Indonesia has an estimated population of more than 237 million [1
], and thus its economic growth should be managed effectively to ensure secure access to food, housing, education, and health. Population growth increases land use demand for agriculture commodities [2
] and results in forest exploitation, particularly impacting communities nearby forested areas. Of particular environmental concern is the illegal practice of forest conversion into agricultural land in forested areas that are easy to access. Such practices result in accelerated soil erosion from increased land exposure [3
] and increased nutrient loss [4
] to rivers and streams.
The Indonesian government issues regulations on social forestry to involve the local community in sustainable forest management. These regulations support local economic growth and provide equity for social welfare, and also maintain and protect forest ecosystem functions [5
]. The community forestry (Hutan Kemasyarakatan, Hkm
) and forestry partnership (Kemitraan Kehutanan, mitra
) regulations have been applied in Tanggamus Regency in Lampung Province. The farmers in this district predominantly plant coffee trees, as well as pepper, cacao, clove, and fruit trees (durian and avocado). Coffee is the largest export from the agricultural and forestry sector in Lampung Province, with a value USD 435,288,000 [6
] and a production of 131,501 tons [7
] in 2014. Furthermore, Indonesia is the fourth largest coffee producer after Brazil, Vietnam, and Colombia [8
Coffee plantations require fertilization to maintain yield and quality. Eleven years ago, chemical fertilizers were not commonly applied in coffee plantations under the management of social forestry in the Tanggamus Regency [9
]. However, due to lowered soil nutrient availability following the conversion of forests to agricultural land [10
], the application of some chemical fertilizers was necessary to increase productivity. In particular, the application of N-fertilizers has been found to increase coffee yield [11
] and improve bean quality [12
]. Stream water in forested areas is typically higher in quality compared with water from rivers in other land use types [13
]. Excessive fertilizer application can cause water quality degradation in rivers and/or reservoirs near agricultural land [14
]. It is therefore necessary to monitor the water quality in nearby rivers and reservoirs in order to determine the impacts of excessive fertilization as a result of social forestry practices.
The links between water quality and land use have been studied in a number of watersheds throughout the world [15
]. A recent study conducted from March to July 2016 in our study area detected clear differences in water quality between the two adjacent watersheds and briefly analyzed relationship between land use and water quality [18
]. The study identified that Sangharus River had higher nitrate (NO3
) while Sekampung Hulu River had higher total suspended solids (TSS), aluminum, and iron. However, the seasonal patterns of river water quality have not yet been investigated in the area. Moreover, detailed analyses for understanding the reason for the differences in water quality between the watersheds have not been conducted. Seasonal climate variability plays an important role in water quality within ecosystems [19
], as rainy and dry seasons can influence river water quality. The dry season has higher total solids (TS) and biochemical oxygen demand (BOD) because of low river discharges and increased industrial wastewater discharges, while in the rainy season a higher NO3
concentration is detected because of high runoff that transports fertilizers [20
In this study, we targeted two adjacent watersheds, Sekampung Hulu and Sangharus, where forested land has been predominantly converted into coffee plantations under social forestry management. The Batutegi Dam is a water supply source for irrigation, drinking water, and nearby power plants, and is located downstream of the rivers in our study area. Therefore, the hydrological characteristics of the rivers can influence the reservoir function [21
]. As social forestry concept has also been adopted in other areas in Indonesia, management of water quality environment under a social forestry system is essential to give information to stakeholders about the sustainable use of mountainous areas. In this study, we aimed to determine the seasonal water quality characteristics through observations spanning one year and to identify the impacts of local fertilizer application on river water quality in the watersheds. In addition, we tried to understand reasons why clear differences in water qualities were observed in the adjacent watersheds. Based on our results, we provide recommendations for effective water quality management in these watersheds.
2. Materials and Methods
2.1. Study Area
The study area is located in the Sekampung Hulu (5°5′38″ S, 104°30′34″ E) and Sangharus (5°15′58″ S, 104°42′56″ E) watersheds in Lampung Province, Indonesia (Figure 1
). The study area of the Sekampung Hulu watershed covers 141.3 km2
, consisting of social forestry (137.6 km2
) and private land (3.7 km2
). The study area of the Sangharus watershed covers 117.2 km2
, and also consists of social forestry (106.7 km2
) and private land (10.5 km2
). With regards to the local geology, the Sangharus watershed consists of 2% sandstones and tuff, 3.7% clay and sand deposits, 62.3% basaltic andesite tuff, and 32% pumice tuff. The Sekampung Hulu watershed consists of 7.2% clay and sand deposits, 1.9% granite, 2.9% schist, 57.8% basaltic andesite tuff, and 30.1% pumice tuff [22
] (Figure 2
The watershed topographies are characterized by mountain ranges and hills at elevations ranging from 282 to 1767 m above sea level. The total annual precipitation in 2016 was 1294 mm [23
]. The precipitation data for the study period (2016) and the 17-year precipitation mean are illustrated in Figure 3
. The study region is defined as climate type Af (rainfall in the driest month is at least 60 mm) based on the Koppen classification [9
]. The study region is located in the tropics and therefore experiences rainy and dry seasons. In 2016, the dry season in the Tanggamus Regency occurred in June–August, while the rainy season occurred in January–May and September–December [24
Based on field observations and land use data analyses [25
] (Figure 2
B), the watersheds were predominantly covered by coffee trees. Commercial trees such as pepper, cacao, clove, rubber, durian, and avocado were also identified. In addition, timber tree species of high economic value, such as mahogany (Swietenia mahagoni
) and sonokeling trees (Dalbergia latifolia
) were found. The land area in the Sekampung Hulu watershed consists of 33.9% agroforestry coffee, 34.3% shade coffee, 25.7% young coffee, 5.8% forests, and 0.3% rivers. The land area in the Sangharus watershed consists of 25.6% agroforestry coffee, 66.3% shade coffee, 3.3% young coffee, 4.6% forests, and 0.2% rivers. The land use of young coffee involves coffee plantation in early growth and has less coverage condition. Shade coffee refers to coffee plantations with shade trees such as Gliricidia sepium, Paraserianthes falcataria,
and others. Agroforestry coffee is a multistory system that consists of coffee plantations with more than five other tree species.
2.2. Water Sampling and Analyses
The water sampling sites in the Sekampung Hulu and Sangharus Rivers were located downstream of the watersheds because of ease of accessibility (Figure 1
). Water samples were collected in 23rd October, 6th and 20th November, and 4th December, 2016. To determine the water quality characteristics for the entire year, we compared water quality data collected in 26th March, 10th and 23rd April, 8th May, and 17th July, 2016 from a previous study [18
]. Water sampling from a previous study had the same locations as our sampling sites. We analyzed 15 water quality parameters, including calcium, potassium, magnesium, sodium, chloride (Cl), NO3
, phosphate (PO4
), sulfate (SO4
), Al, Fe, silicon, water temperature, electric conductivity (EC), dissolved oxygen (DO), and pH. Through the analyses, we can understand the circumstances of water quality in the area and use the information to consider the effects of human activities and natural processes on water quality characteristics. Water quality information can also support recommendations to handle water quality issues in the study area.
Water temperature, EC, DO, and pH were measured on site using a Horiba multi-parameter water quality meter (U-53G, Horiba, Kyoto, Japan), a DO meter (Hanna Instruments HI 9142, Woonsocket, RI, USA), and a bench pH meter (Hanna Instruments HI 2550, Woonsocket, RI, USA), respectively. Other parameters were analyzed according to the available methods and equipment in our laboratory. Ca, K, Mg, Na, Cl, NO3, PO4, and SO4 were measured by ion chromatography (Dionex ICS-1600, Sunnyvale, CA, USA) and Al, Fe, and Si were analyzed by inductively coupled plasma atomic emission spectroscopy (ICPE-9000, Shimadzu, Kyoto, Japan).
2.3. Survey of Local Fertilizer Application
We obtained information regarding fertilizer application in the Sekampung Hulu and Sangharus watersheds via a questionnaire to local farmers because no statistical information related to this aspect was available in the area. In addition, there are many advantages to understanding the local manner of farming activities through direct communication because chemicals contained in fertilizers are a key parameter determining water quality characteristics. The questions were framed to obtain information regarding the amount of fertilizer applied, kinds of fertilizers applied, and the schedule of fertilizer application. Each watershed contains a habitat of approximately 2500 farmers. We surveyed 93 farmers in each watershed based on the total number of farmers, a confidence level of 95%, and a margin of error of 10%. The respondents were categorized as farmers of private land tenure, farmers of HKm, and farmers of mitra. The dominant crop in the study area is coffee. Area size of farmers’ fields ranges 0.25–6 ha with the predominant size being 1–2 ha.
The social forestry farmers selected in the Sekampung Hulu watershed for the survey were grouped as follows: HKm Sinar Harapan, HKm Wana Tani Lestari, Hkm Mandiri Lestari, HKm Bina Wanajaya 1, and HKm Bina Wanajaya 2. The farmers in the Sangharus watershed were grouped as follows: private land tenure, Hkm Sidodadi, HKm Trisno Wana Jaya, HKm Karya Tani Mandiri, HKm Sinar Harapan, and mitra Sumber Rejeki. As Hkm Sinar Harapan is located both in the Sekampung Hulu and Sangharus watersheds, the respondents were surveyed for both watersheds.
2.4. Statistical Analysis
The water quality data and fertilizer application survey were statistically evaluated. We applied an independent samples t-test or a Mann-Whitney U-test based on normality distributions. These statistical analyses were performed to determine the significant difference of water quality in the two rivers and fertilizer application amount in the two watersheds. We conducted a one sample t-test to determine the seasonal variability of water quality. The one sample t-test was conducted to compare a single data observation in the dry season with that of the mean sample in the rainy season in order to determine the significant differences. Statistical analyses were conducted using Statistical Product and Service Solutions (SPSS) 17.0 software [26
]. SPSS is user friendly and widely used throughout the world.
2.5. Uncertainties and Shortcomings of the Study
Water samples were not collected every month at the target sites. Thus, sampling numbers of stream water may not be sufficient to show the level of water concentrations in the watersheds, though differences in water quality characteristics can be understood through our study. Besides this, as the sampling was conducted only downstream because of low accessibility to the mountainous streams, and no observations were conducted along the rivers from middle to upper streams, our research is not able to discuss any trends in water concentrations along the rivers from upstream to downstream in the watersheds.
In addition, the characteristics of seasonal variability of water quality are affected by the climate condition of El Niño or La Niña. Normally, the dry season in the study area is from June to September, but in 2016, the season was shorter, and was from June to August (Figure 3
). Moreover, the application of fertilizer may vary across years depending on farmers’ preference for applying fertilizer and their financial conditions. Thus, climate variability and farmers’ decisions will also affect stream water quality.
To collect information on fertilizer application, the survey was conducted in such places as farmers’ homes, fields, and pathways. Hence, accurate location of all respondents’ land tenure was difficult to identify on the map. This means it is difficult to understand the exact location of farmland to which amounts of fertilizer are being applied. Increasing the number of respondents and surveys to all farming groups will provide more detailed information. Accumulation of knowledge through long-term observation of water qualities and local surveys should be conducted in future for a comprehensive understanding of water quality circumstances in the watersheds.
Our study has revealed seasonal water quality characteristics and possible reasons for the observed characteristics in adjacent two watersheds for the first time. Although the study sites were located close to each other, they showed different water quality characteristics. The human activities of fertilizer application and young coffee plantations, as well as the natural processes of geological characteristics, influenced the differences between the two watersheds. Based on the results, the Sangharus River contained higher amount of nutrients than the Sekampung Hulu River due to higher fertilizer application amounts in the watershed. Moreover, geological characteristics played an important role in the Sangharus River in determining its water quality characteristics because the watershed consisted of higher basaltic andesite tuff compared to the Sekampung Hulu watershed. Seasonal water quality measurements and questionnaire surveys to local farmers revealed that NO3 concentrations in both watersheds were higher in the rainy season to correspond with the annual schedule and total amounts of fertilizer application in the watersheds. Despite the application of fertilizers, NO3 levels remained below the recommended water quality standard. However, Al and Fe levels in stream water exceeded the recommended level for drinking water, which was likely due to soil erosion from improper land management in the Sekampung Hulu watershed.
To protect the environment from the adverse effects of soil erosion and nutrient loss, soil conservation practices should be implemented in the study area such as cover cropping, contour cropping, terracing, and agroforestry. Agroforestry practices in coffee plantations have already been applied in several sites; however, the practice of planting young coffee plantations needs to be implemented for effective soil conservation practices. Moreover, application of soil conservation practices in shade coffee plantations can provide more environmental benefits to reduce surface runoff.
Policy makers are required to develop regulations for a sound water environment based on the different characteristics of the two watersheds. The policies should consider background reasons to determine water quality characteristics in the area. In addition, farmers are recommended to adopt soil conservation practices to prevent sustainable land from experiencing reducing nutrient loss and erosion.
This study was conducted for only a year, with missing information for a five-month duration. A one-year period of research is too short to investigate all aspects of a water environment. Thus, long-term research on water quality should be conducted to understand comprehensive aspects of water characteristics across dry and wet years. In addition, we could not conduct studies on water quality in the upper and middle watersheds due to low accessibility. To determine effective management strategies, further studies on the upper and middle reaches of the watersheds are necessary for a holistic view of the watershed water chemistry characteristics. In addition, the number of respondents in our questionnaire survey was minimal according to the total number of famers in the study area. To increase the accuracy of the information regarding the schedule and the amount of fertilizer applied, the number of respondents in questionnaire survey needs to be higher.
In recent years, new technology of artificial intelligence (AI) and machine learning tools have begun to be used for water quality forecasts [60
]. These tools are very robust; however, for obtaining good results, it is very important to accumulate local information for a water quality database. By conducting our kind of research in ungauged and poorly gauged watersheds continuously, AI and machine learning based analyses can be conducted to implement water resources management, protect fresh water resources, and develop future conservation plans regarding these watersheds.