We reviewed the knowledge available in the literature for all the input, internal, and output fluxes identified in Fig. 2.
Biological N fixation
One input is the biological fixation of N from the atmosphere (flux no. 1 in Fig. 2), which is carried out by specific bacteria. Three types of fixation were mentioned in oil palm plantations: endophytic fixation inside the tissue of a palm colonized by bacteria (e.g., Azospirillum Reis et al. 2000), non-symbiotic fixation which takes place in the litter or soil (e.g., Azobacter Aisueni 1987), and symbiotic fixation in the nodules of the roots of legumes (e.g. Rhizobia). Regarding endophytic fixation, Amir et al. (2001) reported an uptake of fixed N by palm seedlings in the greenhouse following inoculation with Azospirillum bacteria and Om et al. (2009) reported higher leaf protein and chlorophyll content in 280-day-old oil palm plants inoculated with Acetobacter. These results suggested that endophytic fixation is a flux of N input not negligible in oil palm systems, but other studies are necessary to obtain estimates of the magnitude of this flux.
The results regarding non-symbiotic fixation have so far been inconsistent or difficult to replicate in the field (Tinker and Nye 2000
in Corley and Tinker 2003). The magnitude of such inputs from non-symbiotic fixation might be similar to those in tropical forest ecosystems, which are on average 3.3–7.8 kg N ha−1 year−1, with a tendency to increase with temperature, soil moisture, and soil N scarcity (Reed et al. 2011).
Finally, for symbiotic N fixation, recent reviews were done on oil palm plantations (Giller and Fairhurst 2003; Ruiz and López 2014). Most of the quantifications of N fixation were made in Malaysia in the 1980s and 1990s, mostly with P. phaseoloides, and also M. bracteata, Calopogonium pubescens, and Calopogonium muconoides. Two main methods were reported: 15N isotope labeling and deduction from other fluxes with N budget approaches. The estimates of N fixed by legumes were very similar, with an average of 150 kg N ha−1 year−1 over the first 5 years (Broughton et al. 1977; Agamuthu and Broughton 1985; Zaharah et al. 1986). A more recent work reported amounts of N biologically fixed of 0.3 to 34.2 kg N ha−1 in legume covers under oil palm in shoots and litter, but more research would be needed to take into account fixed N in roots (Pipai 2014). However, Giller and Fairhurst (2003) noted that most estimates of fixation are likely to be underestimates, as they were all based on harvested legume plants without taking into account the biologically fixed N continually added to the litter through residue cycling.
The other main N input is via the application of mineral (flux no. 2 in Fig. 2) and organic fertilizers (flux no. 3 in Fig. 2) such as empty fruit bunches and palm oil mill effluent. Several studies were done on fertilizer efficiency and several papers propose fertilizer recommendations, but few data are easily available on actual amounts of mineral and organic fertilizers applied in plantations. The amount of mineral fertilizer applied is very variable and ranges from 48 to 90 kg N ha−1 year−1 for immature palms (Henson 2004; Banabas 2007; Choo et al. 2011) and from 56 to 206 kg N ha−1 year−1 for mature palms (Foster 2003; FAO 2004; Carcasses 2004, unpublished data; Hansen 2007; United Plantations Berhad 2006; Wicke et al. 2008). It seems to be a common practice to reduce or even stop fertilizer application over the 2–3 years before felling (Choo et al. 2011), despite evidence that effects of N fertilizer on yield do not always persist from 1 year to the next (Caliman et al. 1994). The amount of fertilizer applied is adapted over time mainly on the basis of foliar N contents. This amount hence depends indirectly on the age of the palms, the soil and climate conditions, and the planting material which influences the potential yield.
The main types of N fertilizers used in oil palm are urea, containing 46 % of N, used everywhere; ammonium sulfate, 21 % of N, mainly used in Southeast Asia; and ammonium nitrate, 34 % of N, used in Africa and South America (Corley and Tinker 2003; Goh and Härdter 2003; Banabas 2007). The main factors governing the choice of fertilizer type are the availability, e.g., related with legal framework; the cost per unit N, including transport; and the local soil and climate conditions. The choice of the type of fertilizer is critical for N cycling processes and there might be tradeoffs between these selection factors. For instance, urea is less costly than other types, but it may produce high gaseous losses of NH3 in dry conditions (Goh et al. 2003). A common practice is to manually apply the fertilizers in an arc around the palm, using calibrated containers to deliver the required amount to each tree. For immature palms, it is applied close to the palm (Goh et al. 2003; Caliman et al. 2002). For mature palms, application practices vary. Applications can be made manually on the weeded circle, on the edge of the weeded circle, and even on the frond piles where more feeding roots are found and fewer losses may occur through runoff (Banabas 2007). Broadcast mechanical applications by tractors using spreaders with deflectors are now often used where labor is expensive or in short supply (Goh and Härdter 2003). Aerial application is also a developing practice but mainly used on peat soils and steeply sloping areas where mechanical application is not possible (Caliman et al. 2002). It is a common practice to split the application of N fertilizers in 2 or 3 per year, depending on soil type and rainfall distribution, to reduce the risk of nutrient losses. In immature palms, the splitting is usually increased to 4 to 5 applications per year because of the use of various fertilizers that cannot be systematically combined together (Goh et al. 2003; Banabas 2007). The optimal frequency is therefore a compromise between the need to meet nutrient demand, labor cost, risk of nutrient losses, and logistical issues for transport and storage (Goh et al. 2003). Fertilizers are normally applied after rainfall when the soil is wet, especially for urea to limit volatilization, but not during heavy rain periods to avoid losses through leaching, runoff, and erosion. However, there are situations where labor availability is also an important factor which influences the timing of applications (Banabas 2007).
Empty fruit bunches are commonly returned directly to the plantation from the mill after oil extraction, with an addition of supplementary mineral N (Corley and Tinker 2003). A plantation yielding 22 t of fresh fruit bunches per hectare would produce empty fruit bunches for only about 10 % of the mature plantation area. This estimate results from the assumptions that the weight of empty fruit bunches produced is 20 to 25 % of the weight of fresh fruit bunches processed (Corley and Tinker 2003; Redshaw 2003) and that the application rate of empty fruit bunches is 50 t.ha−1 (Redshaw 2003). Thus, there is not enough empty fruit bunches for the whole plantation area and the preferential areas for spreading are those close to the mill and on relatively flat terrain, for reasons of cost and feasibility (Redshaw 2003). Soils with low carbon content are also favored because empty fruit bunch inputs increase their organic matter content (Carcasses 2004, unpublished data). This uneven distribution of empty fruit bunches creates a spatial heterogeneity of organic N input at the plantation scale.
Under immature palms, empty fruit bunches are applied in a single layer immediately around the palms. Annual applications of 15 to 60 t ha−1 are common, and even larger rates of 80 t ha−1 may be used on an 18-month or 2-year cycle (Redshaw 2003). Under mature palms, empty fruit bunches are usually spread in the harvest pathway or in some cases in between palms in the row in order to keep the weeded circle easily accessible for harvest. Application rates of 30 to 60 t ha−1 are common (Banabas 2007; Redshaw 2003). The empty fruit bunches contain from 0.26 to 0.38 % N in fresh matter (0.65 to 0.94 % in dry matter) (Corley et al. 1971; Singh et al. 1982; Singh 1999; Gurmit et al. 1990, 1999; Caliman et al. 2001b). Empty fruit bunch application rates vary widely. Hence, the associated inputs of N are also very variable ranging from 39 to 228 kg N ha−1 year−1 under immature palms and from 78 to 228 kg N ha−1 year−1 under mature palms. In addition to direct application to fields, empty fruit bunches are also used to produce compost, with the advantage of reducing the volume of biomass to transport for field application. Empty fruit bunches are commonly mixed with palm oil mill effluent or urea, and the final N content of compost ranges from 1.5 to 2.7 % in dry matter (Lord et al. 2002; Siregar et al. 2002; Schuchardt et al. 2002
in Redshaw 2003).
Palm oil mill effluent is often spread in the plantations following treatment in ponds. The treatment ponds are designed to decrease biological oxygen demand. Depending on the treatment, palm oil mill effluent contains from 0.92 to 1.2 kg N t−1 (Redshaw 2003; Corley and Tinker 2003; Corcodel et al. 2003; Schmidt 2007). The rate and frequency of application depend mainly on the maximal rate legally allowed and on the application system, but one reported application rate was about 375 t ha−1 year−1 split in three applications (Carcasses 2004, unpublished data). At that rate, the inputs of N generated are rather high at approximately 345 to 450 kg N ha−1 year−1. As for the empty fruit bunches, palm oil mill effluent is spread onto only a small portion of the whole plantation area, dictated by the application system and the distance between the mill and the field. Several application systems are used, such as gravity flow, pipe irrigation with a pump, or application by a tractor with a tanker (Lim 1999; Redshaw 2003).
The N inputs that are the most difficult to quantify and least well known are those from atmospheric (flux no. 4 in Fig. 2) and sediment depositions. At a global scale, production of Nr, such as NH3 and NO
, by lightning and volcanic activity is small (Galloway et al. 1995; Mather et al. 2004), but it may be significant in some oil palm-growing regions. To our knowledge, only measurements of wet deposition have been done in oil palm systems, i.e., for N contained in rain water (possibly including aerosols). Depositions were reported to range from 14.6 to 20 kg N ha−1 year−1 in Malaysia (Agamuthu and Broughton 1985; Chew et al. 1999) and were measured at 8 kg N ha−1 year−1 in Brazil (Trebs et al. 2006).
N inputs also result from the deposition of eroded particles of soil coming from upslope of the system studied. This flux concerns mainly lowland areas where the eroded soil from upper areas accumulates and hence it depends on the local topography. To our knowledge, no specific measurements of N deposition have been done to estimate this input flux in palm plantations. Finally, input of N to ecosystems from weathering of rocks is usually considered to be negligible. However, it is possible that it constitutes a significant input if the geology consists of fine sedimentary rocks (Holloway and Dahlgren 2002), given the intense weathering conditions of oil palm-growing regions.
In summary, N inputs were estimated, in kg N ha−1 year−1, at 150, 0–206, 0–450, 8–20, for biological N fixation, mineral fertilizer, organic fertilizers, and atmospheric deposition, respectively. The results and references are synthesized in Table 1.
A major internal flux is the N uptake from soil by palms, legume cover crops, and other plants, mainly as inorganic N (NH4
+ and NO3
−) (flux no. 5 in Fig. 2). Uptake by plants other than palms and legumes may be significant because it is known to compete with palms and affect fresh fruit bunches production (Corley and Tinker 2003). However, to our knowledge, no measurements of such uptake terms are available. For the legume cover, Agamuthu and Broughton (1985) estimated that 149 kg N ha−1 year−1 was taken up from the soil over the first 3 years of the oil palm cycle. For palms, two main reviews have reported estimates of N uptake (Xaviar 2000; Goh and Härdter 2003), with most of the work done on Dura palms in Malaysia and Nigeria between the 1960s and 1990s. Other work was done more recently on Tenera palms in Sumatra (Foster and Parabowo 2003). In all cases, estimates reported are not direct measurements of N uptake by roots but indirect estimates inferred from a nutrient budget approach. Thus, over the whole growth cycle, the net N uptake is considered to be equal to the N immobilized in the palm, above- and below-ground biomass; the N released in palm residues such as pruned fronds, removed inflorescences, frond bases, dead roots; and the N exported in harvested bunches.
The results reported by Xaviar (2000) and Goh and Härdter (2003) showed that uptake rate mainly depends on the age of the palms, with estimates of 40 kg N ha−1 year−1 for 0 to 3-year-old palms (Tan 1976, 1977) and ranging from 114 to 267 kg N ha−1 year−1 for 3 to 9-year-old palms (Ng 1977; Pushparajah and Chew 1998; Henson 1999; Ng et al. 1999; Ng and Thamboo 1967; Ng et al. 1968; Tan 1976, 1977). However, recent work has resulted in considerably higher estimates of uptake by Tenera palms, up to 272 kg N ha−1 year−1 in 10-year-old palms and even 380 kg N ha−1 year−1 in adult palms (Foster and Parabowo 2003). Both studies considered only above-ground biomass in the budgets. This difference could be explained by the higher yields now obtained with current genotypes (Goh and Härdter 2003). Recent measurements in trials in Indonesia showed uptake rates by above-ground biomass ranging from about 221 to 272 kg N ha−1 year−1, depending on the planting material. In addition to genotype, variability of uptake seems to be linked with soil and climate conditions. For example, uptake was estimated at 149 kg N ha−1 year−1 in Nigerian conditions with a production of 9.7 t of fresh fruit bunches ha−1 year−1 (Tinker and Smilde 1963) and at 191 kg N ha−1 year−1 in Malaysian conditions with a production of 24 t of fresh fruit bunches ha−1 year−1 (Ng and Thamboo 1967; Ng et al. 1968).
N from plant residues to the litter
Another major internal flux is the N contained in plant residues, which goes from the plants to the litter (flux no. 6 in Fig. 2). Residues come from the palms, legume cover crops, and other vegetation. For plants other than palms and legumes, to our knowledge no data is available. For legume cover, Agamuthu and Broughton (1985) estimated an amount of 123 kg N ha−1 year−1 going from the living plants to the litter over the first 3 years under oil palm and Pushparajah (1981) estimated an amount of about 120–160 kg N ha−1 year−1 over the first to the third years and less than 40 kg N ha−1 year−1 over the fourth to the seventh years under rubber trees. In both cases, root turnover was not taken into account. For palms, several residues are distinguished: those produced throughout the crop cycle, mostly in the mature phase such as pruned fronds, removed inflorescences, frond bases, root exudates, and dead roots and those produced only once before replanting, i.e., the whole palm when it is felled.
For pruned fronds, the flux of N depends on the quantity of fronds pruned and their N content. Frond production rate stabilizes after 8–12 years at about 20–24 fronds year−1 (Corley and Tinker 2003). Several publications estimated the annual flux of N going to the litter, with values ranging from 67 to 131 kg N ha−1 year−1 (Redshaw 2003; Carcasses 2004, unpublished data; Turner and Gillbanks 2003; Schmidt 2007). Therefore, this flux is uncertain and the reasons for the variability are not well defined; they may depend on the soil, climate, and planting material which influence frond production and frond weight and on the methods of measurement of N content. For male inflorescences, the flux of N going to the litter has been ignored in most N cycling studies. We found only two estimates, being 6 and 11.2 kg N ha−1 year−1 (Carcasses 2004, unpublished data; Turner and Gillbanks 2003, respectively). These estimates suggest that this flux is lower than the uncertainty of the concomitant N flux via pruned fronds. For frond bases, which rot and fall naturally from the trunk, the only estimate we found was of 3 kg N ha−1 year−1 going to the litter (Carcasses 2004, unpublished data).
For root exudates and transfers into the soil via Mycorrhizae, no estimate of N flux is available to our knowledge. Roots themselves are continuously dying and being replaced by new ones. This death of roots constitutes a flux of N going from the palm to the litter pool and depends on the rate of root turnover and on the N content of roots when they die. Root turnover is very difficult to measure. Corley and Tinker (2003) reviewed several methods to estimate it such as deduction from measurements of soil carbon balance or measurements of the growth of roots after extracting soil cores and refilling the holes with root-free soil. Estimates of average turnover ranged from 1.03 to 11.5 t of dry matter ha−1 year−1 for adult palms (Dufrêne 1989; Lamade et al. 1996; Henson and Chai 1997; Jourdan et al. 2003), and turnover was reported to be zero for 3–4-year-old palms (Henson and Chai 1997). Thus, with an average root N content of 0.32 % of dry matter measured by Ng et al. (1968) in 8–15-year-old palms in Malaysia, the average N flux from root turnover would range from 3.3 to 36.8 kg N ha−1 year−1. (Carcasses 2004, unpublished data) also proposed the value of 7.5 kg N ha−1 year−1 based on data from Henson and Chai (1997). Therefore, this flux is highly uncertain. Moreover, Corley and Tinker (2003) noted that root turnover measured in Malaysia was much lower than that in Africa, which could be explained by the death of a larger part of the root system in Africa during the annual dry season (Forde 1972).
Finally, the estimate of the N contained in the felled palms must take into account above- and below-ground biomasses. Several publications estimated the weight of dry matter of above-ground biomass of old palms at felling and the N content of their different tissues, i.e., trunk, fronds, inflorescences, and frond bases (see for e.g., Corley and Tinker 2003). Some of them reviewed available data to estimate the total N content of palms at felling and reported values ranging from 400 to 577 kg N ha−1 (Khalid et al. 1999a; Redshaw 2003; Schmidt 2007). Fewer studies estimated the below-ground dry matter of palms, but Khalid et al. (1999b) reported a value of 65 kg N ha−1. Therefore, the total N contained in palms at felling and going to the litter has been estimated at 465 to 642 kg N ha−1.
N from the litter to the soil
Another important internal flux is the mineralization or incorporation of N from the litter to the soil (flux no. 7 in Fig. 2). The litter is composed mostly of plant residues but also contains active microorganisms and fauna. To our knowledge, no data is available regarding the decomposition of residues from plants other than oil palm or legumes in the oil palm system.
For legume litter decomposition, Chiu (2004) measured losses of about 70 % of dry matter after about 2–3 months in leaves and stems of P. phaseoloides and M. bracteata. But the net N release follows a slower dynamic due to the immobilization of the N by the microbial fauna and flora involved in decomposition and the partial uptake of the N released by growing legumes. For instance, Vesterager et al. (1995) measured in a pot experiment with P. phaseoloides a net release of about 25 % of the N of the legume litter after 2 months, using a 15N labeling technique. In an oil palm field, Turner and Gillbanks (2003) reported that net N release from legume litter occurred between the 24th and the 30th months after planting.
For palm residues, no data was found for frond bases. For pruned fronds and felled and chipped trunks, Khalid et al. (2000) observed a loss of 50 % of dry matter after 6–8 months and a total decomposition after 12–18 months. For roots, Khalid et al. (2000) observed a loss of 50 % of dry matter after 10 months and a total decomposition after about 25 months. These decomposition rates were considered as approximately linear by Khalid et al. (2000), but Moradi et al. (2014) observed an exponential decrease with a faster decomposition over the first 5 months. Khalid et al. (2000) identified rainfall distribution as the main climatic factor controlling the rate of decomposition and observed that shredded residues decompose faster than un-shredded residues. For empty fruit bunches, when mineral N fertilizer was also added, losses of 50 % of dry matter were reported after 2–3 months (Turner and Gillbanks 2003; Lim and Zaharah 2000; Rosenani and Hoe 1996), and total decomposition occurred within 6 to 12 months (Rosenani and Hoe 1996; Henson 2004; Caliman et al. 2001b). The decrease followed an exponential dynamic (Lim and Zaharah 2000); the decomposition was faster when empty fruit bunches were applied in one layer than in two layers (Lim and Zaharah 2000) and was slower without addition of mineral N (Caliman et al. 2001b). However, for all of these palm residues, the dynamics of N release is more complex than the dynamics of decomposition due to immobilization by the microbial fauna and flora involved in decomposition. For instance, for trunks, Kee (2004) observed that the net release of N occurred only 12 months after felling. For empty fruit bunches, Zaharah and Lim (2000) observed a complete N immobilization over their experimental period of about 8 months, and Caliman et al. (2001b) reported a N release of only 50 % at about 6 months, without adding mineral N.
The last internal flux considered is the mineralization of soil organic N (flux no. 8 in Fig. 2). Only few data are available, and they involve various soil depths, which hampers comparison. Schroth et al. (2000) estimated the net mineralization in the top 10 cm of a central Amazonian upland soil at approximately 157 kg N ha−1 year−1 after 15 years of oil palm production without any N fertilizer inputs. Khalid et al. (1999c) estimated the N mineralization after replanting in Malaysia at about 312 kg N ha−1 year−1 in fields without residues from the previous cycle except dead roots and at about 421 kg N ha−1 year−1 in fields where the palm residues from the previous cycle were left on the soil. Finally, Allen et al. (2015) estimated the N mineralization in the top 5 cm of soil in Sumatra at about 920 kg N ha−1 year−1 in loam Acrisol and up to 1528 kg N ha−1 year−1 in clay Acrisol. However, those measurements were done under more than 7-year-old oil palms established after logging, clearing, and burning of either forest or jungle rubber.
In summary, internal fluxes were estimated, in kg N ha−1 year−1, at 149, 40–380, 0–160, 76–182, and 157–1528, for legume uptake, oil palm uptake, legume residues decomposition, oil palm residues decomposition, and soil N mineralization, and 465–642 for the decomposition of the felled palm at the end of the cycle. The results and references are synthesized in Table 2.
N exported in fresh fruit bunches
A major output is the N contained in fresh fruit bunches and exported during harvest (flux no. 9 in Fig. 2). The N content of the fresh fruit bunches was reported to be around 2.89–2.94 kg N t−1 of fresh fruit bunches in fresh weight (Ng and Thamboo 1967; Ng et al. 1968; Hartley 1988; in Corley and Tinker 2003 and Goh et al. 2003), but some higher values were also reported, as much as 6.4 kg N t−1 fresh fruit bunches (Ng et al. 1999). In general, the fresh fruit bunches production starts at about 2–3 years of age and increases rapidly until leveling off at yields around 10–34 t of fresh fruit bunches ha−1 year−1 after the tenth year (Tinker 1976; Corley and Tinker 2003). Some very high yields were also reported at around 40 t of fresh fruit bunches ha−1 year−1 (Kee et al. 1998). Thus, the yield depends on the age of the palm, but it also differs with the type of planting material, soil, and climate conditions. For instance, yields were reported to be lower in Nigeria (9.6 t FBB ha−1 year−1) than in Malaysia (24 t FBB ha−1 year−1) (Tinker 1976). Therefore, for adult palms more than 10 years old producing 10 to 34 t of fresh fruit bunches ha−1 year−1, we deduced an export of N through harvest of around 30 to 100 kg N ha−1 year−1, consistent with other estimates done for Nigeria (Tinker and Smilde 1963) and Malaysia (Ng and Thamboo 1967; Ng et al. 1968).
Soluble forms of N (NO3
− and NH4
+) can be lost by leaching out of the root zone (flux no. 10 in Fig. 2). Tropical soils may have significant anion exchange capacity and thus retain NO3
− (Rasiah et al. 2003), but such anion exchange capacity is usually not significant within the root zone. As most of the oil palm root activity is located within 1 m depth (Ng et al. 2003; Corley and Tinker 2003) and rainfalls are high in the tropics, this suggests a high potential risk of nutrient leaching under oil palm.
Many studies investigated the losses of N through leaching in plantations and were reviewed by Corley and Tinker (2003) and Comte et al. (2012). Most of the research was done in the 1980s and 1990s in Malaysia. Different plot-scale methods were used, such as lysimetric measurements, suction cup, and soil core sampling, and some studies were done at a larger scale with catchment sampling (e.g., Ah Tung et al. 2009). The age of the palms is one of the main control variables which can be identified. The measured values varied over a wide range, from 1 to 34 % of N applied (Omoti et al. 1983; Foong et al. 1983; Chang and Abas 1986; Foong 1993; Ng et al. 1999; Henson 1999; Ah Tung et al. 2009). Of the fertilizer N applied, 10.9 to 26.5 % was lost with palms less than 4 years old (Foong et al. 1983; Foong 1993) versus 1 to 4.8 % for palms older than 5 years (Foong et al. 1983; Foong 1993; Ah Tung et al. 2009). Only Omoti et al. (1983) reported losses of 34 % of N applied in Nigeria for palms from 4 to 22 years old.
In the conditions studied and despite very large variability, measurements hence showed that high losses through leaching are restricted to the first years of the palms, when the root systems are not fully developed and N inputs from decomposing plant residues are large. Moreover, fertilizer placement may have a significant effect on leaching because of the spatial variability of application rate, rainfall as through fall and stem flow, and N uptake (Banabas et al. 2008; Schroth et al. 2000). However, there is little information about the spatial distribution of NO3
− leaching within the plantation.
N losses through runoff and erosion
N can also be lost through runoff (flux no. 11 in Fig. 2) and erosion (flux no. 12 in Fig. 2) as a solute (NO3
− and NH4
+) or as eroded particles of soil containing N. Corley and Tinker (2003) and Comte et al. (2012) reviewed measurements of N losses through runoff and erosion from oil palm plantations. Research was done in Malaysia from the 1970s to the 1990s (Maena et al. 1979; Kee and Chew 1996) and more recently in Papua New Guinea (Banabas et al. 2008) and Sumatra (Ribka 2014). The main variables studied were the effect of soil type, slope, and spatial heterogeneity resulting from management practices, such as soil cover management. The variability of reported values is less than for leaching, ranging from 2 to 15.6 % of N applied lost through runoff, and from 0.5 to 6.2 % of N applied lost through erosion (Maena et al. 1979; Kee and Chew 1996). Spatial heterogeneity of soil cover seems to have an important effect on losses. Maena et al. (1979) reported losses through runoff of 2 % of N applied in frond piles, but 16 % of that applied in the harvest pathway. Ribka (2014) showed that 10 to 37 t of soil ha−1 year−1 were lost through erosion of bare soil, depending on slope, but this reduced to 2 to 4 t of soil ha−1 year−1 with a standard vegetation cover and the same slopes.
These results indicated that soil cover has a significant effect on both runoff and erosion under oil palm. However, data is lacking concerning the transition between the felling of palms and the early development of young palms when the soil is not yet covered by the legume. Finally, it can be noted that in a given situation, there is a balance between runoff/erosion losses and leaching losses, in which soil permeability plays an important role. For instance, in Papua New Guinea, Banabas et al. 2008 estimated losses through leaching at about 37–103 kg N ha−1 year−1 and negligible runoff, even with a high rainfall of 3000 mm year−1. The authors suggested that the high permeability of volcanic ash soils could favor leaching over runoff.
N gaseous losses
A potentially important gaseous output is the volatilization of NH3 (flux no. 13 in Fig. 2), which can occur directly from the leaves and from soil after fertilizer application, especially urea. Regarding emissions from palm fronds and other vegetation in the system, to our knowledge, no measurements have been reported. For emissions from soil following fertilizer application, several studies were done into urea efficiency under oil palm (e.g. Mohammed et al., 1991) but only a few measured NH3 volatilization. Most of them were done in Malaysia between the 1960s and the 1980s, and they often compared urea and ammonium sulfate, the most commonly used fertilizers in oil palm plantations. Two studies were done in Malaysia using different fertilizer rates (125 and 250 kg N ha−1 year−1) and on different soil types. Reported volatilization rates from urea ranged from 11.2 to 42 % of N applied (14 to 105 kg N ha−1 year−1), and volatilization from ammonium sulfate ranged from 0.1 to 0.4 % of N applied (0.1 to 0.5 kg N ha−1 year−1) (Sinasamy et al. 1982; Chan and Chew 1984). Another experiment was carried out in Peru by (Bouchet 2003, unpublished data) with a lower fertilization rate (85 kg N ha−1 year−1). The study found that 4 to 16 % of N applied in urea was volatilized (3.4 to 13.6 kg N ha−1 year−1), with higher volatilization under vegetation cover and no volatilization from ammonium sulfate. Therefore, given the few studies done and the high variability of the results, the magnitude of losses and the reasons for variations are uncertain. For urea, the highest values were in sandy loam soils with high application rates, and for ammonium sulfate the highest values were in clay soils with high application rates, but they did not exceed 1 % of N applied.
Gaseous emissions of N2O, NO
, and N2 are produced by soil microorganisms, principally through nitrification and denitrification (flux no. 14 in Fig. 2). Tropical soils are considered as important sources of N2O due to rapid N cycling (Duxbury and Mosier 1993). As N2O and NO
emissions are difficult to measure and have a very high variability, very few measurements were carried out in oil palm (Corley and Tinker 2003; Banabas et al. 2008; Banabas et al. 2008; Nelson et al. 2010). Maybe due to the recent growing concern about greenhouse gases emissions, most of the measurements available were done in the 2000s and most of them involved peatlands (e.g. Melling et al. 2007). To our knowledge, only two trials were carried out under oil palm on mineral soils. They focused on N2O emissions and showed very variable results whose average values ranged from 0.01 to 7.3 kg N ha−1 year−1. Emissions tended to decrease with the age of palms and to be higher in poorly drained soils. Potential N2O emissions are high in poorly drained soils due to limited N uptake by plants and conditions that are conducive for denitrification.
The first study showed N2O emissions ranging from 0.01 to 2.5 kg N ha−1 year−1 in Indonesia (Ishizuka et al. 2005). The highest values were reported for young palms while the lowest were reported for old palms. Ishizuka suggested that the high emissions under young palms could result from the low uptake of young palms being concomitant with the application of fertilizer and the fixation of N by the legume cover. Conversely, the low emissions under old palms could result from the higher N uptake by palms and the absence of legume cover. The results also indicated that in this area, the N2O emissions were mainly determined by soil moisture. The second study showed emissions ranging from 1.36 to 7.3 kg N ha−1 year−1 on two different soil types in Papua New Guinea (Banabas 2007). Banabas explained the highest emissions as being related to poor drainage of the soil.
Despite the limited number of measurements in oil palm plantations on mineral soils and the high variability of results, emissions seem to be higher over the first years of the palms. In addition, they seem to be of the same order of magnitude as those under oil palm in peatlands, e.g., average of 1.2 kg N ha−1 year−1 (Melling et al. 2007); under other crops in tropical conditions, e.g., average of 1.2 kg N ha−1 year−1 (Bouwman et al. 2002); and under tropical forest, e.g., average of 3 kg N ha−1 year−1 (Keller et al. 1986). However, data is lacking on the effect of spatial heterogeneity of N2O emission drivers, such as fertilizer application, soil water content, and organic matter content. Moreover, no measurements of NO
and N2 emissions have been reported for oil palm.
In summary, N outputs were estimated at 0–100 and 0.01–7.3 kg N ha−1 year−1 through harvest and N2O emissions, respectively, and in percentage of mineral N applied, 1–34, 2–15.6, 0.5–6.2, and 0.1–42, for leaching, runoff, erosion, and NH3 volatilization, respectively. The largest losses are volatilization of NH3 and leaching of NO3
−. The results and references are synthesized in Table 3.