FORESTRY IDEAS, 2021, vol. 27, No 1 (61): 210–232

FOSTERING THE SUSTAINABILITY OF COMMUNITY FORESTRY PROGRAM: CASE STUDY IN - Christine Wulandari1, Samsul Bakri2*, Melya Riniarti1, and Supriadi Supriadi3 1Forestry Department and Graduate Program of Forestry, Faculty of Agriculture, University of Lampung, Jl. Soemantri Brojonegoro # 1, , 35145, . E-mails: [email protected], [email protected] 2Forestry Department and Graduate Program of Environmental Science, University of Lampung, # 1 Jalan Soemantri Brojonegoro, Bandar Lampung, 35145, Indonesia. *E-mail: [email protected] 3Way Seputih-Sekampung Watershed Management Unit, Ministry of Environment and Forestry of Republic Indonesia, Bandar Lampung, Indonesia.

Received: 26 May 2020 Accepted: 30 June 2021

Abstract Community Forestry Program in Indonesia (called HKm: Hutan Kemasyarakatan) as the scheme for recovering forest degradation from encroachment has been operating for 13 years now, but until nowadays there is no available guideline for fostering its sustainability (SUST). HKm Authority, however, still have 22–24 years remain to foster the program before the scheme due. It is in needing both economic and tree biodiversity indicators of every HKm’s member land as the guidance for sustaining the program. The income as the economic indicator itself is commonly affected by the endogenous and exogenous variables. This research aimed at determining the roles of: (i) endogenous and exogenous variables on income agroforestry yield (INCM), (ii) INCM on tree biodiversity performance (BIODV), and (iii) BIODV on the HKm’s SUST. Data collected by interviewing 230 members of HKm Jaya Lestari located in Way Kanan Regency-Lampung, Indonesia in February-May 2018. These HKm members consisted of four ethnic groups which the two are as the native (Semendonese and Lampungese) that more adaptive to agroforestry cultivation, whereas the remains (Javanese and Sundanese) are the offspring of people moved from Java Island under colonization program by Dutch Administration since 1905. We employed Ordinary Least Square regression postulate model to investigate the first purpose and Loglinear Regression to examine the second and the third purposes. Minitab Version 16 software was ap- plied for the models’ goodness-fits test and parameter optimization at 90 and 95% confident level. The research suggests: (i) endogenous variables with positive effect on INCM were the fami- ly size, participation in extension activities, land holding acreage, the tribes whose Sundanese and Semendonese were higher than Javanese, and with negative effect was the land elevation; whereas the exogenous variable affected negatively the rural facility construction and nursery demonstration plot activities; (ii) the higher INCM the higher BIODV, but (iii) the better BIODV the lower SUST will be. For the sake of pursuing SUST, it recommends to continue fostering HKm members by managing the endogenous variables in order to rise up INCM, and then multiply BI- ODV as well as to broaden HKm members’ awareness on voluntary planting some wooden trees that have been adaptive or endemic at local region such as iron wood (Fagraea fragrans Roxb), cinnamon (Cinnamomum verum J. Presl), and Shorea javanica. Key words: agroforestry, biodiversity, forest recovery, income, sustainability. Fostering the Sustainability of Community Forestry Program: Case Study ... 211

Introduction claimed land by applying agroforestry gradually, forbidden cutting tree woods in- Sustainability still remains a major theme stead of non-wood products such as sap, in every development process since the latex, rattan, coffee fruit, vegetable, food Malthusian pessimism aroused in the ear- crops, etc. This scheme can be consid- ly 19th century (Gleditsch 2021, Meiring ered as an incremental planning instead 2020), including in the agrarian sectors of radical planning due to limited resource (Pawlak and Kołodziejczak 2020, FAO or power. HKm schemes contribution on 2017) that commonly susceptible to the forest recovery can be indicated by the socio-demographical dynamics, market improved coverage forested area in Lam- pricing commodity shock, environmental pung Province. disturbances, as well as macroeconomic According to Seno et al. (2018) in this or international policy changes as in Indo- province there has been forest cover im- nesia (Putra et al. 2021). The phenomena provement from 39,380 to 129,136 ha be- is also experiencing by almost all forest tween 2001 and 2014. This improvement management programs in Indonesia in- occurred particularly in the state own cluding forest recovery from encroach- production forest and protected forest ment called HKm scheme (Wulandari and areas. Whereas the conservation forests Inoue 2018). Forest encroachment itself or national parks were almost free from was rampant during Indonesia underwent encroachment due to very tightly guard- governmental reformation from authori- ing activities. The improvement of forest tarian to democratization regime followed coverage areas, further can contribute to by forest decentralization to local govern- the environmental services performance ment in the period of 1998 to 2001. Ac- including microclimatic amenity, enhance cording to Kelman (2013) during the peri- ecological equilibrium, and at least to re- od, especially in areas controlled by cen- duce infectious or zoonosis diseases in- tral government, such as protected forest cidence such dengue hemorrhagic fever areas, people perceive that they were free (Seno et al. 2018, Mustika et al. 2016), to act without any lawsuit of punishment. malaria (Wigaty et al. 2016), pulmonary Besides, the poverty, meanwhile, accom- tuberculosis (Rosari et al. 2017), pneu- panied by lacking of land holding as well monia (Adhyaksa et al. 2017), and avian as the limited government apparatus for influenza (Rohayati et al. 2018). controlling state own forest region were This significant progress of forest re- the major triggering on the encroachment covery that also contributed by HKm phenomenon. scheme, however, has never been exam- According to Sanudin et al. (2016) ined for sustainability yet. It is important deforestation reached 54.56 % in Lam- to note that the sustainability indicator is pung Province, the province which most dynamic, always changes and even move severely experienced deforestation in In- down. The sustainability HKm program, donesia during reformation and govern- therefore, must be nurtured, managed, mental decentralization period. So that and pursued by fostering some of deter- this province was the first established mined variables. The important and rele- HKm scheme. Under the schem, forest vant ones commonly used are both eco- encroachers recruited as HKm member nomic and tree biodiversity performance for 35 years, permitted to manage their of every parcel of forest land as the indi- 212 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi cators of sustainability of HKm programs ket place (Idayanti et al. 2019). Besides, in Batutegi and Liwa Forest Management the skills and knowledge land acreage, the Unit (Ruchyansyah et al. 2018, Wulandari ownership of hand phone, grocery, fish- et al. 2018). According to some research- pond, and motorbike will become produc- ers both indicators are trade off in a forest tive capital that determines their productiv- ecosystem. That condition was true for ity and will shed out in the form of their in- conventionally forest management in past, come later on. Additionally, the exogenous but not for agroforestry system currently. variable including the availability of rural Agroforestry system actually can afford public investment as well as the social both economic and biodiversity at once. safety net maybe impact to their income. The strata of vegetation are stake, pole, Triggering by some incentive that and seedling can be grown bellow the possibly could be generated by the eco- wood stumpages in it. This stratified tree nomic benefits from miscellaneous agro- trunk system means to compose tree bio- forestry yields, it is normal to expect that diversity on one hand and can afford eco- the farmers will add various tree wood nomic benefit for the farmers (Puspasari species voluntarily on their land, particu- et al. 2017). Among them that can afford larly whenever their income can be pro- a highly economic benefit is for example moted by fostering the HKm authority. In cinnamon (Cinnamomum verum J. Presl), line with this background, it is needed to coffee (Coffea canephora Pierre ex A. build model prediction of farmer income Froehner) or cocoa (Theobroma cacao L.) as the function of both the endogenous for the stake and pole stratum respective- and exogenous variables and then the ly. There are so many vegetation species income needs to predict tree biodiversity of seedling stratum that can afford highly performance as well as the control HKm economic benefit as well as to contribute program sustainability. This research, the biodiversity enhancement, i.e. soy- therefore, was conducted with the aims of: bean, peanut, some vegetables, aromatic (i) Modeling HKm members’ INCM based or medical herbs, etc. (Bakri et al. 2018). on their endogenous and exogenous var- Applying agroforestry system for HKm iables, (ii) Modeling BIODV performance members, therefore, is a prominent way to based on INCM, and (iii) Modeling HKm meet their obligation for forest recovering sustainability program as the function of besides providing for their family income. BIODV. These three series model will be- But, the farmers’ ability certainly will be come reliable tools in planning to foster constrained by some endogenous and ex- HKm Schemes sustainability endeavor in ogenous variables. Their stock of knowl- Indonesia. edge and skills, their physical capital, or productive asset available are the endoge- nous variables that are possibly controlled Material and Methods internally by HKm management. The skills and knowledge level, however, are com- This study consists of field survey and monly affected by their socio-demographic data analyses, from February to May variables including age, sex, family size, 2018. The field survey was conducted education, culture or ethnicity, group in- at Talang Mangga Village, Banjit District, volvement, participation in extension, ac- Way Kanan Regency, Lampung Province, cess to information, and distance to mar- Indonesia (Fig. 1). Data were collected Fostering the Sustainability of Community Forestry Program: Case Study ... 213

Fig. 1. Research location (GWKR 2016). through interviews using questionnaires Gauss 1777–1855 as cited by Lunsford et guidance to 230 HKm members. The clus- al. 2006, Dinov et al. 2008, Ghasemi and ter random sampling method on a popu- Zahediasl 2012, Sang and Hae 2017) if lation of HKm Mangga Jaya group mem- the sample data size is large enough, it bers totaling 697 households and com- will follow a normal distribution regardless posed of 9 sub-groups. The sub-grouping population distribution form. Sang and was based on block area that may consist Hae (2017) and also Ghasemi and Zahe- of one or more villages. The division fol- diasl (2012) have proven a sample size of lowed the list that has been made since >30 is the golden rule to meet normal dis- the beginning of HKm membership was tribution, while according to Lunsford et al. formed. Each sub-group had about 73 to (2006) and Dinov et al. (2008) the golden 76 households as members. Each one number of minimum size is 25 members. was treated as cluster. We randomly drew samples of 25–26 households in each sub-group as poten- Model approach and testing tial respondents so total samples were hypotheses 230 households, and we named this 230 respondents. As to the decision about In accordance with the aim of this research 25–26 respondents as the minimum size there are three measured response vari- of sample per sub-group is intended of ables: 1) farmers’ income of agroforestry having normal data distribution. According yield (INCM), 2) tree biodiversity indices to the Central Limit Theorem (Johann Karl (BIODV), and 3) sustainability of HKm pro- 214 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi gram indices (SUST). As for the surrogate therefore, should be counted on every of three are explained in the following. community development plan included in the fostering to behave more efficiently in earning from agroforestry yields (INCM). Model I: Farmer income The earning, in turn, will be the basic for Farmer INCM is intended to depict their stimulating HKm member to plant wood- welfare under HKm program. In this re- en tree crops other than rubber. INCM search, ECNM is measured by using the modeling functioning as the endogenous total income per family per annum, the in- and exogenous variables of household in come from cultivation on HKm land only, detailed is summarized in Table 1. Model such as latex (rubber sap), coffee bean, I as the INCM prediction is expressed in areca nut, etc. Increased INCM is expect- formula (1): ed to become an important variable in de- INCM = α + α AGE + α SEX + α FMLSZ termining the improvement of BIODV of i o 1 i 2 i 3 i + α VDAM + α GADM + α AJOB + every land parcel which belongs to HKm 4 i 5 i 6 i α CLVTN + α D1_ELS + α D1_JHS + members. BIODV itself is a good indicator 7 i 8 i 9 i α D2_SND + α D2_SMD + α D2_LPG of the forest ecosystem recovery process 10 i 11 i 12 i + α LHLD + α UPLN + α RICEF + particularly agroforestry system instead of 13 i 14 i 15 i α GOAT + α FPOND + α MBIKE + monoculture cropping pattern. 16 i 17 i 18 i α GCERY + α TVOW + α HPOW According to Idayanti et al. (2019), 19 i 20 i 21 i + α EXTN+ α ELVT + α DLND + INCM is significantly influenced by age, 22 i 23 i 24 i α DVLM + α DDST + α RINVT + sex, and number of dependents. While as 25 i 26 i 27 i α SCNET + α NURSY reported by Puspasari et al. (2017), there 28 i 29 i + €, (1) are no real differences between ethnic- i ities in earning income. Setiawan et al. where: INCM is income from agroforestry

(2014), however, proved that in this study yield of respondents; α0 is intercept; α1-29 area Javanese ethnicity was more resil- are parameter Model I; € is residual error ient in increasing their income than ethnic Model I; i=1, 2, …, 230 are respondent Lampung and others in making the living. numbers; the other symbols correspond Meanwhile, in the local forestry institu- to the symbols in Table 1. tion namely KPH Bukit Punggur (2014) The working hypothesis Model I is ex- Strategic Plan document, it is stated that pressed in the following: the HKm members in this research area - Ho: α1 = α2 = α3 = … = α29 = 0 (among are composed of Javanese, Sundanese, the 29 variables enlisted in Table 1, there Lampung, and Semendo ethnic groups. is no variable that affect the family’s in- The first two are immigrants while the oth- come significantly); ers are native who are very adaptive to - H1: α1 ≠ α2 ≠ α3 ≠ … ≠ α29 ≠ 0 (among agroforestry cultivation legacy from their the 29 variables enlisted in Table 1, at ancestor. The ethnicity can be considered least there would be one variable that af- as the surrogate of a traditionally cultural fects the family’s income significantly). bundle that have accumulated day by day The testing hypothesis and optimiza- along history. The different ethnicity can tion parameters process for Model I was be utilized for contrasting skills and local conducted by applying Minitab 16. The knowledge in relation to their productivity Fisher statistic employed to examine the in agroforestry cultivation. This variable, goodness fits test for the Model I at confi- Fostering the Sustainability of Community Forestry Program: Case Study ... 215 - - : who applied agroforestry have miscellaneous yield so that more secure from cropping failure as well as from pricing market shock : will bring other source of income : will earn more access to information and power : will earn more access to information and power : the more person income : the more total land holding income : the more up land in HKm area income : the older lessen their productivity will be : man more powerful in cultivation than woman : Semendonese more adaptive than Javanese : graduate ELS more innovative then who never schooling : Lampungeses the most adaptive in agroforest : graduate JHS more innovative then who ELS : Sundanese less adaptive in agroforest than Ja vanese + + + + + + + Expected Sign in Regression: Brief explanation in Regression: Brief explanation in Expected Sign relation to the income per family - + + + + + - - Scoring =1 if agroforestry; =0 if monoculture =1 if any =1 if administrator of HKm =1 if village adminis trator integer number rational number rational number integer number =1 if man; =0 other =1 if Semendonese; =0 if others =1 if elementary; =0 if others =1 if Lampungnese; =1 if =0 if others =1 if junior high sch.; =0 if others =1 if Sundanese; =0 =1 if if others - - - - ha ha pected sign in regression result. year person Unit or dummy dummy dummy dummy dummy measured data scale categorical _ 2 _JHS _ELS _LPG D _SND SEX 1 1 AGE 2 SMD 2 LHLD AJOB UPLN VADM GADM CLTVN FMLSZ D D D Symbol D

Education level* (0 = never schooling) 8. Dummy elementary school 7. Cultivation applied 6. Additional job 5. Role in HKm group 4. Role in village administration 14. Up land acreage inside HKm area 2. Sex 3. Family size number 12. Lampungnese 9. Dummy junior high school Productive assets* land holding acreage Total 13. Ethnicity* (0 = Javanese tribe) 10. Sundanese 11. Semendonese 11. Group of variable and the variables Social demographic* 1. Age Table 1. The endogenous and exogenous variables of family head their data scoring in model employed as well the ex Table 216 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi ------: the plot will affect positively to the farmers skill : the plot will affect and their income : social safety net also will affect positively to in - : social safety net also will affect come : public investment activities project such as rural time and nega - road improve will make trade off tively impact to agroforestry activities : the higher the more effort and the less income : the higher more effort : the farther distance more time or transporta tion cost needed : the farther distance more time or transporta tion cost needed : the farther distance more time or transporta tion cost needed tive : who participate in extension have more produc : who own TV will have more productive informa : who own will have more productive informa : who own HP tion tion : who owe small grocery will earn more income : who owe motor bike will earn more income : who owe fishpond will earn more income : the more rice field in HKm area income : the more number goat income + + - - - - - + + + + + + + + Expected Sign in Regression: Brief explanation in Regression: Brief explanation in Expected Sign relation to the income per family Scoring =1 if any; =0 other =1 if any; =0 other =1 if any; =0 other rational number rational number rational number rational number =1 if active =0 oth - er =1 if own; =0 other =1 if own; =0 other =1 if own; =0 other =1 if own; =0 other =1 if own; =0 other integer number rational number - - - - ha ha bike bike bike m a.s.l. Unit or measured minutes by minutes by minutes by categorical categorical categorical categorical categorical data scale ELVT EXTN DDST DLND GOAT DVLM RINVT TVOW RICEF HPOW MBIKE SCNET NURSY FPOND GCERY Symbol

Note: * endogenous variables, ** the exogenous variables. 29. Nursery demonstration plot 28. Social safety net 24. Distance to land area 25. Distance to village market 26. Distance to subdistrict centre Beneficiary from rural facility invest - ment** 27. Rural facility investment 23. Elevation of land area 21. HP ownership 21. HP 22. Extension participation Information and physical accessibility* 20. TV ownership 19. Small grocery shop ownership 18. Motorbike ownership 16. Goat number ownership 17. Fish pond ownership Group of variable and the variables 15. Rice field inside HKm area Fostering the Sustainability of Community Forestry Program: Case Study ... 217 dent level of 90 and 95 %. Then to depict ab 16. Gald statistics was employed to every parameter of predictor variables we test the goodness fits of model. Whereas employed the T statistic at confident level the significance of parameter model was of 90 and 95 % as well. tested by employing Wald statistics. Both Gald test and Wald test were employed at confident levels of 90 and 95 %. Model II: Tree biodiversity prediction model Model III: HKm program sustainability As mentioned before Model II is intend- prediction model ed to predict the probability of tree biodi- versity improvement of every land parcel Model III is intended to predict SUST by belonging to HKm members. Model II is using single predictor variable of BIODV. predicted by single variable of INCM. It is The model III is expressed in the formula expected to express that HKm members (3): will add more tree species voluntarily at p(SUSTi)= 1 their land whenever their INCM is risen Ln = up. In order to depict this objective, we 1−= p(SUSTi) 1 employed the Loglinear Regression or Lo- = £0 + £1BIODVi + πi, (3) gistics Regression that expressed in the where: Ln and i are as in formula (2); p(- formula (2): SUSTi)=1 is the probability of respondents th p(BIODVi)= 1 – i succeeded to support HKm program Ln = sustainability; 1–p(SUST) = 1 is the prob- 1−= p(BIODVi) 1 i ability of respondents – ith fail to support = β + β INCM + ξ, (2) 0 1 i i HKm program sustainability; £0 is inter- where: Ln is logarithm operator using the cept; £1 is a parameter Model III; πi is re- natural number (e = 2,718...) as the basis; sidual error of Model III. p(BIODVi) = 1 are the probability of being The working hypothesis Model III is ex- succeeded in improving tree biodiversity pressed in the following: th of the respondent i ; 1–p(BIODVi) = 1 are - H0: £1 = 0 (tree biodiversity perfor- probability of being fail in improving tree mance does not affect the sustainability th biodiversity of respondent i ; β0 is inter- program significantly); cept; β1 is parameter Model II; ξi is residual - H1: £1 ≠ 0 (tree biodiversity perfor- error Model II; i = 1, 2, …, 230 are respon- mance affects the sustainability program dent numbers. significantly). The working hypothesis of Model II is Also in Model III, a positive number of expressed in the following: parameter £1 was expected that would be

- H0: β1 = 0 (family income does not af- produced by optimization process using fect tree biodiversity performance signifi- statistical software of Minitab 16. Gald cantly); statistics was employed to test the Good-

- H1: β1 ≠ 0 (family income affects tree ness fits of the model. Whereas the signif- biodiversity performance significantly). icance of parameter model was tested by

For H1 above, a positive number of pa- employing Wald statistics. Both Gald test rameter β1 was expected that would be and Wald test were employed at confident produced later on through optimization levels of 90 and 95 %. process using statistical software of Minit- 218 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi

Results and Discussion The distribution BIODV is described in Table 3. From this table we can appraise Result in descriptive statistics for forestry workers, who have been striv- ing their effort to recovery forest degra- Very firstly we need to describe three dation (Bakri et al. 2018, Chamberlain et variables that were used as the respond al. 2019). It is the fact that for 11 years variables in the three models regression, HKm’s program operating only did 1.7 % namely [INCM], [BIODV], and [SUST]. or 4 parcels of land covered by the high Having displayed the descriptive statis- BIODV. The dominant is still 54.4 % (125 tics of their respond, the three modeling parcels) – low in tree biodiversity perfor- results will be provided consecutively mance (BIODV). This fact can be treated later on. It is important to note from data as the bench mark in planning to speed up gathered of the field survey, that lowest the forest recovery program under HKm [INCM] was USD 10 and the highest was scheme as well. USD 1,608 per year. Briefly distribution of [INCM] is provided in Table 2. Based on Table 3. Proportion of families based on this term, the dominant class [ICNM] was tree biodiversity. less than USD 300 per year, with the fre- Tree biodi- Fre- Proportion, No quency of 143 families (62.2 %). This fact versity class quency % can be treated as the bench mark in plan- 1 Low 125 54.4 ning to enhance INCM from agroforestry 2 Moderate 101 43.9 earning that have to be fostered by HKm 3 High 4 1.7 scheme. Table 2. Proportion of families based on The third respond variable is classified their income (n=230). as yes or no in supporting HKm’s SUST. The distribution of this data is provided in Income Fre­quen­ Propor- Table 4. It is important to note that the cri- No class, cy tion, % USD/year terion for separating into two classes (yes 1 <300 143 62.2 versus not) is the compliance of HKm’s 2 301–600 38 16.5 members to plant the additional wood- 3 601–900 31 13.5 en tree crops on their land even though 4 901–1200 9 3.9 without any financial assistant from HKm authority or from other sources. In case 5 1201–1400 7 3.0 a member comply is ‘yes’ – to support 6 >1400 2 0.9 the probability of HKm’s sustainability Mean= Minimum Maximum Sd=340 351 =10 =1,608 program. The revers is as ‘no’. This dis- tribution is depicted in Table 4. The table As for [BIODV] is defined by addition displays more than half of the families do the wooden tree crop that have planted not support yet the probability of HKm’s and successful to grow, the number ad- SUST. This fact should be taken into ac- dition of wooden tree then are grouped in count in developing programs to foster it. three classes BIODV namely: low, mod- For the sake of pursuing the guidelines erate, and high. The moderate class is to foster SUST, we need to examine the defined in between 3 to 5 species added, relationship among the three variables while beyond both boundaries are low consecutively in series models regres- and high classes of BIODV respectively. sion. Having known which ones among Fostering the Sustainability of Community Forestry Program: Case Study ... 219

both endogenous and exgenous variables to reject H1 (that also means there is no

(Tabel 1) that significantly affect INCM, enough facts to accept H0). In other words, we then can estimate or predict BIODV Model I is valid or robust predictor model performance. We, furthermore, use the of INCM based on both endogenous and predicted BIODV to examine the SUST. exogenous variables of every family at the The three regressions model is presented study area. In addition, P = 0.000 does not below. mean that this predictor model can predict Table 4. Proportion of families who sup- 100% correctly or without any making port the probability of HKm’s sustainability miss prediction, but only below 1 % and program. maximum 0.04 % (=0.0004) miss predic- Family who Frequen- Proportion, tion. It is important to realize that there No support cy % is no breaking any mathematical rules in 1 Yes 87 37.8 case we write this P = 0.0004 because 2 No 143 62.2 this number is approximately the same with the original number namely P=0.000 as the output of Minitab process. But Result of the regression models P = 0.0004 can help the interpretation Model I is to explain the role of both the about the robustness Model I obtained. endogenous and exogenous variables It tells us that if we applied the model for in determining family income from agro- predicting INCM based on the 29 predic- forestry yield (INCM) in HKm land. The tor (Table 1, the endogenous and exoge- goodness of fits test result for Model I ap- nous) variables to predict 10,000 families, plied is provided in Table 5. Meanwhile the we will get very high precision of predicted optimization parameter of variables affect- INCM, i.e. there will be maximum 4 fam- ing INCM can be seen in Table 6. ilies that miss predicted than the actual By referring to P = 0.000 (Table 5), it ones. Briefly, the confident level of Model means that we have had no enough facts I is more than 99 %. Table 5. Analysis of variance of role independent variables on family income. Source Degree of freedom Sum square Mean square F P Regression 33 18354608 632918 15.60 0.000 Residual error 195 8073640 40571 Total 228 5404.31

Note: S = 201.423; R2 = 69.5 %; R2-adj. = 65.0 %. S = root square of total variances, R2-adj= R2 adjusted to sample size; F = Fisher statistic value; P = Probability to miss prediction. Table 6. Parameter optimization of socioeconomic variables that affected family income.

Predictor Symbol Coeff. SE Coeff. T P Constant - 330 1.24 0.215 Social demography 1. Age AGE -0.413 1.203 -0.34 0.731 2. Sex SEX 27.350 55.049 0.50 0.620 3. Family size number FMLSZ 29.441 9.692 3.04 0.003 4. Role in village administration VADM -0.503 60.294 -0.01 0.993 5. Role in HKm group GADM -855.245 1283.217 -0.67 0.506 220 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi

Predictor Symbol Coeff. SE Coeff. T P 6. Additional job ADJOB -4.385 46.455 -0.09 0.925 7. Cultivation applied CLTVN 841.259 1134.266 0.74 0.459 Education level (0 =never schooling)

8. Dummy elementary school D1_ELS 71.545 54.972 1.30 0.195

9. Dummy junior high school D1_JHS 7.965 57.951 0.14 0.891 Ethnicity (0 = Javanese tribe)

10. Sundanese D2_SND 114.448 41.203 2.78 0.006

11. Semendonese D2_SMD 130.476 48.189 2.71 0.007

12. Lampungese D2_LPG 51.818 1108.392 0.47 0.640 Productive assets 13. Total land holding acreage LHLD 278.203 18.748 14.84 0.000 14. Up land acreage in Hkm area UPLN 9.874 47.755 0.21 0.836 15. Rice field inside HKm area RICEF -64.881 61.776 -1.05 0.295 16. Goat number ownership GOAT -4.154 33.112 -0.13 0.900 17. Fish pond ownership FPOND 18.559 33.580 0.55 0.581 18. Motorbike ownership MBIKE -35.692 37.804 -0.94 0.346 19. Small grocery shop ownership GCERY -24.622 40.105 -0.61 0.540 Information and physical accessibility 20. TV ownership TVOW -12.063 39.797 -0.30 0.762 21. HP ownership HPOW -34.566 45.140 -0.77 0.445 22. Participation in Extension EXTN 166.790 35.622 4.68 0.000 23. Elevation of land area ELVT -0.503 0.175 -2.88 0.004 24. Distance to land area DLND -1.601 5.175 -0.31 0.758 25. Distance to village market DVLM -1.601 5.175 -0.31 0.758 26. Distance to district centre DDIST -4.021 3.797 -1.06 0.291 Beneficiary from rural facility investment 27. Rural facility investment FIVST -171.825 45.531 -3.77 0.000 28. Social safety net SCNET 68.657 63.245 1.09 0.279 29. Nursery demonstration plot NURSY -199.587 40.685 -4.91 0.000

Note: the bold numbers are significant at level <1.0 %; T=the critical value of T statistic.

As for which ones among the 29 vari- position (ELV), rural investment facility ables that affect ICNM significantly, we (FISVT) and nursery development plot must examine their probability to miss (P) project (NURSY). In other words, against in Table 6. It connoted that there are eight these eight variables, there is no enough variables affecting INCM significantly with facts to reject H1 (or all at once to accept

P < 1 % (or confident level more than H0). Additionally, that the last two vari- 99 %) namely the variables of the family ables mentioned happen to be the exog- number size (FMLSZ), the ethnicity where enous variables where the others six are the Sundanese (D1_SND) and Semen- the endogenous variables. Fortunately donese (D1_SMD) compare to Javanese, the effect of the six endogenous ones are land holding acreage (LHLD), participa- in the same direction with the INCM, the tion in extension (EXTN), land elevation better quality of each variable the better Fostering the Sustainability of Community Forestry Program: Case Study ... 221

INCM will be. On the contrary, in case the BIODV. In case INCM at average rises activities of both FISVT and NURSY exist, up as much as one USD per year, BIO- the INCM will reduce significantly. DV will improve by 1.13 as indicated by its As for the 21 variables remain on Odds Ratio and this improvement is very the contrary, there is no enough facts significant that indicated by its P-value = to reject H0 (or all at once to accept H1). 0.000 or with confident level more than All these variables have any significant ef- 99 %. fect to INCM. As mentioned before, Model Developing Model III is motivated by II is intended to explain BIODV based on seeking some clues to compose a guide- INCM. The goodness fits test of Model II line for fostering every HKm member in as well as the significance effect of INCM order to contribute the probability increase on BIODV is provided in Table 7. This of HKm’s SUST. The goodness fits test of table depicts that there is no enough facts Model III as well as the significance effect to reject H1 (and all at once to accept H0). of BIODV as the predictor of SUST is pro- In other words INCM affects significantly vided in Table 8. Table 7. Wald test of Model II: effect of income to BIODV. 95% confident in- Odds Predictor Symbol Coefficient SE Coeff. Z P terval (CI) Ratio Lower Upper Constant - -0.330442 0.200843 -1.65 0.100 - - - Income INCM 0.122602 0.034123 3.59 0.000 1.13 1.06 1.21

Note: Log-Likelihood = -149.873; Test that all slopes are zero: G = 15.685, DF = 1, P-value = 0.000. G = Gald statistic; Z = Coeff./SE coefficient; P = probability to miss of separately predicted

β0 or β1 numbers; P-value = probability to miss of predicted BIODV using ECNM.

Table 8. Wald test of Model III: effect of BIODV to HKm SUST. 95% confident Odds Predictor Symbol Coefficient SE Coeff. Z P interval (CI) Ratio Lower Upper Constant - -0.138836 0.199487 -0.70 0.486 - - - Tree Biodi- BIODV -0.660920 0.275737 -2.40 0.017 0.52 0.30 0.89 versity

Note: Log-Likelihood = -149.637; test that all slopes are zero result: G = 5.801; DF =1; P-value

= 0.016. G = Gald statistic; Z = Coeff./SE coeff. P = probability to miss of separately predicted £0 or £1 numbers; P-value = probability to miss of predicted SUST using BIODV. Table 8 indicates that there is no enough moderate or to high levels, the probability facts to reject H1 (all at once to accept H0). HKm program become sustain (SUST) It means that BIODV is a reliable variable would be reduced by 0.52 as indicates by to predict HKm’s SUST. But unfortunate- its Odds Ratio with confident level is more ly, in case the BIODV improve from low to than 98.3% as indicated by its P=0.017. 222 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi

Discussion the family heads each year, however, the effect is not significant (P = 0.731). This Role of demographic variables on reflects the agricultural work in the region income is resilient enough to change in age of the farmers. So are the effects of the variables Among the demographic variables, only of VADM, GADM, and ADJOB. INCM FMLSZ variable affects INCM, if the oth- of the family heads involved VADM are er variables are constant, a family who commonly lower than those free from that has more than one family member will affair, but the difference is not significant have an increase in income around USD (P = 0.993). It is similar to GDAM, with 29.441 per year. The effect of addition of P = 0.506. Both cases suggest that vol- labour supply (family members) to the re- untary work does not significantly disturb gion is very significant (P = 0.003). This families in making a living. An interesting phenomenon means that the marginal fact there is the effect of ADJOB such as labour productivity in the region is still construction labour, pedicab drivers, etc. positive in stimulating INCM. Because The families having ADJOB tend to lose the core business of HKm authority is to their opportunity to intensify their agricul- release this protected forest from human tural work and have lower income than occupation, it is a must to employ policy those not having ADJOB. to enhance labour skills by extension or In short, FMLSZ is the only demo- training program instead of mobilizing graphic variable that can enhance INCM. labour from outside area (Idayanti et al. In development practice, allocating more 2019). Training in silviculture techniques manpower to this region can be the only cycle from seedling to post harvesting will opportunity available for forest planners to make labour augmented by knowledge increase economic performance as a nec- and technology, in turn will increase their essary condition for enabling sustainable work productivity and then lessen the ex- development besides the biodiversity of cess demand of traditional labour (Supria- any landholding plots. For the sake to find di et al. 2018). This finding is congruence some opportunities in enhancing the eco- with the effect of LHLD that will be elabo- nomic performance, we need to examine rated below. other variables that give significant effects The other variables of demographics, on INCM as also depicted in Table 3. namely: age, sex, role in village admin- istration, role in a HKm group, additonal Effect of education level on income job, and cultivation applied do not affect significantly INCM. As depicted in Table It is normal to expect an educational back- 5, there is no significant effect of SEX, ground influence on people’s productivity as INCM is relatively the same. Similar in economic sectors, and so is it in the to SEX is the variable of CLTVN, a fam- agroforestry sector in the study area. The ily applying an agroforstry pattern indeed higher people’s education background the has a higher INCM than that using mono- higher their innovativeness, that will be culture, but the difference is not significant accompanied by their higher productivity (P = 0.459). reflected in their income. This expecta- Another AGE variable shows that tion, unfortunately, cannot be achieved in INCM will decrease with increasing age of this study. As referring to Table 5, the level Fostering the Sustainability of Community Forestry Program: Case Study ... 223 of education of family heads does not sig- On the contrary, INCM of Sundanese nificantly affect ICNM. Family heads with reaches USD 130.475 per year more than an Elementary School education level of Javanese. The difference is significant

(D1_ELS) have an income of USD 71.545 (P = 0.006 = 0.6 % or <1 %). It seems compared to those who never schooling. that Semendonese people are the most The difference, however, is not significant productive in the study area. The fami-

(P = 0.195 = 19.5 % or > 5 %). The similar ly income of Semendonese (D2_SMD) is effect occurs with family heads catego- USD 114.448 per year more than of Ja- rized as Junior High School (D1_JHSC). vanese. The difference is also significant

Although the family heads of D1_JHS cat- (P = 0.007 = 0.7 % or <1 %). egory have an extra of USD 7.965 per It is interesting to discuss the differenc- year also compared to those, the differ- es in INCM based on the ethnicity analy- ence is not significant (P = 0.891). sis. Setiawan et al. (2014) revealed that In brief, the 3 levels of education back- Lampungese and Semenodenese in the ground of the family heads have no signifi- study area could be regarded as the na- cant impact on ICNM. It can be concluded tive ethnic groups whereas Javanese and that the education has not yet significantly Sundanese were the newcomers. The stimulated innovativeness so that there natives commonly cultivate cash crops, are no significant effects on ICNM either. including coffee, cocoa, cinnamon, and The phenomenon, however, will be very rubber through the agroforestry system. useful for forest planners eager to look Meanwhile, according to Nurhaida et al. for an INCM elevating strategy, at least (2007) both Javanese and Sundanese that the forest planners will not need to migrated from Java Island (or their off- separate the family heads based on the springs) started in 1905 under Dutch Col- education level in conducting extension onization and last recurrent in between activities, etc. The extension program it- 1961 and 1965 under the National Re- self gives a significant effect on INCM, as construction Scheme. Newcomers do not discussed below. do much cash crop cultivation through the agroforestry system. Instead, they do the perennial crops cultivation especially Effect of ethnicity on income rice. Many of them, in fact, have no back- Contrary to the educational level, the eth- ground in any agricultural livelihood at all. nicity affects INCM significantly. The fam- From this background it is easily under- ily heads in the study area belong to 4 stood why Javanese come last in INCM ethnic groups: Javanese, Sundanese, Se- from rubber agroforestry cultivation. Re- mendonese, and Lampungese. In order to versing the argument, we have proven make a comparison of the magnitude ef- why Semendonese (D2_SMD) is the most fect among them, Javanese becomes the productive ethnic group in the study area, reference. The scoring applied for them shown by their INCM. equals to 0 in ICNM modeling (Table 1). We would be trapped in a wrong argu- The results, in Table 5 show that Lam- ment if we insist on using the origin of peo- pungese (D2_LGP) indeed have an aver- ple to explain Sundanese or Lampungese age ICNM of USD 51.818 per year more productivities. Although Sundanese are than Javanese do. But the difference is the newcomers, they commonly live side not significant (P = 0.640 = 64 % > 5 %). by side with Semendonese. Like them, 224 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi the ancestors of Sundanese in West Java agroforestry pattern demonstrated by Se- commonly lived in the upper area of a mendonese or Sundanese should be in- landscape, cultivated rice on terraced troduced to Javanese or Lampungese by land combined with fish ponds and wood a subtle extension methods in order to be plants in their backyard. The agrocomplex accepted without any cross culture barrier. system pattern in Sundanese commu- Additionally, the relevant custom in- nity was found during the field research. cluding norm, habit, satire, allusion, folk- Except for cultivating rice and managing lore and other cultural acceptability or fish ponds, both Sundanese and Semen- heritage from their ancestor were used donese prefer to live in an area biophys- exhaustively in Nurhaida et al. (2007, ically very similar so they commonly live 2011). Because the participation in ex- side by side. Sundanese, however, are tension (EXT) is positively significant in more intensive in adopting skills for the enhancing INCM (Table 5), it is a rational cultural technique of rubber agroforest- way in using up the strategic program in ry in daily work. This background can order to become the strategy for fostering serve as an appropriate explanation why HKm’s SUST through improving BIODV. Sundanese are more productive than All of the culture heritage of every tribe Javanese. A reverse argument perhaps such as hospitality, local norm, habit, local could explain better why the productivity wisdom, should be taken as consideration of Lampungese is not different from Ja- in developing extension programs. vanese as expressed by their average INCM. The implication in developmental Effect of productive asset economic planning in relation to the four ethnic groups planners should take into Among the 8 variables of productive as- consideration whenever they convey the sets (Table 3) only landholdings under technical assistant in extension activities. HKm scheme acreage have a significant This argument is in line with the find- effect on INCM. This claim is connoted by ings of Nurhaida et al (2007) who proved P-value of 0.000 which means this vari- the local language usage in extension able significantly influences the income activities was very effective in enhancing because it has an error below 5 %. Its farmers’ knowledge about the role agro- coefficient is 278.203 which means that forestry on soil and water conservation the family income will increase by USD control. Nurhaida et al. (2011) also con- 278.203 per ha per year on landholding ducted a research at buffer zone of Way plots under HKm scheme. This implies Kambas National Park-Lampung Province that in the study area a landholding is still and reported the use of three tribal lan- limited to achieve the community’s welfare guages (Javanese, Sundanese, Balinese) such as in of Lam- in cartoon media of fable. They were much pung Province (Ruchyansyah et al. 2018). more effective in conveying messages of This constraint can naturally encourage wildlife conservation than in Indonesian. further encroachment if the authority can- Beside the language, the local knowledge not make an incentive policy to slow this about the matter (that had been invento- tendency down. Thus, the challenge for ried before) also to was utilized in the ex- forestry planners is to look for another pro- tension media designed by Nurhaida et al. ductive asset that is possible to leverage (2007, 2011). In Figure 2 is described that INCM without presupposing any need to Fostering the Sustainability of Community Forestry Program: Case Study ... 225 increase land allocation. In relation to this ly economic activities that are possible to challenge, fish pond management, goat improve the economic performance due and poultry development are prospective- to their better economic gains.

a b

c d Fig. 2. Croping pattern in rubber silviculture: (a) Javanese, monoculture, seems deligent to clean weed as their ancessor who custome with annual crops in Java Island; (b) Semendonese almost an ideal agroforestry pattern; (c) Sundanese seems make adaptation to Semendonese; (d) Lampungese seems too litle cultivation.

The effect of information and physical elevation such as found by Zhafira et al. accessibility (2019). The rubber sap productivity will also decrease significantly if cultivated on Among the 4 variables, only the land ele- higher elevation due to the temperature vation gives a significant effect on INCM. suitability for rubber crop growth (Andrian Farmers whose land elevation lies above et al. 2014). 100 m a.s.l. will reduce their income by As for both TVOW and HPOW varia- USD 0.503 per year. Perhaps the farmers bles, they do not significantly affect INCM. must spend more effort or a bigger bud- Farmer participation in extension pro- get to cultivate rubber on land with higher grams, on the contrary, plays a significant 226 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi role in the enhancement of INCM. This et al. (2019) that the improvement of proj- kind of condition is based on the results ect facilities has a negative influence on of Idayanti et al. (2019) and Supriadi et al. the INCM. (2018). Both research results have also As for SCNET program it has no sig- proven that HPOW and TVOW have no nificant effect on INCM. It seems that the effect on the income. money allocation in cash has reduced the It is only the variable of farmer’s par- work spirit of the farmers. But based on ticipation in EXTN among the 6 ones of his research among pine (Pinus merkus- the information and physical accessibility ii Jungh. & de Vriese) sap tappers com- having a positively significant effect on munity at forest plantation in Central Ja- INCM. Farmers who are more active in va-Indonesia, Cahyono (2010) reported EXTN can get an income of USD 166.790 that SCNET in cash assistance had been per year, or more than those who are not. lessen the farmers’ morale on working, This proves that the extension program reduced their productivity and increased is an important source of information for consumption spending. On the contrary the villagers such as found by Wulandari was the research conducted in the com- and Inoue (2018) in West Lampung. The munity of sap damar (Shorea javanica knowledge contained in the extension can Koord. & Valeton) farmers at West Lam- be very valuable in rubber cultivation ac- pung Regency reported by Setiawan et tivities. It is also possible that the social al. (2014). SCNET in cash had stimulat- atmospheric interaction among the villag- ed their spirit in working and had induced ers has been improved by the extension their INCM enhancement. Both Cahyono activities so that the social capital can be (2010) and Setiawan et al. (2014) recom- developed as well. mend that enhance extension program in order to foster the farmer’s spirit to be able to empower themselves gradually escap- Effect of rural facility investment ing from poverty trap. activities These exogenous variables include The role of income on tree biodiversity FIVST, NURSY, and SCNET. The pres- performance ence of FIVST has decreased INCM, farm- ers joining the temporary work have an As seen in Table 7, G statistic is big enough average income of USD 171.85 less than (G = 15.685) with P-value = 0.000. This those who do not. Perhaps the farmers can be equated with P-value = 0.0004 be- spend too much time in the non farm work cause the rounding of 3 digits will be the so they sacrificed the opportunity to work same as P-value = 0.000. So the meaning in the fields, including plant care, tapping of P-value = 0.0004 (or P-value = 0.04 %) sap, processing latex, etc. A similar situ- is that 99.06 % (100 % subtracted by ation also occurs in NURSY. These jobs 0.04 %) of BIODV variations among the are more attractive because the villagers 230 respondents’ land parcels can be ex- can take cash than take money from sell- plained by the sole INCM variable, where- ing rubber latex, coffee bean, cocoa and as the remaining 0.004 % can not, but other agroforestry yields which the prices should be explained by other variables always decrease in the harvest season. that were not applied in this model. In This condition is also proven by Idayanti brief, the precision of this model is very Fostering the Sustainability of Community Forestry Program: Case Study ... 227 high because the confident level is more HKm due to its small missing probability than 99.04 %. of only 0.016 (=1.6 % or less than 5 %). As for how much INCM variable This number tells us that there will be 16 can multiply BIODV, we need to refer to units missing if we use BIODV variable to Odd Ratio obtained, namely 1.13 with predict 1,000 sustainability of HKm Pro- P = 0.000 that can also be equated to gram. Therefore we can rely on this model P = 0.0004 (Table 7). It connotes that ev- to predict the sustainability of HKm. ery farmer’s income increases by an av- We, furthermore, are interested in erage of USD per year, accompanied by knowing how big the change of sustain- increasing of BIODV by 1.13 times. The ability of HKm is in case BIODV rises from increase is around 95 % CI (namely in low to medium or high. Unfortunately, the between 1.01 and 1.26) with the miss- sustainability of HKm will drop to only ing probability being only around 0.06 %. 0.52 as BIODV rises from low to medi- This very small probability tells us that the um or high. This clue is indicated by Odd model built is strong. This fact also justifies Ratio described in Table 5, namely 0.52 that the improvement of BIODV through (CI = 0.30 to 0.89) with P = 0.017, indi­ ­ farmers’ income improvement in this re- cating the model is strong enough. search area is in line with the mission of The reduction in HKm program sustain- HKm Scheme. In other words, HKm Au- ability as a result of the increase in BIODV thority has conducted an amendment pro- gives a signal of a critical phase. It implies gram for rehabilitating the degradation of that farmers, whose BIODV has been ris- protected forests through farmers’ income ing up, need more nurturing from HKm increasement effort. It indicate that the re- authority to support a critical breakthrough forestation program under HKm scheme of a biodiversity development phase es- has been effective as proved by the re- pecially for newly wooden planted crops sponsiveness HKm members to the pro- other than rubber. During that phase the gram of planting some annual and cash farmers must pay much more attention in crops under rubber trees stumpage. It is order to promote the growth of the newly logical that farmers depend on forest’s planted wooden tree other than rubber. biodiversity for their daily needs so they Those plants must survive from weed dis- will manage their forest neatly and con- ruption, pests and borne diseases, poor serve the biodiversity (Wulandari et al. soil fertility, and even from rubber plant 2018). shading. The farmers need to allocate their resources more than when planting rubber in a monocultural way, such as The role of tree biodiversity controlling weeds and pests, applying soil performance on HKm’s program fertilizer, pruning rubber canopies and us- sustainability ing them for green manure. This certain- The BIODV effect is strong enough to be ly will reduce the farmers’ opportunity to the predictor variable of HKm Program earn income from the rubber sap produc- sustainability. As seen in Table 8, this tion. This claim is also in line with FMLSZ claim is proven by Gald statistic results at roles to INCM as expressed in Table 6. 5.801 with P-value = 0.016. It means that HKm authority, therefore, strongly recom- the sole predictor variable of BIODV is re- mended to conduct activities, including liable as the indicator of sustainability of technical assistance through intensifying 228 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi extension activities beside financial aids on voluntary planting some wooden trees in the form of chemical material and crop- is the prerequisite of pursuing SUST. ping facilities for farmers whose BIODV has been rising up; otherwise, BIODV will possibly decrease. Recommendations For the sake of pursuing SUST, it recommends to continue fostering HKm Three consecutive models have been members by managing the endogenous achieved by this research and are cer- variables in order to rise up INCM, and tainly very meaningful tool for developing then multiply BIODV as well as to broad- next programs, particularly the strategies en HKm members’ awareness on vol- in fostering every HKm member in order untary planting some wooden trees that to be able to contribute in pursuing HKm’s have been adaptive or endemic at local program sustainability (SUST). Model III ecoregion such as Fagraea fragrans Roxb is the final point and flash back to use (Bramasto and Sudrajat 2018), Cinnamo- Model II and Model I. At the ultimate ob- mum sp. (Menggala et al. 2019), and Sho- jective that the enhancement of tree bio- rea javanica (Wakhidah et al. 2020). diversity performance (BIODV) has to be able to stimulate the awareness of HKm’s member to add wood tree at every vacant Conclusion space of their lands. The extension activ- ities, including silviculture techniques, are The hypotheses have to be accepted at certainly a strategic way to broaden their confident level more than 95% namely: knowledge if planting some wooden tree (i) the endogenous variables with positive species will get some benefits of improve- effect on the income from agroforestry ment in: shading coffee or cocoa crops, lit- yield (INCM) were the family number size, ter basalt and mulch of organic matter, soil participation in extension activities, land fertility, and the opportunity to make earn- holding acreage, tribes Sundanese and ing from miscellaneous vegetable yields. Semendonese higher than Javanese, and Planting timber tree crops, therefore, can with negative effect was the land elevation; generate these economic incentive for the whereas the exogenous variable affected farmers. On the other side, planting many negatively the rural facility construction species of timber tree crops for HKm au- and nursery demonstration plot activities; thority is the main indicator of succeeding (ii) the INCM effect on tree biodiversity duties in forest recovery program from the (BIODV) enhancement namely BIODV encroachment phenomenon. performance will rise up by 1.13 if ENCM As proved by Model II, this scenario in rises up by USD one per year; and (iii) the fostering SUST through corner stones of BIODV on HKm’s program sustainability improving tree biodiversity (BIODV) will (SUST) will reduce by 0.53. It is clue that be possible as the prerequisite if and only the nurturing phase on HKm member at if the farmer’s income (INCM) can be ris- critical stage. It recommends that contin- en up. The flash back, therefore, we have ue fostering HKm members by managing to exploit the Model I as the basis for en- the endogenous variables in order to rise hancing INCM strategy. It is clear that we up INCM, and then multiply BIODV as well have to use the endogenous variables as to broaden HKm members’ awareness which have characteristic of enabling to Fostering the Sustainability of Community Forestry Program: Case Study ... 229 and being possible to induce INCM. From Lampung University, under The Scheme the six of endogenous variables (Table 4), of Graduate Research Grant, fiscal year only does the extension variable (EXTN) 2018. that meets the two characteristics, where- as the three variables remain do not, namely the family number size, land hold- References ing acreage, and land elevation. Family size number is the endogenous Adhyaksa A., Bakri S., Santoso T. 2017. The variable that does not have meaning for effect of land cover change on infant pneu- fostering SUST. This variable is not pos- monia incidence rata in Lampung Provice, sible to be added or reduced for the sake Journal Sylva Lestari 5(1): 26-34 (in Indo- of enhancing INCM. And so does the ele- nesian) DOI:http://dax.doi.org/10.23960/ vation of land parcel position, there is no jsl526-34 AndrianA., Supriadi S., Marpaung P. 2014. The real act that can be conducted on chang- Effect of Elevation and Slope on Rubber ing the land elevation position across the (Hevea brasiliensis) Production in PTPN III landscape in whole HKm location. Addi- Hapesong Farm of South Tapanuli. Online tionally, although the land holding variable Journal Agroekoteknologi 2(3): 981–989 is possible to be expanded in order to en- (in Indonesian). hance INCM, but adding the allocation of Bakri S., Setiawan A., Nurhaida I. 2018. Coffee land tenure to HKm members is contrary bean physical quality: The effect of climate to the main mission of the HKm Authority change adaptation behavior of shifting up namely to free the protected forest from cultivation area to a higher elevation. Biodi- versitas 19(2): 413–420. the occupation problem. So that the only Bramasto Y., Sudrajat D.J. 2018. Mor- strategic way in fostering SUST through pho-physiological characteristics differenc- INCM enhancement (that later on induc- es of leaves, fruits, and seeds of tembesu ing BIODV) is the promoting extension. (Fagraea fragrans Roxb.) from five popula- In order to achieve SUST, therefore, tions in Western Java and South Sumatra. the extension promotion is a very impor- [Karakteristik morfo-fisiologi daun, buah, tant strategy in fostering of every HKm dan benih tembesu (Fagraea fragrans member. Therefore, even though the ex- Roxb.) dari lima populasi di Jawa Bagian tension materials are focused on silvicul- Barat dan Sumatera Selatan] Jurnal Pe- nelitian Hutan Tanaman 15(1): 1–15. (in ture techniques, as exemplified by Nurhai- Indonesian). Available at: http://ejournal. da et al. (2007, 2011), the implementation forda-mof.org/ejournal-litbang/index.php/ of the activities of each extension program JPHT/article/view/2658/4259 must also consider the ethnicity variables, Cahyono A.S. 2010. The direct impact of cash which in this HKm Sundanese and Se- transfers and human resources investment mendonese need to be distinguished from to household economy surrounding pine for- Javanese or Lampungese. est in Samagede Village [Dampak bantuan langsung tunai dan investasi sumberdaya manusia terhadap ekonomi rumah tangga sekitar hutan pinus di Desa Samagede]. Acknowledgment Jurnal Penelitian Sosial dan Ekonomi Ke- hutanan 7(2): 101–115. (in Indonesian). We would like to offer acknowledgement DOI: 10.20886/jpsek.2010.7.2.%p for single sponsorship for the research Chamberlain D.E., Henry D.A.W., Reynolds C., grant Badan Layanan Umum (BLU) of Caprio E., Amar A. 2019. The relationship 230 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi

between wealth and biodiversity: A test of and Development Projects. Conservation the luxury effect on bird species richness in and Society 11(3): 247–263. the developing world. Global Change Biol- KPH Bukit Punggur 2014. Rencana Pengelo- ogy 25(9): 3045–3055. laan Hutan Jangka Panjang Kesatuan Dinov I.D., Christou N., Sanchez J. 2008. Cen- Pengelolaan Hutan Produksi (KPHP) Mod- tral Limit Theorem: New SOCR applet and el Bukit Punggur Kabupaten Way Kanan demonstration activity. Journal of Statistics Provinsi Lampung Tahun 2015–2024. Gov- Education 16(2): 1–16. Available at: www. ernment of Way Kanan Regency. amstat.org/publications/jse/v16n2/dinov. Lunsford M.L, Rowell G. H., Goodson-Espy T. html 2006. Classroom research: assessment of FAO 2017. The future of Food and Agriculture student understanding of sampling distri- – Trends and Challenges. Food and Ag- butions of means and the central limit theo- riculture Organization, of the UN. Rome. rem in post-calculus probability and statis- 180 p. Available at: http://www.fao.org/3/ tics classes. Journal of Statistics Education i6583e/i6583e.pdf 14(3): 1–28. Available at: https://doi.org/10. Ghasemi A., Zahediasl. S. 2012. Normality tests 1080/10691898.2006.11910587 for statistical analysis: A guide for non-stat- Meiring H.J. 2020. Thomas Robert Malthus, isticians. International Journal of Endocri- naturalist of the mind. Annals of Science nology Matabolisme 10(2): 486–489. 77(4): 495–523. https://doi.org/10.1080/00 Gleditsch N.P. 2021. This time is different! Or 033790.2020.182347 is it? NeoMalthusians and environmental Menggala S.R., Vanhove W., Muhammad D.R.A., optimists in the age of climate change. Hendri J., Speelman S., Van Damme P. Journal of Peace Research 58(1): 177– 2019. Sustainable Harvesting of Cinnamo- 185. DOI: 10.1177/0022343320969785 mum burmannii (Nees & T. Nees) Blume GWKR (Government of Way Kanan Re- in Kerinci Regency, Indonesia. Sustain- gency) 2016. Rencana Program In- ability 11(23): 6709–6726. https://doi. frastruktur Jangka Menengah Bidang org/10.3390/su11236709 Cipta Karya, Kabupaten Way Kanan. Mustika A.A., Bakri S., Wardani D.W.S.R. https://docplayer.info/198157561-Ren- 2016. Land use change in Lampung Prov- cana-program-infrastruktur-jangka-me- ince and its impact on incidence of dengue nengah-rpijm-bidang-cipta-karya-kabu- hemorrhagic fever (DHF). Jurnal Sylva paten-way-kanan-bab-2-profil-kabupat- Lestari 4(3): 35–46 (In Indonesian). DOI: en-way-kanan.html http://dx.doi.org/10.23960/jsl3435-46 Idayanti P., Bakri S., Wulandari C., Yuwono Nurhaida I., Harianto S.P., Syah P. 2007. De- S.B. 2019. Influence of Socio-economic signing of comics as entertainment media Characteristic to Income of Panca Tunggal and communication learning aids. [Meran- Community Forest Group [Karakteristik so- cang media hiburan buku cergam menja- sial ekonomi yang berpengaruh terhadap di media belajar untuk alat bantu komu- pendapatan Kelompok Hutan Kemasyar- nikasi]. Mediator: Jurnal Komunikasi 8(1): akatan Panca Tunggal]. In: Supriatna A., 51–64 (in Indonesian). Available at: http:// Cahyanto T., Widiana A. (Eds) The impor- dx.doi.org/10.29313/mediator.v8i1.1240 tance of Biodiversity and Biotechnology Nurhaida I., Setiawan A., Bakri S., Wiranta to Environmental Conservation [Peman- G.A.B., Syah P. 2011. Development of fable faatan Biodiversitas dan Bioteknologi untuk comics as communication media and envi- Pelestarian Lingkungan]. Fourth Biology ronmental education supplements to sup- National Seminar Proceeding. Universitas port biodiver- Islam Negeri Sunan Gunung Djati Press. sity conservation campaign. [Pengemban- Bandung: 174–180 (in Indonesian). gan komik fabel untuk media komunikasi Kelman C.C. 2013. Governance Lessons from dan suplemen pendidikan lingkungan da- Two Sumatran Integrated Conservation lam rangka kampanye pelestarian keane- Fostering the Sustainability of Community Forestry Program: Case Study ... 231

karagaman hayati di kawasan penyangga (in Indonesia). Taman Nasional Way Kambas-Lampung]. Sang G.K., Hae H.K. 2017. Central limit theo- Bumi Lestari Journal of Environment 11(2): rem: The cornerstone of modern statistics. 2527–6158 (in Indonesian). Available at: Korean Journal of Anesthesiology 70(2): https://ojs.unud.ac.id/index.php/blje/article/ 144–156. view/155 Seno Y., Bakri S., Wardani D.W.R. 2018. Utili- Pawlak K., Kołodziejczak M. 2020. The role zation of Landsat satellite imagery for the of agriculture in ensuring food security in valuation of the state’s forest environmen- developing countries: considerations in the tal services in the dengue fever (DBB) in context of the problem of sustainable food Lampung Province. Journal of Tropical production. Sustainability 12(13): 5488– Forests 6(3): 237–248 (in Indonesian). 5500. https://doi.org/10.3390/su12135488 Setiawan A., Bakri S., Effendi A., Nurhaida I. Puspasari E., Wulandari C., DarmawanA, Banu- 2014. Characterization of parameter mod- wa I.S. 2017. Social economic aspects of el predictions for elevation of community agroforestry system in Forest Community poverty trap in TNBBS buffer area through Work Area (HKm) District West Lampung, fiscal policy intervention. Journal of Society Lampung Province. Journal Sylva Lestari of Culture and politics 27(2): 65–76 (in In- 5(3): 95–103 (in Indonesian). donesian). PutraA.W., Supriatna J., Koestoer R.H., Soe- Supriadi S., Riniarti M., Bakri S. 2018. Rub- silo T.E.B. 2021. Differences in local rice ber productivity on the land of HKm Jaya price volatility, climate, and macroeconom- Lestari, the right Way district Lampung ic determinants in the Indonesian market. Province. Gorontalo: Journal of Forestry Sustainability 13(8): 4465–4486. https:// Research 1(1): 1–6 (in Indonesian). doi.org/10.3390/su13084465 Wakhidah A.Z., Chikmawati T., Purwanto Y. Rohayati R., Muhtarudin M., Bakri S., Suwandi 2020. Homegarden ethnobotany of two J.F. 2018. The impact of climate change and Saibatin Villages in Lampung, Indonesia: land cover on avian influenza incidence in Species diversity, uses, and values. Forest sector 4 livestock poultry (traditional poul- and Society 4(2): 338–357. http://dx.doi. try) in Lampung Province. Proceeding of org/10.24259/fs.v4i2.9720 the 3rd SHIELD International Conference: Wigaty L., Bakri S., Santoso T., Wardani 221–228. Available at: http://repository. D.W.S.R. 2016. Influence of land use lppm.unila.ac.id/11777/1/%5B1%5D%20 change to malaria morbidity index: study THE%203RD_SHIELD_Rohayati%20 case in Lampung Province [Pengaruh pe- et%20al.pdf rubahan penggunaan lahan terhadap ang- Rosari R.S., Bakri S., Santoso T., Wardani ka kesakitan malaria: studi di provinsi lam- D.W.S.R. 2017. The effect of land use pung]. Journal Sylva Lestari 4(3): 1–10 (in toward pulmunary tuberkulosis inci- Indonesian). dence: Case Study in Lampung Province. Wulandari C., BintoroA., Rusita, Santoso T., Jurnal Sylva Lestari 5(1): 71–80. (in Indo- Duryat, Kaskoyo H., Erwin., Budiono P. nesian) DOI:http://dx.doi.org/10.23960/ 2018. Community forestry adoption based jsl1571-80 on multipurpose tree species diversity to- Ruchyansyah Y., Wulandari C., Riniarti M. 2018. wards to sustainable forest management in Silviculture effect in community forestry in ICEF of University of Lampung, Indonesia. KPHL VIII Batutegi to Farmers Income and Biodiversitas 19(3): 1102–1109. Soil Fertility. Journal Sylva Lestari 6(1): Wulandari C., Inoue M. 2018. The Importance 100–106 (in Indonesian). of Social Learning for the Development of Sanudin S., Awang S.A., Ronggo Sadono R., Community Based Forest Management in Purwanto R.H. 2016. Progress of commu- Indonesia: The Case of Community Forest- nity forest in Lampung Province. Jurnal ry in Lampung Province. Manusia dan Lingkungan 23(2): 276–283 Zhafira G., Wulandari C., Rusita R., Bakri S. 232 C. Wulandari, S. Bakri, M. Riniarti, and S. Supriadi

2019. Effect of community forest altitude sity and Biotechnology to Environmental to forest rubber latex production in Way Conservation [Pemanfaatan Biodiversitas Kanan District [Pengaruh ketinggian tem- dan Bioteknologi untuk Pelestarian Ling- pat terhadap produksi getah karet Hutan kungan]. Fourth Biology National Seminar Kemasyarakatan di Kabupaten Way Proceeding. Universitas Islam Negeri Su- Kanan]. In: Supriatna A., Cahyanto T., Wid- nan Gunung Djati. Bandung: 181–184 (in iana A. (Eds) The importance of Biodiver- Indonesian).