Get 20M+ Full-Text Papers For Less Than $1.50/day. Subscribe now for You or Your Team.

Learn More →

A global assessment of the social and conservation outcomes of protected areas

A global assessment of the social and conservation outcomes of protected areas Contributed Paper A global assessment of the social and conservation outcomes of protected areas ∗ ∗∗ J. A. Oldekop, † G. Holmes,‡ W. E. Harris,§ and K. L. Evans International Forestry Resources and Institutions Research Network, School of Natural Resources and Environment, The University of Michigan, Ann Arbor, MI 48109, U.S.A. †School of Biology, Newcastle University, Newcastle-upon-Tyne, NE1 7RU, United Kingdom ‡School of Earth and Environment, University of Leeds, Leeds LS2 9JT, United Kingdom, email g.holmes@leeds.ac.uk §School of Science and Environment, Manchester Metropolitan University, Manchester M1 5GD, United Kingdom ∗∗ Department of Animal and Plant Sciences, The University of Sheffield, Sheffield, S3 7HF United Kingdom Abstract: Protected areas (PAs) are a key strategy for protecting biological resources, but they vary con- siderably in their effectiveness and are frequently reported as having negative impacts on local people. This has contributed to a divisive and unresolved debate concerning the compatibility of environmental and socioeconomic development goals. Elucidating the relationship between positive and negative social impacts and conservation outcomes of PAs is key for the development of more effective and socially just conservation. We conducted a global meta-analysis on 165 PAs using data from 171 published studies. We assessed how PAs affect the well-being of local people, the factors associated with these impacts, and crucially the relationship between PAs’ conservation and socioeconomic outcomes. Protected areas associated with posi- tive socioeconomic outcomes were more likely to report positive conservation outcomes. Positive conservation and socioeconomic outcomes were more likely to occur when PAs adopted comanagement regimes, empowered local people, reduced economic inequalities, and maintained cultural and livelihood benefits. Whereas the strictest regimes of PA management attempted to exclude anthropogenic influences to achieve biological conservation objectives, PAs that explicitly integrated local people as stakeholders tended to be more effective at achieving joint biological conservation and socioeconomic development outcomes. Strict protection may be needed in some circumstances, yet our results demonstrate that conservation and development objectives can be synergistic and highlight management strategies that increase the probability of maximizing both conservation performance and development outcomes of PAs. Keywords: biodiversity, management, new conservation, socioeconomic development, trade-offs Una Evaluacion ´ Global de los Resultados Sociales y de Conservacion ´ de las Areas Protegidas Resumen: Las areas ´ protegidas (APs) son una estrategia clave para la proteccion ´ de los recursos biologicos, ´ pero estas var´ıan considerablemente en su efectividad y son reportadas frecuentemente por tener impactos negativos sobre los habitantes locales. Esto ha contribuido a un debate divisivo y sin resolucion ´ con respecto a la compatibilidad de los objetivos de desarrollo socioeconomico ´ y ecologico. ´ Esclarecer la relacion ´ entre los impactos sociales positivos y negativos y los resultados de conservacion ´ de las APs es esencial para el desarrollo de una conservacion ´ mas ´ efectiva y mas ´ justa socialmente. Realizamos un meta-analisis ´ de 165 APs usando datos de 171 estudios publicados. Evaluamos como ´ las APs afectan al bienestar de los habitantes locales, los factores asociados con estos impactos y significativamente, la relacion ´ entre los resultados socioeconomicos ´ y de conservacion ´ de las APs. Las APs asociadas con resultados socioeconomicos ´ positivos tuvieron una mayor probabilidad de reportar resultados positivos de conservacion. ´ Los resultados positivos, tanto socioeconomicos ´ como de conservacion, ´ tuvieron una mayor probabilidad de ocurrir cuando las APs adoptaron reg´ımenes de co-manejo, les otorgaron poder a los habitantes locales, redujeron la inequidad economica ´ y mantuvieron los Paper submitted December 31, 2014; revised manuscript accepted May 31, 2015. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. Conservation Biology, Volume 30, No. 1, 133–141 2015 The Authors. Conservation Biology published by Wiley Periodicals, Inc. on behalf of Society for Conservation Biology DOI: 10.1111/cobi.12568 134 Social and Conservation Impacts of Protected Areas beneficios culturales y de sustento. Mientras los reg´ımenes mas ´ estrictos de manejo de APs intentaron excluir las influencias antropog´enicas para alcanzar los objetivos de conservacion ´ biologica, ´ las APs que integraron expl´ıcitamente a los habitantes locales como actores tuvieron la tendencia de ser mas ´ efectivos en la obtencion ´ de resultados conjuntos de desarrollo socioeconomico ´ y de conservacion. ´ La proteccion ´ estricta puede ser necesaria en algunas circunstancias, pero nuestros resultados demuestran que los objetivos de desarrollo ydeconservacion ´ pueden ser sin´ergicos. Tambi´en resaltan las estrategias de manejo que incrementan la probabilidad de maximizar tanto al desempeno de la conservacion como a los resultados de desarrollo de las ˜ ´ APs. Palabras Clave: biodiversidad, compensaciones, desarrollo, manejo, nueva conservacion ´ Introduction wider social and economic changes, for example the ef- fects of increased tourism (Holmes & Brockington 2012). Second, how local people experience and respond to An alarming erosion of taxonomic and functional bio- the social impacts of PAs is influenced by socio-political diversity is occurring in half of tropical protected ar- contexts at both local and regional scales (Brockington eas (PAs) (Laurance et al. 2012). The magnitude of & Igoe 2006; Nelson & Agrawal 2008). Third, impacts this decline is directly linked to human mediated habi- of PAs are unevenly distributed. They are felt most in- tat disruption, including land-use change, hunting, and tensely at local rather than national scales and within exploitation of other forest-related resources. These communities along lines of class, gender, ethnicity, and human-induced pressures on PAs and conflict between caste; benefits tend to accrue to the wealthiest and most biodiversity conservation and the needs of local people powerful and costs fall on the weakest and poorest are predicted to increase due to numerous factors, in- (Holmes 2007). cluding market forces and a reduction in distance be- The lack of a global study on the impact of key recur- tween PAs and human population centers (Joppa et al. ring factors affecting PA socioeconomic and conservation 2008; McDonald et al. 2008). Conflicts between local outcomes has resulted in three specific knowledge gaps people and conservation initiatives have generated one (Adams & Hutton 2007; Mascia & Claus 2009): how the of the greatest and longest running debates in conserva- socioeconomic and biodiversity conservation outcomes tion science (Roe 2008). At one end of the spectrum is of PAs are linked to specific social impacts; how social the fences-and-fines approach, which contends that to impacts are influenced by the management and other deliver successful conservation outcomes people must be excluded, even forcibly, from PAs (Brockington & characteristics of PAs; and how social impacts relate to Igoe 2006). Opponents of this approach consider such socioeconomic and biodiversity conservation outcomes, exclusionist protection arrangements ethically troubling given the insufficient performance of many PAs. Address- because they frequently result in PAs having disadvanta- ing these questions is critical for the design of efficient geous social outcomes for local people that ultimately and effective conservation and development interven- result in ineffective long-term conservation outcomes tions that meet both biodiversity conservation targets and socioeconomic needs. (Adams et al. 2004). An increasingly advocated strategy is We conducted a global review and analysis of the sci- that to deliver effective and long-term environmental pro- entific literature that addresses these knowledge gaps re- tection PAs must accommodate the needs of local people garding the principal drivers of the social impacts of PAs so as to secure sustainable livelihoods and enhance their and their consequences for biodiversity conservation. We well-being (Roe 2008). The debate between adherents to used data from 160 terrestrial and marine PAs distributed these two approaches and the importance of considering across six continents and that were representative of all human well-being in conservation remains lively, intense, International Union for Conservation of Nature (IUCN) and unresolved (Soule ´ 2013; Marvier 2014). PA management and governance categories (Supporting A key factor limiting the resolution of this debate is Information). We quantified and determined how PAs’ the insufficient evidence base, which is currently lim- geographical, physical, and management characteristics ited to individual case studies, with few studies specifi- were associated with their social impacts (Table 1). We cally testing causal pathways (e.g., Andam et al. 2010) then evaluated how these geographical, physical, and and lack of a global analysis (Geldmann et al. 2013). management characteristics were associated with over- Available case studies do, however, usefully highlight all PA socioeconomic and biodiversity conservation out- several key issues (Table 1). First, social impacts of PAs comes and assessed whether these two contrasting types take different forms, including economic, livelihood, and of outcomes trade off against each other or are positively cultural impacts, and can result directly from PA poli- cies, such as hunting regulations, or indirectly through associated. Conservation Biology Volume 30, No. 1, 2016 Oldekop et al. 135 Table 1. Protected area (PA) properties and impacts. PA property Definition Justification for inclusion in analysis Protection PA protection categories included in the WDPA Strict protection can increase costs for local people arrangement IUCN categories (I-IV, strictly protected; V-VI & (West et al. 2006). biosphere reserves, sustainable use) Governance entity responsible for PA management (state, Community and co-managed areas may benefit community, co-managed) communities and lead to better conservation outcomes (Berkes 2004). Geographical Africa, Europe, Oceania, North America, Central Regional differences in political contexts and region America, South America, Central Asia, Southeast histories may lead to variations in impacts (Nelson Asia, Southern Asia & Agrawal 2008). Size PA size (km ) included in the WDPA Large PAs may have greater impacts on people (Brockington & Igoe 2006) than small PAs. Biome terrestrial or marine Marine resource governance differs from terrestrial resource governance (Schlager & Ostrom 1992). Social impacts Displacement voluntary or involuntary displacement, including Displacement is an often cited impact but its moves in response to livelihood changes frequency is uncertain (Brockington & Igoe 2006). Monetary increases or decreases in monetary wealth of any PAs may increase or decrease income within section of local communities resulting from the neighboring populations (Andam et al. 2010). existence of a PA Livelihood positive or negative livelihood impacts outside the PAs may restrict non-monetary livelihood activities monetary economy (e.g., subsistence farming, (West & Brockington 2006). hunting, and gathering of natural resources) Cultural impacts on cultural identity or community cohesion, PAs may increase or restrict access to spiritually access to culturally important sites and resources, important sites (Dudley et al. 2009). and aesthetic appreciation of surroundings Compensation acts by PA authorities designed to offset negative Compensation can lessen negative impacts or lead to impacts of PAs positive conservation outcomes (Beazley 2009). Conflict heavy handedness, corruption, or extortion from PA The creation of PAs may lead to direct conflicts staff toward local people and local resistance to between PA authorities and local communities these impacts (Holmes 2013). Empowerment increased control over lives and livelihoods, Empowerment may improve socioeconomic and including control over natural resource conservation outcomes (Karanth 2007). management, or increased land-tenure security Unequal PA impacts differ among sections of neighboring Impacts of protected areas are not felt equally among distribution of communities local people (Holmes 2007). impact Abbreviations: PA, protected area; WDPA, World Database on Protected Areas; IUCN, International Union for Conservation of Nature. the following research areas: environmental sciences and Methods ecology, biodiversity conservation, forestry, sociology, anthropology, government law, ethnic studies, and social Case Study Selection issues. Following Waylen et al. (2010), we also entered We created a database of peer-reviewed articles on the the search terms in Google Scholar and reviewed the social impacts of PAs by conducting systematic searches first 500 results. We purposefully excluded the substan- in ISI Web of Knowledge using the following Boolean tial body of non-peer-reviewed studies on social impacts search terms: topic = (“protected area” OR “reserve” OR of PAs because of potential biases and lack of detailed “national park”) AND topic = (“social impact” OR “cost” statistical analysis within much of this material (Holmes OR “benefit” “eviction” OR “displacement” OR “liveli- & Brockington 2012). Peer-reviewed studies were only hood” OR “compensation” OR “culture” OR “gender” selected if we could clearly identify impacts on local OR “class” OR “caste” OR “indigenous” OR “income” OR communities resulting from a protected area. “community”) AND topic = (“Conservation”) NOT topic Our initial search yielded 1635 studies. Inclusion in our = (“Species”). To minimize bias in our searches, we chose final selection required studies to meet precise criteria keywords that were either neutral (14 of 19 keywords) (Supporting Information). To be included, studies had to or identified general or specific known negative impacts have assessed the impacts of a specific, named PA. The (3 keywords, i.e., displacement, eviction,and cost)and impacts could have resulted directly from PA policies, keywords related to measures of restitution or general such as hunting regulations, or indirectly, such as through positive impacts (2 keywords, i.e., compensation and increased tourism to the PA. Impacts also had to be di- benefit). We refined our search further by focusing on rectly linked by the authors to the presence of a PA and its Conservation Biology Volume 30, No. 1, 2016 136 Social and Conservation Impacts of Protected Areas institutions, rather than to the natural resources cluded that specific biodiversity conservation or socioe- contained within the PA. This meant, for example, that conomic objectives set by PA management authorities ecosystem services provided by a PA were only classified had been met. The conservation objectives related to as a PA benefit if the authors concluded that these a range of ecological attributes from specific species to services would be under serious threat should a PA and components of ecosystems (such as habitat cover or qual- its institutions not be present (e.g., Allendorf 2006). ity). Socioeconomic outcomes referred to any objectives Similarly, negative effects of predation and crop raiding relating to improving or maintaining any aspect of the by wild animals were only considered if the authors social, economic, cultural, or political life of populations provided evidence that the presence of a PA increased residing inside or within 10 km of the PA. Information exposure to this impact (e.g., Ogra 2008). Furthermore, about the IUCN status (i.e., category) of individual PAs because impacts tend to be distributed locally (West & was obtained from the World Database on Protected Ar- Brockington 2006), we included only impacts affecting eas (IUCN & UNEP 2013) or from individual case studies people living within PAs, people living within 10 km when the IUCN status was reported in a study but absent of PA boundaries, or, in the case of mobile peoples, from the database. The great majority of our case study people who customarily used PAs. Finally, we included PAs were single category PAs, but some had multiple cat- only impacts that could be considered as likely having egories. In cases where PAs had more than one category, occurred during the adult lifetime of a local resident we used information presented in the original papers to and excluded all impacts occurring prior to 1950. We identify the management strategy of the particular section excluded national level studies (e.g., Andam et al. 2010) of the PA in which the published study was focused. because they cannot be used to relate the outcomes and The IUCN protected areas in category V have a primary impacts of specific PAs to their characteristics. objective of maintaining conservation values created by Based on these selection criteria, we identified 171 interactions with humans through traditional manage- articles (Supporting Information) that covered 165 in- ment practices, and sub-objectives include providing nat- dividual PAs (inter-rater agreement for the inclusion ural products, conserving agro-biodiversity, and acting of studies Randolph’s free-marginal κ = 0.67) and as models of sustainability. The primary objective of cat- reported information on eight social impacts and on egory VI PAs is conserving ecosystems and traditional socioeconomic and conservation outcomes (Supporting natural resource management systems when conserva- Information). Socioeconomic targets identified in indi- tion and sustainable use are mutually beneficial (Dudley vidual studies were met in 23 PAs, whereas overall neg- et al. 2010). We thus classified IUCN PAs in categories ative socioeconomic outcomes were recorded in six V and VI as sustainable use. We classified PAs in IUCN PAs. These negative outcomes stem from one study categories I-IV, for which objectives are much more fo- (Schmidt-Soltau 2003), which we excluded due to on- cused on biological conservation, as strictly protected. going disputes over its methodology and results (Holmes We also classified biosphere reserves as sustainable use & Brockington 2012). We are not aware of any disputes because although they contain zones of strict protection, regarding the methodology and results of any of the other overall they aim to increase people’s ability to sustainably papers included in our final selection, and this exclusion manage resources while delivering effective nature con- resulted in a final sample of 160 individual PAs. North servation (UNESCO 1996). This classification has been American PAs were represented by three case studies clearly defined (Bridgewater et al. 1996) and was con- and privately protected areas by two case studies. We firmed as appropriate for the seven biosphere reserves thus excluded these cases respectively from regression and five special designation sites. models that included geographical region and manage- Although studies varied greatly in their methods, ment regimes as predictor variables but retained them in from quantitative studies focusing on changes over time all other analyses. We were constrained by data availabil- (Naughton-Treves et al. 2011) to ethnographic accounts ity to use of self-reported outcomes, rather than using (Beazley 2009), most studies focusing on the social im- quantitative data to assess impacts and outcomes. Such pact of PAs used qualitative approaches to analyze spe- self-reported data can, however, be used reliably to as- cific costs and benefits and their distribution among sess the outcomes of protected areas (Hockings 2003; affected populations. We used a subset of studies to Laurance et al. 2012). develop a coding protocol that focused solely on the presence of specific impacts and whether they were dis- tributed unequally among the affected populations (inter- Variable Construction and Coding rater agreement for individual variable coding Cohen’s K = 0.97). We coded PA impacts as being present if We identified potential predictor and response vari- costs or benefits were specifically evaluated and com- ables based on the current state of knowledge about bined data on individual PAs analyzed in more than one the social impacts of PAs (Table 1). To assess potential trade-offs and associations between socioeconomic and paper. Monetary, livelihood, cultural impacts, and the ecological outcomes, we coded whether studies con- unequal distribution of impacts were coded into costs, Conservation Biology Volume 30, No. 1, 2016 Oldekop et al. 137 2 2 benefits, or a mixture of costs and benefits. This approach D and parameter coefficients. Partial D values are used is similar to that used in previous studies in which quan- as a measure of explanatory power in ordinal regressions titative and qualitative socio-ecological data were com- (equivalent to partial r for linear regressions) and were bined (Oldekop et al. 2010; Waylen et al. 2010). Impacts calculated as D² = (null deviance – model deviance) / not mentioned by specific case studies were coded as null deviance (Guisan & Zimmermann 2000). We cross- not reported. This coding system is not based on the validated our bootstrapped results by comparing boot- assumption that these impacts are absent or do not occur; strapped AIC and D values of binomial regressions with rather, we assumed they have not yet been recorded for those generated using the brglm package’s Firth bias a particular location. Including not reported impacts in correction function (Kosimidis 2013), which addresses the analysis allowed us to elucidate specific relationships, issues of near perfect separation in logistic regressions isolate gaps in the literature, and identify further areas of (Heinze & Schemper 2002). research. Results Analyses PA Impacts and Socioeconomic and Conservation Outcomes We performed all statistical analyses in R (R Development Reported positive conservation outcomes were associ- Core Team 2014). Due to the number of potential pre- ated with reported positive socioeconomic outcomes dictor variables and unequal balance in the data, we used (partial D = 0.16). Sustainable-use PAs were more an area-under-the-curve (AUC) corrected random forest (2, 160) likely to report overall positive socioeconomic outcomes analysis in the package party (Hothorn et al. 2014) to than more strictly protected ones (model averaged par- identify key predictor variables that explained substan- tial D = 0.05) (Fig. 1a, Supporting Information), tial variation in overall reported socioeconomic and con- (1, 142) and positive socioeconomic outcomes were more fre- servation outcomes. Random forests is increasingly used quently reported from PAs in which local people experi- to select variables for subsequent use in regression and enced empowerment (model averaged partial D = classification analyses. It robustly handles small data sets (1, 142) 0.11) (Fig. 1b, Supporting Information), positive cultural with a large number of correlated or interacting predictor outcomes (model averaged partial D = 0.11) variables (Geneur et al. 2010). (1, 142) (Fig. 1c, Supporting Information), and relatively fewer We used the relative variable importance values of negative livelihood impacts (model averaged partial D predictor variables from 10,000 trees of our random = 0.07) (Fig. 1d, Supporting Information). Simi- forests analysis to select a subset of predictor variables (1, 142) larly, positive cultural (model averaged partial D = for the construction of multiple regression models that (2, 156) 0.05, Fig. 1e, Supporting Information) and livelihood im- evaluated quantitative predictions for socioeconomic and conservation outcomes (see Supporting Information for pacts (model averaged partial D = 0.06, Fig. 1f, (2, 156) variable importance plots that identify predictor variables Supporting Information) were associated with positive for socioeconomic and conservation outcomes). To en- conservation outcomes. These models also retained em- sure the robustness of our random forest analyses results, powerment (model averaged partial D = 0.01) (2, 156) each analysis was performed in triplicate with random (Fig. 1g, Supporting Information) and PA size (model averaged partial D = 0.02) (Supporting Infor- seeds. Variable importance thresholds were set as the (2, 156) mation), but their explanatory capacity was particularly value above the absolute value of the lowest negative- poor. There was no evidence for regional variation in the scoring variable (Strobl et al. 2009). We followed the reporting of successful conservation or socioeconomic same procedure to identify a set of PA biophysical and outcomes. management characteristics explaining substantial vari- ation in PA social impacts (see Supporting Information for variable importance plots identifying predictors of PA PA Impacts and Management and Physical Characteristics social impacts). We subsequently ran all possible model combina- Governance of PAs was, to some extent, associated tions of predictor variables identified through the ran- with empowerment (model averaged partial D = (1, 148) dom forest analysis using the nnet package’s multinom 0.06), monetary impacts (model averaged partial D (3, 149) function (Ripley & Venables 2014) and controlled for = 0.07), livelihood impacts (model averaged partial quasi-complete separation in parts of our data set by D = 0.06), and the unequal distribution of (2, 134) using bootstrapped samples of our data (n = 100,000) impacts (partial D = 0.09). Co-managed PAs (3, 157) to calculate deviance (D) and Akaike information crite- were associated with more empowerment (Fig. 2a, rion (AIC) values and model parameter coefficients. We Supporting Information), monetary benefits (Fig. 2b, Sup- subsequently selected all models with AIC < 2ofthe porting Information), and less unequal distribution of model with the lowest AIC value and performed model costs (Fig. 2c, Supporting Information) than community- averaging to calculate individual predictor variable partial or state-managed PAs, but they were associated with Conservation Biology Volume 30, No. 1, 2016 138 Social and Conservation Impacts of Protected Areas Figure 1. The proportion of studies reporting positive, negative, or no impact of protected areas on (a-d) socioeconomic and (e-g) conservation outcomes. Socioeconomic outcomes are relative to (a) protection arrangements (SU, sustainable use [IUCN categories V and VI], n = 30; ST, strict protection [IUCN categories I-IV], n = 112), (b) empowerment (ER, empowerment reported, n = 38; EN, empowerment not reported, n = 104), (c) cultural impacts (B, benefits reported, n = 21; CB, costs and benefits reported, n = 7; C, costs reported, n = 14; CN, cultural impacts not reported, n = 114), and (d) livelihood impacts (B, n = 12; CB, n = 34; C, n = 56; LN, livelihood impacts not reported, n = 40). Conservation outcomes are depicted in relation to (e) cultural impacts (B, n = 21; CB, n = 7; C, n = 14; CN, n = 114) and (f) livelihood impacts (B, n = 12; CB, n = 40; C, n = 60; LN, n = 44), and (g) empowerment (ER, n = 42; EN, n = 114). Figure 2. Proportion of studies reporting benefits, reporting costs and benefits, reporting costs, and not reporting outcomes of (a-d) the impact of protected area governance arrangements on (a) empowerment (C, community, n = 11; CM, co-managed, n = 8; S, state, n = 129), (b) monetary impacts (C, n = 11; CM, n = 8; S, n = 130), (c) the unequal distribution of impacts (C, N = 11; CM, N = 8; S, N = 130), and (d) livelihood impacts (C, N = 7; CM, N = 5; S, N = 122) and the impact of region (e-i) on (e) displacement (CAM, Central America, N = 16; SAM, South America, N = 16; EUR, Europe, N = 5; AFR, Africa, N = 57; NCA, north and central Asia, N = 6; SAS, south Asia, N = 33; SEA, Southeast Asia, N = 19; OCE, Oceania, N = 5), (f) empowerment (CAM, N = 15; SAM, N = 16; EUR, N = 5; AFR, N = 54; NCA, N = 4; SAS, N = 33; SEA, N = 16; OCE, N = 5), (g) monetary impacts (CAM, N = 15; SAM, N = 16; EUR, N = 5; AFR, N = 54; NCA, N = 5; SAS, N = 33; SEA, N = 16; OCE, N = 5), (h) livelihood impacts (CAM, N = 13; SAM, N = 14; EUR, N = 5; AFR, N = 48; NCA, N = 4; SAS, N = 30; SEA, N = 16, OCE, N = 4), and (i) conflict (CAM, N = 16; SAM, n = 16; EUR, N= 5; AFR, N = 57; NCA, N = 6; SAS, n = 33; SEA, n = 19; OCE, n = 5). Conservation Biology Volume 30, No. 1, 2016 Oldekop et al. 139 livelihood costs similar to state-managed PAs and more Our results also draw attention to the links between the livelihood costs than community-managed PAs (Fig. 2d, governance of PAs and their outcomes. Comanagement Supporting Information). of PAs by local communities and conservation bodies Geographical region was associated with empower- were typically associated with delivering greater benefits ment (model averaged partial D = 0.18), displace- to local communities than community- or state-managed (1, 148) ment (model averaged partial D = 0.13), monetary PAs. This finding potentially challenges a key justifica- (1, 157) impacts (model averaged partial D = 0.13), liveli- tion for the rise of community-managed protected areas; (3, 149) hood impacts (model averaged partial D = 0.17), that is, they provide more benefits to local people than (3, 134) and conflict (model averaged partial D = 0.08). PAs under other forms of governance (Berkes 2008). (3, 157) People affected by African and southern Asia PAs ex- Local institutional capacity building and project design perienced more displacement (Fig. 2e, Supporting In- are likely to be important for successful and equitable formation) and, together with Central America, less community-based natural resource management projects empowerment (Fig. 2f, Supporting Information) than (Brooks et al. 2012), but communities co-managing PAs other regions, whereas monetary and livelihood impacts alongside other organizations are likely to benefit from varied substantially across regions (Figs. 2g & 2h, Sup- additional institutions that strengthen tenure rights and porting Information). Conversely, African, Southern Asia, participatory decision-making processes while also pro- and Southeast Asian PAs and those in Oceania experi- moting monetary benefits and more equal distribution of enced more conflict than those in other regions (Fig. 2i, these benefits. Supporting Information). Finally, we found that regional context determined Protected area size was retained as an explanatory how PA effects differed among local communities. Al- variable for empowerment (model averaged partial D though national contexts might not always be as im- = 0.03) (Supporting Information), unequal impact portant as community characteristics in determining suc- (1, 148) distribution (model averaged partial D = 0.001, cessful socioeconomic outcomes (Kabra 2009), regional (3, 157) Supporting Information), and livelihood impacts (model differences in the representation and empowerment of averaged partial D = 0.04, Supporting Informa- rural peoples in national politics (Galvin & Haller 2008), (1, 134) tion), but their overall explanatory capacity was poor. differences in economic stability, and the robustness and transparency of national governance can drive di- vergence of PA outcomes in different regions (Nelson & Agrawal 2008). We confirmed that factors affecting positive socioeconomic outcomes of PAs are influenced Discussion by regional characteristics, suggesting that blanket con- We found that PAs in which socioeconomic benefits were servation initiatives are less likely to succeed if they do not reported were also more likely to report positive conser- consider regional socioeconomic and political contexts. vation outcomes, and these socioeconomic benefits were Collectively, our results provide further support for the more likely to arise when PAs were managed to promote involvement of local people as stakeholders in decision- sustainable resource use rather than enforcing stricter making processes, particularly as co-managers of pro- protection of biological resources. Although strict pro- tected areas (such a role may include agreement to more tection may be needed under certain circumstances (e.g., stringent conservation measures). Protected areas are not extreme poaching pressure), our results strongly suggest inevitably linked to negative social impacts for resident that conservation initiatives should consider whether en- or neighboring human populations (Andam et al. 2010), forcing strict protection on the exploitation of natural and the strengthening of resource management and land- resources is essential for protecting biodiversity. Conser- tenure rights can lead to successful livelihood and con- vation targets of PAs were met more often when the servation outcomes (Oldekop et al. 2010; Persha et al. PA empowered local people, improved cultural benefits, 2011; Nolte et al. 2013). and decreased livelihood costs. Furthermore, trade-offs We defined levels of protection and governance between positive conservation and human development arrangements according to IUCN management categories outcomes were not inevitable, rather the two outcomes but could not take into account any variation in precise were often (65% of cases) compatible. Indeed, we found arrangements within a management category or in that conservation outcomes were predicted by socioeco- how effectively management policies are implemented nomic outcomes more reliably than they were predicted (Dudley 2008). These variations may have affected how by the physical and management characteristics of PAs. social costs and benefits were perceived, managed, and reported and created additional noise in our data The later certainly play a role in determining positive set. The paucity of studies reporting conservation and conservation outcomes of PAs, but our results provide socioeconomic outcomes and the focus of studies we evidence that the attention given to them by conserva- tion initiatives should not come at the expense of the considered on specific targets meant we could measure socioeconomic outcomes of PAs. outcomes only as unidimensional variables. Our inclusion Conservation Biology Volume 30, No. 1, 2016 140 Social and Conservation Impacts of Protected Areas of eight social impacts helped capture some of the S4 - S5), tables containing bootstrapped model parame- nuances related to different socioeconomic outcomes. ters, coefficients, and confidence intervals for all regres- Our measure of conservation outcome, however, sion analyses (Appendix S6 - S13), and a list of the articles did not account for multiple outcomes or trade-offs included in our analysis (Appendix S14) are available on- where protection and management arrangements might line. The authors are solely responsible for the content be targeted toward or benefit some focal taxa or and functionality of these materials. Queries (other than ecosystem components but not others. Finally, sample absence of the material) should be directed to the corre- size limitations also meant we could not evaluate the sponding author. effect of different types of ecosystems on socioeconomic or conservation outcomes. Therefore, future studies should aim to elucidate some of the nuances between Literature Cited protection arrangements, subsistence use, ecosystems, Adams WM, Aveling R, Brockington D, Dickson B, Elliott J, Hutton J, regional factors, and multiple conservation outcomes. Roe D, Vira B, Wolmer W. 2004. Biodiversity conservation and the Our study moves debates on the social impact of pro- eradication of poverty. Science 306:1146–1149. tected areas and its relevance for nature conservation Adams WM, Hutton J. 2007. People, parks and poverty: Political ecology forward in three significant ways. First, we have provided and biodiversity conservation. Conservation and Society 5:147–183. Allendorf T. 2006. Residents’ attitudes toward three protected areas in a novel, global analysis showing a positive association southwestern Nepal. Biodiversity and Conservation 16:2087–2102. between the socioeconomic and biodiversity conserva- Andam KS, Ferraro PJ, Sims KRE, Healy A, Holland MB. 2010. Protected tion outcomes of PAs; these two objectives thus need areas reduce poverty in Costa Rica and Thailand. Proceedings of the not be considered as conflicting. Second, we found that National Academy of Sciences USA 107:9996–10001. sustainable-use PAs were more likely to report successful Beazley K. 2009. Interrogating notions of the powerless Oustee. Devel- opment and Change 40:219–248. socioeconomic outcomes than more strictly protected Berkes F. 2004. Rethinking community-based conservation. Conserva- areas. Taken together these two results suggest that sus- tion Biology 18:621–630. tainable use PAs can perform as well for conservation as Berkes F. 2008. Community conserved areas: Policy issues in historic those with stricter management regimes. Finally, we have and contemporary context. Conservation Letters 2:19–24. provided evidence that PA initiatives aiming to deliver Bridgewater P, Phillips A, Green M, Amos B. 1996. Biosphere Reserves and IUCN System of Protected Area Management Categories. ANCA, joint positive socioeconomic and conservation outcomes Canberra. should consider specific regional socioeconomic and po- Brockington D, Igoe J. 2006. Evictions for conservation: A global litical contexts, support co-management arrangements overview. Conservation and Society 4:424–470. that promote empowerment of local people alongside Brooks JS, Waylen KA, Borgerhoff-Mulder M. 2012. How national con- other institutions, reduce inequalities in the distribution text, project design, and local community characteristics influence success in community-based conservation projects. Proceedings of of these benefits, and help maintain cultural and liveli- the National Academy of Sciences USA 109:21265–21270. hood benefits from local PAs. Dudley N, Editor. 2008. Guidelines for applying protected area manage- ment categories. IUCN, Gland. Dudley N, Higgins-Zogib L, Mansourian S. 2009. The links between pro- Acknowledgments tected areas, faiths, and sacred natural sites. Conservation Biology 23:568–577. Dudley N, Parrish JD, Redford KH, Stolton S. 2010. The revised IUCN We thank R. Freckleton and W. Pitchers for statistical protected area management categories: The debate and ways for- advice and the regional and handling editors and two ward. Oryx 44:485–490. anonymous reviewers for helpful comments on an earlier Galvin M, Haller T. 2008. People, protected areas and global change: version of this manuscript. J.A.O. is supported through Participatory conservation in Latin America, Africa, Asia and Europe. NCCR, Bern. an EU FP7 Marie Curie International Outgoing Fellowship Geldmann J, Barnes M, Coad L, Craigie ID, Hockings M, Burgess ND. (FORCONEPAL) and G.H. acknowledges support from 2013. Effectiveness of terrestrial protected areas in reducing habitat the Leverhulme Trust’s Early Career Fellowship scheme. loss and population declines. Biological Conservation 161:230–238. Genuer R, Poggi JM, Tuleau-Malot C. 2010. Variable selection using Supporting Information random forests. Pattern Recognition Letters 31:2225–2236. Guisan A, Zimmermann NE. 2000. Predictive habitat distribution models A global map of protected areas contained in the IUCN in ecology. Ecological Modelling 135:147–186. Heinze G, Schemper MA. 2002. Solution to the problem of separation and UNEP world database on protected areas and how in logistic regression. Statistics in Medicine 21:2409–2419. our sample of studies compares with the information con- Hockings M. 2003. Systems for assessing the effectiveness of manage- tained within the database (Appendix S1), the selection ment in protected areas. BioScience 53:823–832. path for research articles and case studies (Appendix S2), Holmes G. 2007. Protection, politics and protest: Understanding resis- the number of protected areas associated with specific tance to conservation. Conservation and Society 5:184–201. Holmes G. 2013. Exploring the relationship between local support and impacts and socioeconomic and conservation outcomes the success of protected areas. Conservation and Society 11:72–82. within our sample (Appendix S3), representative variable Holmes G, Brockington D. 2012. Protected areas – What people say importance plots for random forest analysis (Appendix about well-being. Pages 160–172 in Roe D, Elliott J, Sandbrook C, Conservation Biology Volume 30, No. 1, 2016 Oldekop et al. 141 Walpole M, editors. Linking biodiversity conservation and poverty Nolte C, Agrawal A, Silvius KM, Soares-Filho BS. 2013. Governance reduction. Wiley, London. regime and location influence avoided deforestation success of pro- Hothorn T, Hornik K, Strobl C, Zeilis A. 2014. Package ‘party’: A tected areas in the Brazilian Amazon. Proceedings of the National laboratory for recursive partitioning. Available from http://cran.r- Academy of Sciences USA 110:4956–4961. project.org/web/packages/party/party.pdf (accessed June Ogra MV. 2008. Human–wildlife conflict and gender in protected area 2014). borderlands: A case study of costs, perceptions, and vulnerabilities IUCN and UNEP. 2013. The World Database on Protected Areas from Uttarakhand (Uttaranchal), India. Geoforum 39:1408–1422. (WDPA), UNEP-WCMC, Cambridge, United Kingdom. Available Oldekop JA, Bebbington AJ, Brockington D, Preziosi RF. 2010. Under- from http://www.protectedplanet.net (accessed between May 2012 standing the lessons and limitations of conservation and develop- and May 2013). ment. Conservation Biology 24:461–469. Joppa LN, Loarie SR, Pimm SL. 2008. On the protection of “pro- R Development Core Team. 2014. R: A Language Environment for Sta- tected areas”. Proceedings of the National Academy of Sciences tistical Computing. R Foundation for Statistical Computing, Vienna. USA 105:6673–6678. Ripley B, Venables W. 2014. Package ‘nnet’: Feed-forward neu- Kabra A. 2009. Conservation-induced displacement: A comparative ral networks and multinomial log-linear models (http://cran.r study of two Indian protected areas. Conservation and Society project.org/web/packages/nnet/nnet.pdf). 7:249–267. Roe D. 2008. The origins and evolution of the conservation-poverty Karanth K. 2007. Making resettlement work: The case of In- debate: A review of key literature, events and policy processes. dia’s Bhadra Wildlife Sanctuary. Biological Conservation 139:315– Oryx 42:491–503. 324. Schlager E, Ostrom E. 1992. Property-rights regimes and natural re- Kosimidis I. 2013. Package ‘brglm’: Bias reduction in binomial- sources: A conceptual analysis. Land Economics 68:249–262. response generalized linear models. Available from http://cran.r- Schmidt–Soltau K. 2003. Conservation-related resettlement in Central project.org/web/packages/brglm/brglm.pdf (accessed June 2014). Africa: Environmental and social risks. Development and Change Laurance WF, et al. 2012. Averting collapse in tropical forest protected 34:525–551. areas. Nature 489:289–294. Soule ´ M. 2013. The “new conservation”. Conservation Biology 27:895– Marvier M. 2014. New conservation is true conservation. Conservation 897. Biology 28:1–3. Strobl C, Malley J, Tutz G. 2009. An introduction to recursive parti- Mascia MB, Claus CA. 2009. A property rights approach to un- tioning: Rationale, application, and characteristics of classification derstanding human displacement from protected areas: The and regression trees, bagging, and random forests. Psychological case of marine protected areas. Conservation Biology 23:16– Methods 14:323–348. 23. UNESCO. 1996. Biosphere reserves: The Seville strategy and the statu- McDonald RI, Kareiva P, Forman RTT. 2008. The implications tory framework of the world network. UNESCO, Paris. of current and future urbanisations for global protected areas Waylen KA, Fischer A, McGowan PJK, Thirgood SJ, Milner Gulland EJ. and biodiversity conservation. Biological Conservation 141:1695– 2010. Effect of local cultural context on the success of community- 1703. based conservation interventions. Conservation Biology 24:1119– Naughton-Treves L, Alix-Garcia J, Chapman CA. 2011. Lessons about 1129. parks and poverty from a decade of forest loss and economic growth West P, Brockington D. 2006. An anthropological perspective on some around Kibale National Park, Uganda. Proceedings of the National unexpected consequences of protected areas. Conservation Biology Academy of Sciences USA 108:13919–13924. 20:609–616. Nelson F, Agrawal A. 2008. Patronage or participation? Community- West P, Igoe J, Brockington D. 2006. Parks and people: The social based natural resource management reform in Sub-Saharan Africa. impact of protected areas. Annual Review of Anthropology 35:251– Development Change 39:557–585. 277. Conservation Biology Volume 30, No. 1, 2016 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Conservation Biology Wiley

A global assessment of the social and conservation outcomes of protected areas

Loading next page...
 
/lp/wiley/a-global-assessment-of-the-social-and-conservation-outcomes-of-3beUfAI02O

References (46)

Publisher
Wiley
Copyright
© 2016, Society for Conservation Biology
ISSN
0888-8892
eISSN
1523-1739
DOI
10.1111/cobi.12568
pmid
26096222
Publisher site
See Article on Publisher Site

Abstract

Contributed Paper A global assessment of the social and conservation outcomes of protected areas ∗ ∗∗ J. A. Oldekop, † G. Holmes,‡ W. E. Harris,§ and K. L. Evans International Forestry Resources and Institutions Research Network, School of Natural Resources and Environment, The University of Michigan, Ann Arbor, MI 48109, U.S.A. †School of Biology, Newcastle University, Newcastle-upon-Tyne, NE1 7RU, United Kingdom ‡School of Earth and Environment, University of Leeds, Leeds LS2 9JT, United Kingdom, email g.holmes@leeds.ac.uk §School of Science and Environment, Manchester Metropolitan University, Manchester M1 5GD, United Kingdom ∗∗ Department of Animal and Plant Sciences, The University of Sheffield, Sheffield, S3 7HF United Kingdom Abstract: Protected areas (PAs) are a key strategy for protecting biological resources, but they vary con- siderably in their effectiveness and are frequently reported as having negative impacts on local people. This has contributed to a divisive and unresolved debate concerning the compatibility of environmental and socioeconomic development goals. Elucidating the relationship between positive and negative social impacts and conservation outcomes of PAs is key for the development of more effective and socially just conservation. We conducted a global meta-analysis on 165 PAs using data from 171 published studies. We assessed how PAs affect the well-being of local people, the factors associated with these impacts, and crucially the relationship between PAs’ conservation and socioeconomic outcomes. Protected areas associated with posi- tive socioeconomic outcomes were more likely to report positive conservation outcomes. Positive conservation and socioeconomic outcomes were more likely to occur when PAs adopted comanagement regimes, empowered local people, reduced economic inequalities, and maintained cultural and livelihood benefits. Whereas the strictest regimes of PA management attempted to exclude anthropogenic influences to achieve biological conservation objectives, PAs that explicitly integrated local people as stakeholders tended to be more effective at achieving joint biological conservation and socioeconomic development outcomes. Strict protection may be needed in some circumstances, yet our results demonstrate that conservation and development objectives can be synergistic and highlight management strategies that increase the probability of maximizing both conservation performance and development outcomes of PAs. Keywords: biodiversity, management, new conservation, socioeconomic development, trade-offs Una Evaluacion ´ Global de los Resultados Sociales y de Conservacion ´ de las Areas Protegidas Resumen: Las areas ´ protegidas (APs) son una estrategia clave para la proteccion ´ de los recursos biologicos, ´ pero estas var´ıan considerablemente en su efectividad y son reportadas frecuentemente por tener impactos negativos sobre los habitantes locales. Esto ha contribuido a un debate divisivo y sin resolucion ´ con respecto a la compatibilidad de los objetivos de desarrollo socioeconomico ´ y ecologico. ´ Esclarecer la relacion ´ entre los impactos sociales positivos y negativos y los resultados de conservacion ´ de las APs es esencial para el desarrollo de una conservacion ´ mas ´ efectiva y mas ´ justa socialmente. Realizamos un meta-analisis ´ de 165 APs usando datos de 171 estudios publicados. Evaluamos como ´ las APs afectan al bienestar de los habitantes locales, los factores asociados con estos impactos y significativamente, la relacion ´ entre los resultados socioeconomicos ´ y de conservacion ´ de las APs. Las APs asociadas con resultados socioeconomicos ´ positivos tuvieron una mayor probabilidad de reportar resultados positivos de conservacion. ´ Los resultados positivos, tanto socioeconomicos ´ como de conservacion, ´ tuvieron una mayor probabilidad de ocurrir cuando las APs adoptaron reg´ımenes de co-manejo, les otorgaron poder a los habitantes locales, redujeron la inequidad economica ´ y mantuvieron los Paper submitted December 31, 2014; revised manuscript accepted May 31, 2015. This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. Conservation Biology, Volume 30, No. 1, 133–141 2015 The Authors. Conservation Biology published by Wiley Periodicals, Inc. on behalf of Society for Conservation Biology DOI: 10.1111/cobi.12568 134 Social and Conservation Impacts of Protected Areas beneficios culturales y de sustento. Mientras los reg´ımenes mas ´ estrictos de manejo de APs intentaron excluir las influencias antropog´enicas para alcanzar los objetivos de conservacion ´ biologica, ´ las APs que integraron expl´ıcitamente a los habitantes locales como actores tuvieron la tendencia de ser mas ´ efectivos en la obtencion ´ de resultados conjuntos de desarrollo socioeconomico ´ y de conservacion. ´ La proteccion ´ estricta puede ser necesaria en algunas circunstancias, pero nuestros resultados demuestran que los objetivos de desarrollo ydeconservacion ´ pueden ser sin´ergicos. Tambi´en resaltan las estrategias de manejo que incrementan la probabilidad de maximizar tanto al desempeno de la conservacion como a los resultados de desarrollo de las ˜ ´ APs. Palabras Clave: biodiversidad, compensaciones, desarrollo, manejo, nueva conservacion ´ Introduction wider social and economic changes, for example the ef- fects of increased tourism (Holmes & Brockington 2012). Second, how local people experience and respond to An alarming erosion of taxonomic and functional bio- the social impacts of PAs is influenced by socio-political diversity is occurring in half of tropical protected ar- contexts at both local and regional scales (Brockington eas (PAs) (Laurance et al. 2012). The magnitude of & Igoe 2006; Nelson & Agrawal 2008). Third, impacts this decline is directly linked to human mediated habi- of PAs are unevenly distributed. They are felt most in- tat disruption, including land-use change, hunting, and tensely at local rather than national scales and within exploitation of other forest-related resources. These communities along lines of class, gender, ethnicity, and human-induced pressures on PAs and conflict between caste; benefits tend to accrue to the wealthiest and most biodiversity conservation and the needs of local people powerful and costs fall on the weakest and poorest are predicted to increase due to numerous factors, in- (Holmes 2007). cluding market forces and a reduction in distance be- The lack of a global study on the impact of key recur- tween PAs and human population centers (Joppa et al. ring factors affecting PA socioeconomic and conservation 2008; McDonald et al. 2008). Conflicts between local outcomes has resulted in three specific knowledge gaps people and conservation initiatives have generated one (Adams & Hutton 2007; Mascia & Claus 2009): how the of the greatest and longest running debates in conserva- socioeconomic and biodiversity conservation outcomes tion science (Roe 2008). At one end of the spectrum is of PAs are linked to specific social impacts; how social the fences-and-fines approach, which contends that to impacts are influenced by the management and other deliver successful conservation outcomes people must be excluded, even forcibly, from PAs (Brockington & characteristics of PAs; and how social impacts relate to Igoe 2006). Opponents of this approach consider such socioeconomic and biodiversity conservation outcomes, exclusionist protection arrangements ethically troubling given the insufficient performance of many PAs. Address- because they frequently result in PAs having disadvanta- ing these questions is critical for the design of efficient geous social outcomes for local people that ultimately and effective conservation and development interven- result in ineffective long-term conservation outcomes tions that meet both biodiversity conservation targets and socioeconomic needs. (Adams et al. 2004). An increasingly advocated strategy is We conducted a global review and analysis of the sci- that to deliver effective and long-term environmental pro- entific literature that addresses these knowledge gaps re- tection PAs must accommodate the needs of local people garding the principal drivers of the social impacts of PAs so as to secure sustainable livelihoods and enhance their and their consequences for biodiversity conservation. We well-being (Roe 2008). The debate between adherents to used data from 160 terrestrial and marine PAs distributed these two approaches and the importance of considering across six continents and that were representative of all human well-being in conservation remains lively, intense, International Union for Conservation of Nature (IUCN) and unresolved (Soule ´ 2013; Marvier 2014). PA management and governance categories (Supporting A key factor limiting the resolution of this debate is Information). We quantified and determined how PAs’ the insufficient evidence base, which is currently lim- geographical, physical, and management characteristics ited to individual case studies, with few studies specifi- were associated with their social impacts (Table 1). We cally testing causal pathways (e.g., Andam et al. 2010) then evaluated how these geographical, physical, and and lack of a global analysis (Geldmann et al. 2013). management characteristics were associated with over- Available case studies do, however, usefully highlight all PA socioeconomic and biodiversity conservation out- several key issues (Table 1). First, social impacts of PAs comes and assessed whether these two contrasting types take different forms, including economic, livelihood, and of outcomes trade off against each other or are positively cultural impacts, and can result directly from PA poli- cies, such as hunting regulations, or indirectly through associated. Conservation Biology Volume 30, No. 1, 2016 Oldekop et al. 135 Table 1. Protected area (PA) properties and impacts. PA property Definition Justification for inclusion in analysis Protection PA protection categories included in the WDPA Strict protection can increase costs for local people arrangement IUCN categories (I-IV, strictly protected; V-VI & (West et al. 2006). biosphere reserves, sustainable use) Governance entity responsible for PA management (state, Community and co-managed areas may benefit community, co-managed) communities and lead to better conservation outcomes (Berkes 2004). Geographical Africa, Europe, Oceania, North America, Central Regional differences in political contexts and region America, South America, Central Asia, Southeast histories may lead to variations in impacts (Nelson Asia, Southern Asia & Agrawal 2008). Size PA size (km ) included in the WDPA Large PAs may have greater impacts on people (Brockington & Igoe 2006) than small PAs. Biome terrestrial or marine Marine resource governance differs from terrestrial resource governance (Schlager & Ostrom 1992). Social impacts Displacement voluntary or involuntary displacement, including Displacement is an often cited impact but its moves in response to livelihood changes frequency is uncertain (Brockington & Igoe 2006). Monetary increases or decreases in monetary wealth of any PAs may increase or decrease income within section of local communities resulting from the neighboring populations (Andam et al. 2010). existence of a PA Livelihood positive or negative livelihood impacts outside the PAs may restrict non-monetary livelihood activities monetary economy (e.g., subsistence farming, (West & Brockington 2006). hunting, and gathering of natural resources) Cultural impacts on cultural identity or community cohesion, PAs may increase or restrict access to spiritually access to culturally important sites and resources, important sites (Dudley et al. 2009). and aesthetic appreciation of surroundings Compensation acts by PA authorities designed to offset negative Compensation can lessen negative impacts or lead to impacts of PAs positive conservation outcomes (Beazley 2009). Conflict heavy handedness, corruption, or extortion from PA The creation of PAs may lead to direct conflicts staff toward local people and local resistance to between PA authorities and local communities these impacts (Holmes 2013). Empowerment increased control over lives and livelihoods, Empowerment may improve socioeconomic and including control over natural resource conservation outcomes (Karanth 2007). management, or increased land-tenure security Unequal PA impacts differ among sections of neighboring Impacts of protected areas are not felt equally among distribution of communities local people (Holmes 2007). impact Abbreviations: PA, protected area; WDPA, World Database on Protected Areas; IUCN, International Union for Conservation of Nature. the following research areas: environmental sciences and Methods ecology, biodiversity conservation, forestry, sociology, anthropology, government law, ethnic studies, and social Case Study Selection issues. Following Waylen et al. (2010), we also entered We created a database of peer-reviewed articles on the the search terms in Google Scholar and reviewed the social impacts of PAs by conducting systematic searches first 500 results. We purposefully excluded the substan- in ISI Web of Knowledge using the following Boolean tial body of non-peer-reviewed studies on social impacts search terms: topic = (“protected area” OR “reserve” OR of PAs because of potential biases and lack of detailed “national park”) AND topic = (“social impact” OR “cost” statistical analysis within much of this material (Holmes OR “benefit” “eviction” OR “displacement” OR “liveli- & Brockington 2012). Peer-reviewed studies were only hood” OR “compensation” OR “culture” OR “gender” selected if we could clearly identify impacts on local OR “class” OR “caste” OR “indigenous” OR “income” OR communities resulting from a protected area. “community”) AND topic = (“Conservation”) NOT topic Our initial search yielded 1635 studies. Inclusion in our = (“Species”). To minimize bias in our searches, we chose final selection required studies to meet precise criteria keywords that were either neutral (14 of 19 keywords) (Supporting Information). To be included, studies had to or identified general or specific known negative impacts have assessed the impacts of a specific, named PA. The (3 keywords, i.e., displacement, eviction,and cost)and impacts could have resulted directly from PA policies, keywords related to measures of restitution or general such as hunting regulations, or indirectly, such as through positive impacts (2 keywords, i.e., compensation and increased tourism to the PA. Impacts also had to be di- benefit). We refined our search further by focusing on rectly linked by the authors to the presence of a PA and its Conservation Biology Volume 30, No. 1, 2016 136 Social and Conservation Impacts of Protected Areas institutions, rather than to the natural resources cluded that specific biodiversity conservation or socioe- contained within the PA. This meant, for example, that conomic objectives set by PA management authorities ecosystem services provided by a PA were only classified had been met. The conservation objectives related to as a PA benefit if the authors concluded that these a range of ecological attributes from specific species to services would be under serious threat should a PA and components of ecosystems (such as habitat cover or qual- its institutions not be present (e.g., Allendorf 2006). ity). Socioeconomic outcomes referred to any objectives Similarly, negative effects of predation and crop raiding relating to improving or maintaining any aspect of the by wild animals were only considered if the authors social, economic, cultural, or political life of populations provided evidence that the presence of a PA increased residing inside or within 10 km of the PA. Information exposure to this impact (e.g., Ogra 2008). Furthermore, about the IUCN status (i.e., category) of individual PAs because impacts tend to be distributed locally (West & was obtained from the World Database on Protected Ar- Brockington 2006), we included only impacts affecting eas (IUCN & UNEP 2013) or from individual case studies people living within PAs, people living within 10 km when the IUCN status was reported in a study but absent of PA boundaries, or, in the case of mobile peoples, from the database. The great majority of our case study people who customarily used PAs. Finally, we included PAs were single category PAs, but some had multiple cat- only impacts that could be considered as likely having egories. In cases where PAs had more than one category, occurred during the adult lifetime of a local resident we used information presented in the original papers to and excluded all impacts occurring prior to 1950. We identify the management strategy of the particular section excluded national level studies (e.g., Andam et al. 2010) of the PA in which the published study was focused. because they cannot be used to relate the outcomes and The IUCN protected areas in category V have a primary impacts of specific PAs to their characteristics. objective of maintaining conservation values created by Based on these selection criteria, we identified 171 interactions with humans through traditional manage- articles (Supporting Information) that covered 165 in- ment practices, and sub-objectives include providing nat- dividual PAs (inter-rater agreement for the inclusion ural products, conserving agro-biodiversity, and acting of studies Randolph’s free-marginal κ = 0.67) and as models of sustainability. The primary objective of cat- reported information on eight social impacts and on egory VI PAs is conserving ecosystems and traditional socioeconomic and conservation outcomes (Supporting natural resource management systems when conserva- Information). Socioeconomic targets identified in indi- tion and sustainable use are mutually beneficial (Dudley vidual studies were met in 23 PAs, whereas overall neg- et al. 2010). We thus classified IUCN PAs in categories ative socioeconomic outcomes were recorded in six V and VI as sustainable use. We classified PAs in IUCN PAs. These negative outcomes stem from one study categories I-IV, for which objectives are much more fo- (Schmidt-Soltau 2003), which we excluded due to on- cused on biological conservation, as strictly protected. going disputes over its methodology and results (Holmes We also classified biosphere reserves as sustainable use & Brockington 2012). We are not aware of any disputes because although they contain zones of strict protection, regarding the methodology and results of any of the other overall they aim to increase people’s ability to sustainably papers included in our final selection, and this exclusion manage resources while delivering effective nature con- resulted in a final sample of 160 individual PAs. North servation (UNESCO 1996). This classification has been American PAs were represented by three case studies clearly defined (Bridgewater et al. 1996) and was con- and privately protected areas by two case studies. We firmed as appropriate for the seven biosphere reserves thus excluded these cases respectively from regression and five special designation sites. models that included geographical region and manage- Although studies varied greatly in their methods, ment regimes as predictor variables but retained them in from quantitative studies focusing on changes over time all other analyses. We were constrained by data availabil- (Naughton-Treves et al. 2011) to ethnographic accounts ity to use of self-reported outcomes, rather than using (Beazley 2009), most studies focusing on the social im- quantitative data to assess impacts and outcomes. Such pact of PAs used qualitative approaches to analyze spe- self-reported data can, however, be used reliably to as- cific costs and benefits and their distribution among sess the outcomes of protected areas (Hockings 2003; affected populations. We used a subset of studies to Laurance et al. 2012). develop a coding protocol that focused solely on the presence of specific impacts and whether they were dis- tributed unequally among the affected populations (inter- Variable Construction and Coding rater agreement for individual variable coding Cohen’s K = 0.97). We coded PA impacts as being present if We identified potential predictor and response vari- costs or benefits were specifically evaluated and com- ables based on the current state of knowledge about bined data on individual PAs analyzed in more than one the social impacts of PAs (Table 1). To assess potential trade-offs and associations between socioeconomic and paper. Monetary, livelihood, cultural impacts, and the ecological outcomes, we coded whether studies con- unequal distribution of impacts were coded into costs, Conservation Biology Volume 30, No. 1, 2016 Oldekop et al. 137 2 2 benefits, or a mixture of costs and benefits. This approach D and parameter coefficients. Partial D values are used is similar to that used in previous studies in which quan- as a measure of explanatory power in ordinal regressions titative and qualitative socio-ecological data were com- (equivalent to partial r for linear regressions) and were bined (Oldekop et al. 2010; Waylen et al. 2010). Impacts calculated as D² = (null deviance – model deviance) / not mentioned by specific case studies were coded as null deviance (Guisan & Zimmermann 2000). We cross- not reported. This coding system is not based on the validated our bootstrapped results by comparing boot- assumption that these impacts are absent or do not occur; strapped AIC and D values of binomial regressions with rather, we assumed they have not yet been recorded for those generated using the brglm package’s Firth bias a particular location. Including not reported impacts in correction function (Kosimidis 2013), which addresses the analysis allowed us to elucidate specific relationships, issues of near perfect separation in logistic regressions isolate gaps in the literature, and identify further areas of (Heinze & Schemper 2002). research. Results Analyses PA Impacts and Socioeconomic and Conservation Outcomes We performed all statistical analyses in R (R Development Reported positive conservation outcomes were associ- Core Team 2014). Due to the number of potential pre- ated with reported positive socioeconomic outcomes dictor variables and unequal balance in the data, we used (partial D = 0.16). Sustainable-use PAs were more an area-under-the-curve (AUC) corrected random forest (2, 160) likely to report overall positive socioeconomic outcomes analysis in the package party (Hothorn et al. 2014) to than more strictly protected ones (model averaged par- identify key predictor variables that explained substan- tial D = 0.05) (Fig. 1a, Supporting Information), tial variation in overall reported socioeconomic and con- (1, 142) and positive socioeconomic outcomes were more fre- servation outcomes. Random forests is increasingly used quently reported from PAs in which local people experi- to select variables for subsequent use in regression and enced empowerment (model averaged partial D = classification analyses. It robustly handles small data sets (1, 142) 0.11) (Fig. 1b, Supporting Information), positive cultural with a large number of correlated or interacting predictor outcomes (model averaged partial D = 0.11) variables (Geneur et al. 2010). (1, 142) (Fig. 1c, Supporting Information), and relatively fewer We used the relative variable importance values of negative livelihood impacts (model averaged partial D predictor variables from 10,000 trees of our random = 0.07) (Fig. 1d, Supporting Information). Simi- forests analysis to select a subset of predictor variables (1, 142) larly, positive cultural (model averaged partial D = for the construction of multiple regression models that (2, 156) 0.05, Fig. 1e, Supporting Information) and livelihood im- evaluated quantitative predictions for socioeconomic and conservation outcomes (see Supporting Information for pacts (model averaged partial D = 0.06, Fig. 1f, (2, 156) variable importance plots that identify predictor variables Supporting Information) were associated with positive for socioeconomic and conservation outcomes). To en- conservation outcomes. These models also retained em- sure the robustness of our random forest analyses results, powerment (model averaged partial D = 0.01) (2, 156) each analysis was performed in triplicate with random (Fig. 1g, Supporting Information) and PA size (model averaged partial D = 0.02) (Supporting Infor- seeds. Variable importance thresholds were set as the (2, 156) mation), but their explanatory capacity was particularly value above the absolute value of the lowest negative- poor. There was no evidence for regional variation in the scoring variable (Strobl et al. 2009). We followed the reporting of successful conservation or socioeconomic same procedure to identify a set of PA biophysical and outcomes. management characteristics explaining substantial vari- ation in PA social impacts (see Supporting Information for variable importance plots identifying predictors of PA PA Impacts and Management and Physical Characteristics social impacts). We subsequently ran all possible model combina- Governance of PAs was, to some extent, associated tions of predictor variables identified through the ran- with empowerment (model averaged partial D = (1, 148) dom forest analysis using the nnet package’s multinom 0.06), monetary impacts (model averaged partial D (3, 149) function (Ripley & Venables 2014) and controlled for = 0.07), livelihood impacts (model averaged partial quasi-complete separation in parts of our data set by D = 0.06), and the unequal distribution of (2, 134) using bootstrapped samples of our data (n = 100,000) impacts (partial D = 0.09). Co-managed PAs (3, 157) to calculate deviance (D) and Akaike information crite- were associated with more empowerment (Fig. 2a, rion (AIC) values and model parameter coefficients. We Supporting Information), monetary benefits (Fig. 2b, Sup- subsequently selected all models with AIC < 2ofthe porting Information), and less unequal distribution of model with the lowest AIC value and performed model costs (Fig. 2c, Supporting Information) than community- averaging to calculate individual predictor variable partial or state-managed PAs, but they were associated with Conservation Biology Volume 30, No. 1, 2016 138 Social and Conservation Impacts of Protected Areas Figure 1. The proportion of studies reporting positive, negative, or no impact of protected areas on (a-d) socioeconomic and (e-g) conservation outcomes. Socioeconomic outcomes are relative to (a) protection arrangements (SU, sustainable use [IUCN categories V and VI], n = 30; ST, strict protection [IUCN categories I-IV], n = 112), (b) empowerment (ER, empowerment reported, n = 38; EN, empowerment not reported, n = 104), (c) cultural impacts (B, benefits reported, n = 21; CB, costs and benefits reported, n = 7; C, costs reported, n = 14; CN, cultural impacts not reported, n = 114), and (d) livelihood impacts (B, n = 12; CB, n = 34; C, n = 56; LN, livelihood impacts not reported, n = 40). Conservation outcomes are depicted in relation to (e) cultural impacts (B, n = 21; CB, n = 7; C, n = 14; CN, n = 114) and (f) livelihood impacts (B, n = 12; CB, n = 40; C, n = 60; LN, n = 44), and (g) empowerment (ER, n = 42; EN, n = 114). Figure 2. Proportion of studies reporting benefits, reporting costs and benefits, reporting costs, and not reporting outcomes of (a-d) the impact of protected area governance arrangements on (a) empowerment (C, community, n = 11; CM, co-managed, n = 8; S, state, n = 129), (b) monetary impacts (C, n = 11; CM, n = 8; S, n = 130), (c) the unequal distribution of impacts (C, N = 11; CM, N = 8; S, N = 130), and (d) livelihood impacts (C, N = 7; CM, N = 5; S, N = 122) and the impact of region (e-i) on (e) displacement (CAM, Central America, N = 16; SAM, South America, N = 16; EUR, Europe, N = 5; AFR, Africa, N = 57; NCA, north and central Asia, N = 6; SAS, south Asia, N = 33; SEA, Southeast Asia, N = 19; OCE, Oceania, N = 5), (f) empowerment (CAM, N = 15; SAM, N = 16; EUR, N = 5; AFR, N = 54; NCA, N = 4; SAS, N = 33; SEA, N = 16; OCE, N = 5), (g) monetary impacts (CAM, N = 15; SAM, N = 16; EUR, N = 5; AFR, N = 54; NCA, N = 5; SAS, N = 33; SEA, N = 16; OCE, N = 5), (h) livelihood impacts (CAM, N = 13; SAM, N = 14; EUR, N = 5; AFR, N = 48; NCA, N = 4; SAS, N = 30; SEA, N = 16, OCE, N = 4), and (i) conflict (CAM, N = 16; SAM, n = 16; EUR, N= 5; AFR, N = 57; NCA, N = 6; SAS, n = 33; SEA, n = 19; OCE, n = 5). Conservation Biology Volume 30, No. 1, 2016 Oldekop et al. 139 livelihood costs similar to state-managed PAs and more Our results also draw attention to the links between the livelihood costs than community-managed PAs (Fig. 2d, governance of PAs and their outcomes. Comanagement Supporting Information). of PAs by local communities and conservation bodies Geographical region was associated with empower- were typically associated with delivering greater benefits ment (model averaged partial D = 0.18), displace- to local communities than community- or state-managed (1, 148) ment (model averaged partial D = 0.13), monetary PAs. This finding potentially challenges a key justifica- (1, 157) impacts (model averaged partial D = 0.13), liveli- tion for the rise of community-managed protected areas; (3, 149) hood impacts (model averaged partial D = 0.17), that is, they provide more benefits to local people than (3, 134) and conflict (model averaged partial D = 0.08). PAs under other forms of governance (Berkes 2008). (3, 157) People affected by African and southern Asia PAs ex- Local institutional capacity building and project design perienced more displacement (Fig. 2e, Supporting In- are likely to be important for successful and equitable formation) and, together with Central America, less community-based natural resource management projects empowerment (Fig. 2f, Supporting Information) than (Brooks et al. 2012), but communities co-managing PAs other regions, whereas monetary and livelihood impacts alongside other organizations are likely to benefit from varied substantially across regions (Figs. 2g & 2h, Sup- additional institutions that strengthen tenure rights and porting Information). Conversely, African, Southern Asia, participatory decision-making processes while also pro- and Southeast Asian PAs and those in Oceania experi- moting monetary benefits and more equal distribution of enced more conflict than those in other regions (Fig. 2i, these benefits. Supporting Information). Finally, we found that regional context determined Protected area size was retained as an explanatory how PA effects differed among local communities. Al- variable for empowerment (model averaged partial D though national contexts might not always be as im- = 0.03) (Supporting Information), unequal impact portant as community characteristics in determining suc- (1, 148) distribution (model averaged partial D = 0.001, cessful socioeconomic outcomes (Kabra 2009), regional (3, 157) Supporting Information), and livelihood impacts (model differences in the representation and empowerment of averaged partial D = 0.04, Supporting Informa- rural peoples in national politics (Galvin & Haller 2008), (1, 134) tion), but their overall explanatory capacity was poor. differences in economic stability, and the robustness and transparency of national governance can drive di- vergence of PA outcomes in different regions (Nelson & Agrawal 2008). We confirmed that factors affecting positive socioeconomic outcomes of PAs are influenced Discussion by regional characteristics, suggesting that blanket con- We found that PAs in which socioeconomic benefits were servation initiatives are less likely to succeed if they do not reported were also more likely to report positive conser- consider regional socioeconomic and political contexts. vation outcomes, and these socioeconomic benefits were Collectively, our results provide further support for the more likely to arise when PAs were managed to promote involvement of local people as stakeholders in decision- sustainable resource use rather than enforcing stricter making processes, particularly as co-managers of pro- protection of biological resources. Although strict pro- tected areas (such a role may include agreement to more tection may be needed under certain circumstances (e.g., stringent conservation measures). Protected areas are not extreme poaching pressure), our results strongly suggest inevitably linked to negative social impacts for resident that conservation initiatives should consider whether en- or neighboring human populations (Andam et al. 2010), forcing strict protection on the exploitation of natural and the strengthening of resource management and land- resources is essential for protecting biodiversity. Conser- tenure rights can lead to successful livelihood and con- vation targets of PAs were met more often when the servation outcomes (Oldekop et al. 2010; Persha et al. PA empowered local people, improved cultural benefits, 2011; Nolte et al. 2013). and decreased livelihood costs. Furthermore, trade-offs We defined levels of protection and governance between positive conservation and human development arrangements according to IUCN management categories outcomes were not inevitable, rather the two outcomes but could not take into account any variation in precise were often (65% of cases) compatible. Indeed, we found arrangements within a management category or in that conservation outcomes were predicted by socioeco- how effectively management policies are implemented nomic outcomes more reliably than they were predicted (Dudley 2008). These variations may have affected how by the physical and management characteristics of PAs. social costs and benefits were perceived, managed, and reported and created additional noise in our data The later certainly play a role in determining positive set. The paucity of studies reporting conservation and conservation outcomes of PAs, but our results provide socioeconomic outcomes and the focus of studies we evidence that the attention given to them by conserva- tion initiatives should not come at the expense of the considered on specific targets meant we could measure socioeconomic outcomes of PAs. outcomes only as unidimensional variables. Our inclusion Conservation Biology Volume 30, No. 1, 2016 140 Social and Conservation Impacts of Protected Areas of eight social impacts helped capture some of the S4 - S5), tables containing bootstrapped model parame- nuances related to different socioeconomic outcomes. ters, coefficients, and confidence intervals for all regres- Our measure of conservation outcome, however, sion analyses (Appendix S6 - S13), and a list of the articles did not account for multiple outcomes or trade-offs included in our analysis (Appendix S14) are available on- where protection and management arrangements might line. The authors are solely responsible for the content be targeted toward or benefit some focal taxa or and functionality of these materials. Queries (other than ecosystem components but not others. Finally, sample absence of the material) should be directed to the corre- size limitations also meant we could not evaluate the sponding author. effect of different types of ecosystems on socioeconomic or conservation outcomes. Therefore, future studies should aim to elucidate some of the nuances between Literature Cited protection arrangements, subsistence use, ecosystems, Adams WM, Aveling R, Brockington D, Dickson B, Elliott J, Hutton J, regional factors, and multiple conservation outcomes. Roe D, Vira B, Wolmer W. 2004. Biodiversity conservation and the Our study moves debates on the social impact of pro- eradication of poverty. Science 306:1146–1149. tected areas and its relevance for nature conservation Adams WM, Hutton J. 2007. People, parks and poverty: Political ecology forward in three significant ways. First, we have provided and biodiversity conservation. Conservation and Society 5:147–183. Allendorf T. 2006. Residents’ attitudes toward three protected areas in a novel, global analysis showing a positive association southwestern Nepal. Biodiversity and Conservation 16:2087–2102. between the socioeconomic and biodiversity conserva- Andam KS, Ferraro PJ, Sims KRE, Healy A, Holland MB. 2010. Protected tion outcomes of PAs; these two objectives thus need areas reduce poverty in Costa Rica and Thailand. Proceedings of the not be considered as conflicting. Second, we found that National Academy of Sciences USA 107:9996–10001. sustainable-use PAs were more likely to report successful Beazley K. 2009. Interrogating notions of the powerless Oustee. Devel- opment and Change 40:219–248. socioeconomic outcomes than more strictly protected Berkes F. 2004. Rethinking community-based conservation. Conserva- areas. Taken together these two results suggest that sus- tion Biology 18:621–630. tainable use PAs can perform as well for conservation as Berkes F. 2008. Community conserved areas: Policy issues in historic those with stricter management regimes. Finally, we have and contemporary context. Conservation Letters 2:19–24. provided evidence that PA initiatives aiming to deliver Bridgewater P, Phillips A, Green M, Amos B. 1996. Biosphere Reserves and IUCN System of Protected Area Management Categories. ANCA, joint positive socioeconomic and conservation outcomes Canberra. should consider specific regional socioeconomic and po- Brockington D, Igoe J. 2006. Evictions for conservation: A global litical contexts, support co-management arrangements overview. Conservation and Society 4:424–470. that promote empowerment of local people alongside Brooks JS, Waylen KA, Borgerhoff-Mulder M. 2012. How national con- other institutions, reduce inequalities in the distribution text, project design, and local community characteristics influence success in community-based conservation projects. Proceedings of of these benefits, and help maintain cultural and liveli- the National Academy of Sciences USA 109:21265–21270. hood benefits from local PAs. Dudley N, Editor. 2008. Guidelines for applying protected area manage- ment categories. IUCN, Gland. Dudley N, Higgins-Zogib L, Mansourian S. 2009. The links between pro- Acknowledgments tected areas, faiths, and sacred natural sites. Conservation Biology 23:568–577. Dudley N, Parrish JD, Redford KH, Stolton S. 2010. The revised IUCN We thank R. Freckleton and W. Pitchers for statistical protected area management categories: The debate and ways for- advice and the regional and handling editors and two ward. Oryx 44:485–490. anonymous reviewers for helpful comments on an earlier Galvin M, Haller T. 2008. People, protected areas and global change: version of this manuscript. J.A.O. is supported through Participatory conservation in Latin America, Africa, Asia and Europe. NCCR, Bern. an EU FP7 Marie Curie International Outgoing Fellowship Geldmann J, Barnes M, Coad L, Craigie ID, Hockings M, Burgess ND. (FORCONEPAL) and G.H. acknowledges support from 2013. Effectiveness of terrestrial protected areas in reducing habitat the Leverhulme Trust’s Early Career Fellowship scheme. loss and population declines. Biological Conservation 161:230–238. Genuer R, Poggi JM, Tuleau-Malot C. 2010. Variable selection using Supporting Information random forests. Pattern Recognition Letters 31:2225–2236. Guisan A, Zimmermann NE. 2000. Predictive habitat distribution models A global map of protected areas contained in the IUCN in ecology. Ecological Modelling 135:147–186. Heinze G, Schemper MA. 2002. Solution to the problem of separation and UNEP world database on protected areas and how in logistic regression. Statistics in Medicine 21:2409–2419. our sample of studies compares with the information con- Hockings M. 2003. Systems for assessing the effectiveness of manage- tained within the database (Appendix S1), the selection ment in protected areas. BioScience 53:823–832. path for research articles and case studies (Appendix S2), Holmes G. 2007. Protection, politics and protest: Understanding resis- the number of protected areas associated with specific tance to conservation. Conservation and Society 5:184–201. Holmes G. 2013. Exploring the relationship between local support and impacts and socioeconomic and conservation outcomes the success of protected areas. Conservation and Society 11:72–82. within our sample (Appendix S3), representative variable Holmes G, Brockington D. 2012. Protected areas – What people say importance plots for random forest analysis (Appendix about well-being. Pages 160–172 in Roe D, Elliott J, Sandbrook C, Conservation Biology Volume 30, No. 1, 2016 Oldekop et al. 141 Walpole M, editors. Linking biodiversity conservation and poverty Nolte C, Agrawal A, Silvius KM, Soares-Filho BS. 2013. Governance reduction. Wiley, London. regime and location influence avoided deforestation success of pro- Hothorn T, Hornik K, Strobl C, Zeilis A. 2014. Package ‘party’: A tected areas in the Brazilian Amazon. Proceedings of the National laboratory for recursive partitioning. Available from http://cran.r- Academy of Sciences USA 110:4956–4961. project.org/web/packages/party/party.pdf (accessed June Ogra MV. 2008. Human–wildlife conflict and gender in protected area 2014). borderlands: A case study of costs, perceptions, and vulnerabilities IUCN and UNEP. 2013. The World Database on Protected Areas from Uttarakhand (Uttaranchal), India. Geoforum 39:1408–1422. (WDPA), UNEP-WCMC, Cambridge, United Kingdom. Available Oldekop JA, Bebbington AJ, Brockington D, Preziosi RF. 2010. Under- from http://www.protectedplanet.net (accessed between May 2012 standing the lessons and limitations of conservation and develop- and May 2013). ment. Conservation Biology 24:461–469. Joppa LN, Loarie SR, Pimm SL. 2008. On the protection of “pro- R Development Core Team. 2014. R: A Language Environment for Sta- tected areas”. Proceedings of the National Academy of Sciences tistical Computing. R Foundation for Statistical Computing, Vienna. USA 105:6673–6678. Ripley B, Venables W. 2014. Package ‘nnet’: Feed-forward neu- Kabra A. 2009. Conservation-induced displacement: A comparative ral networks and multinomial log-linear models (http://cran.r study of two Indian protected areas. Conservation and Society project.org/web/packages/nnet/nnet.pdf). 7:249–267. Roe D. 2008. The origins and evolution of the conservation-poverty Karanth K. 2007. Making resettlement work: The case of In- debate: A review of key literature, events and policy processes. dia’s Bhadra Wildlife Sanctuary. Biological Conservation 139:315– Oryx 42:491–503. 324. Schlager E, Ostrom E. 1992. Property-rights regimes and natural re- Kosimidis I. 2013. Package ‘brglm’: Bias reduction in binomial- sources: A conceptual analysis. Land Economics 68:249–262. response generalized linear models. Available from http://cran.r- Schmidt–Soltau K. 2003. Conservation-related resettlement in Central project.org/web/packages/brglm/brglm.pdf (accessed June 2014). Africa: Environmental and social risks. Development and Change Laurance WF, et al. 2012. Averting collapse in tropical forest protected 34:525–551. areas. Nature 489:289–294. Soule ´ M. 2013. The “new conservation”. Conservation Biology 27:895– Marvier M. 2014. New conservation is true conservation. Conservation 897. Biology 28:1–3. Strobl C, Malley J, Tutz G. 2009. An introduction to recursive parti- Mascia MB, Claus CA. 2009. A property rights approach to un- tioning: Rationale, application, and characteristics of classification derstanding human displacement from protected areas: The and regression trees, bagging, and random forests. Psychological case of marine protected areas. Conservation Biology 23:16– Methods 14:323–348. 23. UNESCO. 1996. Biosphere reserves: The Seville strategy and the statu- McDonald RI, Kareiva P, Forman RTT. 2008. The implications tory framework of the world network. UNESCO, Paris. of current and future urbanisations for global protected areas Waylen KA, Fischer A, McGowan PJK, Thirgood SJ, Milner Gulland EJ. and biodiversity conservation. Biological Conservation 141:1695– 2010. Effect of local cultural context on the success of community- 1703. based conservation interventions. Conservation Biology 24:1119– Naughton-Treves L, Alix-Garcia J, Chapman CA. 2011. Lessons about 1129. parks and poverty from a decade of forest loss and economic growth West P, Brockington D. 2006. An anthropological perspective on some around Kibale National Park, Uganda. Proceedings of the National unexpected consequences of protected areas. Conservation Biology Academy of Sciences USA 108:13919–13924. 20:609–616. Nelson F, Agrawal A. 2008. Patronage or participation? Community- West P, Igoe J, Brockington D. 2006. Parks and people: The social based natural resource management reform in Sub-Saharan Africa. impact of protected areas. Annual Review of Anthropology 35:251– Development Change 39:557–585. 277. Conservation Biology Volume 30, No. 1, 2016

Journal

Conservation BiologyWiley

Published: Feb 1, 2016

Keywords: ; ; ; ; ; ; ; ; ;

There are no references for this article.