pergunta5

(1)

When lessons from population models and local ecological knowledge

coincide – Effects of ﬂower stalk harvesting in the Brazilian savanna

Isabel B. Schmidt

a,b,⇑

_{, Tamara Ticktin}

a

Botany Department and Ecology, Evolution and Conservation Biology Program (EECB), University of Hawai‘i at Ma¯noa, 3190 Maile Way, #101, 96822 Honolulu, HI, USA

b

Ibama – Instituto Brasileiro de Meio Ambiente e Recursos Naturais Renováveis, SCEN Trecho II Ed. Sede do Ibama, 70818-900 Brasília, DF, Brazil

a r t i c l e

i n f o

Article history:

Received 23 September 2011

Received in revised form 14 March 2012 Accepted 18 March 2012

Available online 23 June 2012 Keywords: Policy Management Handicrafts Wet grasslands Fire Cerrado Jalapão Sustainable use

a b s t r a c t

Sustainable harvest of non-timber forest products (NTFPs) can play an important role in biodiversity con-servation and livelihoods. However, harvesting policy intended to promote concon-servation are frequently either ineffective or too complicated. Successful policies should consider ecological impacts, local ecolog-ical knowledge and management practices, but NTFP policies are rarely based on these elements. Syngo-nanthus nitens (Eriocaulaceae, ‘golden-grass’) is one of the most valuable NTFPs from the Brazilian savanna. The handicrafts made from this species’ ﬂower stalks are traditional to the Jalapão region, Tocantins state, but have expanded over a much larger area in recent years. We combined ethnoecolog-ical interviews, seed phenology surveys over a large geographethnoecolog-ical area and harvest experiments in nine sites over 3 years to assess local ecological knowledge and management of golden-grass and its long-term effects on population dynamics. Although handicrafting activities are rapidly expanding, local ecological knowledge associated with harvest or management has not been transferred or created outside of Jalapão. Matrix population models illustrate that harvest according to traditional management practices had no impact on golden-grass population dynamics. Earlier harvest of golden-grass, as practiced by new artisans, leads to population decline due to plant uprooting. Local policies for golden-grass harvest are consistent with traditional management, limit the timing but not the quantity of harvest, and are appro-priate over a wide geographical scale. Golden-grass and other wild harvested species with similar char-acteristics hold high potential to help conserve threatened habitat.

1. Introduction

The extraction of non-timber forest products (NTFPs) has been regarded as an alternative for sustainable development with much enthusiasm as well as criticism over the past three decades ( Bel-cher and Schreckenberg, 2007). For NTFPs to improve local liveli-hoods and promote conservation, it is essential to identify management and marketing practices that minimize ecological im-pacts and increase revenues (Varghese and Ticktin, 2008). Many public policies regarding NTFP harvest, although developed with good intentions, require high technical expertise and/or economic investments, and therefore frequently fail to be implemented effectively by local communities. Inappropriate regulations can generate conﬂicts and instigate unsustainable resource exploita-tion. Successful regulations can only be implemented by consider-ing both ecological impacts, and local ecological knowledge and

management practices (Laird et al., 2010). However, policy for few NTFPs is actually based on both elements.

Local ecological knowledge and the traditional management practices derived from it are developed over time as a result of observation, as well as from trial and error processes. They are passed down by individuals and communities (Berkes, 2008) and can affect NTFP harvest sustainability (Varghese and Ticktin, 2008) and yield (Ticktin and Johns, 2002). Local ecological knowl-edge and traditional management may vary widely among com-munities across a landscape (Gaoue and Ticktin, 2010; Ghimire

et al., 2008; Varghese and Ticktin, 2008). When the commercial

sale of NTFPs expands to areas without a history of use, and local ecological knowledge and management is not transferred, adapted or newly developed, this can pose challenges for sustainable har-vest (Ticktin and Johns, 2002). However, if and how local ecological knowledge spreads is rarely considered. In addition, although many NTFPs are harvested over large geographical scales, most studies on harvesting impacts are geographically restricted ( Tick-tin, 2004).

Matrix population models (Caswell, 2001) are frequently ap-plied to assess NTFP harvest impacts (see review by Schmidt

et al. (2011)). By assessing long-term population growth rates

http://dx.doi.org/10.1016/j.biocon.2012.03.018

⇑ Corresponding author at: Ibama – Instituto Brasileiro de Meio Ambiente e Recursos Naturais Renováveis, SCEN Trecho II Ed. Sede do Ibama, Prevfogo, Bl. C, 70818-900 Brasília, DF, Brazil. Tel.: +55 61 33161818.

E-mail address:[email protected](I.B. Schmidt).

Contents lists available atSciVerse ScienceDirect

Biological Conservation

(2)

(lambda, k), and carrying out prospective (sensitivity and elasticity analysis) and retrospective analyses (life table response experi-ments, LTREs) it is possible to identify the effects of harvest on plant populations. Previous research has illustrated that the har-vest of reproductive structures, including propagules and ﬂowers, usually has a low impact on the population dynamics of perennial species (Emanuel et al., 2005; Lamont et al., 2001; Ticktin, 2004). Conversely, harvest that increases adult mortality can severely de-crease population growth rates of plants (e.g., Soehartono and

Newton, 2001), including perennial herbs (e.g., Ghimire et al.,

2008; Law, 2007). However, most matrix model studies of NTFPs

lack comparisons between populations under different harvesting levels (Schmidt et al., 2011), which limits their use in designing management plans. In addition, very little information is available on the impacts of harvesting tropical herbs, regardless of the plant-part harvested (but see Jimenez-Valdes et al., 2010; Ticktin and Johns, 2002).

The sustainable use of NTFPs has the potential to be an impor-tant conservation strategy in the Brazilian savanna, the Cerrado, a biodiversity hotspot where 80% of the original vegetation has been disturbed or converted to large pastures and monocultures, and less than 3% is inside protected areas (Klink and Machado, 2005). Syngonanthus nitens (Bong.) Ruhland (Eriocaulaceae), known as ‘golden-grass’ capim-dourado), is a highly valuable NTFP that oc-curs throughout the Cerrado but is especially harvested in the Jala-pão region in Tocantins state, where it has represented the main income source for hundreds of families since the late 1990s (Schmidt et al., 2007). This perennial herb’s bright ﬂower stalks are sewed with buriti palm (Mauritia ﬂexuosa) young-leaf strips to make handicrafts, including basketry and jewelry (Sampaio et al., 2008).

Local ecological knowledge from one community in Jalapão that has the longest history of handicrafting (Mumbuca), maintains that golden-grass should only be harvested ‘late’ – after mid-Septem-ber, when flower stalks are bright and dry (‘ripe’) – since this pre-vents plant uprooting during harvest and increases handicraft brightness. This traditional management practice, although not di-rectly based on concern for seed production, coincides with seed production phenology, and prevents negative impacts of harvest-ing on individual survival and reproduction (Schmidt et al., 2007). After a long negotiation process with about a dozen Jalapão communities, the Tocantins environmental agency, Naturatins, established formal harvesting regulations for the Jalapão region in 2004 based on traditional management and data on seed pro-duction phenology from a 1-year study (Schmidt et al., 2008). The regulations state that (i) harvest can only be performed after September 20th, when flower stalks are dry and seeds have been produced; (ii) flowerheads must be cut and dispersed in gathering areas just after harvesting; (iii) only harvesters who register with Naturatins are allowed to harvest.

However, since the 2000s, golden-grass handicrafting has ex-panded to new regions of the Cerrado, and therefore into commu-nities with no previous experience harvesting this species. Due to the increased harvesting pressure, the government harvest regula-tions for the Jalapão region were extended to the whole state in 2007 (Portaria-Naturatins 362/2007). This expansion of the regula-tions was not based on ecological research or local knowledge, and upset some harvesters, who maintain that golden-grass ﬂowering periods vary across the state, so that one state-wide rule is not appropriate.

We used a combination of ethnoecological interviews, matrix population models and regional surveys of golden-grass seed pro-duction to assess the potential for sustainable harvest of golden-grass ﬂower stalks. Speciﬁcally we addressed the following ques-tions: (1) Have the local ecological knowledge and management practices from Jalapão been passed onto new harvesters and/or

has new local ecological knowledge developed? (2) What are the effects of Jalapão traditional management (‘late’ harvest) on gold-en-grass population dynamics? (3) What are the potential effects of earlier harvesting of ﬂower stalks on golden-grass population dynamics?

2. Methods

2.1. Study region – Tocantins state and the Jalapão region in the Brazilian savanna context

The Cerrado is the second largest biome in Brazil and originally occupied more than two million km2 (Furley, 2004; Oliveira and Marquis, 2002). Tocantins state has large conserved areas both in-side and outin-side of protected areas and plays an important role in conservation of the Cerrado. Within this state, the Jalapão region is one of the best conserved areas in the Cerrado (Silva and Bates, 2002). Most of golden-grass handicrafting takes place in the core of Jalapão region in the municipalities of Mateiros and São Félix. Ja-lapão human population density is low (<1 inhabitant/km2_{), the}

lo-cal economy is based on subsistence agriculture, extensive cattle raising (Seplam, 2003), and recently, tourism and golden-grass handicrafts. The Cerrado rainfall is highly seasonal; mean rainfall in Jalapão is 1700 mm, 90% of which falls between October and April, and mean annual temperature is 27 °C (Seplam, 2003). 2.2. Study species

Golden-grass is a clonal, polycarpic, perennial rosette-forming herb that grows to about 4 cm in diameter (Giulietti et al., 1996). Flowering occurs once a year from July to August, with each plant producing 1–10 flower stalks, with a capitulum flower (or flower-head). Flowerheads bear 30–60 wind-dispersed seeds, each about 0.9 mm long. Seed germination rates in laboratory conditions are high (>85%,Schmidt et al., 2008); seeds retain viability for less than 1 year in the laboratory and field conditions (Schmidt, 2011).

Golden-grass occurs in wet grasslands, on organosoils with high plant diversity and endemism (Munhoz and Felfili, 2006). In Jala-pão, local harvesters set fire to wet grasslands during the dry sea-son to promote golden-grass flowering in the following dry seasea-son. Fire is also used to promote resprouting of native grasses for cattle. The present, anthropogenic, fire return interval in Jalapão wet grasslands is 2–3 years (Schmidt et al., 2007). Although it has not been investigated in the Jalapão region specifically, the natural fire return in the Cerrado may vary from 1 to 9 years, depending on the vegetation physiognomy (Miranda et al., 2010). Golden-grass is harvested from community-owned areas, private lands and pro-tected areas. The land tenure system and presence of large, non-inhabited areas allow more than one community to manage the same areas, often generating conflicts related to fire and golden-grass management.

Golden-grass handicrafting began in Mumbuca community in the 1930s. By 2000, it had expanded to all of the core Jalapão re-gion (Schmidt et al., 2007), becoming a trademark of the Jalapão, and even Tocantins state. Handicrafts are sold locally as well as in all the main Brazilian cities, and exported to several countries. 2.3. Assessing local ecological knowledge and management of golden-grass in Tocantins state

In June and July 2008, we visited communities in nine munici-palities in Tocantins state, outside of Jalapão region, where gold-en-grass handicrafts have recently become economically important. In each community, we conducted semi-structured interviews and focus groups (up to 12 people) with artisans to

(3)

ac-quire information on golden-grass local ecological knowledge and management practices. At least ﬁve harvesters/artisans were inter-viewed per community. We compared these results to data from similar interviews conducted in the core of Jalapão region from 2002 to 2010.

2.4. Seed production surveys

To characterize golden-grass seed production phenology across Tocantins state, 20 flowerheads were collected by 22 volunteers every 15 days, in each of nine municipalities, as well as in the core of Jalapão region during the flowering-fruiting period in 2008 (see Fig. 1, N = 2476 flowerheads). Harvesting experiments (see below) showed that seed production phenology is a good indicator of flower stalk dryness and therefore rates of uprooting golden-grass rosettes during harvest.

2.5. Impacts of harvest on golden-grass population dynamics To assess the demographic impacts of harvesting golden-grass flower stalks, we established permanent plots in nine sites in the core of Jalapão region, ranging from 3 to 27 km apart, and moni-tored annual rates of survival, growth, and reproduction of gold-en-grass. In each site, plots were randomly assigned to harvest and control (no harvest) treatments and local harvesters harvested all flower stalks in the plots, according to traditional management practices (i.e., ‘late’ harvest – after mid-September). Since tradi-tional management involves harvesting golden-grass 1 year after fire, all experimental harvest was performed in plots that had been burned during the previous year. All study sites were chosen in col-laboration with local harvesters, are considered good harvesting areas and had been burned by local harvesters the year before the experiments were established.

For sites 1–3, which were managed by Mumbuca community, we established 20 0.25 0.25 m plots per treatment (harvest and control), which were harvested in 2003 and monitored from 2003 to 2004 (harvest year 1). Plots in the remaining sites (4–9) were part of a larger experiment that included experimental burns

(Schmidt, 2011), and were managed by four local communities

from the core of Jalapão region. At each of these sites, we estab-lished 0.5 0.5 m study plots in 2006. Experimental harvests were carried out in 2006 (harvest year 2, 8–20 plots per treatment per site) and 2008 (harvest year 3, 4–8 plots per treatment per site). Plots were experimentally burned in 2007 and 2009. The number of individuals used to build the population models in each site and year are relatively similar between harvested and control treatments (Fig. 2). Analyses using the same number of individuals per treatment and year did not result in any signiﬁcant changes to the results, so we maintained all individuals monitored in the anal-yses presented here.

In each plot, all golden-grass rosettes were tagged and at each annual census, in June–July, they were classified into five life-stages based on ramet size and size of the genet (or clump): (S) ‘Seedlings’ –single rosettes <2 cm in diameter; (SR) ‘Small Ra-mets’ – rosettes <2 cm that are part of a larger genet. These plants do not flower. Individuals P2 cm can flower, and were classified into three adult classes since adult ramet survival varied signifi-cantly with genet size (I.B.S, unpublished results): (A) ‘Adults’ – single rosettes; (AR1) ‘Adult Ramets 1’ – adults in genets with 2–4 ramets and (AR2) ‘Adult Ramets 2’ – adults in genets with more than 4 ramets. The number of small ramets was low, so we used only one small ramet stage to avoid increasing sampling error

(Moloney, 1986). We monitored a total of 7744 plants over the

study period. Apart from the aforementioned differences in sur-vival among ramets in genets of different sizes (AR1 and AR2 indi-viduals) there was no other evidence of density dependence.

Fig. 1. Tocantins municipalities color-coded according to the local ecological knowledge and knowledge about state harvesting regulations related to golden-grass among interviewed artisans and harvesters. Interviews and seed surveys were performed in these municipalities in 2008.

(4)

For each of the nine sites and three harvest years, we built stage-based matrix models (Caswell, 2001). Since golden-grass ro-settes can recruit into all stages by sexual or clonal reproduction within a year interval (Fig. A1), we used vital rates (survival, growth and reproduction,Tables A1 and A2) instead of matrix tran-sitions in our analyses (Franco and Silvertown, 2004). When neces-sary, we added 0.01 to vital rates to ensure matrix irreducibility (Caswell, 2001). We calculated 95% confidence intervals for each kvalue by bootstrapping the observed annual individual transition among stages, using 5000 iterations, with replacement. We consid-ered k with non-overlapping confidence intervals to be signifi-cantly different.

We assessed the demographic impacts of traditional harvesting of golden-grass ﬂower stalks by comparing deterministic k, elastic-ity and LTRE results (Caswell, 2001) from control and harvested populations. These deterministic k do not represent long-term pop-ulation growth rates since they are based only on demographic rates 1 year after ﬁre. We also calculated stochastic lambda (ks)

using a Markov chain, which adds memory to the system (Caswell, 2001). In our case the Markov chain had two states that alternated periodically: ‘burned’ and ‘one-year-after-ﬁre’ states (Appendix B). We ran the stochastic simulation for 100000 years, and discarded the ﬁrst 2000 iterations to remove transient dynamics from the long-term simulation.

The six matrices built from the same sites in harvest year 2 (Table B2) and harvest year 3 (Table B3) of this study composed two levels of the ‘one-year-after-ﬁre state’, each level (year) having 50% chance of occurring. Within each state and level, the six matri-ces (sites) had equal probability of occurrence. The burned state was composed of six matrices built from populations in sites 4–9 from 2009 to 2010, experimentally burned in September 2009 (Schmidt, 2011). To avoid confounding the effects of harvest and

rainfall, we did not use the matrices from 2007 to 2008 (the other year with experimental burning monitored) since this year had exceptionally low rainfall and this signiﬁcantly affected golden-grass vital rates (Schmidt, 2011).

To assess the effects of early-harvesting, i.e., before flower stalks are dry and seeds have been produced, we conducted two ‘early’ experimental harvests, on August 8th and September 8th, and two traditional experimental harvests on September 20th and 21st in 2006 with seven experienced harvesters. Each harvester collected flower stalks for 5 min, we then weighed the flower stalks harvested and counted the number of rosettes uprooted. We sim-ulated the effects of early-harvest on golden-grass population dynamics using the control matrices from sites 4 to 9 by reducing fertility values to zero (since flower stalks are harvested before the seeds mature) and incrementally increasing mortality of flowering individuals from 5% to 40%. These simulated early harvest matrices were included in the biennial fire stochastic simulations described above.

We used R, version 2.11.1 (R Development Core Team, 2011) for all analyses, and the ‘popbio’ package (Stubben and Milligan, 2007) for matrix analyses.

3. Results

3.1. Local ecological knowledge and management of golden-grass All harvesters from the core of Jalapão stated that they harvest after mid-September. These harvesters were the only ones to state that harvesting after mid-September increases the brightness of golden-grass ﬂower stalks and handicrafts, and therefore their quality. Moreover, they were the only ones who directly associated

Fig. 2. Deterministic population growth rates (k) and 95% bootstrap confidence intervals for golden-grass populations in nine study sites, for 3 years, in Jalapão region, Tocantins, Brazil. Numbers inside bars indicate the number of plants considered to build each matrix;_{indicate a significant difference (non-overlapping confidence intervals)}

(5)

early harvesting with an increased chance of uprooting adult plants. They also maintained that ﬁre promotes increased ﬂowering.

Harvesters from outside the core of Jalapão had less ecological knowledge of golden-grass and knowledge decreased with increas-ing distance from the core of Jalapão (Fig. 1). Harvesters from the four municipalities closer to the core of Jalapão, had harvested golden-grass for 1–6 years and all stated that fire promotes flower-ing. Despite being aware of the state regulations for harvesting, they mainly harvested early – between August and mid-September – and none of them cut the flowerheads after harvest. These har-vesters associated the prohibition of harvesting before September 20th only with decreasing plant uprooting during harvesting.

In the five remaining municipalities further away from the Jala-pão region, harvest of golden-grass began only after 2006. All har-vesters stated no prior knowledge about the harvesting rules and carry out very early harvests – between June and August, as soon as the flower stalks are ‘‘long enough’’. These harvesters perceived fire mostly as a threat to golden-grass. We observed flower stalks recently harvested with flowers (no seeds) and uprooted rosettes. Although harvesting the same species, many of these harvesters were not sure if they were harvesting the ‘‘real’’ golden-grass, since they perceive their handicrafts to be not as shiny as those from the Jalapão region.

3.2. Golden-grass seed production across Tocantins state

Across all regions included in our seed survey (Fig. 1), the num-ber of seeds per ﬂowerhead increased from August 1st to August

15th and remained stable until September 15th, after which point it decreased. This indicates that seed production occurs mostly in August, that most seed dispersal happens only after mid-Septem-ber, and that this is consistent across all our study regions (Fig. C1).

3.3. Effects of traditional harvesting on golden-grass dynamics Experimental harvesting performed according to traditional harvest practices (‘late’ harvest) had no signiﬁcant impact on k. Among the 15 paired comparisons between harvested (H) and con-trol (C) populations within a same site and year, only two showed signiﬁcant differences in k, both in harvest year 2, where in site 6 kC> kHand in site 8 kC< kH(Fig. 2). Irrespective of harvest, k varied

among sites within years, but there were no sites that had consis-tently higher or lower k. Both harvested and control populations varied even more greatly among years, with all populations with k> 1 in harvest years 1 and 2 and mostly k < 1 in year 3 (Fig. 2).

Traditional harvest did not signiﬁcantly affect elasticity values of any vital rate type (Fig. 3) or plant stage. For all populations, sur-vival had the highest elasticity values. Elasticity of k to perturba-tions of other vital rates was lower and varied across years: k was more elastic to fertility, clonal reproduction and growth in harvest year 1 than in the other 2 years (Fig. 3). The three adult stages had the highest elasticity values (mean: 0.29, 0.33 and 0.34 for the Adult, Adult Ramet 1 and Adult Ramet 2 stages respec-tively); elasticity values per stage varied across sites, irrespective of treatment. Elasticity of k to variation in seedling (S) vital rates was higher in harvest year 1 than other years (mean: year 1: 0.14; year 2: 0.025; year 3: 0.09).

(6)

LTRE contributions to differences in k between harvested and control populations also varied across sites, and especially years. There was no consistent pattern in LTRE contributions that could be attributed to harvest. The only exception was the overall nega-tive LTRE contribution of sexual fertility in harvest year 1 (Fig. 4), i.e., harvested populations had lower fertility rates than control populations and these differences contributed to decreases in k of harvested populations. In that year, golden-grass ﬂowerheads were not cut to be left in the ﬁeld after harvest, and the sexual fer-tility had higher elasticity values in this year compared to the other studied years (Fig. 3).

3.4. Effects of early-harvest on golden-grass dynamics

Based on early-harvest experiments, we estimate that 10–15% of the ﬂowering plants were uprooted in early August, 3–10% in early September and 0–2% in mid-September. The number of ro-settes uprooted varied among harvesters and for the same

har-vester in different sites. The amount of rosettes uprooted at each harvest trail was determined by the date of harvesting and was not related to the quantity of stalks obtained.

Our simulation of golden-grass dynamics based on a biennial ﬁre return interval indicates that the long-term growth rate is slightly above unity for both control (unharvested) and tradition-ally managed (harvested) populations. Early-harvesting, even at low levels of ﬂowering plant mortality (e.g. <10%), causes ksto drop

below 1 (Fig. 5).

4. Discussion

When NTFPs have high economic value, commercial harvest of-ten expands rapidly out of regions with a history of traditional use (Shackleton et al., 2009). This is often aided by non-governmental and governmental organizations aiming to improve local liveli-hoods and promote sustainable resource use. However, if and

Fig. 4. Life table response experiment (LTRE) contributions of vital rates to differences in k between control and harvested populations of golden-grass in Jalapão region, in nine sites. Negative contribution values indicate that differences in the vital rates contributed to decreased k in the harvested populations compared to the control (reference matrix). Positive life table response experiments indicate that the vital rates contributed to increased k in harvested populations in comparison to control populations.

(7)

how the local knowledge associated with harvest and/or manage-ment is transferred, is rarely examined.

For golden-grass, the traditional knowledge and practice of har-vesting only after mid-September are essential for sustainability. The knowledge that harvesting only after seed production prevents uprooting rosettes and improves the quality of the handicrafts was once restricted to experienced harvesters (Schmidt et al., 2007) and has now spread throughout the core of Jalapão region. The high economic value of the products likely motivated this knowl-edge exchange among communities (Berkes, 2008).

On the other hand, most harvesters outside of the core of Jala-pão region have no knowledge of the relationships between the timing of harvest and the risk of uprooting rosettes, nor of ﬁre management or other factors that may inﬂuence golden-grass pop-ulations. This indicates that local ecological knowledge and man-agement practices have neither been spread to these areas nor developed independently. Golden-grass harvesting is less than 10 years old in these regions, a short time for new local ecological knowledge to develop (Berkes, 2008). Additionally, handicrafting has mostly spread through courses promoted by state and munici-pal agencies, which did not consider harvesting and management methods.

This lack of local ecological knowledge and management prac-tices is problematic, since it leads to ‘early’ harvest, which de-creases k considerably, even at low levels of adult mortality from uprooting. In contrast, all our analyses suggest that the traditional ‘late’ harvest of golden-grass ﬂower stalks has little effect on pop-ulation dynamics. The timing of harvest has been shown to affect the impacts of harvest in other systems (Delvaux et al., 2010; Joyal, 1996) and strongly determines the ecological impacts of golden-grass harvesting.

The effects of golden-grass harvesting are consistent with the NTFP literature: when ﬂower stalks were harvested after seed pro-duction (‘late’ harvest), it is similar to fruit-harvest, which is sus-tainable even at high intensities for many perennial plants (e.g., Emanuel et al., 2005; Shankar et al., 1996). In addition, the state harvest policy requires that golden-grass seeds be left in the envi-ronment after harvesting, which is comparable to systems where harvesters facilitate plant regeneration (e.g.,Mooney, 2007; Ticktin et al., 2003).

In contrast, early harvesting eliminates sexual reproduction and causes adult mortality. This is similar to other perennial herbs, where limits for sustainable harvest of reproductive structures be-fore seed dispersal are low (Baltzer et al., 2002; Lamont et al., 2001). The negative impacts of early-harvest on golden-grass pop-ulations are mostly due to the increased mortality of adults, which has a high elasticity value. Other perennial herbaceous species can also tolerate only low levels of adult mortality (Ghimire et al.,

2008; Law, 2007; Mooney, 2007; Nault and Gagnon, 1993). This

is also consistent with the observation of experienced Jalapão har-vesters that golden-grass ﬂower production decreases after early-harvest is performed. Our experimental early-harvests illustrated that the earlier the ﬂower stalks are harvested, the greater the mortality due to uprooting. Adult mortality levels of 30–40% may occur if harvest is carried out as early as June, as was reported by new har-vesters. Our models suggest that harvest practices that lead to more than 10% mortality of adults are likely unsustainable.

While variation in plant population dynamics is expected across geographic ranges (Jongejans et al., 2010), our ﬁnding that the tim-ing of golden-grass seed production is similar across a wide region suggests that the present state policy on the timing of harvest is appropriate since it prevents early-harvesting. In addition, the pol-icy that ﬂowerheads must be cut and left in the harvesting area to disperse the seeds, a recommendation derived from an ecological study, not local ecological knowledge, may play a key role in some years, even if sexual production is only sporadically important for k. The impact of depleting seeds might be severe in years when the elasticity sexual reproduction is higher (Fig. 3). The combination of moderate elasticities and high negative LTRE contributions for sex-ual reproduction observed in harvest year 1 (Fig. 4) suggest that golden-grass harvesting maybe potentially unsustainable under these conditions (Zuidema et al., 2007).

Other analyses have shown (Schmidt, 2011) that k of golden-grass populations significantly decrease with decreasing total rain-fall. For our stochastic simulations of the effects of different types of harvest under a biennial fire regime, we considered only years with rainfall close to and above the historical mean in the region. A more frequent occurrence of drier years would likely exacerbate the negative effects of early harvesting, as shown in other systems (Martinez-Ramos et al., 2009). The combined effects of variation in fire return interval and rainfall on golden-grass populations are discussed elsewhere (Schmidt, 2011). Although our results suggest biennial fires – which Jalapão harvesters maintain is ideal for gold-en-grass management – allow for the persistence of goldgold-en-grass populations, the impacts on other species and on soil erosion need to be assessed.

4.1. Implications for conservation

Our results suggest that the potential for sustainable golden-grass handicrafting is high. The combination of traditional manage-ment of experienced harvesters and existing state policy promote harvest sustainability, and limit the timing of harvest but not har-vest levels. As such, they do not limit income generation nor ex-clude communities with lower resources; they are also appropriate for the geographical scale over which harvest is occur-ring. This high potential for sustainable harvest is likely the case for other NTFP that share these characteristics. This includes species for which the timing of harvest is more important than the quan-tity removed – for example the many species whose dried fruit parts are harvested and used in handicrafts. In general, this may be the case for species for which the harvest negatively affects only low elasticity vital rates (Zuidema et al., 2007).

This situation contrasts with many NTFPs that can be harvested sustainably only at low levels or long harvesting intervals (Ghimire et al., 2008; Guedje et al., 2007).

Fig. 5. Stochastic population growth rates (ks) for golden-grass populations subject

to no harvest (control), late harvest (according to traditional practices, after flower stalks are dry), and early-harvest (before flower stalks are dry). Early-harvest was simulated by increasing rates of flowering adult mortality caused by flower stalks harvest. All simulations assume a biennial fire return interval.

(8)

Early harvesting of golden-grass, however, is unsustainable. The harvest of flower stalks during the flowering period has been pointed out as a cause of decline in other Eriocaulaceae species, including Syngonanthus spp. populations in Brazil (Bede, 2006; Giulietti et al., 1996). The increase in sales reported outside of the core-Jalapão area suggests that harvesting pressure is increas-ing in the regions where this unsustainable early harvest is preva-lent, due to lack of local knowledge. Early harvesting is carried out by harvesters in two situations. One involves new harvesters in re-gions where golden-grass handicrafting is spreading, who have no prior local ecological knowledge of golden-grass or information on state regulations. The second situation that motivates early harvest is increased harvesting pressure in regions where handicrafts have been economically important for many years. Here harvesters have local ecological knowledge and/or information on state regula-tions, but competition is high, wet grasslands are open access re-sources, and law enforcement is not effective. As with other NTFP systems, harvesting practices are influenced by numerous factors, including land tenure (Varghese and Ticktin, 2008), com-peting uses (Gaoue and Ticktin, 2009) and economic pressures (Baldauf et al., 2007; Zhang et al., 2008). As elsewhere, it is clear that sustainable use of economically important resources is more likely when community and/or governmental rules are continually enforced (Shahabuddin and Rao, 2010). Creating economic incen-tives for sustainable harvest is another approach that can foster conservation (Guariguata et al., 2011). In this specific case, this could be facilitated by the fact that golden-grass handicrafts made from ripe flower stalks and those made with ‘‘unripe’’ flower stalks can be easily detected by experienced artisans and buyers.

Establishing harvest policies for NTFPs is a challenge because le-gal requirements often prevent communities with less resources from beneﬁtting, and the extraction of resources is then continued by local or regional elites (Laird et al., 2010). The sustainable use of Cerrado products by local communities is one strategy to promote the conservation of this threatened biome and presents an alterna-tive to the economic model established for the development of the region (Klink and Machado, 2005). Golden-grass provides a good example of an NTFP with high potential for sustainable harvest, which can be achieved by combining local ecological knowledge, ecological studies, and public policy.

Acknowledgments

We thank the golden-grass harvesting communities for sharing knowledge and field assistance; T. Dias for the visit to Krahô area; E. Bonfatti, F. Borghetti, S. Caldas and UnB/Termobiology labora-tory. I.B. Figueiredo, K.F. Pellizzarro, C.P. Borges, J. Pereira, N. Barb-osa, A. Fidelis, R. Viana, M.B. Sampaio, A.E.O. Santos, F.N.G. Cardoso, A.B.C. Gonçalves, M. Alexandre, C.S. Moreira and the JSP staff for field assistance; L. Mandle, A.B. Sampaio, O. Gaoue for R assistance and discussions. Ibama, Capes/Fubright, SGP-UNDP, UH-Botany, GSO-UH, EECB-UH and IFS for financial support; Naturatins for the permits and fire brigade. C. Daehler, D. Drake and two anony-mous reviewers for comments in the previous versions of the manuscript.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.biocon.2012.03. 018.

References

Baldauf, C., Hanazaki, N., Reis, M.S.d., 2007. Caracterização etnobotânica dos sistemas de manejo de samambaia-preta (Rumohra adiantiformis (G. Forst)

Ching – Dryopteridaceae) utilizados no sul do Brasil. Acta Botan. Brasil. 21, 823– 834.

Baltzer, J.L., Reedkie, E.G., Hewlin, H.L., Taylor, P.D., Boates, J.S., 2002. Impact of ﬂower harvesting on the salt marsh plant Limonium carolinianum. Can. J. Bot. 80, 841–851.

Bede, L.C., 2006. Alternativas para o uso sustentado de sempre-vivas: efeitos do manejo extrativista sobre Syngonanthus elegantulus Ruhland (Eriocaulaceae). Instituto de Ciencias Biologicas, Universidade Federal de Minas Gerais, UFMG, Belo Horizonte, p. 184.

Belcher, B., Schreckenberg, K., 2007. Commercialization of non-timber forest products: a reality check. Develop. Policy Rev. 25, 355–377.

Berkes, F., 2008. How local knowledge develops: case from the West Indies. In: Berkes, F. (Ed.), Sacred Ecology. Routledge, New York, NY, pp. 203–224. Caswell, H., 2001. Matrix Population Models – Construction, Analysis, and

Interpretation,, Sinauer Associates, Sunderland, Massachusetts.

Delvaux, C., Sinsin, B., Dammea, P.V., 2010. Impact of season, stem diameter and intensity of debarking on survival and bark re-growth pattern of medicinal tree species, Benin, West Africa. Biol. Conserv. 143, 2664–2671.

Emanuel, P.L., Shackleton, C.M., Baxter, J.S., 2005. Modelling the sustainable harvest of Sclerocarya birrea subsp. caffra fruits in the South African lowveld. Forest Ecol. Manage. 214, 91–103.

Franco, M., Silvertown, J.W., 2004. A comparative demography of plants based upon elasticity of vital rates. Ecology 85, 531–538.

Furley, P., 2004. Tropical savannas. Prog. Phys. Geogr. 28, 581–591.

Gaoue, O.G., Ticktin, T., 2009. Fulani knowledge of the ecological impacts of Khaya senegalensis (Meliaceae) foliage harvest in Benin and its implications for sustainable harvest. Econ. Bot. 63, 256–270.

Gaoue, O.G., Ticktin, T., 2010. Effects of harvest of nontimber forest products and ecological differences between sites on the demography of African mahogany. Conserv. Biol. 24, 605–614.

Ghimire, S.K., Gimenez, O., Pradel, R., McKey, D., Aumeeruddy-Thomas, Y., 2008. Demographic variation and population viability in a threatened Himalayan medicinal and aromatic herb Nardostachys grandiﬂora: matrix modelling of harvesting effects in two contrasting habitats. J. Appl. Ecol. 45, 41–51.

Giulietti, A.M., Wanderley, M.G.L., Longhi-Wagner, H.M., Pirani, J.R., Parra, L.R., 1996. Estudos em ‘‘sempre vivas’’: taxonomia com ênfase nas espécies de Minas Gerais, Brasil. Acta Botan. Brasil. 10 (2), 329–383.

Guariguata, M.R., García-Fernández, C., Nasi, R., Sheil, D., Herrero-Jáuregui, C., Cronkleton, P., Ndoye, O., Ingram, V., 2011. Timber and non-timber forest product extraction and management in the tropics: towards compatibility? In: Shackleton, S., Shackleton, C., Shanley, P. (Eds.), Non-Timber Forest Products in the Global Context. Springer-Verlag, Berlin, Heidelberg, pp. 171–188. Guedje, N.M., Zuidema, P.A., During, H., Foahom, B., Lejoly, J., 2007. Tree bark as a

non-timber forest product: the effect of bark collection on population structure and dynamics of Garcinia lucida Vesque. Forest Ecol. Manage. 240, 1–12. Jimenez-Valdes, M., Godinez-Alvarez, H., Cavallero, H., Lira, R., 2010. Population

dynamics of Agave marmorata Roezl. under two contrasting management systems in Central Mexico. Econ. Bot. 64, 149–160.

Jongejans, E., Jorritsma-Weink, L.D., Becker, U., Dostál, P., Milden, M., Kroon, H., 2010. Region versus site variation in the population dynamics of three short-lived perennials. J. Ecol. 98, 279–289.

Joyal, E., 1996. The palm has its time: an ethnoecology of Sabal uresana in Sonora, Mexico. Econ. Bot. 50, 446–462.

Klink, C.A., Machado, R.B., 2005. Conservation of the Brazilian Cerrado. Conserv. Biol. 19, 707–713.

Laird, S.A., McLain, R.J., Wynberg, R.P. (Eds.), 2010. Wild Product Governance: Finding Policies That Work for Non-Timber Forest Products. Earthscan, Washington, DC.

Lamont, B.B., Marsula, R., Enright, N.J., Witkowski, E.T.F., 2001. Conservation requirements of an exploited wildﬂower: modelling the effects of plant age, growing conditions and harvesting intensity. Biol. Conserv. 99, 157–168. Law, W., 2007. Ecological and evolutionary impacts of harvesting the Himalayan

snow lotus (Saussurea laniceps Maxim. and S. medusa Hand-Mazz.). In: Program in Evolution, Ecology, and Population Biology. Washington University, St. Louis, Missouri, p. 166.

Martinez-Ramos, M., Anten, N.P.R., Ackerly, D.D., 2009. Defoliation and ENSO effects on vital rates of an understorey tropical rain forest palm. J. Ecol. 97, 1050–1061. Miranda, H.S., Neto, W.N., Neves, B.M.C., 2010. Caracterização das queimadas de Cerrado. In: Miranda, H.S. (Ed.), Efeitos do regime do fogo sobre a estrutura de comunidades de Cerrado: resultados do Projeto Fogo. Ibama, Brasília, DF, Brazil, pp. 23–34.

Moloney, K.A., 1986. A generalized algorithm for determining category size. Oecologia 69, 176–180.

Mooney, E.H., 2007. Genetic and Evolutionary Consequences of Harvest in American Ginseng, Panax quinquefolius L. (Araliaceae). Department of Biology, West Virginia University, Morgantown, p. 192.

Munhoz, C.B.R., Felﬁli, J.M., 2006. Floristic of the herbaceous and subshrub layer of a moist grassland in the Cerrado Biosphere Reserve (Alto Paraiso de Goias), Brazil. Edinburgh J. Bot. 63, 343–354.

Nault, A., Gagnon, D., 1993. Ramet demography of Allium tricoccum, a spring ephemeral, perennial forest herb. J. Ecol. 81, 101–119.

Oliveira, P.S., Marquis, R.J. (Eds.), 2002. The Cerrados of Brazil. Columbia University Press, New York.

R Development Core Team, 2011. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria.

(9)

Sampaio, M.B., Schmidt, I.B., Figueiredo, I.B., 2008. Harvesting effects and population ecology of the buriti palm (Mauritia ﬂexuosa L. f.; Arecaceae) in the Jalapão region, Central Brazil. Econ. Bot. 62, 171–181.

Schmidt, I.B., 2011. Effects of Local Ecological Knowledge, Harvest and Fire on Golden-Grass (Syngonanthus nitens, Eriocaulaceae), A Non-Timber Forest Product (NTFP) Species from the Brazilian Savanna. Botany Department/ Ecology, Evolution and Conservation Biology Program, University of Hawai’i at Manoa, Honolulu.

Schmidt, I.B., Figueiredo, I.B., Scariot, A.O., 2007. Ethnobotany and effects of harvesting on the population ecology of Syngonanthus nitens (Bong.) Ruhland (Eriocaulaceae), a NTFP from Jalapão Region, Central Brazil. Econ. Bot. 61, 73–85.

Schmidt, I.B., Figueiredo, I.B., Borghetti, F., Scariot, A.O., 2008. Produção e germinação de sementes de ‘‘capim dourado’’, Syngonanthus nitens (Bong.) Ruhland (Eriocaulaceae): implicações para o manejo. Acta Botan. Brasil. 22, 37– 42.

Schmidt, I.B., Mandle, L., Ticktin, T., Gaoue, O.G., 2011. What do matrix population models reveal about the sustainability of non-timber forest product (NTFP) harvest? J. Appl. Ecol. 48, 815–826.

Seplam, 2003. Plano de manejo do Parque Estadual do Jalapão, p. 131.

Shackleton, S.E., Shackleton, C.M., Wynberg, R., Sullivan, C.A., Leakey, R.R.B., Mander, M., McHardy, T., Den Adel, S., Botelle, A., Du Plesis, P., Lombard, C., Laird, S., Cunningham, A.B., O’Regan, D.O., 2009. Livelihood trade-offs in the commercialization of multiple-use NTFP: lessons from marula (Sclerocarya birrea subsp. caffra) in Southern Africa. In: Shaanker, R.U., Hiremath, A., Joseph, G., Rai, N.D. (Eds.), Non-Timber Forest Products: Conservation, Management and Policy in the Tropics. Ashoka Trust for Research in Ecology and the Environment (ATREE), Bangalore, India, pp. 139–173.

Shahabuddin, G., Rao, M., 2010. Do community-conserved areas effectively conserve biological diversity? Global insights and the Indian context. Biol. Conserv. 143, 2926–2936.

Shankar, U., Murali, K.S., Shaanker, R.U., Ganeshaiah, K.N., Bawa, K.S., 1996. Extraction of non-timber forest products in the forests of Biligiri Rangan Hills, India 3. Productivity, extraction and prospects of sustainable harvest of Amla Phyllanthus emblica (Euphorbiaceae). Econ. Bot. 50, 270–279.

Silva, J.M.C., Bates, J.M., 2002. Biogeographic patterns and conservation in the South American Cerrado: a tropical savanna hotspot. Bioscience 52, 225–234. Soehartono, T., Newton, A.C., 2001. Conservation and sustainable use of tropical

trees in the genus Aquilaria II. The impact of gaharu harvesting in Indonesia. Biol. Conserv. 97, 29–41.

Stubben, C., Milligan, B., 2007. Estimating and Analyzing Demographic Models Using the popbio Package in R. Journal of Statistical Software 22. <http:// www.jstatsoft.org/>.

Ticktin, T., 2004. The ecological implications of harvesting non-timber forest products. J. Appl. Ecol. 41, 11–21.

Ticktin, T., Johns, T., 2002. Chinanteco management of Aechmea magdalenae: implications for the use of TEK and TRM in management plans. Econ. Bot. 56, 177–191.

Ticktin, T., Johns, T., Xoca, V.C., 2003. Patterns of growth in Achmea magdalenae (Bromeliaceae) and it potential as a forest crop and conservation strategy. Agric. Ecosyst. Environ. 94, 123–139.

Varghese, A., Ticktin, T., 2008. Regional variation in non-timber forest product harvest strategies, trade, and ecological impacts: the case of Black Dammar (Canarium strictum Roxb.) use and conservation in the Nilgiri Biosphere Reserve, India. Ecol. Soc. 13, 11.

Zhang, L., Brockelman, W.Y., Allen, M.A., 2008. Matrix analysis to evaluate sustainability: the tropical tree Aquilaria crassna, a heavily poached source of agarwood. Biol. Conserv. 141, 1676–1686.

Zuidema, P.A., Kroon, H., Werger, M.J.A., 2007. Testing sustainability by prospective and retrospective demographic analyses: evaluation for palm leaf harvest. Ecol. Appl. 17, 118–128.