The Role of Agroforestry to Improve Farming System Sustainability in Upper Brantas Watershed

The upper Brantas Watershed area must be conserved, especially for ecological function. On the other hand, the farmers need income to support their livelihoods. So that, the farmers prefer cultivating vegetables in highlands and on land with slopes more than 30 degrees. It will make the agricultural systems tend to be unsustainable in the long run so that it can affect the ecological function of the watershed. This research aimed to analyze the effect of agroforestry techniques on farming sustainability in the upper Brantas Watershed. The location was determined purposively in a farming system with and without an agroforestry system in Batu City. This study uses quantitative and qualitative approaches with survey techniques by field observation and interviews with the farmers and relevant stakeholders. The results showed that the application of conservation farming with the concept of agroforestry will support the farming system‘s sustainability, especially on the ecological aspect.


Introduction
Upper Brantas Watershed is a very strategic area in East Java Province because of its function as an area to store rainwater and as a provider of water for various uses, including for power plant, local water company (PDAM -Perusahaan Daerah Air Minum), irrigation, industry, and other needs.Watershed areas also produce environmental services such as producing oxygen, absorbing carbon, preventing erosion, storing groundwater, comfort, beautiful scenery, and producing food.The Brantas Watershed is one of the 4 watersheds in the Brantas River basin, in addition to the Tengah Watershed, the Ringin Bandulan Watershed, and the Kondang Merak Watershed.Brantas River area is the second largest river area on Java Island.The Brantas River is approximately 320 km long and has an area of approximately 14,103 km2 (1).The upper Brantas watershed area in Batu City is an area where the environmental function must be conserved.On the other hand, people's livelihoods must also be considered, even more the growing population in Batu City, which is increasing by 1.14% per year (2).Population growth pushes an increase in demand for land for residential purposes, as well as hotels and various building facilities, considering Batu City as a tourist city and for agricultural land.Farmers grow a variety of commodities, including vegetables, even on land with steep slopes (more than 30 degrees), which should apply conservation techniques if planted with seasonal crops.Farmers plant vegetables because vegetables are short-lived and have high economic value, also the climate in the highlands is suitable for vegetable commodities.In the other places in Upper Citarum Watershed (UCW) West Java, Indonesia, farmers cultivate vegetables at the steep topography.Most farmers in the UCW grow seasonal crops, in particular, rice and vegetables (3).The determinants of farmer participation and intensity in the application of conservation depend on the ability to finance, land tenure and utilization, and the physical condition of the land.In the future, conservation priorities must be more directed at a vegetative approach through developing synergistic perennial crops with particular efforts to convert vegetables on sloping lands to perennial crops (3).
On the other hand, farmers plant vegetables under the trees, that called agroforestry.Agroforestry is the combination of trees and crops in spatial or temporal arrangements that results in greater structural and functional complexity compared to monoculture production (4,5).Agroforestry systems are defined as landscape units comprising land uses that combine aspects of agriculture and forestry, including the agricultural use of trees (6).Agroforestry intercropping systems can provide multiple ecosystem services in regions where agricultural systems are becoming more intensive (7).These research objectives are to identify farmers perceptions about conservation farming and agroforestry also to analyze the effect of agroforestry techniques on farming sustainability in the upper Brantas Watershed.

Research Approach and Sampling Method
This research used a quantitative approach to analyze farmer's perceptions and the application of conservation farming to farming sustainability.The research locations were in Upper Brantas Watershed, specifically in Sumberbrantas Village, Bumiaji District, Batu City, and Tawangargo Village, Karangploso District, Malang Regency, East Java.Sumberbrantas was chosen because it is one of the upland vegetable farming areas which not apply an agroforestry system.The farmers in Tawangargo have implemented an agroforestry system on their land.Thirty-five vegetable farmers in Upper Brantas Watershed were randomly selected as respondents.The data were collected through a combination of in-depth interviews and survey techniques, by interviewed the farmers, and by field observation.

Data Analysis Method
This research uses the Structural Equation Model -Partial Least Square (SEM-PLS) through the SmartPLS 3.0 application.The reason for choosing this analysis approach is based on the data/sample features and the moderation and mediation analysis.Based on (8), SEM consists of inner and outer model analyses, which examine the relationships between independent and dependent variables and relationships between latent constructs and their observed pointers.PLS focuses on variance analysis, which could be done using Smart PLS.Therefore, this approach is selected for the present study.The independent variables in this study consisted of three, namely mechanical conservation (X1), cultivation conservation (X2), and vegetation conservation/agroforestry (X3).The dependent variables are the sustainability of farming, from ecological (Z1), economic (Z2), and socio-cultural (Z3) aspects.The application of conservation farming is an intervening variable between perceptions and sustainability of farming.

General Overview and Respondent Characteristics
The research location is at Sumberbrantas Village, Bumiaji District, Batu City, Tawangargo Village, Karangploso District, Malang Regency.Sumberbrantas Village is located on the slopes of Arjuno Mountain with a temperature of around 12-24℃, an elevation of 1.400-1.700masl, and a rainfall of 875-3000 mm/year.The area of this village is around 5.156 hectares, with the majority of land use being state forest (4.666,9 ha) and agricultural land (318,32 ha).Most of the villagers are farmers (60 percent of the total population in Sumberbrantas Village.Horticultural commodities such as potatoes, carrots, and mustard greens are the agricultural potential of this village (9).In addition, there is a coffee IOP Publishing doi:10.1088/1755-1315/1299/1/0120123 plantation area of around 4 ha with a total production is 2,5 tons (Kecamatan Bumiaji Dalam Angka 2020).Tawangargo Village is also located on the slopes of Arjuno Mountain with an elevation of 700-1000 masl and the total of land area is 650 hectares.The research location, which is in Tawangargo Village specifically, was carried out around UB Forest.UB Forest is an educational forest that was previously managed by Perhutani (10).Some of the agricultural potentials on this is plantation commodities such as coffee, mahogany, and pine, horticultural commodities such as carrots, cabbage, chayote, chilies, leeks, and food crops such as corn and cassava.Agricultural activities by farmers in this forest use a rental system with a cost is around Rp250.000/ha/year.They were allowed to cultivate between the stands (e.g., pine and mahogany); it is a form of plantation-crop combination (10).The condition of respondents at the research location can be seen from several characteristics, as shown in Table 1.Living in rural areas, most respondent farmers only have primary education (48.57percent), with an average number of household members of 3 people (60 percent).After completing their primary school education, their parents invited them to run a farm on the land, so their experience was quite long and most of them had reached more than 25 years (71.42 percent).Therefore, many farmers are over 50 years old (45.71 percent).Only a few farmers were under age 35 (11.44 percent), and they reported that they previously had other professions outside of agriculture.(11), CA and CR values should be higher than 0.70, and this study found the values to be in an acceptable range because all of the results show higher than 0.70.The AVE values must exceed the 0.50 threshold to examine the convergent validity (12).The value requirement for testing reliability is using the composite reliability value.The questionnaire is said to have good composite reliability if the value is >0.7, and the indicator is consistent if the Cronchbach Alpha value is >0.6 (13).
The reliability testing in this study showed that the composite reliability value for all variables had scored >0.7, which means that the indicators for variables X, Y, and Z can fulfill composite reliability.The largest composite reliability value is at Y2, namely 0.870.Meanwhile, the highest Cronbach alpha value is 0.981 and has exceeded 0.6, which means the questionnaire meets consistency reliability.Therefore, the questionnaire in this study can be used because its values fulfill the specified requirements.
Table 2 shows that all of the variables have AVE values greater than 0.70, so it can be said that the data in this model is valid.Fornell Larcker and heterotrait -monotrait (HTMT) ratio was assessed to test the discriminant validity.The result in Table 3 shows that the values are more significant than the correlations among the variables.The validity test requirements for SEM-PLS are if the indicator has a loading factor value of >0.5 and a p-value of <0.001, then the model can meet the requirements for convergent validity (13).Apart from that, the convergent validity test can be determined through the average variance extracted (AVE) value for each variable in the model and is said to be valid if the value is >0.50 (14).
Another form of convergent validity test can also be seen from the average variance extracted (AVE) value.In this study, the AVE value is valid because the X1 to Z3 values are >0.50.Another validity test is to test discriminant validity by comparing loading and cross-loading values.If the loading factor is more than cross-loading, discriminant validity is fulfilled.Apart from that, discriminant validity also compares the square root values (AVEs) or the square root of average variance with the latent variable in question with other latent variables (13).
Based on the results of the AVES test, the loading value is greater than the cross-loading so that the discriminant validity can be fulfilled.Table 3 shows that the square root value of the AVE variable X1 is (0.522), X2 (0.707), 0.858), and Z3 (0.970) is greater than the correlation value between the variables so that the validity in this study shows that the discriminant criteria are good or meet the discriminant.

Structural Equation Modeling
The PLS-SEM findings in Table 4 show that perception of mechanical conservation (X.1) has a significant effect on farming sustainability from an ecological aspect (Z.1), intervening by application of mechanical conservation (Y.1) (p = <0.10).Terracing, one of the indicators in the perception of mechanical conservation, is a vital support for agricultural sustainability (15).Besides that, the perception of vegetation conservation/ agroforestry (X.3) has a significant effect on farming sustainability from an ecological aspect (Z.1), intervening by the application of vegetation conservation/ agroforestry (Y.3) (p = <0.05).One of the indicators in the perception of vegetation conservation is terrace strengthening plants, which is an effort to maintain the terraces so that erosion can be minimized (16).The resulting output of this study can be seen in Figure 1.The Mechanical Conservation variable (X1) affects the Mechanical Conservation Implication (Y1) with a value of 0.056, so there is a significant and positive influence on the Conservation Implementation variable with an error rate of 5 percent.This result shows that the relationship between conservation and the implementation of mechanical conservation has a significant influence.The influence of farmers' perceptions regarding mechanical conservation and its implementation can be seen in the creation and erosion prevention activities carried out by farmers with soil and surface mechanization.Some of the activities carried out include the implementation of reforestation activities for plants, terracing, and crop rotation for crops.
The variable Perspective of Cultivation Conservation (X2) does not affect implementing Cultivation Conservation (Y2) with a value of 0.091.Based on the results of this data analysis, a figure of 0.091 was obtained, which at the 5 percent level does not significantly affect mechanical conservation.
The variable Perception of Agroforestry (X3) influences the Implementation of Agroforestry Conservation (Y3) with a value of 0.057.Based on the results of this analysis, it is known that there is a positive and significant influence between these two variables.At an error rate of 5 percent, the results obtained are significant.This is because agroforestry is a component of farming that involves some integration between existing seasonal crops and various types of woody plants, which benefits farmers in the research location.The main aim of agroforestry is to provide maximum and sustainable results in ensuring the continuity and improvement of life for the existing community so that environmental sustainability can be maintained and the community's needs at the research location are met.The results IOP Publishing doi:10.1088/1755-1315/1299/1/0120126 of (17) show that community involvement in optimizing agroforestry conservation has an ecological and economic impact and also helps the community with the forest products they own.
The Mechanical Conservation Implementation variable (Y1) influences the Sustainability of Ecological Aspects (Z1) with a value of 0.029.Based on this value with an alpha level of 5 percent, these two variables have a positive and significant relationship.The implementation of mechanical conservation is based on the results of research, one of which is carried out by using activities to create soil through existing terraces or parking lots so that plants are more fertile and fallen leaves become natural fertilizer for plants.Apart from that, plant maintenance is also carried out through mechanical controls (18).Of course, we still pay attention to ecological and environmental principles, such as making drainage, building retaining loose material for plants, and making terracing traps for plants.
The Mechanical Conservation Implementation variable (Y1) does not affect the Sustainability of Economic Aspects (Z2) with a value of 0.165.The research results showed that at an alpha value of 5 percent, there was no significant and influential relationship between the two variables.The creation of agricultural mechanization in farming, especially in tools and technology, requires quite a lot of costs, so in a sustainable economic aspect, it is not yet able to provide the economic feasibility that farmers hope for.Investment through agricultural mechanization and minimizing existing costs does not meet the economic sustainability of farming, so special efforts need to be made periodically.
The Mechanical Conservation Implementation (Y1), Cultivation Implementation (Y2), and Agroforestry Implementation (Y3) variable has no effect on the social and cultural aspect sustainability variable (Z3) with a value of 0.571, 0.390 and 0.289 (at the 5 percent).One of the indicators used in the socio-cultural aspect of this research is gender equality in work in the agricultural sector.This is in line with research by ( 19) that gender had no significant impact on on farmland management behavior, attitude, and knowledge.
The Cultivation Implementation variable (Y2) has no effect on the ecological aspect sustainability variable (Z1), with a value of 0.344 at a significance level of 5 percent.This is because many cultivation implementations that are carried out do not meet the ecological aspects in terms of controlling pests and diseases.The uses of livestock grazing as weed control and proper land clearing will have a big impact on the existence of forests, especially for farmers in villages.The protection of cultivation and implementation of activities is only a regional (national) problem but is already a world (global) problem (20).The Cultivation Implementation variable (Y2) does not affect the economic aspect sustainability variable (Z2) at a significance level of 5% because it has a value of 0.618.This is because conservation investment accounted for only a small portion of household income (19).The Cultivation Implementation variable (Y2) has no effect on the social and cultural aspect sustainability variable (Z3) at the 5 percent significance level with a value of 0.390.
The Agroforestry Implementation variable (Y3) partially has no effect on the ecological aspect sustainability variable (Z1) with a value of 0.618.However, it has a direct and significant effect through the mediating variable Implementation of Agroforestry Conservation (Y3).Perception of the community regarding forest management can be determined from the level of knowledge about the forest and the function of the forest in their lives.(21) stated that public perception will be good if people understand that they depend on forest biological resources for their living and that these resources are managed sustainably.Some people do not understand the use of the area due to a lack of knowledge about the importance of forest management.The lack of public knowledge about forest management is due to a lack of information and outreach, including counseling related to community forest management (22).However, in fact, in the field, agroforestry management has been quite good for ecological aspects in the community at the research location.
The Agroforestry Implementation variable (Y3) does not affect the sustainability of economic aspects (Z2) with a value of 0.647 at a significance level of 5 percent.This is because the economic implementation of agroforestry has not been able to increase household income, and the education of farmers' children is still in the low category.In this case, the economic aspect is reviewed based on community understanding regarding the collection of forest products and the economic benefits of agroforestry management.Non-timber forest product (NTFP) collection activities as an effort to maintain forest sustainability and increase the strengthening of economic development have not been implemented optimally.This is in line with research by (23), who stated that the economic value resulting from the use of NTFPs is much greater than wood and does not cause forest damage, so it will not result in the loss of the functions and service value of the forest.NTFPs are considered important to continue to be developed, considering that the productivity of wood from natural forests is decreasing (24).Apart from that, NTFPs in their use have advantages over wood products, so NTFPs have great prospects in their development.However, it has not been able to improve the welfare of farmers economically.Based on the R2 test in this study, the values for variables Y1, Y2, Y3, Z1, Z2, and Z3 were obtained respectively at 0.659, 0.701, 0.897, 0.556, 0.550, and 0.691.This shows that the results of this research are between 0 and 1, which states that the relationship between variables has high accuracy.If have high accuracy results, namely close to 1, the R-squared test value will be better, so this result study has a high accuracy relationship.

Conclusion
The results showed that the perception of mechanical and vegetation conservation has a significant effect on farming sustainability from an ecological aspect.The application of conservation farming with the concept of agroforestry also will support the farming system sustainability, especially in the ecological aspect.
(a) Conventional farming system (b) Agroforestry system Figure 1.Research location at Sumberbrantas Village (a) and Tawangargo Village (b)

Figure 2 .
Figure 2. SEM-PLS result output Note: Perception of Mechanical Conservation (X.1), Perception of Cultivation Conservation (X.2), Perception of Vegetation Conservation/ Agroforestry (X.3), Application of Mechanical Conservation (Y.1), Application of Cultivation Conservation (Y.2), Application of Vegetation Conservation/ Agroforestry (Y.3), Ecological Aspect (Z.1), Economic Aspect (Z.2), Socio-cultural Aspect (Z.3).The Cultivation Implementation variable (Y2) does not affect the economic aspect sustainability variable (Z2) at a significance level of 5% because it has a value of 0.618.This is because conservation investment accounted for only a small portion of household income(19).The Cultivation Implementation variable (Y2) has no effect on the social and cultural aspect sustainability variable (Z3) at the 5 percent significance level with a value of 0.390.The Agroforestry Implementation variable (Y3) partially has no effect on the ecological aspect sustainability variable (Z1) with a value of 0.618.However, it has a direct and significant effect through the mediating variable Implementation of Agroforestry Conservation (Y3).Perception of the community regarding forest management can be determined from the level of knowledge about the forest and the function of the forest in their lives.(21)stated that public perception will be good if people understand that they depend on forest biological resources for their living and that these resources are managed sustainably.Some people do not understand the use of the area due to a lack of knowledge about the importance of forest management.The lack of public knowledge about forest management is due to a lack of information and outreach, including counseling related to community forest management(22).However, in fact, in the field, agroforestry management has been quite good for ecological aspects in the community at the research location.The Agroforestry Implementation variable (Y3) does not affect the sustainability of economic aspects (Z2) with a value of 0.647 at a significance level of 5 percent.This is because the economic implementation of agroforestry has not been able to increase household income, and the education of farmers' children is still in the low category.In this case, the economic aspect is reviewed based on community understanding regarding the collection of forest products and the economic benefits of

Table 1 .
Descriptive statistics of farmer respondents.We have used Cronbach Alpha (CA) and Composite Reliability (CR) values to measure the reliability of the data.According to

Table 2 .
Result of measurement model.

Table 3 .
Discriminant validity (latent variable correlation and square root of AVE).

Table 4 .
(25)tesis construct Source: Primary data processed (2023) The R-squared test shows what proportion of the response variable can be explained by the predictor variable or to test how much influence the exogenous variable has on the endogenous variable.(25)saidthat if the R2 value shows a value between 0 and 1, if it gets closer to 1, then the relationship between these variables has a high prediction or a perfect relationship.