Genecological zones and selection criteria for natural forest populations for conservation:the case of Boswellia papyrifera in Ethiopia

Abayneh Derero?Ade fi res Worku?Habtemariam Kassa

Introduction

Forest genetic resources are important to humans and to life on Earth in general,but they are under continuous threat due to forest cover reduction and degradation,competition for other land-uses,climate change,ecosystem modi fi cation,overharvesting and spread of invasive species(FAO 2014).Efforts to conserve forest genetic resources have so far involved identifying the level of threat by preparing the Red List for species(IUCN 2012),understanding species genetic diversity and structure(Derero et al.2011;Ayele et al.2011;Addisalem et al.2016),and introducing tools to monitor trends in genetic diversity(Graudal et al.2014).Furthermore,priority-setting exercises based on criteria and indicators,designing systematic approaches such as genecological zoning(Graudal et al.1997),and establishing protected area network and other conservation mechanisms have been developed.

As a broader strategy,introducing a system of genecological zoning,which considers natural vegetation,edaphic conditions,climate,elevation and barriers to gene fl ow,is helpful for managing genetic resources systematically(Graudal et al.1997;St Clair et al.2005).Then forest populations can be prioritized for conservation in each zone through the use of useful criteria(Rotach 2005;FAO 2000).

In Ethiopia,genecological knowledge remains limited even though there was an attempt to designate tree seed zones to guide tree seed collection and distribution(Aalb?k 1993).In addition,population selection through objectively measurable criteria and indicators has also been lacking.

The target species,Boswellia papyrifera,is a small deciduous tree that belongs to the family Burseraceae.In Ethiopia,the species is mainly found at altitudes from 950 to 1800 m a.s.l.inCombretum–Terminaliawoodlands and wooded grasslands,one of the major natural vegetation types in Ethiopia mainly occurring along the western escarpments of the Ethiopian highlands(Friis et al.2010).The species thrives best in warm,moist lowlands compared with moist or dry highlands(Ogbazghi et al.2006).Although little has been known about its reproductive biology and genetic structure,a recent analysis of the genetic structure of the species employing microsatellite markers identi fi ed geographic structuring of the genetic variation of the species in Ethiopia(Addisalem et al.2016).

The species is an important source of oleo-gum resin,called frankincense or olibanum,which in the Horn of Africa has numerous traditional,ceremonial and medicinal uses(Lemenih and Teketay 2003).Five-yeardata(2009–2013)from the Ethiopian Revenues and Customs Authority shows that the annual foreign exchange revenues generated from the export of gums and resins was 11.82 million(±1.24)USD.However,this revenue is very low compared with other commercial crops during the same period,such as sesame(368.57±66.5 million USD/year).In addition,the net revenue from sesame production is higher when compared with the forest land-use that includes commercial nontimber forest products(Dejene et al.2013).

TheB.papyriferadominated forest and woodlands of Ethiopia are under severe threat owing to the agricultural expansion,uncontrolled livestock grazing and recurrent droughts(Lemenih et al.2014).The problem is exacerbated by high seedling mortality,poor recruitment(Negussie et al.2008)and high adult mortality(Groenendijk et al.2012).If the current trend continues unchecked,the country will lose the key environmental role played by the forest in protecting against deserti fi cation,conserving biodiversity and provisioning other products and services.

Amidst developments in the area,regional authorities have called for help in identifying forest areas that should be given priority for conservation.Thus,decision makers at various levels need speci fi c information on which forest areas should be given priority for conservation while also maintaining the expansion of commercial agriculture for export.Then the questions are,what approach and methodology should be used to select viable populations for conservation,and what information on genetic structure on adaptive traits is lacking.Therefore,the present study was conducted to quickly collect information about the forest structure in one of the forests in the extensiveCombretum–Terminaliawoodland and reclassify the woodland into subunits with the main objective of selecting natural populations for conservation.By also engaging key stakeholders through a multi-criteria methodology,we aimed to incorporate their knowledge and concerns in the development of tools for guiding decisions based on scienti fi c evidence.

Materials and methods

Designation of genecological zones

We hypothesized that species genetic variation correlates with environmental gradient,and since there is a marked rainfall gradient within theCombretum–Terminaliavegetation zone,reclassi fi cation of the vegetation into moisture zones would be a sound approach to identify genecological zones.Thus,map of theCombretum–Terminaliawoodlands and wooded grasslands was obtained in a shape fi le format from the project ‘Vegetation and climate change in eastern Africa(VECEA)’,and annual rainfall surface data were obtained from World Clim(www.worldclim.org,version 1).Then the annual rainfall data were clipped usingCombretum–Terminaliaboundary,and the rainfall values were reclassi fi ed into three moisture zones(above 1400,900–1400 mm,and below 900 mm).Finally,the moisture zones were designated as genecological zones.

Vegetation survey

Site description

Vegetation was surveyed in a selected forest area at Lemlem Terara in MetemaWoreda of North Gondar Zone in Amhara Regional State, northwestern Ethiopia(12°38.42′–12°41.46′N,36°15.94′–36°20.42′E)from 769 to 889 m a.s.l.The mean annual rainfall at Metema is 924.2 mm,and the mean annual temperature is 32.98°C(Wale et al.2012).

Sampling design and data collection

Information on vegetation was a basic input to identify realistic criteria and indicators for conservation.B.papyrifera,like many other species,is not uniformly distributed in the forest at Lemlem Terara.Hence,we targeted segments of the forest dominated byB.papyriferaand laid a total of 15 circular plots(each with a 15-m radius)on three 2-km-long transects.The consecutive plots were 500 m apart.In each plot,geographical location,altitude,tapping history,health and disturbance condition,dbh and height of all woody species were recorded.In each plot,two cohorts were identi fi ed,i.e.,trees(dbh≥2 cm)and saplings(dbh＜2 cm,height≥ 1 m).Seedlings(individuals with height＜1 m)were counted in circular subplots with 7.5-cm radius.

In addition,three quadrats,each 50 m×20 m,were laid along a randomly selected transect(with a bearing of 90°).The distance between two quadrats was 1 km.However,the second quadrat,where nomads kept their cattle,was heavily disturbed,and had only four adult trees.Since this disturbance does not re fl ect natural conditions,the quadrat was avoided for data collection and analysis.Data as similar to the types collected in the circular plots were collected in the two quadrats.Here seedlings were counted in subplots of 5 m×5 m.

Identifying and weighting criteria and indicators

A multi-criteria decision analysis method(MCDA),a widely applied model in forest management(Mendoza and Martins 2006),with three steps was followed to identify and weight criteria and indicators to select viable populations for conservation.

Step 1Identify criteria and indicators,based on globally agreed and widely employed criteria and indicators for sustainable forest management,to use in selecting a forestforconservation among severalcandidates(Casta?eda 2000;FOREST EUROPE et al.2011).

Step 2Conduct a focus group discussion involving six research scientists and a gum and resin production expert to deliberate on the set of multiple criteria and indicators to fi ne-tune the list of criteria and indicators(C&I).

Step 3Of the three multi-criteria decision-making(MCDM)methodologies described by Mendoza and Prabhu(2000),the ‘rating method’was used for C&I assessment.The rating method directly assigns weights explicitly to each decision element by distributing 100 points,which is the sum of all the weights.The relative weight,wji,for indicatoriunder criterionj,was computed following the method of Mendoza and Prabhu(2000),

The rating involved 43 respondents:professionally 28 were foresters and the remaining 15 were nonforesters,and in terms of responsibility,22 were researchers and the remaining 21 were nonresearchers.Respondents were allowed to add a criterion and/or an indicator under each criterion whenever they believed that there was a need to do so.Those C&I seconded by at least fi ve respondents were considered for inclusion.Corrections were made whenever weights did not add to 100 by adding equal points to each entry.The ratings by the 43 experts of the single qualitative criterion,the fi ve quantitative criteria and 20 indicators were grouped into two(1)along the line of profession(from foresters and nonforesters)and(2)along the line of responsibility(from researchers and nonresearchers).The data on the ratings of criteria and indicators were then analyzed using descriptive statistics and a multivariate test in a generalized linear model.During the multivariate test,both profession and responsibility were considered as fi xed effects.The results from this analysis were presented as multivariate tests by responsibility and profession and also as tests of between-subjects effects on each criterion and indicator.

Results and discussion

Genecological zones

We mapped three moisture zones within theCombretum–Terminaliawoodland and designated them as genecological zones,which are referred to as(1)DryCombretum–TerminaliaWoodland Zone,(2)MoistCombretum–TerminaliaWoodland Zone and(3)WetCombretum–TerminaliaWoodland Zone(Fig.1).The altitude in the dry zone varied the most,from 380 to 2410 m(mean=1110±418 m).The altitude in the moist area ranged from 410 to 2270 m(mean=1080±403 m);the wet zone varied from 750 to 2100 m(mean=1450±248 m).

Fig.1 Moisture zones in the Combretum–Terminalia woodlands and wooded grasslands(CTW)in Ethiopia

The zones can serve as sampling frames for provenance research and identifying and selecting viable natural populations ofBoswellia papyriferaand other species for conservation.One of the environmental variables with high correlations with adaptive traits elsewhere is rainfall,and according to Wright(1976),trees from moister regions grow faster,have smaller seeds,are less deeply rooted and have greener foliage.The zonation approach based on environmental gradients such as in rainfall and elevation will be very instrumental in light of the absence of mapped genetic variation of adaptive traits(St Clair et al.2005),and adaptive variation is generally manifested by the relationship between measured traits and environmental gradients(White et al.2007).Because techniques based on DNA and isozymes developed to understand species genetic structure are insuf fi cient for providing information regarding adaptive traits(Ko nig 2005),we need to work with a zonation system based on environmental differences.

Vegetation status at Lemlem Terara

Species diversity and population structure

The total area of the sampled plots was 1.26 ha,and mean altitude was 847 ± 36 m(769–889 m).Most of the plots had moderate slope(5–20%),29% of the plots were fl at(0–3%),and the remaining 21%were steep(30–48%).There were 405 trees per ha and 35 woody species in the forest:30 species were recorded within the 1.26 ha sampled plots(Fig.2); fi ve tree species(Adansonia digitata,Combretum collinum,Ficus thoningi,Lonchocarpus laxifl orusandZiziphus spina-christi)were found outside the sample plots.The Shannon diversity index was 2.01,and the evenness value was 0.62,indicating a moderate level of diversity.Another survey in the same area identi fi ed 87 vascular plants,of which 40 were woody species(Wale et al.2012).In theCombretum–Terminaliawoodland and wooded grassland,199 species,subspecies and varieties of woody species were reported previously,of which 81 are speci fi c to this vegetation type and do not occur in the remaining vegetation types of the country(Friis et al.2010).

Fig.2 Adult,sapling and seedling populations of woody species at Lemlem Terara,North Gonder Zone,Amhara Regional State

The overall mean dbh of the species was 16.9±9.5 cm,ranging from 2.5 cm forCombretum adenogoniumto 63.5 cm forSterculia setigera.The majority(58.5%)of the trees measured had a mean dbh of 10–25 cm,25.1%had dbh less than 10 cm and the remaining 16.4%had dbh above 25 cm.

Speci fi cally,the measured dbh ofBoswellia papyriferaranged from 8 to 49.5 cm.The majority of individuals(75.5%)were medium-sized(10–25 cm),22.7%were larger trees(＞25 cm),and only a few(1.9%)represented small trees(＜10 cm).In addition,theB.papyriferainventory data obtained from MetemaWoreda Of fi ce of Agriculture from 723 1-ha sample plots indicated that 87% of the individuals had a dbh of 10–30 cm,and only 8.7 and 4.3%had dbh less than 10 and above 30 cm,respectively.Studies at some populations of the species in Eritrea,Sudan and elsewhere in Ethiopia indicated that medium-sized trees dominate the populations, and small trees(dbh＜8 cm)were generally not found(Ogbazghi et al.2006;Negussie et al.2008;Lemenih et al.2007).

Forest disturbance

In almost all the plots(93%),grazing and expansion for crop production happened to be one of the major anthropogenic threats for regeneration and healthy species population structure,and only one plot was not affected by either farming or grazing damages.Major forest land-use changes were mainly caused by the expansion of sesame and sorghum farms.The other major cause of forest disturbance and large-scale degradation was forest fi re;53% of the plots had at least one fi re during the 1–3 years before the survey.The possible reasons for forest fi res include farmland expansion,harvesting honey,aim to initiate grass growth and reduce livestock parasites such as ticks,ground preparation for fi rst tapping ofBoswelliatrees by gum collectors,and facilitating free movement of people and livestock by chasing snakes and wild animals(Eshete et al.2005).

Table 1 List of criteria and their relative mean weight,as rated by 43 experts and listed in decreasing order of rank

Table 2 List of indicators,their respective descriptions and mean relative weights under each criterion,as rated by 43 experts and listed in deceasing order of rank

Other disturbances recorded were small-scale;among 29 small-scale disturbances,52%were tree falls due to cuts and the other 48%were mainly windfalls(two cases were falls assisted by insect attack and one tree was over-topped).While such small-scale disturbances,or canopy gaps,in closed high forests may favor the growth of some tree species(Tesfaye et al.2010),their effect on regeneration cannotbeoveremphasizedinCombretum–Terminaliawoodlands,which are open forests.

Regeneration and recruitment

Of the very few saplings(10 ind.·ha-1)in the forest,75%belonged to aCrotonspecies,and the rest wereBalanites aegyptiaca,Dichrostachys cinereaandMaytenussp.In addition,there were 3314 seedlings/ha belonging to 24 different woody species,andB.papyriferaranked 12th in seedling density.Pterocarpus lucens,Lonchocarpus laxifl orusandStereospermums cuntianumwere the three mostabundant seedlings in the forest fl oor(Fig.2).Though some studies reported evidence that tapping ofB.papyriferatrees for frankincense production reduces fruitand seed production signi fi cantly(Rijkers et al.2006;Eshete et al.2012),our data did not reveal any regeneration problem for the species,which might be due to the stand having been at a resting stage for consecutive years and thus helped the trees to recuperate.Our data showed that there was no apparent natural regeneration problem in general,but recruitment of seedlings to sapling and tree cohorts was seriously hampered by disturbances from grazing and fi re.Similarly,a study on several forest populations in northern Ethiopia indicated that seedlings ofB.papyriferawere abundant on forest fl oors,but saplings failed to recruit owing to damage from fi re and grazing(Negussie et al.2008;Groenendijk et al.2012).In stark contrast,a high density of seedlings and saplings ofB.papyriferain less-disturbed forests was reported recently(Addisalem et al.2016).

Population selection criteria

Five rating criteria were listed by the 43 experts for selection of viable populations ofB.papyrifera(Table 1).They were adapted from the list of the globally agreed and widely used criteria for sustainable forest management

Table 3 Analysis of C&I relative weights given by the respondents,n=43

Effect Criteria Hypothesis df Error df Sig. F Hypothesis df Error df Sig.Responsibility 3.679a 4 36 0.013 4.784a 18 22 0.000 Profession 0.326a 4 36 0.859 1.126a 18 22 0.391 Responsibility×Profession 2.409a 4 36 0.067 0.491a 18 22 0.935 Indicators F

aExact statistic(Casta?eda 2000;FOREST EUROPE et al.2011).The list of criteria and indicators with their relative weights are presented in Tables 1 and 2,respectively.

Table 5 Tests of between-subjects effects on the ratings of the criteria and indicators

Table 5 Tests of between-subjects effects on the ratings of the criteria and indicators

C&I Code of C&I Sources of variation Between researchers and non-researchers Between foresters and non-foresters Mean Square F P value Mean square F P-value Criteria Health C1 1562.0 14.16 0.001 10.2 0.09 0.763 Forest cover C2 329.6 4.02 0.052 20.0 0.24 0.624 Productive C3 31.8 0.45 0.508 0.0 0.00 0.994 Biological C4 93.3 1.83 0.184 49.5 0.97 0.330 Socioeconomic C5 301.0 4.06 0.051 0.5 0.01 0.937 Indicator Fire C1I1 8366.0 52.68 0.000 493.7 3.11 0.086 Biotic C1I2 3337.7 23.79 0.000 249.3 1.78 0.190 Defoliation C1I3 1126.2 9.06 0.005 43.1 0.35 0.559 Forest area C2I1 6.1 0.06 0.810 94.6 0.90 0.349 Boswellia trees C2I2 36.9 0.67 0.417 8.1 0.15 0.703 Diameter C2I3 151.0 3.27 0.078 11.0 0.24 0.628 Growing stock C2I4 234.9 3.83 0.057 68.9 1.13 0.295 Sapling C2I5 44.0 0.48 0.494 0.9 0.01 0.922 Yield C3I1 60.9 0.11 0.743 2088.2 3.74 0.060 Other products C3I2 52.5 0.11 0.739 380.4 0.82 0.372 Quality C3I3 226.4 1.00 0.323 686.0 3.04 0.089 Genetic variation C4I1 393.1 3.11 0.086 33.7 0.27 0.609 Habitat C4I2 662.0 8.69 0.005 355.4 4.66 0.037 Composition C4I3 2.7 0.02 0.900 182.9 1.10 0.301 Threatened C4I4 18.6 0.32 0.573 0.3 0.00 0.946 Wood C5I1 41.7 0.20 0.653 1605.0 7.88 0.008 Feed C5I2 25.7 0.18 0.678 1.8 0.01 0.913 Medicinal C5I3 42.8 0.32 0.577 46.7 0.34 0.561 Gums and resins C5I4 272.1 1.00 0.323 537.4 1.98 0.168 Other bene fi ts C5I5 2.1 0.02 0.875 131.8 1.57 0.218

Ratings of criteria and indicators

The individuals engaged in rating varied greatly in their judgments of the C&I as depicted in Table 3.The lowest variability and hence the highest level of consistency among individual judgments was recorded for the criterion ‘Biological diversity’and for the indicators ‘Number ofB.papyriferatrees/ha’, ‘Diameter distribution of adultB.papyrifera’and ‘number of threatened species’.The highest variability among the respondents was recorded for ‘Forest ecosystem health and vitality’from the list of the criteria and for ‘Frankincense yield’from the group of indicators.

Multivariate tests on the fi ve criteria revealed statistically signi fi cant differences among the researchers and the nonresearchers(p＜0.05)but not between foresters and nonforesters(Table 4).The tests of between-subjects effects showed that,among the criteria compared,only the criterion‘Health and vitality’,differed signi fi cantly different(p＜0.05)between the researchers and non-researchers(Table 5).Similarly,the multivariate test on the twenty indicators revealed statistically signi fi cantdifferences among the researchers and the nonresearchers(p＜0.05),but not between foresters and nonforesters(Table 4).The tests of between-subjects effects showed that,of the indicators compared,habitat diversity,defoliation,forest damage by biotic agents and forest damage by fi re differed signi fi cantly among researchers and nonresearchers(p＜0.05),but not between foresters and nonforesters(Table 5).Foresters and nonforesters were thus consistent because there was no signi fi cant difference between the two groups in the multivariate tests.The results also indicate that involving researchers and no-researchers in MCDA can provide a more balanced view with respect to selecting populations for conservation because the two groups differed signi fi cantly in the multivariate tests.

Policy framework and implementation

The overwhelming majority of the respondents(98%)agreed that the surrounding land-use is the single most important factor to consider,as commercial farms and human settlement are expanding in the woodlands in recent years.This result implies that identifying conservation sites in areas designated for large-scale agricultural development would not be practical forB.papyriferaconservation.The lack of policy implementation tools and poor law enforcement were highlighted by 10 of the respondents as one of the underlying causes for the worsening forest conditions and lack of habitat protection in the area.Sidelining proper implementation of forest policy and law,wildlife policy and law,the environmental impact assessment and other relevant legal frameworks in favor of agricultural expansion especially undermines efforts toward sustainable resource management.In addition,the distribution of the gum and resin resources to communities without considering production capacity and lack of clear forest area demarcation are resulting in con fl icts among user groups.Further,the following factors were also identi fi ed by respondents as important policy issues that need to be addressed:lack of proper institution for overseeing policy implementation,unclear property rights,low bene fi t to primary producers in the value chain,uncontrolled grazing systems,and limited experience in designating areas for conservation in the dry lowland areas.

Conclusion and recommendations

The categorization of the region into three genecological zones is a sensible approach for delineating conservation zones for all the tree species in the extensiveCombretum–Terminaliawoodland.Hence,conservation efforts forB.papyriferaand any other species in this dry forest can bene fi t from the current classi fi cation.And the analysis of the forest patch at Metema was instrumental in establishing candidate conservation criteria and indicators by determining the species composition,diversity and the forest structure and dynamics.Thus,population selection indices can be computed by giving priority to those high-rated indicators under the six criteria,namely health and vitality,cover and population structure,productive function,biological diversity,and socioeconomic bene fi ts.The criteria and indicators can be applicable to more species with some modi fi cations to context.The level of consistency of different groups in judging different C&I indicates that participatory MCDA is a powerful tool for sound decisionmaking by involving relevant professionals from various institutions at various capacities irrespective of professional differences because we did not observe professional biases in judgments.

We propose that the selected and weighed C&I be used to select populations within a given genecological zone by producing population selection indices based on biophysical and socioeconomic data for each forest.It will be prudent to evaluate carefully the policy environment especially with respect to land-use plans and related development programs of the government in this regard.Hence,we recommend that land-use and concerns related to development policy be discussed and addressed at a higher level which in most cases appears to be an absolute prerequisite before any exercise to select populations.Once decisions on site selections have been made,conservation ofB.papyriferapopulations should incorporate conservation principles into sustainably managing forests,establishing a network of conservation stands in the various genecological zones and promoting parkland agroforestry in areas where forest areas are being converted to agricultural fi elds.

AcknowledgementsAll scientists and experts who participated in the focus group discussion and in the re fi nement and weighing of the criteria and indicators are appreciated.This work was fi nanced by CIFOR through its Community Forestry Project in Ethiopia funded by the Austrian Development Agency(Project No.2008/03).The authors are thankful to the People and Government of Austria.They are also grateful to the Central Ethiopia Environment and Forestry Research Center,which supported the study by covering the staff time of the lead author.

Aalb?k A(1993)Tree seed zones for Ethiopia.Forestry Research Center/National Tree Seed Project,Addis Ababa,p 120

Addisalem AB,Bongers F,Kassahun T,Smulders MJM(2016)Genetic diversity and differentiation of the frankincense tree(Boswellia papyrifera(Del.)Hochst)across Ethiopia and implications for its conservation.For Ecol Manage 360:253–260

Ayele TB,Gailing O,Finkeldey R(2011)Assessment and integration of genetic,morphological and demographic variation inHagenia abyssinica(Bruce)J.F.Gmel to guide its conservation.J Nat Conserv 19:8–17

Casta?eda F(2000)Criteria and indicators for sustainable forest management:international processes,current status and the way ahead.FAO,Rome,p 13

Dejene T,Lemenih M,Bongers F(2013)Manage or convertBoswelliawoodlands?Can frankincense production payoff?J Arid Environ 89:77–83

Derero A,Gailing O,Finkeldey R(2011)Maintenance of genetic diversity in Cordia Africana Lam.,a declining forest tree species in Ethiopia.Tree Genet Genomes 7:1–9

Eshete A,Teketay D,Hulten H(2005)The socio-economic importance and status of populations ofBoswellia papyrifera(Del.)Hochst in Northern Ethiopia:the case of North Gondar Zone.For Trees Livelihoods 15:55–74

Eshete A,Teketay D,Lemenih M,Bongers F(2012)Effects of resin tapping and tree size on the purity,germination and storage behaviour ofBoswellia papyrifera(Del.)Hochst seeds from Metema District,northwestern Ethiopia.For Ecol Manage 269:31–36

FAO(2000)Annex 6:de fi nitions and basic principles of sustainable forest management in relation to criteria and indicators.Italy,Rome,p 7

FAO(2014)The state of world’s forest genetic resources.Italy,Rome,p 276

FOREST EUROPE,UNECE,FAO(2011)State of Europe’s Forests 2011.Status and trends in sustainable forest management in Europe.Ministerial conference on the protection of forests in Europe,Forest EUROPE Liaison Unit Oslo,Norway,p 337

Friis I,Demissew S,Breugel PV(2010)Atlas of the potential vegetation of Ethiopia.The Royal Danish Academy of Sciences and Letters,Denmark,p 307

Graudal L,Kjaer E,Thomsen A,Larsen AB(1997)Planning national programmes for conservation of forest genetic resources.Technical Note No.48,Danida Forest Seed Centre,Humlebaek,Denmark,p 54

Graudal L,Aravanopoulos F,Bennadji Z,Changtragoon S,Fady B,Kj?r ED,Loo J,Ramamonjisoa L,Vendramin GG(2014)Global to local genetic diversity indicators of evolutionary potential in tree species within and outside forests.For Ecol Manage 333:35–51

Groenendijk P,Eshete A,Sterck FJ,Zuidema PA,Bongers F(2012)Limitations to sustainable frankincense production:blocked regeneration,high adult mortality and declining populations.J Appl Ecol 49:164–173

IUCN(2012)IUCN red list of threatened species.Version 2012.2.http://www.iucnredlist.org.

Ko nig AO(2005)Provenance research:evaluating the spatial pattern of genetic variation.In:Geburek T,Turok J(eds)Conservation and management of forest genetic resources in Europe.Arbora Publishers,Zvolen,pp 535–565

Lemenih L,Teketay D(2003)Frankincense and myrrh resources of Ethiopia.II.Medicinal and industrial uses.SINET Ethiop J Sci 26:161–172

Lemenih M,Feleke S,Tadesse W(2007)Constraints to smallholders production of frankincense in Metema district,North-western Ethiopia.J Arid Environ 71:393–403

Lemenih M,Arts B,Wiersum KF,Bongers F(2014)Modelling the future ofBoswellia papyriferapopulation and its frankincense production.J Arid Environ 105:33–40

Mendoza GA,Martins H(2006)Multi-criteria decision analysis in natural resource management:a critical review of methods and new modelling paradigms.For Ecol Manage 230:1–22

Mendoza GA,Prabhu R(2000)Multiple criteria decision making approaches to assessing forest sustainability using criteria and indicators:a case study.For Ecol Manage 131:107–126

Negussie A,Aerts R,Gebrehiwot K,Muys B(2008)Seedling mortality causes recruitment limitation ofBoswellia papyriferain northern Ethiopia.J Arid Environ 72:378–383

Ogbazghi W,Bongers F,Rijkers T,Wessel M(2006)Population structure and morphology of the frankincense treeBoswellia papyriferaalong an altitude gradient in Eritrea.J Drylands 1:85–94

Rijkers T,Ogbazghi W,Wessel M,Bongers F(2006)The effect of tapping for frankincense on sexual reproduction inBoswellia papyrifera.J Appl Ecol 43:1188–1195

Rotach P(2005)In situ conservation methods.In:Geburek T,Turok J(eds)Conservation and management of forest genetic resources in Europe.Arbora Publishers,Zvolen,pp 535–565

St Clair JB,Mandel NL,Vance-Borland KW(2005)Genecology of Douglas fi r in Western Oregon and Washington.Ann Bot 96:1199–1214

Tesfaye G,Teketay D,Fetene M,Beck E(2010)Regeneration of seven indigenous tree species in a dry Afromontane forest,southern Ethiopia.Flora Morphol Distrib Funct Ecol Plants 205(2):135–143

Wale HA,Bekele T,Dalle G(2012)Floristic diversity,regeneration status,and vegetation structure of woodlands in Metema Area,Amhara National Regional State,Northwestern Ethiopia.J For Res 23(3):391–398

White TL,Adams WT,Neale DB(2007)Forest genetics.CAB International,Wallingford,p 682

Wright JW(1976)Introduction to forest genetics.Academic Press,New York,p 463