area in which we sought to avoid bias was the Credibility criterion.Within all sectors there are groups and individuals that endorse and use each certification system.Therefore, we made conservative estimates of credibility.We considered a system as credible only if authors reported that over 75% of a sampled population thought the system was credible.
Conversions to binary values differed as a consequence of study methodology.For quantitative studies that used continuous rankings to compare certification systems, we followed the author's metrics for what constituted a pass or fail.For example, in Mater et al. (2002), foresters ranked FSC and SFI along a gradient from 0 to 7 for 54 SFM "elements" (synonymous with indicators as defined here).The rank of 5 was taken as the cutoff for failing to meet an SFM goal.For qualitative studies, when authors discussed the failure of a certification to meet an SFM goal, we assigned a value of 0. Values of 1 were assigned when authors said that a certification standard performed well.
We calculated the average percentage of indicators that met SFM goals under each criterion with these numerical values: (1) The score for each criterion (CI) was the summation of the CI indicators (i) that met SFM goals, divided by the total number of indicators that met and did not meet SFM goals (j) per criterion (x).Percentage values were not calculated for indicators that lacked data for all three certification systems (all indicators can be found in Appendix 1).The average indicator scores for each criterion were plotted in radar plots (Figs.2-4).

Literature review
Nine peer-reviewed journal articles and 26 grey literature sources met our criteria.Of the 368 peerreviewed articles identified, 98% were discarded.
Excluded literature often referenced certification systems superficially or not at all.Many papers reviewed the SFM industry generally, or provided criteria and indicators or monitoring techniques that could be used to evaluate SFM goal achievement under any regulatory environment.
Within the excluded articles, 41 examined a single certifier, mainly FSC (investigated in 71% of references.CSA-SFM was investigated in 8% of references, SFI in 29%).The three most common study types were surveys of the forest industry or consumers (37%), reviews (22%), and evaluations of the wording of FSC principles and criteria (12%).The remaining papers were forest industry case studies that focused on FSC (5 papers) and SFI (1 paper), modeling studies of SFI forests (3 papers), studies of economic impacts of certification (3 papers), and other studies (4 papers).Only 11 of the 41 articles evaluated a certifier with qualitative or quantitative criteria and indicators (Table 1).The criteria or indicators were inconsistent between studies, and none were used to compare systems.

Themes, criteria, and indicators used in comparisons
Three themes were extracted from the comparative literature, and all related to the functioning of certification systems in markets and forestry operations.Product tracking and claims (Metafore 2006), and quality of forest management (ÉEM 2007) were drawn from the literature.System function was based on subjective groupings of variables that describe how forestry companies use certifications.Each theme had at least one criterion, and each criterion had between 4 and 14 indicators (Appendix 1).A total of 77 indicators were extracted from the 35 sources that met our search criteria.
The number of studies examining each indicator was highly variable (Fig. 1), though the methods used to compare systems were consistent.Only one reference based conclusions on data collected in forests (Sverdrup-Thygeson et al. 2008).Twothirds of papers based comparisons on analyses of the wording of certification system principles and criteria (e.g., Hickey et al. 2005).The remaining third evaluated certification systems by surveying perceptions of workers in the forest industry (e.g., Mater et al. 2002) or consumers (Perera et al. 2008).FSC performed poorly under repeatability and consistency, stakeholder participation, and credibility (Fig. 2).CSA-SFM was the intermediate certifier, performing poorly under ecological issues, socioeconomic issues, applicability, credibility, stakeholder participation, and labeling system (Fig. 3).SFI was the most variable, performing poorly under credibility, ecological issues, stakeholder participation, labeling system, public input, repeatability, certification, applicability, and socioeconomic issues (Fig. 4).

Criterion 2.3: Public input
FSC, CSA-SFM, and SFI met 100% ± 0%, 97% ± 3%, and 43% ± 6% of indicators, respectively (Appendix 1, Figs. 2-4).By 2005, all certification systems encouraged public participation.FSC and CSA-SFM have encouraged public participation since their inception, though there is disagreement about the quality of CSA-SFM's process.Hansen et al. (2006) stated that CSA-SFM public input for developing standards is limited because it is based on a selected group of stakeholders, but Abusow (2006) states that public landowner considerations ensures a rigorous public participation process.After its 2003 revisions, SFI improved by addressing training, outreach, and procurement for private land suppliers (Abusow 2006, CEPT 2008a).
Criterion 2.4: Repeatability and consistency FSC, CSA-SFM, and SFI met 54% ± 19% 92% ± 8%, and 61% ± 17% of indicators, respectively (Appendix 1, Figs. 2-4).Clarity of all three standards has improved through time.FSC failed in communicating procedures for standard setting, certification, and accreditation.CSA-SFM began providing clear and rigorous procedures for standard setting, certification, and accreditation after 2002.SFI scored poorly under indicators of annual monitoring and field assessments because http://www.ecologyandsociety.org/vol16/iss1/art3/Criterion 2.5: Adaptability FSC and SFI met 100% ± 0% of indicators, while CSA-SFM met 93% ± 6% (Appendix 1, Figs. 2-4).Very few papers examined the indicators of adaptability, and indicators were not consistent across studies or time periods.The only measure represented across all time periods and certification systems was whether systems required forest managers/owners to be committed to continuous improvement of forest management.
Criterion 2.6: Applicability FSC, CSA-SFM, and SFI met 100% ± 0%, 71% ± 13%, and 67% ± 14% of indicators, respectively (Appendix 1, Figs. 2-4).FSC is an international organization that certifies forests with standardized forest management requirements across multiple countries and all forest types and firm sizes.CSA-SFM and SFI are North American based organizations that certify nationally and regionally.
For CSA-SFM and SFI, forest size needs to be considered because both score low under these indicators.CSA-SFM could be applied to small forest sizes but was not reported as accrediting them (Oliver 2004).The size of forests certified by SFI needs to be explored (Oliver 2004, Hickey et al. 2005).
Theme 3: Quality of forest management Criterion 3.1: Ecological issues FSC, CSA-SFM, and SFI met 90% ± 4%, 59% ± 12%, and 62% ± 10% of indicators, respectively (Appendix 1, Figs. 2-4).All three certifications scored poorly for exotic species indicators because exotic species are allowed to varying degrees (Cashore et al. 2004).FSC performed the best but failed under several indicators.Mater et al. (2002) found that of several ecological indicators, natural regeneration was ranked in the lowest range for SFI and FSC, and soil conservation and protection were also found lacking in FSC forests.Hickey et al. (2005) found that some FSC certified companies use erosion control to protect and conserve soil, but they have limited soil conservation policies otherwise.While FSC is supposed to prohibit use of pesticides and genetically modified organism (GMO) trees, prior to 2002 there were instances where exceptions were made (Ozinga 2001).
CSA-SFM scored poorly under six indicators.Prior to 2002, SFM criteria violations included unfettered clear-cutting, and use of pesticides and GMO trees.Tan (2003) reported that CSA-SFM methods resulted in poor preservation of aquatic habitat.CSA-SFM failed under wildlife habitat preservation because it limits preservation to those habitat areas that are already restricted by government (ÉEM 2007), suggesting a complete failure to protect areas of high conservation values (Tan 2003, CEPT 2008a).SFI's shortcomings included the permissible use of exotic species, acceptance of GMO trees, and allowance of conversion of natural forests to http://www.ecologyandsociety.org/vol16/iss1/art3/plantations (Cashore et al. 2004).SFI has improved in the realms of clear-cutting and logging, soil conservation, and pesticide allowance.There is disagreement in the quality of water resource management in the SFI system.Mater et al. (2002) found that managers felt water resources were maintained comparable to FSC, but Hickey et al. (2005) point out that monitoring of resources does not equate to SFM.Criterion 3.2: Social/economic issues FSC, CSA-SFM, and SFI met 91% ± 3%, 69% ± 19%, and 46% ± 16% of indicators, respectively (Appendix 1, Figs. 2-4).FSC failed under the indicator of "Promoting sustainability with the public and staff" (ÉEM 2008).CSA-SFM was not perceived as protecting indigenous peoples' rights (Ozinga and Krul 2004).Little other information exists for CSA-SFM because it was evaluated under only five of seven indicators.SFI fell short under the categories of workers' rights (Meridian Institute 2001a, Ozinga and Krul 2004), protecting cultural and historical areas (Meridian Institute 2001a), indigenous peoples' rights (Mater et al. 2002, BC Market Outreach Network 2008), local benefits and opportunities (Meridian Institute 2001a, Mater et al. 2002), and promoting sustainability (Mater et al. 2002).

DISCUSSION
The strengths of FSC differed from those of the PEFC endorsed certifiers.FSC performed better regarding ecological and social issues, and showed the greatest amount of consensus among studies (as signified by the standard error in Fig. 2).The strengths of CSA-SFM and SFI were in maintaining forest productivity to ensure the economic longevity of a firm, not in addressing social or ecological issues.Consensus about the performance of these two systems was less than that for FSC (Fig. 2).
Each system received a different amount of study, which made it difficult to make generalizations from this meta-analysis.FSC and SFI were compared in similar numbers of studies, but CSA-SFM was examined the least (Fig. 1).FSC also received the most attention in the literature that examined just one system (Table 1).These conclusions about SFM performance are important, but they have several caveats that are linked to the methods used to compare systems.
The main methods of analysis involved examining the wording of certifier principles and criteria, and analyzing surveys.Wording and survey analyses are both limited as tools for assessing certification systems because the data they generate are based on perceptions, not on empirical evidence of on-theground impacts.

Analyses of wording
Analyses of wording are qualitative, do not generate original data, and use a predetermined set of SFM criteria.Authors outlined elements required to meet SFM criteria and then examined certification system principles and criteria to determine whether any were missing (Oliver 2001, Ozinga 2001, Cashore et al. 2004, Abusow 2006).Analyses of wording are appropriate for a criterion (and its indicators) such as "Chain-of-custody" because success can be determined based on written protocols.Other indicators, such as "Protects genetic and biodiversity", should be measured by field data.It should not be assumed that the wording of principles or criteria translates into tangible impacts.
Bias may be created if the authors favor one point of view, such as industrial or NGO (Kneeshaw et al. 2000).For example, the "Applicability" criterion preferred systems that were international in scope rather than regional.This created a bias towards FSC.It should not be assumed that an international system has greater impacts than a regional system.However, other biases towards FSC from the analyses of wording may be reasonable given its clear standpoint on human rights and ecological issues.The 10 principles of FSC focus specifically on human and indigenous peoples' rights, biodiversity, GMOs, and other issues.CSA-SFM and SFI do not address these issues as explicitly (Meridian Institute 2001a, McDermott et al. 2008).

Surveys
While surveys generate new data, those data are based on perceptions and do not represent measurable variables in forests.Some studies rated certifiers across a set of common criteria relative to one another (e.g., Mater et al. 2002, Newsom et al. 2006).Others collected data through surveys about audit processes (Wilson et al. 2001, Hickey et al. 2005) or about forest product user opinions (Perera et al. 2008).For example, Mater et al. (2002) worked http://www.ecologyandsociety.org/vol16/iss1/art3/with state and university forest managers and owners to rank the FSC and SFI systems after going through the certification feasibility audits of both systems.These methods are appropriate for criteria such as certification or stakeholder participation, which are measured by indicators that are based on user experience.

Advancements in comparative analyses
To understand the actual impact of SFM, social or ecological data must be collected in field studies.Certification systems have integrated new ideas and norms into forestry management and public engagement with forestry, but little is known about their field performance (Tikina and Innes 2008).Only one study drew its conclusion from fieldcollected data, which looked at tree retention in Sweden (Sverdrup-Thygeson et al. 2008).We propose that the conclusions from analyses of wording and surveys presented here can be used to generate hypotheses about how each certification system performs.
Each certification system can be viewed as an unintentional experiment to reach SFM goals.All certification systems are loosely based on the same definition of sustainable development.Sustainable development was originally defined as "economic development that meets the needs of the present generation without compromising the ability of future generations to meet their needs" (WCED 1987).This definition has been adapted for business as the preservation of ecological, social, and other non-business resources while business proceeds (WBCSD 2000).The different focuses of FSC and the PEFC endorsed systems may be due to different interpretations of SFM (Cashore et al. 2004, Vogel 2008).
Analyses of wording could link the interpretation of SFM inherent in a system's principles or criteria to a hypothesis that can be field tested.For example, Suzuki and Olson (2007) hypothesized that the wording of FSC's principles and criteria support the conservation of biodiversity, but they did not test their hypothesis.Future research could build on this by examining the wording of other systems for how they deal with conservation areas.Then the number and impact of conservation areas in FSC, SFI, and CSA-SFM forests could be measured and compared.

Social indicators
Many social indicators have been proposed (Table 1).We provide certification impacts on Aboriginals as an example of how existing frameworks could guide future research.Aboriginal groups are prominent indicators of certifier success in the studies that examined a single certification system (Hickey andInnes 2008, Kant andBrubacher 2008; Table 1).For example, Hickey and Innes (2008) included several indicators of Aboriginal rights and traditions, including economic considerations.Gale and Gale ( 2006) also created a framework (without assessing a certification system) that included Aboriginal rights as a criterion in addition to broader social criteria of employment realities and subsidies paid from public funds.While impacts on Aboriginals were a common indicator of socioeconomic impacts in the comparison literature, they were under-represented.In our analysis of comparison studies we found that indigenous groups were absent from the "Credibility" criterion, where Aboriginal approval of certification could be a meaningful indicator.For example, Kant and Brubacher (2008) surveyed Canadian Aboriginals and found that certification systems did not meet their expectations of environmental values, Aboriginal community inclusion in decision-making, respect of Aboriginal treaty rights, or economic opportunities and development.Future research could be based on existing indicators, and we suggest that existing data sets could be mined for information to compare certification system impacts on Aboriginal communities, such as has already occurred for comparisons of different nations (Ebeling and Yasue 2009) and tree sizes (Cerutti et al. 2008; Table 1).

Ecological indicators
Although "Ecological issues" was one of the most studied criteria, too little is understood about actual impacts on forest ecosystems.For example, of the 18 studies that assessed whether wildlife habitat was protected, none examined field data.Future research must draw on existing models and monitoring data sets, as well as new field data to advance our understanding of ecological impacts.
Several modeling studies show that both FSC and SFI have positive or neutral impacts on ecological variables (Table 1).Models of SFI hydrological http://www.ecologyandsociety.org/vol16/iss1/art3/impacts predict that SFI will reduce sediment loads by as much as 50% compared to forestry practices that do not leave riparian buffers (Azevedo et al. 2005b).Coarse woody debris was predicted to be three times greater in FSC forests than in conventional forests (Ranius et al. 2003), and SFI may have no negative effects on habitat suitability for birds (Azevedo et al. 2005a).Unfortunately, there seems to be insufficient monitoring of riparian zones, wetlands, and site damage to test hypotheses empirically (Hickey et al. 2005).Existing data sets may present opportunities to compare different certification systems.
Data sets of governments and firms certified under different systems are being mined for studies of the impacts of FSC.We found that FSC's impacts were studied using government data sets in Africa, and in North and Central America (Newsom et al. 2006, Cerutti et al. 2008, Ebeling and Yasue 2009).Future study could include other certification systems that are endorsed by the PEFC label.Firm data sets were used in studies that were excluded from our analysis to infer ecological impacts by looking for improvements in management practices that are required through the certification process (Gullison 2003, Newsom et al. 2006).However, data sets have also been used to examine ecological indicators.Using regeneration data collected by the forestry company, Kukkonen et al. (2008) found that neotropical forest regeneration was lower in certified forest compared to uncertified forest due to greater logging disturbance prior to certification.

Several theoretical indicators have been developed
for SFM forests.Bio-indicators include carabid beetles and spiders (Pearce and Venier 2006), rodents (Pearce and Venier 2005), and birds (Venier and Pearce 2004).One of the benefits of bioindicators is that they can be studied across years, which avoids bias created by inter-annual population variability (Pearce and Venier 2005).Other areas of focus for SFM performance include climate change impacts like carbon sequestration (Ogden and Innes 2009), amount of fragmentation (Brown et al. 2001), soil variables (Cline et al. 2006), and genetic diversity (Geburek and Konrad 2008).While we found some of these ecological indicators (e.g., soils) were included in analyses, many others were not, such as climate change or fragmentation.
It is likely that the impacts of certification systems manifested most in the first 17 years since FSC certification began.Only by knowing whether these systems induce positive changes, and the effect sizes of those changes, in variables that are adversely affected by conventional forestry (e.g., biodiversity, employment), can these systems be effectively compared (Vogel 2008, Auld et al. 2008).
The superiority of FSC over SFI and CSA-SFM may not hold up after a more rigorous examination.It remains to be seen whether the wording of these systems is related to improvements in forest and human health and well-being.Only assessment of ecological data will determine whether all certification systems are created equal, regardless of their underpinnings.If, after more than 10 years of certification, no distinct benefits exist between competing systems, then arguments supporting FSC as the premier certifier out of these three certification systems will be undercut.

CONCLUSIONS
Despite their existence for more than a decade, little is known about how well forest certification systems achieve their SFM goals.FSC, CSA-SFM, and SFI have been compared on the basis of the wording of their criteria and indicators or on user surveys.As such, we found a strong consensus that FSC certified forests achieve higher levels of sustainable forest management compared to CSA-SFM or SFI.However, empirical comparisons based on ecological or socioeconomic field data are lacking.Empirical data collection in forests must be used to field test hypotheses about the merits of different certification systems.Only by understanding the effectiveness at which different certification systems meet SFM goals can consumers select the products that most effectively advance sustainable forestry goals.

Fig. 2 .
Fig. 2. Radar plot of the mean (± SE) percent of indicators meeting sustainable forest management goals within each criterion for the Forest Stewardship Council.Means are solid black lines, and dashed lines represent one SE.

Fig. 3 .
Fig. 3. Radar plot of the mean (± SE) percent of indicators meeting sustainable forest management goals within each criterion for the Canadian Standards Association -Sustainable Forestry Management Standard.Means are solid black lines, and dashed lines represent one SE.

Fig. 4 .
Fig. 4. Radar plot of the mean (± SE) percent of indicators meeting sustainable forest management goals within each criterion for the Sustainable Forestry Initiative.Means are solid black lines, and dashed lines represent one SE.

Table 1 .
Criteria and indicators from studies that examined a single sustainable forest management certification system and which were excluded from comparative analysis (see Methods for details).Variables used to measure indicators or criteria are omitted.
(but no comparison)Identify and map areas of importance to Aboriginal people Ensure that Aboriginal people have access to areas of importance to them Protect areas of importance to Aboriginal people (con'd)