1 Introduction

Oil palm smallholders in Indonesia, the world’s largest palm oil producing country, manage around 40% of the total oil palm area. Approximately two thirds of these smallholders are independent, meaning that they are not formally associated with large plantations and therefore do not receive direct support (Jelsma et al., 2019). Smallholders typically cultivate small landholdings, often around 2 hectares (Daemeter Consulting, 2015), and due to low fertilizer use, they often achieve yields that are below the attainable yield (Rhebergen et al., 2018; Sugianto et al., 2023). In oil palm, the attainable yield is commonly defined as 70% of the yield that could theoretically be achieved under local weather and soil conditions when crops grow without nutrient limitations and weeds, pathogens, and insect pests are effectively controlled (Monzon et al., 2021). However, average yields in smallholder fields are well below the attainable yield. For example, Monzon et al. (2023) found large yield gaps for independent smallholders, with average fresh fruit bunch (FFB) yield representing only 42% of the attainable yield. Narrowing the current yield gap could contribute to higher incomes for farmers (Monzon et al., 2023), while potentially curbing the need to clear new land for oil palm cultivation and associated encroachment of fragile ecosystems and greenhouse gas emissions (Austin et al., 2017). This is on condition that measures are in place to prevent further land expansion driven by higher yields and profits (Dalheimer et al., 2021).

Previous studies on oil palm smallholders in Indonesia have shown that proper nutrient management, in addition to other management practices such as good planting material, proper harvesting, and soil management, is essential to achieve good yields (e.g., Lim et al., 2023; Sugianto et al., 2023; Woittiez et al., 2017). Other studies have focused on smallholder replanting strategies and use of good planting material (Zhao et al., 2023), smallholder harvesting practices (De Vos et al. 2023a), and soil management practices such as integration with cover crops and livestock (Álvarez et al., 2024). This study focuses specifically on nutrient management because this was identified as a key factor explaining yield gaps by Sugianto et al. (2023). In the context of oil palm, good nutrient management entails applying a balanced combination of the right amount of nutrients at the right place and time (Lim et al., 2023; Tiemann et al., 2018). However, independent oil palm smallholders in Indonesia often apply little or no fertilizer and prefer to use more affordable fertilizer sources that are often imbalanced in relation to plant nutrient requirements (Lim et al., 2023; Mettauer et al., 2021; Woittiez et al., 2019). This behavior can ultimately lead to severe nutrient deficiencies, especially for potassium (K), nitrogen (N), and phosphorous (P), and consequentially low FFB yields (Sugianto et al., 2023; Woittiez et al., 2019). Thus, improving plant nutrition in smallholder oil palm fields is relevant in the context of, on the one hand, the need to increase palm oil production on existing plantations without encroaching on natural ecosystems, and on the other hand, the need to increase yields and profit to benefit smallholders and local communities. Understanding the causes for poor nutrient management in smallholder oil palm fields is essential to enable crop intensification through proper changes in policies, institutions, and research and extension programs.

Yet, there is limited knowledge about the factors that enable and constrain fertilizer use in smallholder oil palm fields. Most studies focus on the impact of good nutrient management on yield, while underlying barriers for adoption of good management practices remain unknown. Several studies have shown that oil palm smallholders in Sub-Saharan Africa and Southeast Asia often had limited agronomic training and/or enough resources to purchase fertilizer and hire labor to perform plantation maintenance (Degli Innocenti & Oosterveer, 2020; Essono et al., 2023; Rhebergen et al., 2018). Other factors can also influence fertilizer use in smallholder fields. For example, Somnuek & Slingerland (2018) found that oil palm farmers in Thailand did not apply fertilizer when the FFB price was too low. In another study, Sahara et al. (2017) found that fertilizer use by smallholders in Indonesia was constrained by their limited access to formal credit, because they often lacked land titles, which serve as collateral. Moreover, they rarely record cash flows, which made it difficult for banks to determine their credit worthiness (Bronkhorst et al., 2017). Finally, De Vos et al. (2023a) found that among RSPO certified smallholders across Indonesia, former ‘plasma’ smallholders (i.e., formerly related to a large oil palm plantation) and farmers with a transmigration background had better access to fertilizer because of their membership to farmer associations or cooperatives. While these studies offer valuable insights, sample sizes are often small and from a single location. Additionally, data on fertilizer use and yield frequently rely on farmers’ recollections of past-year practices and yield, rather than systematically recorded applications and yields over multiple years.

An important research gap remains in understanding which (combinations of) factors shape diverse fertilizer strategies among farmers. Addressing this research gap, this study aims to provide an in-depth and systematic assessment of agronomic, socio-economic, and institutional factors which collectively influence farmers’ decisions on fertilizer use. The main research question is: ‘Which factors enable or constrain oil palm smallholder fertilizer use, and how do farmers respond to constraints?’ To answer this, we analyzed three years of field data collected from 1200 smallholders in Indonesia, complemented by insights from semi-structured interviews with 40 farmers.

2 Materials and methods

2.1 Research sites and data collection

We collected the data in six provinces in Indonesia: Riau (RI), Jambi (JB), South Sumatra (SS), West Kalimantan (WK), Central Kalimantan (CK), and East Kalimantan (EK). We refer to the sites using their respective provinces. Sites were selected based on the research projects’ collaboration with local NGOs who facilitated and engaged in data collection. Trained NGO staff conducted the different surveys and the farmer diary. At the time of the research, the NGOs did not provide training or financial support with regards to nutrient management to the farmers. We followed 200 independent oil palm smallholders in each site, collecting data on management practices, input use and price, yield, and FFB sale. To ensure representativeness, we excluded smallholders with very small (< 0.1 ha) or very large fields (> 25 ha), immature (< 3 years) or very old palms (> 25 years), and intercropped fields (see Monzon et al., 2023). Likewise, we only included smallholders growing oil palm in mineral soils.

We used different research methods to assess factors influencing fertilizer management decisions in independent oil palm smallholder fields (Table 1). First, in Dec 2019, we conducted a baseline survey across ca. 200 smallholders per site to collect data on farmers’ socio-economic background, plantation characteristics, and agronomic management (see Monzon et al., 2023). Secondly, we used GPS and drones to map the borders of their oil palm fields (one per farmer) and determined associated size and data on palm density, plantation age, and soil type. This dataset is referred to as the land survey. Third, we collected data on FFB yield and fertilizer use and associated costs starting in Jan 2020 until Dec 2022 via a ‘farmer diary’. Fertilizer prices were calculated based on pooled costs of fertilizer across sites. We did not examine differences per supplier (see Lim et al., 2023). In the farmer diary, the ca. 1200 selected smallholders noted the timing of all management activities, such as applying fertilizer, harvesting, and maintenance, as well as associated amounts, costs, and FFB yields per harvest. Fourth, we selected 10 farmers from the list of 200 farmers in four sites (RI, SS, WK, CK) for qualitative interviews. We used a stratified random sampling strategy, selecting farmers who applied fertilizer during the study period as well as those who did not, as reported in the farmer diary. We asked questions about topics like changes in fertilizer use over the past years, relations with fertilizer suppliers, access to credit, transportation and logistical issues, government support and subsidies, fertilizer prices, sources of knowledge, and observations on palm and soil health. Additionally, we interviewed one fertilizer supplier at each site whom we asked questions about the supply and demand for fertilizer, distribution issues, credit and payment arrangements, and additional services for farmers. Lastly, we used insights from the previous methods to design a final fertilizer survey focused on fertilizer practices, preferences, and constraints for ca. 200 respondents in all six sites, using the application ODK collect (https://getodk.org). In the fertilizer survey, we asked respondents whether they applied fertilizer, and if not, we asked for the reasons. We also asked for the main constraints; source of knowledge on fertilizer; where respondents bought fertilizer; what kind of credit scheme (if any) they used; availability of fertilizer types; what kind of fertilizer they preferred if they were not cash constrained; quality of fertilizer; and access to and use of organic fertilizer.

Table 1 Overview of databases collected for assessing drivers of fertilizer management in smallholder fields
Full size table

2.2 Data quality control

For the baseline and land survey, as well as the farmer diary, we screened the data to correct any erroneous data entries based on protocols for acceptable ranges for each variable. For example, for FFB yield, we excluded respondents who reported over 35 t FFB ha− 1 y− 1, as it may reflect cases in which FFB was pooled from multiple fields. In case of missing data or doubtful data entries, local partner NGOs followed up by personal visits to the respondents to verify data. A comprehensive description of the study sites, field selection procedures, and quality control measures can be found in Monzon et al. (2023). The final dataset for the baseline and land survey, and the farmer diary included 958 respondents. For the fertilizer survey, we excluded respondents who indicated that they did not make decisions on fertilizer use, for example, when they left all fertilizer management decisions to other family members. The final dataset from the fertilizer survey included 1004 respondents. After excluding respondents who never applied fertilizer to their oil palm fields, the dataset used for analyzing fertilizer practices included 721 respondents who reported fertilizer use. Of this subset, data from 337 respondents were available for further analysis of the amount and types of fertilizer applied.

2.3 Data analysis

First, we explored farmers’ characteristics (gender, age, education, farmer group membership, relative dependency on oil palm for annual household income, and labor source), plantation characteristics (total size of oil palm area, palm age), and fertilizer and FFB prices (Table 2). To get an overview of variation in fertilizer use per site per year, we counted the total number of respondents per site applying fertilizer that year (any type or amount), as reported in the farmer diary. We examined fluctuations in average FFB and fertilizer price including four most frequently used fertilizer types (urea; NPK phonska and other NPK blends with low K concentration; compound fertilizer with higher K concentration, hereafter referred to as NPK+; and MOP), across three years based on pooled data from six sites as reported in the farmer diary.

Table 2 List of variables included in the study
Full size table

Second, we used the dependent variable ‘fertilizer use’ (yes / no) to assess factors that influence fertilizer use. Fertilizer use was based on data from the farmer diary wherein farmers reported whenever they applied fertilizer, with farmers reporting fertilizer use at least once being categorized as fertilizer use: yes.

The binary fertilizer use indicator does not provide information on the amount or nutrient content of the fertilizer that was applied. Therefore, for participants who applied fertilizer during the study, we introduced an additional dependent variable ‘nutrient score’ to analyze the factors influencing the amount and composition of fertilizer used. This nutrient score was calculated based on the amount of N, P, and K applied, relative to nutrients required to achieve attainable yield. To calculate nutrient scores, we compared the NPK inputs with the associated nutrient requirements for each of these nutrient components to achieve the attainable yield. For example, a nutrient balance of 25% for N meant that this field was nutrient deficient for N for 75%. The nutrient requirement was estimated following Lim et al. (2023) based on crop modeling, local weather, palm age, and soil type, and average nutrient concentration per ton of FFB. Aggregate nutrient balances per site are reported in Lim et al. (2023). Next, we scored each field depending on how the associated applied fertilizer compared to the nutrient requirement. The scoring was done separately for each nutrient, allocating 0–3 points: 0–25% (0 points), 25–50% (1 point), 50–75% (2 points), and 75–100% (3 points). Thus, a field can have a minimum score of 0 points (no fertilizer applied) and a maximum total score of 27 points for having applied sufficient N, P and K to meet nutrient requirements over the three years of our study. For simplicity, relative scores were calculated, taking 27 points as 100%. For calculation of the nutrient score, we used an estimation by Lim et al. (2023) of annual nutrient application rates and yields per field per year (2020, 2021, 2022). For this we used the data on fertilizer use and harvested FFB collected via the farmer diary, together with the field size and nutrient concentration associated with each fertilizer source. We excluded organic material, such as empty fruit bunches and manure, because only 6% of our respondents applied organic fertilizer during the study period. We included only respondents with nutrient scores > 0 in follow-up analysis of explanatory factors to make sure we only used data from respondents who applied fertilizer during the study time (n = 337).

Our nutrient score can be considered a useful analytical indicator of nutrient management in smallholder fields and its impact on yield. Indeed, we found a positive relationship between nutrient score and the FFB yield expressed as percentage of the attainable yield, which is based on estimations made by Lim et al. (2023), based on data from the farmer diary (Supplementary Fig. S1). We used random forest regression (RandomForest package in R; Breiman, 2001) to quantify the relative importance of each factor on fertilizer use (yes / no) and nutrient scores. We ran each random forest regression for pooled data across sites. We further explored the relationships between the most important factors identified via the random forest analyses using follow-up statistical tests (Supplementary Table S1). If a variable ranked among the top three in the Random Forest model did not show a significant effect on the dependent variables in follow-up statistical tests, we conducted additional site-specific analyses. All quantitative data were analyzed in RStudio and plotted using GGPlot (GGPLot package; Wickham, 2016).

Finally, to better understand how and why factors influence fertilizer use, we examined farmers’ experiences with challenges and enabling factors in fertilizer application by integrating qualitative interview data with findings from the survey on fertilizer practices. The qualitative dataset consisted of 44 verbatim transcribed interviews conducted in SS, RI, CK, and WK. We analyzed these interviews using qualitative data analysis software (Atlas.ti 22, version 22.2.5.0.2023; Friese, 2019), employing a thematic coding approach (Gibbs, 2012). This involved coding transcripts line by line, assigning descriptive codes to capture the essence of paragraphs or sentences. Similar codes were then grouped into broader code groups to identify overarching themes relevant to our research questions, such as responses to fertilizer cost increases, personal relations with fertilizer suppliers, conditions of credit systems, or logistical issues. To strengthen our qualitative findings, we used data from the fertilizer survey (e.g., reasons for not applying fertilizer, perceived key constraints, fertilizer suppliers, application routines, credit use, and repayment systems) to assess the prevalence of themes and verify patterns.

3 Results

3.1 Socio-economic background and fertilizer use

Farmers in our study generally had small oil palm areas (average: 2 ha), with an average plantation age of 15 years. Their FFB yield was low, ranging across sites between 12 (JB) and 17 t ha− 1 y− 1 (CK), with an overall average of 14 t ha− 1 y− 1, which is well below the average attainable yield of 34.4 t FFB ha− 1 reported by Monzon et al. (2023). Oil palm accounted for more than 50% of the total annual household income for half of our respondents, except in WK, where most farmers derived their income from alternative sources, including plantation labor, off-farm employment, or other crops. Most of our respondents (80%) sold their FFB through FFB collectors rather than directly to mills. Whereas harvesting was often done through hired labor, farmers mostly applied fertilizer themselves. Most farmers (70%) said that they never followed any training in oil palm nutrient management.

Nearly half (47%) of the smallholders applied fertilizer on their oil palm fields, at least once, between Jan 2020 and Dec 2022. There was a decline in number of farmers applying fertilizer from 47% (2020) to 29% (2022), although this was not consistent across sites. For example, the number of farmers applying fertilizer declined in RI, JB, CK, EK, increased in SS, and remained stable in WK. Notably, in JB, farmers stopped using fertilizer altogether in 2022. Farmers received an average price of 1885 IDR kg− 1 FFBFootnote 1, while average fertilizer costFootnote 2 was 5728 IDR kg1. The difference in fertilizer versus FFB price became larger over time, meaning that smallholders would need to sell more FFB to cover the cost of the same amount of fertilizer (Fig. 1).

Fig. 1
Fig. 1
Full size image

Frequency of farmers applying fertilizer per site over three years. Fertilizer includes urea, compound fertilizer containing nitrogen, phosphorous, and potassium (NPK), and muriate of potassium (MOP). Also shown are average fresh fruit bunch (FFB) and fertilizer prices for each year (circles and triangles, respectively). The analysis was based on farmer diary data collected from 958 farmers across six sites: Riau (RI), South Sumatra (SS), Jambi (JB), West Kalimantan (WK), Central Kalimantan (CK), and East Kalimantan (EK). Average FFB-to-fertilizer price ratio was 2.8 (2020), 2.9 (2021), and 3.4 (2022)

Our nutrient score averaged 11% across all smallholders, ranging across sites from 1% (JB) to 20% (CK), suggesting widespread nutrient imbalances across sites (Fig. 2). Smallholders in our study mostly applied urea and NPK (locally called ‘Phonska’), with 30% of farmers applying urea and 32% applying NPK at least once during the study period. (Supplementary Fig. S2). Both urea and NPK used to be subsidized for farmers by the Indonesian governmentFootnote 3 (up until July 2022), so it was not surprising to find that farmers preferred to use these less-expensive sources of fertilizers. Conversely, farmers rarely used sources of fertilizer richer in K, such as MOP and NPK+. Only 18% of farmers applied MOP and 14% applied NPK + at least once during the study period. Farmers from WK who belonged to a farmer group that collectively saved for and procured fertilizer were more likely to apply fertilizer richer in KFootnote 4.

Fig. 2
Fig. 2
Full size image

Range in relative nutrient scores across smallholder fields in each site. The scores were calculated based on farmer diary data on fertilizer use collected from 958 farmers across six sites over three years (2020–2022), and field-specific nutrient requirements estimated following Lim et al. (2023). Boxes represent the 25th and 75th percentiles while bars show the 10th and 90th percentiles. Horizontal lines within boxes are the median values. Red dots represent mean values; black dots represent individual fields. Letters indicate statistically significant differences (p < 0.05). See captions of Fig. 1 for site names

Farmers showed a preference towards compound fertilizer over a combination of straight fertilizer types. In the fertilizer survey, 41% of the respondents indicated that, if they had no availability or financial restrictions, they would still buy NPK fertilizer only and no additional fertilizer types. Most respondents (96%) said that fertilizer in their site was generally reliable, and just 12% said they ever encountered ‘fake’ fertilizer, although 90% of the respondents admitted not knowing how to recognize it. Farmers who reported having used fake fertilizer said that they observed that sometimes fertilizer did not dissolve well, and they presumed that this was fake fertilizer. Only 6% reported application of organic material (empty fruit bunches (EFB), bunch ash, manure, or household waste). Reasons mentioned for low use of organic amendments were the limited availability and high costs of both the material and transportation.

3.2 Explanatory causes for variation in fertilizer use and nutrient score

Dependency on oil palm for total annual household income, total palm area, farmer age, and farmer group membership were the most important variables explaining fertilizer use (Fig. 3a). However, we found differences per site. In SS, there was a positive correlation for fertilizer use and dependency on oil palm, but not for the other sites. In CK and EK, larger total palm areas positively corresponded with fertilizer application. Farmer age did not have a significant impact on fertilizer application across sites. For half of the sites (SS, WK and CK) farmers belonging to a farmer group applied more fertilizer, but no relation was found for the other sites. Notably, the Random Forest model achieved an Out-of-Bag (OOB) error rate of 43%, indicating low predictive performance. This means that the factors included in this model did not have a strong general impact on fertilizer use, but their effect was site specific, and other factors not included in this study may have been more important.

Fig. 3
Fig. 3
Full size image

Importance of variables from random forest analysis for predicting fertilizer use (a) and nutrient scores (b). Relative importance was calculated by comparing the relative contribution of each variable in relation to the variable with greatest explanatory power. For fertilizer use (categorical) the Out-of-Bag (OOB) error is given as a measure of prediction error; for nutrient scores (continuous) the r2 is given to predict goodness of fit of the model. Analysis was based on baseline survey and farmer diary data from 958 (fertilizer use) and farmer diary and fertilizer survey data from 337 (including only respondents with nutrient scores > 0) smallholder fields pooled across the six sites in Indonesia. See Table 2 for the list of variables included in the analysis

Fertilizer supplier and credit usage explained most of the variation in nutrient scores, although r2 of the model was low (Fig. 3b). According to the fertilizer survey, farmers obtained fertilizer from different suppliers, including village kiosksFootnote 5 (31%), town shopsFootnote 6 (25%), FFB collectors (20%), cooperatives (14%), farmer groups (18%), and door-to-door salesmen (6%). The nutrient score was higher for farmers who obtained fertilizer from FFB collectors or farmer groups compared to a town shop, village kiosk, or door-to-door salesmen (Supplementary Fig. S3). In turn, these farmers had better access to credit provided by those same FFB collectors or farmer groups, in contrast to other fertilizer suppliers (Fig. 4).

Fig. 4
Fig. 4
Full size image

Relative frequency credit usage per fertilizer supplier. Analysis was based on fertilizer survey data from 721 farmers across six sites

3.3 Enabling and constraining factors for fertilizer management

We further explored enabling and constraining factors influencing fertilizer management using data collected via interviews with farmers and a survey focused on fertilizer use.

3.3.1 Affordability

More than half of the farmers in our study (53%) did not apply fertilizer between 2020 and 2022. In the fertilizer survey, cost was identified as the main constraint for fertilizer management (52%), followed by availability (28%), and knowledge (16%). Labor availability, infrastructure and quality were mentioned by less than 2% of the respondents as main constraint. Among those who applied fertilizer in the past (46%), nearly all attributed their decision to stop applying fertilizer to costs. Interviewed farmers confirmed this finding, and 68% of the interviewees mentioned the impact of the exclusion of oil palm from the subsidy fertilizer program in 2022 (seven months prior to the interviews). Although some interviewees said that subsidized fertilizer had not been available in their location previously, for the majority this policy change, in combination with the global increase of fertilizer prices and general household expenses, made fertilizer less affordable. Interviewees said that increased costs disrupted their fertilizer routine. Now, they said that they only applied fertilizer when money was available rather than following a pre-determined schedule. For example, one interviewee said:

"What can we do? We apply what we can. But the problem with the current fertilizer price is…we will just buy what we need now, today. So, we do not store fertilizer. When we want to apply, well on that day we buy it. This means that sometimes on a certain day we have money, but the fertilizer is not available."

This finding was confirmed by the survey, which showed that less than half of the respondents applied fertilizer according to a fixed schedule (40%), while for others, application timing depended on their economic situation (31%), availability at their supplier (15%), weather conditions (5%), or signs of nutrient deficiencies (9%).

Interviewees suggested that farmers responded differently to the cost increase. One interviewee said that the increased FFB price sufficiently compensated for the increased fertilizer price:

“The fertilizer price increased, but the FFB price increased too. If the FFB price was still low like before, fertilizer would not be affordable.”

However, most considered that the fertilizer price increase was no longer compatible with the increased FFB price. To cope, some farmers turned to alternative nutrient sources. One-third of the survey respondents who stopped applying fertilizer resorted to applying only organic material, such as manure and empty fruit bunches. For example, one interviewee explained:

"I don’t apply fertilizer anymore. I don’t have the money. I use money to buy food, and for that I even don’t have enough. I just apply EFB and cow manure."

Interviewees also said that manure and EFB were not easily available and expensive, and difficult to transport, especially for empty fruit bunches which need to be transported by truck from the mill to farmers’ fields.

Others completely stopped applying nutrients. For example, one person said:

"I am a poor farmer. For us, the situation has changed after stopping the fertilizer program. We hardly apply fertilizer now; it is very expensive. We cannot fertilize anymore. Fertilizer from shops is not affordable."

Several respondents expressed concerns that their inability to apply sufficient fertilizer would lead to lower yields. Some already observed that their fruit bunches were getting smaller, and they had fewer bunches per harvest. For example, one interviewee said:

"Previously, I applied fertilizer, I had a good yield. I used to get 6–7 tons from 500 palms per harvest; I harvested every two weeks. But now, I don’t fertilize, and I only get one ton, it continues to decrease. I lengthened the harvest interval to 21 days. A large difference."

3.3.2 Fertilizer supplier and credit

For those who continued to use fertilizer, the type and dose applied was mainly influenced by supplier type and access to credit. Interviewees indicated that different fertilizer suppliers each had their advantages and disadvantages (Table 3). Village kiosks were considered to be convenient due to their proximity and delivery services to farmers’ fields, which is relevant for farmers who often do not own large vehicles to transport fertilizer bags. However, only 9% of the survey respondents purchasing from village kiosks used loans or payment arrangements and most had to pay upfront. Another disadvantage was that village kiosks typically offered a limited choice of brands and had small stock, whereas larger shops in towns generally had a more complete stock and wider choice of fertilizer options. Loans or payment arrangements were more frequent in town shops, with 23% of respondents buying here using credit. A response from one farmer suggests that fertilizer needs to be purchased regularly and/or farmers need to have an established trust relationship or family ties to village or town show owners to access credit:

"In general, you can’t buy fertilizer on credit from the shop, but it depends on your client relation, or perhaps the shop owner is family. I can obtain fertilizer on credit from the shop because I am a regular customer. Basically, we trust each other, and I pay off the fertilizer in three months’ time, because I place a new order. I pay 50% upfront."

Table 3 Overview of characteristics, services and prices per fertilizer type, based on qualitative data from 40 interviews
Full size table

Another option for farmers to purchase fertilizer was via FFB collectors, who often provided loans, either in cash or in kind, which could be repaid via deductions from FFB sale. Indeed, 69% of the respondents who obtained fertilizer from their FFB collector used credit (Fig. 4). Another advantage was that FFB collectors sometimes delivered fertilizer directly to farmers’ fields. Similar to village kiosks, the disadvantages of buying fertilizer from FFB collectors were the limited choice of brands and small stock. Interviewees explained that one advantage of taking loans from FFB collectors was that they trusted them, and payment arrangements were usually more flexible compared to payment arrangements available in shops. Typically, loans from FFB collectors had to be paid off within 2–3 months, but this period could be extended when farmers had financial difficulties, needed cash for other expenses such as school fees or festivities, or in times of a low yield season. For example, an interviewee who bought fertilizer from the FFB collector said:

"I often have difficulties paying for fertilizer. So, I will make an offer to the FFB collector: please buy fertilizer for me. It depends how long it takes to pay off the loan, sometimes 2–3 months. When the harvest is good… but if we need cash for other purposes, such as school fees…then we have to reduce the amount per payment."

Another option to buy fertilizer on credit was via a farmer group (70% of respondents who bought via farmer group used credit), but this was only possible in WK and SS. There was a farmer group in WK whose members collectively bought fertilizer. The group also had a communal credit system in place. Similarly, there was a farmer group in SS, headed by an FFB collector, who also distributed fertilizer on credit for the members of his group. There was a cooperative in CK, but it did not provide credit. Lastly, fertilizer can also be bought from door-to-door salesmen, who passed through the villages selling ‘left-over’ fertilizer from fertilizer or plantation companies from the back of trucks. Although their fertilizer was perceived to be less expensive compared to other fertilizer suppliers, it was also regarded to be of lesser quality.

3.3.3 Availability and logistics

In addition to financial constraints, survey respondents identified limited availability of fertilizer (28%) as a constraint for fertilizer use. In the fertilizer survey, half of our respondents said that fertilizer was not always timely available. Interviewees confirmed that they sometimes bought whatever was available from their local supplier. Interviewed village kiosk owners explained that farmers preferred to buy compound fertilizer from well-known brands, but that some NPK brands with higher K content, as well as fertilizer types that were relatively in low demand such as boron, were not always in stock at larger distribution centers. In interviews, farmers said that they had difficulties recognizing nutritional needs of their palm. For example, one person said:

"I pay attention to signs in the leaves, but I don’t know much on that. I just follow the recommendations from my farmer group."

Indeed, 45% of the survey respondents said that they relied on their farmer group for knowledge about fertilizer.

Transportation was not identified as a key constraint in the ranking exercise in the survey, but interviewees suggested that transportation of fertilizer to farmer fields was sometimes hampered by poor road conditions, heavy rains, and remote locations of some fields, as explained by a fertilizer kiosk owner:

"We can only transport fertilizer when the road is good, when the road is going uphill, we need to use a motorcycle. For some farmers I can bring the fertilizer directly to their field, but not for others."

Interviewees particularly mentioned logistical challenges for obtaining empty fruit bunches, as the roads towards their field were often too narrow for trucks to enter.

4 Discussion

Sufficient and balanced use of fertilizer is vital to close the yield gap for oil palm smallholders and increase their profit. Yet, we found that more than half of the farmers in our study sites did not apply any fertilizer between Jan 2020–Dec 2022 (Fig. 1), and those who did apply fertilizer often applied amounts that are well below those needed to meet plant requirements (Fig. 2). Furthermore, formerly subsidized fertilizer sources were rich in N and P, such as urea and NPK phonska, whereas farmers seldom used fertilizer types that are richer in K, such as NPK + or MOP. As a result, potassium has become a major yield-limiting factor for oil palm smallholders in Indonesia (Lim et al., 2023; Sugianto et al., 2023; Thoumazeau et al., 2024). A limitation of this study is that it is based on a nutrient score that is based on crop modelling rather than measured nutrient deficiencies via leaf or soil analyses. The actual nutrient deficiencies in the farmers’ fields may also have been impacted by local soil conditions, and may change over time due to yield fluctuations. Therefore, our nutrient score indicator should be interpreted as an approximation of nutrient deficiencies although the scores significantly correlate with yield. Moreover, previous studies using the same database have shown strong correlations between FFB yield, nutrient application, and leaf tissue analysis (Sugianto et al., 2023; Lim et al., 2023).

Our study went further than these previous studies, exploring the factors that influence fertilizer use, showing that 52% of farmers considered costs to be the main bottleneck for applying fertilizer, followed by availability (28%), and knowledge (16%). Yet, low fertilizer use was not merely a result of a lack of capacity or low awareness among farmers. From the case studies conducted in RI, SS, WK and CK, we found structural constraints, including a lack of credit options; limited availability due to low stock capacities of suppliers in the villages, as well as distribution problems at higher levels; transportation difficulties due to poor road conditions; and perceived incidence of ‘fake’ fertilizer. This is consistent with previous findings reported for oil palm smallholders in Indonesia and elsewhere (Essono et al., 2023; Rhebergen et al., 2018; Sahara & Kusumowardhani , 2017; Somnuek & Slingerland, 2018), and other perennial crops such as cocoa (Hoffmann et al., 2020). Our respondents indicated that financial constraints had become particularly stringent since subsidized fertilizer became unavailable in 2022. Prior to that, the rise in global fertilizer prices due to geopolitical events (Alexander et al., 2023), and comparably smaller increase in FFB prices, already increased the perceived risk of investing in fertilizer, with the frequency of farmers using fertilizer dropping from 47% (2020) to 29% (2022). This issue remains relevant given the continued link between fertilizer prices and geopolitical instability.

Respondents in our study did not explicitly use the term ‘risk’ to describe their challenges in accessing fertilizer, but they did refer to both fluctuations in yield, as well as fertilizer and FFB prices to explain why they considered fertilizer too expensive. One respondent coined this saying: “if the FFB price is unstable, fertilize application is irregular.” Although this was not explicitly stated by respondents, farmers’ reluctance to invest in fertilizer may partly be explained by the nature of perennial crops which are semi-permanent and will produce some fruits regardless of fertilizer applications, albeit well below attainable yield. This means that fertilizer application can be postponed in times of financial difficulties or uncertain markets, although this will have consequences for long-term yield, which is something our interviewed farmers seemed to be aware of. In our previous study on harvesting practices, we found that low yields in turn lead to longer harvesting intervals, further exacerbating low yields (De Vos et al. 2023a). Yet, interviewees also explained that in context of high living costs larger sums of cash were preferably spent on education, household needs, and family celebrations rather than fertilizer, perhaps because the benefits of spending money on such things materialize quicker than increased FFB yield would, and there are limited options to pay for such expenses on credit. Priorities are possibly different for farmers with annual crops, who strongly depend on a successful harvest season for their income, and the effects of fertilizer inputs and yield materialize on a shorter term. For example, Rizzo et al. (2023) found that rice and maize farmers in Indonesia depended for > 75% on these crops for their income, while our oil palm farmers on average depended for 50–75% on oil palm, and often had additional sources of income. Rice and maize farmers in Indonesia are also supported by strong extension services and have access to subsidized fertilizer, seeds, and chemicals, leading to comparably higher yields in relation to oil palm smallholders.

Farmers who obtained fertilizer from FFB collectors or via a farmer group had significantly higher nutrient use and nutrient scores, which can be partly attributed to better access to credit, although their nutrient scores were still below what is required for optimal yields. Notably, farmers from WK who were members of a particular farmer group that also provided fertilizer more often used fertilizer rich in K. Our study highlights the importance of using informal credit as a strategy to continue fertilizer use. As Sahara et al. (2017) pointed out, oil palm smallholders generally have limited access to formal credit, because banks require complex administration, including records of yields and production costs, and formal land titles as collateral. These requirements are particularly difficult to meet for independent oil palm smallholders who frequently lack documentation of land legality and tend not to record yields and expenses (De Vos et al. 2023b). In this context, our study shows that informal credit, as provided by FFB collectors and farmer groups, can be a solution for farmers to continue fertilizer use (Fig. 4). We found that the advantage of informal credit arrangements was that it allowed for more flexible payment arrangements tailored to farmers’ economic situations. A disadvantage, as shown by Sahara et al. (2017), could be that interest rates for informal credit are generally higher compared to banks, but a report by Daemeter Consulting (2016) mentions that FFB collectors usually have zero-interest rates, because the repayment flows via deductions from FFB sales. Although we did not collect quantitative data on interest rates, indeed interviewees mentioned a zero-interest rate, although some said they could not afford the reduction of revenues from FFB sales either. Another disadvantage may be that credit relations usually tie farmers to their FFB collector, limiting freedom to choose where to sell FFB (Watts et al., 2021), but our interviewees did not describe this practice as a disadvantage. In fact, given the central role of FFB collectors and farmer groups in access to fertilizer and credit, they are well placed to facilitate knowledge distribution among farmers, when FFB collectors and farmer group leaders are trained in good nutrient management.

How to improve oil palm smallholder access to fertilizer via policy? An option could be to re-install the previous subsidy program but tune it to favor the use of sources of fertilizer that are better suited for oil palm, as described by Lim et al. (2023). Alternatively, others have proposed to provide funds directly to farmers via ‘farmer cards’, which farmers can use to buy agricultural inputs at registered agri-input suppliers (Alta et al., 2021). Yet, our study shows that in 2020 and 2021, when farmers could still obtain nationally subsidized fertilizer and prices were relatively low, half of our respondents still did not apply fertilizer. Hence, while credit and incentives programs may help, smallholders also need access to technical information to inform their nutrient management decisions. Reaching this end would require government policies to support smallholders, and strengthening current extension services, assistance from nearby large plantations, addressing fertilizer distribution problems, and development of simple, easily accessible recommendations/tools to inform nutrient management. NGOs or government extension institutions could also play an active role in organizing farmers to obtain access to fertilizer, for example via collective saving and organizing transport, as well as with providing training in good nutrient management.

5 Conclusion

Smallholders in our study applied little or no fertilizer, leading to nutrient deficiencies and low FFB yields. In context of globally high fertilizer prices and disappearance of subsidized fertilizer, costs were important constraints to apply fertilizer. Other constraints included limited fertilizer availability, and poor infrastructure. Informal credit, based on flexible repayment arrangements, as provided by FFB collectors and farmer groups, helped farmers to continue fertilizer use. To conclude, enhancing fertilizer use among oil palm smallholders requires policy interventions to increase knowledge on good fertilizer management among smallholders and fertilizer suppliers, as well as policies to facilitate access to credit, and timely availability of good quality fertilizers. This is vital for supporting smallholder incomes, promoting sustainable palm oil production, and optimizing existing oil palm plantations to reduce the incentive for further cropland expansion and deforestation.