Genesis of an indigenous social-ecological landscape in eastern Panama

Knowledge of the interplay between ecological and social influences in the context of land-use decision-making is sparse. To help fill this gap, we conducted participatory land-cover mapping in an indigenous territory of eastern Panama to identify factors that influenced household land-use decisions. The map illustrated a mosaic of land cover dominated by pasture. Primary discourse on influences from 35 semistructured interviews with landowners, women, and youth emphasized economic concerns, such as subsistence, and social-cultural factors, such as reticence to abandon traditional agriculture. Multivariate analysis showed that timing of family settlement helped determine proportion of forest cover, and place of origin helped determine proportion of short fallow for agriculture. Cultural norms and economic opportunities inform gendered perspectives; women perceived internal social-cultural influences and men perceived external and ecological influences on the land. Giving consideration to subsistence, traditional land uses, social organization, and women's perspectives could inform future communal reforestation.

A social-ecological approach to land-use studies enables the consideration of diverse drivers, including historical factors, local perceptions, and gendered perspectives.Recent research recognizes that landscapes are a product of history (Rhemtulla and Mladenoff 2007, Gray et al. 2008, Moran 2010).For example, differential settlement histories between two groups of indigenous Peruvian Asháninka partly explained market integration: the group more involved in the market had arrived later from an area of high colonization (Peralta and Kainer 2008).Venezuelan Barí's settlement patterns have influenced the landscape, with sedentarization and market integration leading to deforestation (Behrens et al. 1994).In the Maya Biosphere Reserve, farmers who had land in their previous settlements were those who practiced more agricultural intensification after migration (Carr 2005).
Land-use change studies must address indigenous resource users' perceptions, which are influenced by their worldviews and inform decision-making (Vanclay 2003, Leonard et al. 2013, Meyfroidt 2013).These include the services forest dwellers perceive to derive from forests (Castillo et al. 2005).In a mangrove deforestation study in Cameroon, for example, lack of flooding and deforestation risk perception meant that mangrove protection would be limited (Munji et al. 2014).Diverse perceptions of causes of deforestation influenced degrees of interest in conservation in Los Tuxtlas, Mexico (Durand and Lazos 2008).
Gender can impact perceived influences on the landscape.Households' livelihood strategies can be determined by gender relations comprising consignment of women's duties to household chores, limited access of females to schooling, and fewer opportunities for women to leave the home or migrate out of the community (Ellis 1998).Distinct livelihoods entail gendered relationships with the environment (Leach et al. 1995); women and men can view the landscape differently depending upon the values they derive from resources, which are determined in part by gendered labor divisions (Etongo and Glover 2012).
Thus, land-use changes result from explicitly linked ecological and social interactions (Redman et al. 2004).To consider diverse influences on land-use decisions and contribute case study data to knowledge on global deforestation, we adopted a socialecological approach to the study of landscape-level land-use change.The objectives of our study were to (1) identify factors that have influenced land-use decisions leading to low forest cover in the indigenous Emberá lands of Piriatí in eastern Panama, and (2) determine the influences of history and gendered perspectives on the landscape following emergence of these factors during data collection.http://www.ecologyandsociety.org/vol20/iss4/art37/In Panama, indigenous peoples have practiced traditional shifting cultivation for thousands of years (McKay 1990, as cited in Fischer andVasseur 2000).The eastern Panamanian landscape was previously old-growth forest or a highly advanced successional stage (Araúz et al. 1973, McKay 1984), but has become mosaics of land uses and secondary forests (Wali 1993, St-Laurent et al. 2013, Vergara-Asenjo et al. 2015).This transition must be understood in the context of historical factors like Emberá migration, Pan-American Highway expansion, hydroelectric dam construction, and small-scale colonist encroachment (Herlihy 1985;Wali 1989Wali , 1993;;Horton 2006;St-Laurent et al. 2013).Emberá men are traditionally responsible for hunting, physical tasks in agriculture, dealing with outsiders, and handling money, while women take care of the household (Herlihy 1986, Kane 1986, Kane 1994).When men are those making land-use decisions, female perspectives can be excluded, with potential outcomes on the resultant landscape.
Our objective was developed jointly with indigenous community leaders, who highlighted concern over deforestation in their lands and their desire to incorporate reforestation in a land-use management plan.Our participatory approach reflects the call to conduct research that is meaningful for locals and informed by local knowledge (Grove and Burch 1997, Scoones 1999, Smith 2003, Evans et al. 2006, Chazdon et al. 2009).

Study site
The Piriatí communal lands (tierras colectivas) comprise 3867 ha in the Alto Bayano watershed, ~100 km east of Panama City along the Pan-American Highway (Fig. 1).Piriatí was formed in the early 1970s following relocation of Bayano River inhabitants after hydroelectric dam construction.The watershed has since witnessed an influx of colonists from western provinces searching for land (Herrera 1986, Wali 1989, 1993, St-Laurent et al. 2013).The village has 497 people, 117 families, and 56 landowners who practice swidden agriculture and cattle ranching.The eastern communal lands, Catrigandí, form an area outside the village where Emberá live with Latino farmers (campesinos).The lands are communal, i.e., belonging to everyone while individual landowners manage each parcel, and were legally recognized by the Panamanian government in 2014 (see Wickstrom 2003, Herrera 2012, Runk 2012 for a discussion of land rights struggles).

Participatory mapping of community lands
Participatory maps communicate facets of the landscape relevant and important to locals (Smith 2003).In 2012, participatory mapping of land cover in Piriatí was begun following the methods of Potvin et al. (2006).In a workshop, 45 landowners discussed appropriate land cover classes, including secondary forest, tall and short fallow, and pasture.They drew their parcels and associated land cover on a base map derived from 2012 RapidEye® satellite images of the region (Vergara-Asenjo et al. 2015).We validated this map in a participatory manner (see Appendix 1).

Household interviews
To quantify the relationship of land cover in the plot to household demographic data and factors perceived to influence land-use decisions, we conducted a participatory wealth ranking (see Appendix 1) and 35 semistructured interviews in 2013.Interviews were held individually with male landowners who had obtained land at the time of settlement (n=9), male land inheritors (n=7), nonlandowning or non-Emberá male immigrants who came after the first waves of settlement (n=4), women (wives of landowners and inheritors, n=8), and youth (n=7).We discussed current and past land-use and management practices, factors influencing land-use/land cover, socioeconomic characteristics, land tenure, and means of subsistence.Interviews were coded in two iterations.Three canonical correspondence analyses compared demographic household characteristics, and interviewees' perceived social and ecological influences to the landscape.Four linear discriminant analyses (LDA) tested which economic, social-cultural, political, and ecological influences mentioned by interviewees (explanatory factors) could discriminate among interviewees' genders, excluding youth (see Appendix 2).We explored gendered differences in perceptions with eight female participants who showed interest in a focus group discussion (see Appendix 1).

Participatory mapping of communal lands
We found a high concentration of pasture in Catrigandí and of fallow land on lands recently allotted to new families north of the community (Fig. 2).Forest was concentrated to the north, furthest Fig. 2. Participatory map of the community land (tierras colectivas) of Piriatí-Emberá.http://www.ecologyandsociety.org/vol20/iss4/art37/from the Highway.Pasture accounted for 38% of the communal lands, short and tall fallow for 48%, and secondary forest for 11% which was labeled as such because participants alleged that forest was selectively logged before resettlement.The proportion of pasture was highly variable across parcels, ranging from 0 to 100%.

Demographic influences
Canonical correspondence analysis (CCA) produces two canonical axes that represent linear combinations of explanatory variables (e.g., demographics).Vectors visually represent the degree to which explanatory variables account for variation in response variables (i.e., land cover).CCA does not test significant differences (see Appendix 2).In the CCA examining the relationship between household demographics and land cover in the plot, cumulative proportion of variance in land cover explained by the two canonical axes was 86.7%; the first axis explained 47.9% (Fig. 3, see Appendix 2).The explanatory variable with the highest loading on the first canonical axis was number of people living at home, with an interset correlation coefficient of 0.5116, followed by location (0.4628), number of people available to help (−0.4517), and wealth ranking (−0.3703).

Fig. 3.
Biplot of canonical correspondence analysis between land cover and demographic characteristics at the household level.The axes are linear combinations of the demographic explanatory variables in italics (e.g., # Elders).The relationship between the explanatory variables and the axis can be estimated by drawing a perpendicular line from the tip of the vector to each axis.The relative importance of the explanatory variables on the canonical axes can be estimated by the length and position of the vectors.The bold arrows position the land cover dependent variables (e.g., % pasture) on the 2D space to help visualize the degree to which the explanatory variables explain the variation in the dependent variables.Origin represents whether the interviewee's family is originally from the Darién province (value of 1; otherwise value of 0).Location represents whether the interviewee lives closer to the Highway (Site 1; value of 0) or further (Site 2; value of 1).# Help represents the number of people available to help work the land.
On the second canonical axis number of people living at home (0.2162) and number of people available to help (−0.3872)had the highest loadings.We found a greater percentage of pasture, used for livestock or rental to others, on plots of interviewees who were wealthier and had more elders living at home.Those plots with a greater percentage of short fallow, associated with agricultural land, had more people living at home and were from the Darién.Those plots with more tall fallow had a greater number of people available to help, and those with more secondary forest tended to be from Site 2.

Perceived influences
Thirty-three interviewees referenced the desire, necessity, or advantages of reforestation within the community or communal lands; 28 explicitly expressed their own desire to reforest.The greatest proportion of interviewees referred to economic (versus social-cultural, political, or ecological) factors that affect their land-use decision-making or are perceived to influence others' decisions (n=33, 94%).Landowners (n=17, 100%) and women (n=8, 100%; Table 1) had the greatest proportion of interviewees who mentioned such factors.Of these, potential income (n=30, 86%), subsistence (n=29, 83%) and availability of resources and labor (n=29, 83%) were most mentioned.Economic influences explained forest cover in the communal lands, partly through encouraging felling and planting of trees such as espavé (Anacardium excelsum), cocobolo (Dalbergia retusa), and teak to sell to external buyers.Historical timber extraction by outsiders and colonist encroachment contributed to lack of forest, and restricted access to seeds has limited reforestation of harvestable wood.Campesinos who rent or buy plots of communal land have felled the forest and/or put pasture in their parcels.Meanwhile, limited land availability means forest reserves are left in parcels as land to cultivate in the future.http://www.ecologyandsociety.org/vol20/iss4/art37/For people there's no other option for sustenance, you know?To get their nourishment [cultivation] is the only form of work they have.-a youth Economic factors helped explain high pasture cover.Traditionally, livelihoods were subsistence based but now indigenous farmers are members of a market economy seeking to buy goods and services.Cattle ranching or renting land to ranchers is considered a less time-consuming, less physically demanding way to earn extra income given limited resources and labor.Others are deterred from pasture conversion due to the monetary investment and limited land availability, which would mean having insufficient land for cultivation or for children.
The only solution that will let me support my children or send them to university is to have a few more animals and in case of hardship I can sell . . .-a landowner Twenty interviewees (57%) referred to the social-cultural factors affecting land uses, of which women comprised the greatest proportion (n=7, 88%; Table 2).Tradition was the factor most mentioned (n=10, 29%).Traditional uses justify reforestation of harvestable tree species, like espavé for dugout canoes or cocobolo for handicrafts.Presence of traditional medicine in forests discourages some landowners from deforesting.Traditional shifting cultivation accounts for forest presence, since it involves replanting seeds of naturally growing trees, fallowing land to enable tree regrowth, leaving trees, and planting trees as borders.Social-cultural influences on pasture levels were also mentioned.Influenced by observing campesinos with pasture, some consider it a lucrative alternative to subsistence agriculture.Weak internal laws (local, traditional rules) and lack of social organization, due to perceived lack of communication and inclusion of people outside the community in communal activities, have encouraged the selling of land to campesinos who have converted land to pasture.Some are reticent to sell or rent out land because it violates internal communal laws.
Even though you don't know how to read, you're working with cows so you have money . . ., [the campesino] told me.-a landowner Fourteen interviewees (40%) indicated political factors affecting land uses (Table 3).Of these, public policies was most mentioned, including land endowment, lack of support of indigenous populations, and governmental timber extraction (n=11, 31%).The 1975 Agreement of Majecito, between community members and the government, is one policy that establishes pasture as a legitimate communal land use.Before the government granted legal land title, some people sold parcels to colonist cattle ranchers, lest the government re-appropriate their lands first.A second CCA used interviewees' perceived economic, socialcultural and political influences on land uses and compared them to land cover in their parcels (Fig. 4).The cumulative proportion of variance in land cover explained by the two canonical axes was 66.2%, with the first axis explaining 35.9%.The explanatory variables with the highest loadings on the first axis were presence of external organizations (0.6638), population increase (0.4108), availability of resources and labor (−0.5034), and lack of social organization (−0.3283).The highest loadings on the second axis were population increase (0.3162), subsistence (0.3118), potential income (−0.2528),and local politicians (−0.2145).We found a greater percentage of tall fallow in the plots of interviewees who referred more to the influence of presence of external organizations on land use.External organizations were nongovernmental organizations that have introduced reforestation, like the Global Brigades and Peace Corps.A greater percentage of secondary forest was found on plots of those who mentioned population increase.More short fallow for agriculture was found on plots of those who referred to availability of resources and labor and proximity to outsiders, and more pasture on that of those who referred to local politicians' influence.In the last decade, a local politician rented approximately eight plots and established cattle pastures managed by indigenous landowners.
Nineteen (54%) interviewees, the greatest proportion of whom were landowners (n=11, 65%), referred to ecological factors affecting land uses (Table 4).Of these, soil characteristics (n=5, 14%) and distance from house to plot (n=5, 14%) were most mentioned.Ecological considerations justified pasture conversion partly because elder landowners claimed that land in Piriatí is less fertile than their place of origin (as cited in Wali 1993).Crop http://www.ecologyandsociety.org/vol20/iss4/art37/disease of yam (ñame, Dioscorea alata), a once important cash crop (Herlihy 1986), and invasion of "canal grass" (paja canalera, Saccharum spontaneous) have encouraged some landowners to forsake cultivation for cattle ranching.Distance to plot was invoked by landowners who require pasture for horses used to access remote plots.Topography was invoked by a landowner who was unable to have cattle on hilly terrain, and restoration of natural beauty/state was invoked as a motivator for future reforestation.
In the CCA that compared ecological influences on land uses mentioned by interviewees to land cover in the parcel (Fig. 5), cumulative proportion of variance in land cover explained by the two canonical axes was second highest of the three CCAs at 83.9%; the first axis explained 55%.The explanatory variables with the highest loading on the first axis were distance to plot (0.3959), soil characteristics (−0.4643), and topography (−0.3541).On the second axis, the highest loadings were crop disease (0.2797), distance to plot (−0.4396), and effects on river (−0.3042).Plots with a greater percentage of pasture tended to belong to interviewees who suggested the influence of crop disease, while plots with a greater percentage of tall fallow tended to belong to those who referred to distance to plot.Those plots with more short fallow tended to belong to those who mentioned topography and natural beauty/state.
The economic LDA showed that mention of off-farm employment was a significant predictor of female interviewees; presence of external organizations and markets were most significant predictors of males (Table 5).Jacknife cross-validated classification showed that 60.7% of women and men were classified correctly.The social-cultural LDA showed that population and social organization were significant predictors of females, and influence of outsiders' (colonists') worldviews was a predictor of males (60.7% correct classification).From the political LDA, property title was a significant predictor of females, and public policies best predicted males (71.4% correct classification).The ecological LDA showed effects on animals as a predictor of females, and natural beauty/state, effects on river, heat/shade and topography were the most significant predictors of males (64.3% correct classification).

Demographics and history
Piriatí's pasture-dominated landscape is a product of a web of social-ecological factors informing individual household decisions, as suggested by the considerable variation in pasture proportion across individual parcels.The positive relationship of wealth and number of elders with a plot's proportion of pasture supports the hypothesis that households further in their life cycle have resources to invest in cattle ranching (Perz 2001, Walker et al. 2002).Households with no elders in Piriatí tended to be younger, formed by children after marrying and establishing their own nuclear families (as occurred traditionally; see Herlihy 1986).The association of number of people at home with proportion of short fallow suggests that households use agriculture as a livelihood strategy shaped by the presence of more dependents.Increased household size has encouraged deforestation in rural Latin America by necessitating conversion to cropland for consumption or sale of surplus (Carr 2005, Mena et al. 2006).In a study of land-use/land cover change in Mexico, population pressure was associated with presence of pasture and agriculture, possibly for household consumption (Ellis et al. 2010).The analysis also illustrates settlement history's legacy on the communal landscape.A group of Emberá first settled in the Bayano from the Darién province following Pan-American Highway expansion into eastern Panama in the 1950s (Araúz et al. 1973, Pastor 1985) to have better access to urban markets and open areas for agriculture (Herlihy 1986, Wali 1993).Some Emberá settled along the Bayano River, where they practiced swidden agriculture.After dam construction, they eventually settled in two communities in the watershed, including Piriatí (Wali 1989(Wali , 1993)).Insecurity over land rights meant that resettled Emberá did not engage in intensive agriculture (Wali 1989).They initially practiced minimal cattle ranching due to lack of experience with the practice (Wali 1993).
Our analysis shows that the original households in Piriatí, whose lands are not concentrated in a specific area, tend to have more customary land use, i.e., more forest and less pasture, than more recent settlers.As Emberá along the Bayano were deciding where to resettle, the Piriatí group leader is said to have sought out Emberá who wanted to create a community based on shared traditional values.These original households tend to have traditional houses and participate more in communal activities.Furthermore, those who originally came from the Darién province tend to have more short fallow than those who were not from that province, which can be explained by their greater engagement in traditional agricultural activities as opposed to cattle ranching.The Darién province, not entirely traversed by the Highway, includes two Emberá-Wounaan indigenous reserves (comarcas).Communities in Darién are therefore more isolated from Latino influences.Thus degree of adherence to traditional worldviews and agriculture based on landowners' history apparently resulted in differential land-use practices.Given the community's objective of reforestation, organizing a more unified community that engages in traditional, communal activities can perhaps reinforce forest conservation.

Tradition and subsistence
Rural livelihood strategies constitute bundles of activities that enable survival and security, including subsistence or commercial agriculture, hunting and gathering, timber harvesting, selling of arts and crafts, day labor, and city jobs (de Sherbinin et al. 2008).Once in the market, indigenous peoples become increasingly reliant on market products (Godoy 2001).Livelihood strategies are then negotiated to incorporate these new needs.Natural resources like timber and nontimber forest products provide an avenue for income generation in rural Latin America, becoming a form of insurance or "livelihood buffer" for wellbeing (de Sherbinin et al. 2008).Livelihood profiles can be highly heterogeneous between and within indigenous villages, however.Among the Honduran Tawahka Sumu, those who generated the most income from forest products were those least financially dependent upon forest goods, given diversified income sources (McSweeney 2002).
Livelihood strategies respond to environmental, socioeconomic, and political circumstances (Nygren 2000).They form as a result of coping behavior, risk management, and market opportunities, negotiated by social institutions like kin networks and gender norms (Ellis 1998).Influence of social-cultural institutions given externally imposed factors has been found in numerous indigenous settings.In southwest China, renegotiation of belief in sacred forests enabled forest restoration, following government policies like sedentarization of swidden cultivation (Liu et al. 2000, as cited in Xu et al. 2009).Increased agroforest tea prices encouraged indigenous Akha farmers in China to use social institutions like ecological knowledge to network with the tea industry (Ahmed et al. 2010).Livelihoods of indigenous Wounaan of eastern Panama, a neighboring group with whom there is some intermarriage (Herlihy 1986), display a similar proclivity for cultural values, which encourage traditional basket weaving but also nontraditional shrimping in tune with norms of independence (Runk et al. 2007).http://www.ecologyandsociety.org/vol20/iss4/art37/Interview results illustrate the importance of cultural worldviews on the landscape.Despite ubiquitous emphasis on economic factors, community members often said conversion to pasture would divert land from traditional subsistence agriculture.
Well, I plan on continuing to work the land [after putting pasture], because it's my everyday sustenance, no? -a landowner There was a steady allusion to the role of tradition even when not explicitly articulated.Often community members explained their decisions by stating, "This is what we do" or "This is what we have always done".Importance of tradition on the landscape was especially evident in the difficulty some landowners had in accounting for forest reserves in their parcels.This struggle to justify a practice elders claim was traditionally carried out to ensure sustainability of the forest and its cultural resources, and to preserve land for future cultivation, suggests tradition's understated role in guiding land-use decisions, albeit at times through convention more than active decision-making.
Ultimately, however, the focus of interviewees on income, subsistence, and resources, and the prevalence of pasture in Piriatí, suggest that social-cultural concerns have become secondary to economic concerns in land-use decision-making and, therefore, genesis of the landscape.The context in which institutions like internal laws were mentioned was largely one of ineffectiveness; social institutions have not been harnessed to ensure livelihoods like in aforementioned indigenous circumstances.Cattle raising or selling of agricultural surplus, both drivers of deforestation in Piriatí, were justified as the optimal livelihood strategy that enables subsistence.In the Bayano, indigenous and colonist communities displayed similar patterns of timber extraction based on proximity to market, despite ethnic differences (Simmons 1997).In a context of limited resources, historical natural resource exploitation (Wali 1993), and exclusionary public policies-like granting of private land title for productive use (Wali 1993) and urban rather than rural economic growth and investment (Runk 2012)-community members perceive little practicality in long-term cultural preservation at the expense of immediate survival.Any future reforestation plans must necessarily value subsistence needs in order to be relevant to villagers' concerns.

Gendered perspectives
Our LDA suggests that Piriatí's women are more aware of internal social-cultural factors that shape land uses than men, who are more aware of external and ecological constraints.Focus group discussion and interviews showed that this gendered divide can be explained by: women's greater social role within the community, women's decreased participation in cultivation due to new technologies and economic opportunities in timber extraction, and women's less frequent interaction with outsiders compared to men due to cultural norms.
Men in Piriatí make land-use decisions, at times informed by women's judgment.Women were often unable to describe in detail land-use and management practices.They help with sowing and harvesting but not land clearing, and sometimes cook for men who are working (as occurred traditionally; see Araúz et al. 1973, Herlihy 1986, Kane 1986).Likewise, in Bolivia there was a subtle gendered subdivision of tasks within particular shared activities (Paulson 2003), and Guatemalan women are not responsible for the most important household decisions (Taylor et al. 2006).This limited involvement of women in cultivation may entail less ecological knowledge and less awareness of ecological influences on land-use decisions.Generally, women are more engaged in community groups, perhaps entailing a keener awareness of social aspects of community life.Women's social role helps explain why, in Mexico, women were more concerned with pollution effects, and men with the threat of deforestation.Specifically, women were those looking after sick children (Arizpe et al. 1996).Men in Piriatí alluded more to the influence of external organizations on forest presence through reforestation, suggesting that they are more aware of the ecological rather than social effects of such organizations.
Gendered labor divisions change over time.Women spend less time in fields than previously, in part due to increased emphasis on timber extraction rather than agriculture, which yields slower financial returns.Timber extraction is a male enterprise, as it requires camping away for weeklong stretches.Women have not become caretakers of farms following men's shift to incomegenerating labor, as has occurred elsewhere (see Razavi 2003), due to a general sense of farming's decreased value.Rather, increased timber extraction has heightened women's roles in the social domain as housekeepers and has thereby influenced women's livelihoods and perceived influences on the land, reflecting the coproduction of land use and livelihood (see McCusker and Carr 2006).
With agriculture it's not the same because you can go yourself, take your kids, leave them in a ranchito there.You're helping.But there [in timber extraction] it's another method.-a woman Economic changes-including new technology, like machine huskers, vendors who come to the community, and stores that sell goods-have decreased the need to work the lands or fish, of which the latter was once an important subsistence activity (Herlihy 1986).Similarly, Emberá women in the Darién spend less time on the land, due to the presence of stores that limit the need for household gardens and sugarcane production, less fishing and gathering associated with diminished resources, and an increasing view of the forest as an unsuitable place for women (Colin 2013).The result is continued consignment of women's roles in the household and less interaction with the land than men.Araúz et al. (1973) argued that the household responsibility of Emberá women gave them power to help with household decisions and an advantage in the home and society.Gendered social roles influence who has power to make decisions, and gendered perspectives of land-use decision makers can shape the subsequent landscape.
Women's lower interaction with outsiders perhaps explains why they referred more to internal social influences, like social organization and population increase, while more men referred to the external influence of outsiders, external markets, and public policies on land-use decisions.Emberá men increasingly engage with the outside world as leaders who represent households and communities (Kane 1986(Kane , 1994)).Meanwhile, women say they are discouraged from leaving the community, as the city is considered dangerous and there is a fear that women will marry nonindigenous men.Therefore, as also suggested in the case of the Chilean Mapuche (Vergara and Barton 2013), because women have stayed in the community, they are more familiar than men http://www.ecologyandsociety.org/vol20/iss4/art37/with communal matters.In Sri Lanka, women's greater involvement in the community compared to men has translated to greater traditional knowledge, which women apply to subsistence activities, thereby contributing to the local system's sustainability (Wickramasinghe 2004).Men's and women's relative degrees of interaction with the external world form part of the context that influences interaction with the land.
Men, informed by ecological and external economic and political concerns, are those with power and knowledge to make land-use decisions.Future reforestation could therefore benefit from inclusion of female perspectives that encompass social-cultural considerations, given that social organization and tradition may discourage pasture conversion.

CONCLUSION
A social-ecological perspective that considers local context avoids generalized prescriptions of deforestation irrelevant on the ground (Ostrom and Cox 2010).The Piriatí landscape is a result of individual household land-use decisions constrained by ecological land characteristics like crop disease, household socioeconomics like wealth, social-cultural communal context like weak internal laws, and broader political and economic circumstances like government policies and resource availability.Decisions are principally informed by subsistence concerns, thus accounting for pasture dominance, but are also influenced by traditional norms and settlement history, which has led to land cover mosaics across the landscape.Gendered perspectives mediate awareness of influences on the landscape.Any future communal reforestation efforts must address these diverse concerns in order to be effective.
Responses to this article can be read online at: http://www.ecologyandsociety.org/issues/responses.php/7897

Fig. 1 .
Fig. 1.Location of the indigenous community of Piriatí-Emberá in eastern Panama.

Fig. 4 .
Fig. 4. Biplot of canonical correspondence analysis between land cover (dependent variables) and social-cultural, economic, and political influences on land use (explanatory variables), as mentioned by interviewees.Influences are in italics, while percent land cover is represented by bold arrows.a = Availability of resources and labor.b = natural resource use by outsiders.c = encroaching frontier settlement.d = off-farm employment opportunities.e = population increase.f = potential income.g = presence of external markets.h = proximity to outsiders.i = influence of outsiders' worldview.j = subsistence.k = lack of social organization.l = presence of external organizations.m = tradition.n = internal laws.o = public policies.p = lack of property title.q = local politicians.
of landowner respondents n=17.‡ Total number of female respondents n=8.§ Total number of youth respondents n=7."Total" is the percentage of respondents in each category that referred to ecological factors as influences.

Fig. 5 .
Fig. 5. Biplot of canonical correspondence analysis between land cover and ecological perceptions of influences on land use, as mentioned by interviewees.Ecological explanatory variables are in italics, while percent land cover dependent variables are represented by bold arrows.Soil characteristics includes the soil fertility and water content of the soil.

Table 1 .
Economic factors that influence land-use decisions, as mentioned by interviewees.

Table 2 .
Social-cultural factors that influence land-use decisions, as mentioned by interviewees.Total number of youth respondents n=7."Total" is the percentage of respondents in each category that referred to social-cultural factors as influences.
‡ Total number of female respondents n=8.§

Table 3 .
Political factors that influence land-use decisions, as mentioned by interviewees.Total number of youth respondents n=7."Total" is the percentage of respondents in each category that referred to political factors as influences.
‡ Total number of female respondents n=8.§

Table 4 .
Ecological factors that influence land-use decisions, as mentioned by interviewees.

Table 5 .
Correlations of influences on land uses with discriminant functions.† Separate linear discriminant analyses were conducted for economic, social-cultural, political, and ecological factors, with female and male interviewees as groups.The most significant predictors are shown. †