GRAPHICAL ANALYSIS OF ANNUAL CROP RESPONSE TO FERTILISER APPLICATION

Transcription

1 Agricltral Systems 9 (1982) GRAPHCAL ANALYSS OF ANNUAL CROP RESPONSE TO FERTLSER APPLCATON H. VAN KEULEN Centre for Agrobiological Research (CABO), Wageningen, and Centre for World Food Stdies (SOW), Wageningen, Amsterdam, The Netherlands SUMMARY The effect of fertiliser application on crop yield is described by the interactive effects of two relations: that between fertiliser application and fertiliser ptake (the recovery of fertiliser) and that between ntrient ptake and yield. t is arged that the latter relation is invariable for a given combination of crop species and plant ntrient, as illstrated for nitrogen on bnded rice. Virtally all variability in application-yield relations reslts from differences in the parameter vales characterising the relation between ntrient application and ntrient ptake. The niqeness of the ptake-yield relation permits its se in the determination of ntrient recovery from measred application-yield data. The se of this procedre to formlate fertiliser recommendations for specific sitations is discssed. NTRODUCTON n the agricltral prodction process, the individal farmer is constantly faced with the problem of allocating highly variable resorces in a meaningfl way. Attempts have often been made to describe this process in the form of so-called prodction fnctions, which express the yield obtained (Y) as a fnction of inpts sed: Y =f(x,y, z,.... ) where x,y, z represent prodction factors, like the amont of water, minerals or labor. t has become increasingly clear, however, that sch attempts are only sccessfl in very simple sitations. n somewhat more complex systems there is no niqe soltion to the prodction fnction, since there are many bio-technically 113 Agricltral Systems X/82/ f$02 75 Applied Science Pblishers Ltd, England, 1982 Printed in Great Britain

2 114 H. VAN KEULEN feasible inpt mixes that will prodce a certain yield, and it is qestionable whether these are all agronomically feasible at the same time. An alternative way of treating the problem is that of developing dynamic prodction models, that are based on detailed knowledge of crop properties and conditions nder which they grow. A nmber of sch models have been developed (van Kelen, 197 5; de Wit et al., 1978), bt their degree of sophistication is too high to permit application in a more general framework. A short ct between both approaches may be fond by sing an hierarchical approach in which the yield is, on the one hand, considered as a dependent variable, governed by crop species and environmental conditions, and, on the other hand, as an independent variable, dictating a feasible inpt combination which will lead to realisation of that predetermined level (van Kelen & de Wit, 1981). The reqired inpts may, in a schematical way, be distingished in fieldwork and material inpts, the latter being again sbdivided in yield-protecting materials (biocides) and yield-increasing materials (water, ntrients, etc.). This paper specifically deals with the determination of the reqirement for the plant ntrient nitrogen for different yield levels. FERTLSERS AND PLANT PRODUCTON Apart from water and light energy, plants reqire inorganic ions to bild strctral organic material. n the corse of time more and more elements have been shown to be indispensable for proper plant growth. This paper mainly deals with nitrogen which was one of the first elements to be recognised as essential and still maintains a special position, owing to the qantities in which it is needed, its high mobility in the soil-plant-atmosphere system and, to some extent, its price. The latter argment is not (yet) decisive in most well-developed contries, bt it is a serios impediment for the introdction of more fertilisers in many less-developed contries. t is therefore important that the fertiliser that is being applied is sed as efficiently as possible. n this respect it is important to realise that an increased economic yield, as a reslt of fertiliser application, reqires, first, that the applied element is being taken p by the crop and, secondly, that it is tilised to prodce the demanded plant parts. Lack of response to fertiliser application may ths be de to the fact that the element has not been taken p, becase it was applied at the wrong time, in the wrong place or in the wrong form, or that after ptake it did not express itself in increased economic yield. The latter may be inhibited by other growth-limiting factors sch as shortage of water or mineral elements, other than the one applied. The common way of presenting the reslts of fertiliser experiments showing the yield response as a fnction of fertiliser applied (Fig. 1) makes it impossible to distingish between these two possible reasons for a lack of response. The explanatory vale of sch experiments, and hence their seflness for extrapolation of reslts, greatly improves when they are accompanied by chemical analysis of the harvested

3 GRAPHCAL ANALYSS OF CROP RESPONSE TO FERTLSER APPLCATON 115 grain yield t.ha x Mojosari Psakanegara 2 N-application Fig kg ha The relation between nitrogen application and grain yield for bnded rice at two locations on Java (sorce: smnadji eta/., 1973). material. Determination of its composition enables the calclation of the amont of the element taken p by the plant and its sbseqent distribtion. Presentation of the reslts When, in fertiliser trials, both the yield and the chemical composition of the harvested material have been determined, interpretation of the data is facilitated when the reslts are presented graphically as sggested by de Wit (1953), and illstrated in Fig. 2. The Figre consists of three graphs. n the first graph (a) the relationship is given between the economic yield and the total amont of the element taken p by the crop. The second graph (b) expresses the ptake of the element as a fnction of the amont applied. The third graph, constrcted from the other two, throgh the elimination of the ptake, is identical to the one in Fig. 1, giving the relation between fertiliser application and yield. The yield-ptake crve The example given here, which is representative for the majority of experiments involving nitrogenos fertilisers, shows that a proportional relation exists between yield and ptake at low ptake vales. This linearity indicates that, nder conditions where nitrogen is in limited spply, its concentration in the tisse eventally reaches a minimm level, beyond which frther diltion or remobilisation from vegetative strctres is not possible. Each nit of N taken p reslts in a constant amont of yield being prodced. An extensive analysis of yield-ptake relations for a nmber of crops (van Kelen, 1977; van Kelen & van Heemst, 1981) has shown that the slope of the initial part of the crve is crop-specific and, to a large extent, independent of environmental conditions. Nmerical vales

4 116 H. VAN KEULEN grain yield (a). ~x-- /~., X x Mojosari Psakanegara N-application N -ptake kg ha-1 N-application Fig. 2. The relation between total nitrogen ptake and grain yield (a), that between nitrogen application and nitrogen ptake(b) and that between nitrogen application and grain yield (c) for bnded rice at two locations on Java (lsmnadji eta/., 1973). are: 70 kg of grain per kg N taken p for small grains, 100 kg of tber dry matter per kg N for potatoes and 80 kg of pre sgar per kg N for sgar beet. At higher levels of ptake the crve deviates from linearity, reflecting higher concentrations of the element in the tisse, both in the economic plant parts and in the crop resides, at harvest. Finally, the crve levels off, indicating that the element nder consideration is no longer a constraint for nrestricted growth. The level of the platea is determined by the growth factor in short spply and is, in the 'potential growth' sitation, a fnction of the available solar energy dring the plant's growth period. The yield-ptake relation is, in general, independent of the type of fertiliser and the method of application, provided that the fertiliser does not change other growing conditions than the one governed by its main acting element and is not applied so late that a period of serios shortage fs followed by one with abndant availability (de Wit, 1953; van Kelen, 1977). The application-ptake relation The relation between fertiliser application and total ptake by the crop is, in the

5 GRAPHCAL ANALYSS OF CROP RESPONSE TO FERTLSER APPLCATON 117 example presented in Fig. 2, a straight line over the fll range of applications. Again this example is representative for the majority of fertiliser experiments involving nitrogenos fertilisers (van Kelen, 1977; van Kelen & van Heemst, 1981). Of corse, when very large amonts of nitrogen fertiliser are applied, the capacity of the vegetation to absorb and synthesise the element may become limiting, reslting in discontinities in the application-ptake relation (Figs 3(d) and 4(b)) A 4 y A y )(.. x--xx; :fo; 0 12 o/o "'o x peat soil 9 / o dry sandy soil 6 (a) (b) 150 ~.. 73 " (c) 0,)<..--r.---x- X 1961 o./"'"/ ~ 75_ 150 A 8 ij A (d).,.., x/x"" "" o./. ~ ~~ (e) \ 07~ x shallow water table well-drained \ i X~ i,40x 0 \ 150 x,_,.. x'/ ""'x x/ x withot vetch vetch in rotation 60 ~"--:~ 100 ~~0" ~ " 120 A Fig. 3. The relation between nitrogen ptake ( U) and yield (Y) and that between nitrogen application (A) and nitrogen ptake. (a) Permanent pastre, The Netherlands (van Steenbergen, 1977) Yin tons dry matter per hectare; U and A in kg N ha- 1 (b) Winter wheat, West Germany (Kohnlein, 1972) Yin tons of grain per hectare (15% moistre), U and A in kgnha- 1. (c)permanent pastre, The Netherlands (Oostendorp, 1964). Yin tons ciry matter per hectare, U and A in kgnha- 1 (d) Winter wheat, The Netherlands (Sieben, 1974). Yin tons of grain per hectare (15% moistre), U and A in kgnha- 1. (e) Flooded rice, USA (Williams et a/., 1972) Yin tons of grain per hectare, U and A in kg N ha- 1.

6 118 H. VAN KEULEN 4 y A 8 y 4 60 (a) sa ~ 100 1~\ \.4Q,17 X \ \ x broadcast placed 150 (c) ~ ~ y 300 A 8 y.26\ 120 A A 4 (b) x ~ '\X~,12\,53'1 (d).~ - - ~ x x dry season '71 wet season'71 /' osing~~x~m~ o spltt ll.52. (3 Fig. 4. The relation between nitrogen ptake ( U) and yield (Y) and that between nitrogen application (A) and nitrogen ptake. (a) Winter wheat, The Netherlands (Lehr, 1950) Yin tons of grain per hectare, U and A in kg N ha - 1. (b) Unimproved pastre, Mali (Penning de Vries et al., 1981.) Yin tons dry matter per hectare, U and A in kg N ha - 1 (c) Bnded rice, Java, ndonesia (smnadji & Sysmiati, 1976) Yin tons of grain per hectare (15% moistre), U and A in kg Nha - 1. (d) Bnded rice, Sri Lanka (Nagarajah et a!., 1975) Yin tons of grain per hectare, U and A in kg N ha- 1 Sch a straight line is characterised by two parameters, the intercept with the ptake axis and the slope with respect to the vertical. The first parameter represents the inherent fertility of the soil for the element concerned, the latter the recovery fraction of the applied fertiliser. Both parameters show widely varying vales. The ptake at zero fertiliser application is, on the one hand, a soil characteristic, governed by the qantity and qality of the organic matter present in the soil. t is, on the other hand, inflenced by environmental conditions, notably soil temperat'..lre and soil moistre conditions, which govern the rate of decomposition of the organic material and the ltimate fate of the nitrogen mineralised or immobilised <;lring this process. Finally, management has an inflence on its vale, throgh the level of soil reclamation, crop rotation and previos fertiliser applications. n the framework of this paper it is impossible to treat all these aspects exhastively bt some examples are presented in Fig. 3. Figre 3(a) relates to permanent pastre in the Netherlands and shows the inflence of soil type on N absorption in the absence of fertiliser application. The sandy soil, with a relatively low organic matter content, spplied ± 160 kg N ha- 1 year- 1, the high vale being the reslt of the application of farmyard manre on all experimental plots, whereas, from the peat soil, with a

7 GRAPHCAL ANALYSS OF CROP RESPONSE TO FERTLSER APPLCATON 119 high organic matter content, almost 300kgha - 1 year- 1 was extracted. The inflence of environmental factors is illstrated in Figs 3(b) and 3(c), referring to an experiment with winter wheat in West Germany and to permanent pastre in the Netherlands, respectively. n the winter wheat, complete interception of winter rainfall by protective shelters increased the ptake of nitrogen in the non-fertilised sitation by ± 30%, presmably becase leaching of the mineralised nitrogen beyond the root zone was prevented. Figre 3(c) shows the combined effect of temperatre and soil moistre. Spring in 1961 was relatively warm, condcive to rapid mineralisation and relatively dry, again preventing losses by leaching and possibly denitrification. Conditions were reversed in the spring of 1962, which was cold and wet, ths leading to a redction of over 50% in availability of native nitrogen. An aspect of management is illstrated in Fig. 3(d) where land improvement by drainage of a field of winter wheat increased the availability of nitrogen in the nonfertilised sitation from abot 30 to abot 70 kg N ha - 1, primarily throgh redction of denitrification losses, bt possibly also by enhancing root activity. The inflence of crop rotation is shown in Fig. 3(e), referring to rice, grown in the US, where incorporation of a fall-sown crop of vetch into the soil increased the ptake of nitrogen from natral sorces by abot 50% These examples may serve to illstrate that the actal amont of nitrogen from natral sorces that will be available to the crop dring its growth cycle may vary widely. Prediction of that amont reqires detailed knowledge of soil characteristics, environmental conditions and management practice for any given sitation. The proportion of the applied nitrogen fertiliser that is taken p by the vegetation is partly inflenced by the same conditions affecting the level of nitrogen ptake at zero-fertiliser application, since processes that render the element navailable for ptake by the plant act in identical ways on native nitrogen and on nitrogen applied as fertiliser. This phenomenon is easily recognised in the examples presented in Fig. 3 where, in all bt one case, higher ptake of soil N is associated with a higher recovery of the applied fertiliser. For fertiliser application on temperate pastre the same conclsion was reached by Brockman et al. (1971). Apart from this, fertiliser recovery may be inflenced by the type of fertiliser sed, as illstrated in Figs 4(a) and (b) for two contrasting sitations. The winter wheat crop of Fig. 4(a), growing in the reclaimed polders of the Zyderzee, absorbed the nitrate fertiliser 2 5 times as efficiently as the ammoniacal fertiliser. This is the reslt of ammonia losses de to volatilisation on the lime-rich gritty clay soil (ph 8 2) in these polders. The natral vegetation of Fig. 4(b ), growing on a heavy clay soil in the Sahelian region, tilised the nitrate fertiliser far less efficiently than rea. Here the major case of losses is denitrification, dring temporary flooding of the area de to rn-on dring high-intensity rain showers. Fertiliser efficiency may also be inflenced by timing and method of application. The effect of method of application is illstrated in Fig. 4(c), referring to bnded rice

8 120 H. VAN KEULEN growing on the island of Java. Placement of rea fertiliser as mdballs, at transplanting, directly into the redced soil layer, prevents the transformation of ammonim into nitrates and the sbseqent loss throgh denitrification. When rea is broadcast onto the layer of standing water, nitrification takes place in the aerobic environment of the water, after which the nitrates, by mass flow or diffsion, enter the redced soil layer where rapid denitrification follows; hence a mch lower availability of nitrogen for the vegetation. Split application of the total amont of nitrogen fertiliser generally leads to higher recoveries since the rate of nitrogen ptake by the vegetation is higher at later growth stages, reslting in shorter residence times of the element in the soil soltion and hence less losses. This phenomenon is illstrated in Fig. 4(d) referring to bnded rice from Sri Lanka. (By the natre of the experimental techniqe employed to stdy the effect of split fertiliser applications, the recovery fractions are based on one application rate only here.) The recovery fractions determined in a single experiment in a particlar year may not be sed indiscriminately to illstrate the losses of fertiliser nitrogen. n some cases (denitrification, volatilisation) the complement of the recovered fertiliser may be irreversibly lost from the system. n other sitations, however, part of the originally added fertiliser may have been incorporated temporarily in the soil organic matter and may become available in sbseqent years. Conclsions From the discssion presented in this section it may be conclded that most of the variability in fertiliser application-yield relationships for a given crop is the reslt of variations in the application-ptake relation. n the examples of Figs. 3 and 4, widely different vales of the two parameters characterising the latter relation reslt, in each case, in a niqe ptake-yield fnction. Casal relationships for a particlar type of crop response to fertiliser application can only be established when all three relations can be constrcted, i.e. when yield, as well as element content of the crop, is determined: and knowledge of those case-and-effect relations is necessary in order to jdge the scope for improvement and to take the appropriate measres. APPLCATON OF THE THEORY FOR FERTLSER RECOMMENDATONS Analysis of past experiments n the previos section the conclsion was drawn that, for proper interpretation, fertiliser experiments shold be followed by chemical analysis of the varios plant parts. That, however, does not imply that the reslts of past experiments cannot be pt to se. Crves as presented in qadrant (c) of Fig. 2 are available in considerable qantities at least for the major crops in a given region. t was arged in the section above on the yield-ptake crve that the initial part of the ptake-yield relation is a

9 GRAPHCAL ANALYSS OF CROP RESPONSE TO FERTLSER APPLCATON 121 crop constant, while the platea level may be calclated from environmental conditions when it is assmed that the potential growth sitation exists (van Kelen, 1976). This implies that the ptake-yield crve for a given combination of crop and environment may be constrcted when climatic data are available in sfficient detail (van Heemst et al., 1978). Combination of the application-yield crve with the theoretical ptake-yield fnction reslts in the application-ptake crve (Fig. 5). (c) grain yield 8 t ha- 1 (a) X~----;;..- X~ ~---- X~--- X~ N -ptake N~--a~pp_li_ca~t_io_n ~--~---4~ ~~--~ r ~0 100 :1 kg ha- 1 """ kgha N-application (b) ~~:: X l ~: X"' Fig. 5. Graphical illstration of the method sed to determine application-ptake relations (b) from measred yield-application relations (c). The same procedre was applied to some of the data presented by Fagi & De Datta (1981) on flooded rice grown at RR in the Philippines. The nitrogen-response crves for the April and Jne harvests of the variety R 28 are prodced in Fig. 6(c). The (hypothetical) ptake-yield crve (Fig. 6(a)) was constrcted with an initial slope of 70 kg grain (at 15 % moistre content) per kilogram of nitrogen absorbed, while the potential grain yield was calclated by the procedre described by van Kelen (1976) employing long term-average radiation and temperatre data for Los Banos. Combination of the graphs in Figs. 6(c) and 6(a) yields the points given in Fig. 6(b). The average recovery fractions for the two trials are 0 36 and 0 26 which compare favorably with the experimentally determined vales of 0 40 and 0 24, respectively (Fagi, 1977). No actal data on

10 122 H. VAN KEULEN nitrogen ptake are given, so that these cannot be directly compared. Similar reslts are obtained from other experiments, ths adding to the procedre described. The available experiments pertaining to nitrogen application to bnded rice for varios experimental stations on the island of Java in ndonesia were analysed in this way, which yielded a reasonable nmber of recovery fractions per location. The nmerical vales obtained show a wide variation, as cold be expected from the evidence presented above in the section on the application-ptake relation. t 10 grain yield (c) 2 N-application kg ha-1 50 N -ptake kg ha (b) kg ha N-application Fig. 6. Application-ptake relations (b) for two plantings of the variety R28 grown at RR, The Philippines, determined from measred yield-application relations (Fagi & De Datta, 1981). trned ot, however, that, for a similar growing period (that is, either the wet or the dry season), and a given location, the recovery fractions show a normal distribtion (Fig. 7). n the light of the argments presented above it may be conclded that the expectation vale is mainly determined by the method of application and its timing, whereas the deviations are largely the reslt of varying environmental conditions: variations in amont and pattern of rainfall, temperatre flctations, etc. Variations between locations mst be attribted to differences in both environmental and soil conditions and mst therefore be taken at face vale (van Kelen, 1977).

11 GRAPHCAL ANALYSS OF CROP RESPONSE TO FERTLSER APPLCATON r~cov~ry fraction K WS OS cmla\ ivt' r~lat ivt' f reqt'ncy ,. Fig. 7. Cmlative relative freqency distribtion of recovery fractions of nitrogenos fertilisers, applied to bnded rice in Mojosari (East Java, ndonesia) in the dry (DS) and the wet (WS) season. Fertiliser recommendations The analysis presented so far may now be sed to provide fertiliser recommendations for specific sitations. The procedre is illstrated in Fig. 8, again for bnded rice: the starting point is the ptake-yield crve for a given combination of weather and rice variety (van Kelen, 1976). For the application-ptake relation se may be made of local knowledge abot yield expectations withot fertiliser application, which yields the intercept with the ptake axis as may be dedced from the examples given in Fig. 3. This is the most difficlt part of the analysis since non-predictable weather conditions may affect that vale appreciably. However, no matter in which way fertiliser recommendations are generated, an estimate of the N spply from natral sorces has to be provided. An analysis as described in the previos section provides the slope of the ptake-application crve. For the decision whether or not to apply fertiliser at all, the linear part of the ptake-yield crve may be sed, in combination with the most nfavorable conditions in terms of recovery, which for bnded rice is ± 10 % This implies that a price ratio between fertiliser nitrogen and rice grain of less than 7 (1 kg N applied = 0 1 kg N taken p = 7 kg rice grain prodced) makes N application profitable. To decide on the actal amont to be applied, the expected recovery fraction may be sed. When the actal recovery fraction in an individal case wold be mch higher than the one expected, the farmer wold be tempted to apply too mch fertiliser, which, in the most nfavorable sitation, wold lead to yield depressions de to lodging or increased ssceptibility to pests and diseases and that at higher costs of inpts. The recovery fraction to be applied for a certain region may

12 124 H. VAN KEULEN (c) grain yield (a) kg ha N-application 100 N ptake (b) Fig. 8. N application Graphical illstration of the method to determine optimm fertiliser application rates. be determined, therefore, as the expectation vale mins twice its standard deviation, which redces the risk of a higher actal vale to less than 10 % Combination of the ptake-yield relation with the application-ptake relation derived in this way permits the constrction of the application-yield crve for specific sitations (Fig. 8(c)). The optimm amont of fertiliser to be applied may now be determined by combining this crve with the prevalent ratio between otpt and inpt ({3 in Fig. 8(c)). At the point where the slopes are eqal, the marginal application rate is fond. The procedre otlined here leads to the se of a price ratio, which leaves sfficient incentives for the farmer to apply nitrogen, while the risk of overfertilisation is minimised. APPLCATON N THE AGRCULTURAL SYSTEMS APPROACH For application of the analysis in the framework of the hierarchical approach, discssed by van Kelen & de Wit (1981), in principle a similar procedre is sed. When the desired or expected yield has been determined on the basis of crop properties and environmental conditions, the nitrogen reqirement can be estimated from the ptake-yield crve. The next step is the assessment of the natral

13 GRAPHCAL ANALYSS OF CROP RESPONSE TO FERTLSER APPLCATON 125 soil fertility; that is, the amont of nitrogen available from natral sorces. Both soil properties and environmental conditions inflence that amont. Finally, the expected recovery of applied fertiliser has to be estimated. This is a fnction of method and time of application bt is also inflenced by the reclamation level of the soil: for instance, mch larger losses are expected nder conditions where temporary waterlogging may occr as a reslt of insfficient drainage capacity. For a qantitative assessment of the reqired parameters, simlation models are being developed (Seligman & van Kelen, 1981), bt these are as yet hardly sfficiently rigid to permit application nder widely varying conditions. t is to be expected, therefore, that for the time being se mst be made of (semi-) empirical relations, based on analogy. REFERENCES BROCKMAN, J. S., ROPE, C. M. & STEVENS, M. T. (1971). A mathematical relationship between nitrogen inpt and otpt in ct grass swards, J. Br. Grass/d. Soc., 26, F AG, A. M. (1977). Environmental factors affecting fertiliser nitrogen efficiency in flooded tropical rice. MSc Thesis, UPLB, Los Banos, Philippines. FAG, A.M. & DE DATTA, S. K. (1981). Environmental factors affecting nitrogen efficiency in flooded tropical rice. Fert. Res., 2, HEEMST, H. D. J., VAN KEULEN, H. & STOLWJK, H. (1978). Potential, gross and net prodction of Netherlands agricltre. Verst. landbowk. Onderz. (Agric. Res. Rep.) 879 (Dtch, with English smmary), Pdoc, Wageningen. lsmunadj, M. & SSMYAT, D. (1976). Stdies on fertilizer nitrogen efficiency in lowland rice on an acid latosol in ndonesie. Paper presented at RR Annal Conference, April 12-15, SMUNADJ, M., ZULKARNAN,., PRAWROSAMUDRA, A. & YAZAWA, F. (1973). Prodctivity of some major Java soils, with special reference to yield and nitrogen ntrition oflowland rice. Contr. Centr. Res. nst. Agric. Bogor, No.7, 17pp. KEULEN, H. VAN (1975). Simlation of water se and herbage growth in arid regions. Simlation Monographs, Pdoc, Wageningen, 176 pp. KEULEN, H. VAN (1976). A calclation method for potential rice prodction. Contr. Centr. Res. nst. Agric. Bogor. No. 21, 26pp. KEULEN, H. VAN (1977). Nitrogen reqirements of rice, with special reference to Java. Contr. Centr. Res. nst. Agric. Bogor. No. 30, 67 pp. KEULEN, H. VAN & DE WT, C. T. (19~1). A hierarchical approach to agricltral prodction modelling. Proc. ASA workshop, Environmental mpact of Agricltre, ASA, Vienna. (n press). KEULEN, H. VAN & HEEMST, H. D. J. VAN (1981). Crop response to the spply of macro ntdents. Verst. landbowk. Onderz. (Agric. Res. Rep.), 916, Pdoc, Wageningen. KoHNLEN, J. (1972). Einflss der winterlichen Stickstoffasswaschng af die Stickstoffversorgng des Getreides. Unterschngen aflehmiger Parabranerde in Schleswig-Holstein. Landw. Forschng, 25, LEHR, J. J. (1950). Differences in the recovery of nitrogen from ammoniacal and nitrate sorces in the reclaimed polders of the Zyder Zee. Plant and Soil, 2, NAGARAJAH, S., JAUFFER, M. M. M. & WLLENBERG, S.M. (1975). Timing of nitrogen application-ts effect on nitrogen tilization and protein content of rice. Plant and Soil, 42, Oosi'ENDORP, D. (1964). Stikstofbemesting en grasgroei in het voorjaar. Landbowk. Tljdschr., 76, PENNNG DE VRES, F. W. T., KRUL, J. M. & VAN KEULEN, H. (1981). Prodctivity ofsahelian rangelands in relation to the availability of nitrogen and phosphors from the soil. 'Nitrogen cycling in West African ecosystems', SCOPE/UNEP nternational Nitrogen Unit, Royal Swedish Academy of Sciences, Stockholm, pp

14 126 H. VAN KEULEN SELGMAN, N. G. & VAN KEULEN, H. (1981). PAPRAN: A simlation model of annal pastre prodction, limited by rainfall and nitrogen. n Simlation of nitrogen behavior of soil-plant systems1(frissel, M. J. & van Veen, J. A. (Eds.). Pdoc, Wageningen. SEBEN, W. H. (1974). Over de invloed van de ontwatering op de stikstoflevering en de opbrengst van jonge zavelgronden in de Jsselmeer-polders. Van Zee tot Land No. 51. STEENBERGEN, T. VAN (1977). nvloed van grondsoort en jaar op het effect van stikstofbemesting op de graslandopbrengst. 'Stikstof, 85, WLLAMS, N. A., MoRE, M.D. & RcKMAN, J. R. (1972). Brning verss incorporation of rice crop resides. Agron. J., 64, WT, C. T. DE (1953). A physical theory on placement of fertilizers. Versl. landbowk. Onderz. (Agric. Res. Rep.) 59 4, Staatsdrkkerij, 's-gravenhage. WT, C. T. DE eta/. (1978). Simlation of assimilation, respiration and transpiration of crops. Simlation Monographs. Pdoc, Wageningen.