Agroforestry Systems 6, 49-62, (1988) © Martinus Nijhoff Publishers, Dordrecht Printed in the Netherlands 49

M o d e l l i n g agrofores try s y s t e m s o f c a c a o (Theobroma cacao) with laurel (Cordia alliodora) and poro (Erythrina poeppigiana) in C o s t a R i c a III . C y c l e s o f organic mat ter and nutr ients*

H.W. FASSBENDER, 1 L. ALP[ZAR, 2 J. HEUVELDOP, 3 H. Ft~LSTER 4 and G. ENRIQUEZ 5 i Technical Forestry Faculty, Gbttingen, Federal Republic of Germany "Coffee Research Institute of Costa Rica, San JosO, Costa Rica 3 Agroforestry Project, CA TIE/GTZ (Tropical Agricultural Research and Training Center/Gesellschaft./fir Technische Zusammenarbeit), Turrialba, Costa Rica. Now in GTZ-Headquarters, Eschborn, Federal Republic of Germany 4Faculty of Forest Sciences, G6ttingen, Federal Republic of Germany J CA TIE, Turrialba, Costa Rica

Key words: Theobroma cacao, shade trees, Erythrina poeppigiana, Cordia alliodora, modelling, organic matter, nitrogen, phosphorus, potassium, calcium, magnesium

Abstract, Models for cycles for organic matter and nutrients element (N, P, K, Ca and Mg) are presented for the agroforestry systems of cacao (Theobroma cacao) with Cordia alliodora or Erythrina poeppigiana in Turrialba, Costa Rica.

For the models, system reserves (soil, humus, vegetation divided into leaves, branches, stems, fine roots, fruits) and transference between compartments (production and decom- position of litter residues) inputs (fertilizer, rainfall) and outputs (harvests) of the system are considered.

The implications of the models are discussed in detail. Aspects of net primary production in the systems studied are considered. N fixation is calculated on the basis of balances. Analysis of soil water showed high

variations that coincided with rainfall patterns and pruning of the E. poeppigiana.

Introduction

The management and improvement of agroforestry systems should be based on a satisfactory understanding of their structures and functions. This is a complex task since these systems are complicated and a great many aspects must be taken into consideration [8].

To facilitate this task, the use of models for organic matter and nutrient cycles have been proposed [1, 2, 3, 8, 9, 10, 11, 13, 14, 17]. The models

*For part I see Vol. 4, No. 3, 1986 For part II see this issue

50

provide a simplified representation of the reality, but demonstrate the most conspicuous aspects of the system through their complete, integral and holistic character [8,9, 10, 11].

Much emphasis has been placed on developing modelling techniques in the last few years [8, 15, 17]; however, until now, their use for partially or completely describing agroforestry systems, has been limited [1, 2, 3, 8, 9, 10, 11, 14, 17, 18, 19]. The basic concepts of models for organic matter and nutrient cycles in the agroforestry systems of cacao (Theobroma cacao) with shade trees, such as C. alliodora and E. poeppigiana, have previously been described by Alplzar et al. [2]. Based on data taken from the Central Experiment on CATIE, Turrialba, Costa Rica, these authors have published values for organic matter and nutrient reserves [2, 9]. More recently, Heuveldop et al. [14] have presented data for cacao and timber yields from these systems, as well as parameters for plant residue production and decomposition.

In this article, models for organic matter and nutrient (N, P, K, Ca and Mg) cycles will be presented for the agroforestry systems T. cacao under the shade trees C. alliodora or E. poeppigiana. The corresponding implications are also discussed.

M a t e r i a l s a n d m e t h o d s

Details of the Central Experimen t at CATIE, Turrialba, Costa Rica, in which the systems were studied, have already been described [1, 2, 5]. The following values were considered for preparing models (all biomass values are oven-dry weights): - - Above ground biomass of T. cacao (after 4.5 years in 1982), C. alliodora

after 5 years, in 19.82 and E. poeppigiana (as a function of 2 prunings in the 5th year (1982) with respective stems measurements) [2]

- - R o o t biomass, irrespective of species at age 4.5 years [2]. - - N u t r i e n t reserves (N, P, K, Ca and Mg) in the vegetation based on

chemical analyses of samples (leaves, branches, stems, roots, fruits) [2]. - - Organic matter and nutrients (N total, P total exchangeable K, Ca and

Mg) in soil at the beginning of the experiment (1977) [2]. - - Cacao production (beans and husks) during the seven years of measure-

ments (1979-1985) [14]. - - Production of vegetative residues, such as tranference of organic matter

and nutrients from the biotic components to the soil, measured over 4 experimental years (Nov. 1981-Oct. 1985) [14]. The decomposition rate is assumed to be the same as the deposition rate [14].

51

- - Nutrient input with rainfall was measured in the vicinity of the experi- ment during two years [12].

- - Rate of fertilization with N, P, K applied to the experiment [2, 5].

Results and discussion

Models for organic matter

Fig. 1 shows the models for organic matter in the agroforestry system T. cacao associated with C. alliodora or E. poeppigiana in Turrialba, Costa Rica.

The two systems under study show different characteristics in their orga- nic matter cycles. Total organic matter accumulation (aerial biomass, fine roots, humus) during the 5 experimental years, has reached 50.3 t.ha-' for T. cacao/C, alliodora and 39.1 t.ha-' for T. cacao/E, poeppigiana. Unfor- tunately, data for large root biomass are unavailable since destructive harvesting could not be permitted.

The difference in organic matter accumulation is found especially in the stems (C. alliodora 23.7; E. poeppigiana 9.3 t.ha -~ ). In the case of cacao there are comparable values for biomass and different shade trees have not affected the crop's development [2].

Values for fruits are shown in brackets in Fig. 1 to indicate that cacao fruit biomass is not always present but is transitory and the data refers to the removal of biomass as an annual mean for seven harvest. When interpreting

. • BRANCHES 4.7~ ~6;11 STEMS 23,71

IN(3'IEME/~IT

2.97 S 4.04 1.71

2.80 BEANS 0 63 II.~))

8T I FINE RO01~; 4.19 I

"07 'J ,,TrER 4.4s J I QECOMPO~TION RESPIRATION i ?

SOIL 168.3 I ZO7 ~ t O-19cm 70.97

/ I S - ~ ~ 96.32

30-49cm 41.03

! RESIOUES E. POF-PPIGIANA 21.90 I ~

4.27 LEAVES 3.27

H BRANCHES 9.31 STEMS 9.32

INCJRF.,kENT

3.91 T- CACAO B.37 ] I LJ[AVES 2.83 BRANCHES . 3.03 I STEMS 2.~ S ?.7~ ~ tLTS) IHAm~ST

HUSKS LO2 FIM[ ROO'T~ 1.781

8.90 LITTER ?.OB

DEOOM~(~ITION 890

I I RESPI~TI]q

SOIL 1911.4 ~ 1 ?

/ O-IScm 91.74

I5-30cm 74.98

30-45cm 41.6S

Fig. 1. Models for organic matter cycles for agroforestry systems T. cacao/C, alliodora and T.

cacao/E, poeppigiana in Turrialba, Costa Rica (reserves t.ha 1, transferences t . h a - L a - ] ) .

52

cacao harvests as exportation from the system, it should be noted that in many cacao plantations, the husk is left on the ground and its recycling represents a source for new humus material. Under experimental conditions, the husks have been removed from the plots. The exportation indices for the harvests in this study thus include all the material produced. The following values were found:

Biomass (t.ha ~) Harvest (t .ha-~.a -~) Expor ta t ion index %

T. cacao/C, alliodora 50.3 1.50 2.98 T. cacao/E, poeppigiana 39.1 1.73 4.42

Results for residue production and decomposition express the dynamics of the systems. It can be concluded, that the system with E. poeppigiana is characterized by faster rates for organic matter recycling than that with C. alliodora.

In an attempt to calculate net primary production in the systems under study, the addition of the following values has been considered: - - annual harvests (means of 6 years) - - residue deposition (leaves and branches, means for 4 years) - - annual fine root renewal diameter ( < 20 mm) - - t h e annual increase in wood in the stems, of C. alliodora measured

between the 5th and 7th years represents 27% of the biomass present in the 5th year.

- - increase in wood reserves (stems and branches) was estimated as 25% of the biomass in each of the other two species (T. cacao and E. poep- pigiana). Results obtained for annual net primary production (t.ha -I.a - l ) are as

follows:

1", cacao/ T. cacao/ C. alliodora E. poeppigiana

Harvests 1.50 - 1.73 - Residues 3.78 2.69 3.91 4.79 Increase in stems + branches 1.71 7.68 1.39 4.56 Roots 4.19 4.78 Sum 21.55 21.26

These values for net primary production are fairly high compared to those obtained in other agroecosystems [8, 15] but do not reach the levels obtained in coffee ( Coffea arabica/C, alliodora and C. arabica/E, poeppigiana) systems [lo].

53

RESIDUES C. ALLIODORA 57.1 LEAVES

BRANCIf.S STEMS --~ !CACA°

43.3 LEAVES BRANCHES Sll[MS FRU115

[ FINE 8D01~

114"8" I LITTER MINERAiJSATION

114.8_i SOIL

I 0-15 Cm 15-30cm

30-45¢m

5.0 I INPUT

233.0 ~ 9S.I 43, i 94.8

96.8 55.8 32.?' IO.S

(1~3)

38.5 J, 76. I

7991

3337

2769

1185

loo;'UT

12.)BEANS .,mS "~ 2HUSKS

1 Um',~E 219,1 120

FERTILIZER

~SIDUES l 1127

52.8

174.9 o I

MINF-RALF~TION

174.9

E.POEPPIGIAN.__.___~A 50. !62.SIj~/,INPUT

IOIA LEAVES 503 BRANCHES 110.8 STEMS

- 102.8 '~--~ 1~ CACAO 5~S LEAVES

BRANCHES 31. '1 STEMS 12.5 ~ ~i~ZEF.~S FRUI~ 1~.71

IQ5 HUSKS FINE ROOTS Z20l ~ 22.0

LITrER III.S I UPTAKE

SOIL 9555 273.9

0-15 ¢m 3780

I5-30cm 3676

30-4Scm 2100

[%,UT

Fig. 2. Nitrogen cycles for agroforestry systems T. cacao/C, alliodora and T. cacao/E, poep- pigiana in Turrialba, Costa Rica (reserves kg.ha -I, transferences kg.ha-l.a-I),

Nitrogen cycles

Figure 2 shows models for nitrogen cycles in the systems studied. Total N accumulation values for vegetation show fairly well defined

tendencies for each system studied, but notable differences do not exist. Including the humus layer, T. cacao/C, alliodora accumulated 446 kg N.ha- i compared to 494 kg.ha- i for I". cacao/E, poeppigiana. Reserves in the peren- nial crop and the shade tree were higher for the second system. Shade tree N reserves were higher for both systems. In both systems the N-reserves in the shade trees were higher than in the other biotic components.

In C. alliodora, the stems and leaves show greatest accumulation, whereas the leaves and branches of E. poeppigiana accumulate most nitrogen.

The humus layer is a transitory reserve for nutrients, with plant residues in the process of mineralization and humification. The values encountered depend on the season in which they are taken and the phenology of the plant. Here, once again, the T. cacao/E, poeppigiana association exceeded (111.5 kg N.ha- 1 ), that of T. cacao/C, alliodora (76.1 kg N.ha-i ).

Soil mineral reserves are high, and for the association with C. alliodora represent 95% of the system total, and with E. poeppigiana 92%. Absolute quantities of soil N differ markedly between the two systems (7991 and 9555 kg N.ha- l ) , respectively due to pre-experiment soil management.

Values for N accumulation in fruits (Fig. 2) are in brackets to show that these correspond to the total annual quantity removed from the system. In the case of cacao, fruit biomass is very dynamic, ripe pods are harvested

54

every 2 weeks, therefore flowering, fruiting and maturation is a continual process.

N transfer from plants to soil with litter shows marked differences. Mean values were 174.9 kgN.ha- l .a -l for T. cacao/E, poeppigiana and 114.8 kg N.ha- l .a -~ for T. cacao./C, alliodora [14].

To interpret the importance of transfer of N and other nutrients in plant residues, the values can be compared with N accumulation and exportation in the harvests. Thus, the following values can be obtained:

T. cacao/C, alliodora T. cacao/E, poeppigiana Residues (kg.ha-I .a - t ) 114.8 174.9 Biomass (kg.ha-~ ) 446.4 498.8 % Recycled 25.7 35.0

These figures imply that for T. cacao/C, alliodora 25.7% of the biomass is circulated within the' system annually. In the case of T. cacao/E, poeppigiana the value is higher (35.0%) showing an accelerated nitrogen cycle.

Based on the observation that the humus layer in both systems studied remains more or less constant [14] it has been suggested that the nitrogen is mineralized over the course of the year in which it is deposited on the soil.

N extraction with harvest (means for 7 years) have reached the following values (kg N.ha- ' . a -~) [14]:

Beans Husks Total T. cacao/C, alliodora 12.1 7.2 19.3 T. cacao/ E. poeppigiana 15.2 10.5 25.7

Comparing recycling with N removal in the systems: Heuveldop et al. [14] have indicated that recycling is 6.8 and 5.9 times higher than exportation for T. cacao/E, poeppigiana and T. cacao/ C. alliodora respectively. Thus recycl- ing maintains the nitrogen cycle.

Other parameters for interpretation were based on a comparison of N extraction with reserves in the biomass or fertilization which gives the following values:

T. cacao/C, alliodora T. cacao/E, poeppigiana Biomass (kg.ha i) 446.4 498.8 Harvest (kg.ha-l .a -I) 19.3 25.7 % Exported 4.3 5.2

N exportation with harvest represents 4.3% and 5.2% respectively for T. cacao associated with C. alliodora and E. poeppigiana; so that N loss is relatively low.

On the other hand, it should be noted that N fertilization (120kg

55

N.ha- l .a -1) is quite high compared to removal at harvest. However it is difficult to interpret the effect of applied fertilizer since information on losses through leaching and volatilization are unavailable.

During the first five years of measurements, the following values for total extraction were obtained for the systems with or without shade legumes (kg N.ha-1):

T. cacao/C, alliodora T. cacao/E, poeppigiana N in plant biomass and humus 446 499 Harvests (4 years) 77 104 Total extraction 523 603

The N reserves in the soils are unchanged (Unpublished data). The dif- ference in accumulation over these five years is 80 kg N.ha -~ , equivalent to 18kg N.ha-~.a -t. This can be considered as an estimate for annual N fixation within the system with E. poeppigiana.

This approximation is lower than that found for E. poeppigiana associated with coffee (Coffea arabica) and other agroforestry systems [11].

The soil in the experimental site is characterized by a high organic matter and nitrogen content [2]:

Depth (cm) Humus (%) Total N (%) C/N 0-15 4.32 0.21 11.9

15-30 3.07 0.16 11.1 30-45 1.87 0.11 9.9

According to the analysis of Martinez and Blasco [16] on the forms of N in the experimental soil, 97.6% N is organic, of which amino acids and amino sugars represent 25.1% and 4.2% respectively. Soluble NH4-N and NO3-N values (NaCI 1 N) represent 2.4% and 0.4% respectively of total-N. From these N fractions, and as a function of the processes of mineralization and interchange, N fixation in the soil, and additions from rainwater, variable quantities of N appear in the soil solution. In the experiment, samples of water were taken using lysimeters from depths of 0-15 and 15-30cm, 0.5 and 1.0m from the E. poeppigiana trunks. Chemical analysis (total N, NO3-N and NH4-N) should show seasonal variations in soil N. Available values indicate an increase in concentrations after the dry season (Feb. to May) coinciding with the weeks after pruning (Fig. 3).

Apparently the dilution effect of rainfall on soil regulates, in the first place, N content in soil solution, as found by Fassbender [7] in other soils in Costa Rica, under different ecological conditions and different crops.

In a later phase of the experiment, lysimeter capsules were grouped around two selected shade trees in order to compare values for total N in soil

56

Prul~q Prunmq pruning

t , 1 o. o i

E

4~- 350-

3O0 ~

250-- E E 200--

I00 -~ 5o_ ~

0

[ • ] Rainfoll (ram)

w ~ , ' y , ' a , ~ I ~ , o l ' o' ,~' l ' l , i N ,~, ';'j ,

Fig. 3. Total N and NO3-N variations in soil water as a function of E. poeppigiana pruning and rainfall (rag.l-~ ).

solution. Weekly samples of water were mixed to form monthly samples analysis. The means in Table 1 refer to 20 lysimeter capsules placed at two difference depths. In the root zone of the leguminous trees, higher N values were found under non-legume tree (23.2 and 9.3 mg N.I -l, respectively). This implies a difference due to N fixing by E. poeppigiana.

Phosphorus cycles

Figure 4 shows phosphorus cycle models for the systems under study. Total amounts accumulated in the vegetation are comparable for the

systems, including the humus layer. T. cacao/C, alliodora shows 50.3kg P.ha- l and T. cacao/E, poeppigiana 51.4 kg P.ha- ~. Reserves are found more in the shade trees (28.9kg P.ha -~ for C. alliodora and 26.3 kg P.ha -~ for E. poeppigiana) especially in the branches or stems. Differences between sys- tems for T. cacao are small.

Compared to other nutrients, especially nitrogen, values for P accumula- tion in biomass prove to be relatively small.

Total P soil reserves are high and represent nearly all the P in the system 99%. This shows P stability in the soil. Available P which may gradually be accumulated in the vegetation is a very small fraction of total P.

In order to evaluate P exportation with harvests, beans and pericarp (husk) of the fruit have been considered separately (Heuveldop et al., Fig. 4). Values for P exported from the system are relatively low (4.0 and 4.3 kg

57

Table 1. Nitrogen concentration in soil water for shade tree E. poeppigiana in asociation with T. cacao (Total N mg.1 -I)

Mar. Apr. May Jun. Jul. Aug. Sept. Oct. Nov. Dec. Jan. Feb. Annual mean

~15cm soil depth E. poeppigiana 12.1 8.6 16.4 3.5 42.1 75 .0 54 .5 26.0 9.3 10.7 19.3 7.9 25.6 C. alliodora 10.0 6.0 10.0 9.3 15.0 18.6 10.0 11.3 3.6 1.2 11.4 5.0 9.3

15-30cm soil depth E. poeppigiana 9.3 8.6 17.9 22.1 36.4 52.9 2 8 . 5 2 3 . 4 13.6 12 .4 12.1 3.6 20.1

C. alliodora 5.7 5.2 6.4 21.4 15.7 15.7 19 .0 16.3 2.2 1.6 5.0 2.2 9.7

P ha-].a -], for T. cacao/C, alliodora and T. cacao/E. Poeppigiana respec- tively) for the systems as a whole, but high for the cacao crop and for plants in an agroforestry system as such. Thus, comparing extraction in harvests with vegetative biomass accumulation, the following values are obtained:

T. cacao/C, alliodora T. cacao/E, poeppigiana Biomass (kg.ha -1) 50.3 51.4 Harvest (kg.ha- I.a-~ ) 4.0 4.3 % Exported 7.9 8.4

Values published by Heuveldop et al. [14] show that P concentrations in plant residues differ considerably. Values for all the litter material from T.

d w4PUT 0.2 ] INPUT O.

8" 9 I 113 RESIDUES6.5 I 1'~ I [, ~l~g_.~..~ 26.3 LEAVES Z8 ~ LEAVF.3

ERANOIF..S 8.0 STEMS II .8 . • . • BRANCHES 13.0 J

STEMS ~.S

_ ~ I.CACA0 38 LEAVES x.4 BRANCH(S 3.9 S'IIEMS 2.5 FRUITS 1431 3.0EE~

L3 HUSKS 2.3

~ LITTER 8.5 i UPTAKE

S01L 3243 !--29.3 °22" 6 29.3

O-15©m 1003 ~ R

| 15-30cm 1193

30-45 cm 1045

7~-~? I I" caea°ee~x~(s izas2s.3

3.6 2.9 Er~_NS FRUII~ {40) LIHU~

4.z I' 4.z

LITTER ~ l O N 4.9 I UPTAKE kI/~RALI~I~TIO N

SOIL 3564 I lSJi 0.18¢m 1139 29,3 88

FERTILIZER IS-30cm 1191

30-4Scm 1264

I OU.~ PUT

LEAVES

STEMS

i FINE R~)~

Fig. 4. Phosphorus cycles of agroforestry systems T. cacao/C, alliodora and T. cacao/E. poeppigiana in Turrialba, Costa Rica (reserves kg.ha -~, transferences kg.ha-~.a-t).

58

cacao/E, poeppigiana are lower than from the T. cacao/C, alliodora associa- tion. As a result, although more or less comparable amounts of residues are produced (Fig. 1), the rates of P transfer for the two systems differ greatly (Heuveldop et al. 14, Fig. 4). The T. cacao/C, alliodora system showed values 13.9kgP.ha-t.a -j whilst that of T. cacao/E, poeppigiana showed 8.8kg P.ha -j .a -~. Based on the observation that decomposition and mineraliza- tion of residues is accomplished in one year, respective values have been included in the model.

Adding values for P harvested and accumulated following values are found (kg.ha-l.a -l Fig. 4).

T. cacao/C, alliodora T. cacao/ E. poeppigiana Aerial biomass 25.7 20.3 Root biomass 4.2 2.3 Harvest 4.0 4.3 Total 34.9 26.9

Application of phosphorus with fertilizer (29.3 kg P.ha- ' .a -~ ) almost com- pensates for extraction by the agroforestry association. However, changes in the applied P through fixation to forms unavailable to the plants may be expected. According to Fassbender [6] soils in the experiment fix 62.2% of available P leaving forms less available to plants (A1-P 67% Fe-P 26%). Thus the utilization of P from fertilizer may be limited.

Potassium cycles

Using nutrient element cycles as a basis, the model for potasium is shown in Fig. 5.

Values for K in the vegetation which show defined tendencies and marked differences between systems. The shade trees showed comaparable K reser- ves 243kg K.ha -1 for C. alliodora and 237kg K.ha -l for E. poeppigiana). Greatest accumulation was found in the roots for T. cacao/E, poeppigiana and in the stems for C. alliodora. Differences between systems are found with the cacao trees (98.8 kg K.ha-i under C. alliodora and 49.1 kg K.ha-t under E. poeppigiana).

Accumulation of K in the biomass is high compared to total K in the systems under study. 61% (T. cacao/C, alliodora) and 69% (T. cacao/E. poeppigiana) of system totals are located in the soil. Annual absorption rates of 198kg.ha-l.a-1 (for T. cacao/C, alliodora association) and 158 kg.ha- 1.a-1 (T. cacao/E, poeppigiana) imply a noticeable reduction in soil K.

In the models in Fig. 5, values for K in the fruit biomass are in brackets to indicate a dynamic biomass which, finally leaves the system. To evaluate

59

251 INPUT RESIDUES C. ALLIODOR A 243.4 LSO'& .T.T]

35 BRANCHES 56.0 STEMS 109.7

~ . 2 TCACAO 9 8 . 8 ~ LEAVES 30.9 BRANCHES 46.4

~ STEMS ' 21.8 [ FRUITS t ZSA) I ~ '

I FiNE ROOTS a,4 I' 2/.4

~6. 8 BEAI, I.,~ 21.6 HUSKS

6 5 ~ I ' LITTER 9.8 l,

F

MINERALISAT~N UPTAKE 1&75

SOIL 577 65.5 OqScm 307 33 O

FERTLIZER IS-3Ocm 165

3,O-45cm lOS

l%PUT

2.5 J INPUT RESIDUES E.POEP~ANA 237.1

21.8 LEAVES 42.5 1 ~ RR~¢HES ~zs.s | STEMS 7o8 /

53.6

N NERALE~AT~N

I 53.6

'T.CACAO 49.1 1.6 26.6 LEAVES

BRANCHES STEMS 14D ,NO,TS

' FIkE R0013 17.1' ~- IZ'.'.Tj

LITTER 12.0 I UPTAKE

S01L 713 ~ . _ j i ~ . o

O-Bern 385 33.0

15-30¢m 202

30-45¢m 125

19~ Ht~(S

Fig. 5. Potassium cycles for agrofOrestry systems T. cacao/C, alliodora and T. cacao/E. poeppigiana in Turrialba, Costa Rica (reserves kg.ha -], transferences kg.ha -~ .yr-I).

this exportation, beans and pericarp (husk) of the fruits are treated separate- ly (Fig. 4). K removal with the husks is higher than with beans. Comparing extraction with harvests and total biomass gives a high exportation index for K (Fig. 5).

T. cacao/C, alliodora T. cacao/E, poeppigiana Biomass (kg.ha -~) 372.7 315.9 Harvest (kg.ha-~ .a - ~ ) 28.4 26.9 % Exported 7.6 8.5

From values published by Heuveldop et al. [14] it can be deduced that plant residue K content differes between the species involved in the agroforestry systems. As a function of amounts of litter produced, T. cacao/C, alliodora and T. cacao/E, poeppigiana respectively deposit 65.5 and 53.6 kg K.ha -l on the soil. The resulting recycling index (residues: biomass) is 18% and 17% respectively for T. cacao/C, alliodora and T. cacao/E, poeppigiana.

Comparing K values in harvests with recycling with plant residues, the following values (kg.ha-l .a - l) are obtained (Fig. 5).

T. cacao/C, alliodora T. cacao/E, poeppigiana Harvests 28.4 26.9 Recycling 65.5 53.6 Recycling: harvest 2.3 2.0

Thus recycling exceeds exportation by more than double. However the systems' annual K requirements are high compared to other

60

~S

c~

~J

0

c~

L~

~s

~j

E~

L~

L~

~J

~S

~N

°i

61

parameters mentioned. Of the exchangeable K, there is an accumulation especially in the biomass and a constant loss from the soil. Applied fertilizer (33.0 kg K.ha - l .a -~) does not manage to compensate for soil losses.

The situation is aggravated considering the K leaching which has not yet been studied.

Calcium and magnesium cycles

Results obtained from the Ca and Mg inventory in the T. cacao/C, alliodora and T. cacao/E, peoppigiana agroforestry systems are summarized in Table 2. Percentage distribution by compartment is as follows:

Total (kg.ha -I) Ca 437.0 Mg 157.4

Total (kg.ha - t ) Ca 433.5 Mg 118.4

Percentage in T. cacao C. aHiodora Humus Roots 23 48 19 10 27 51 13 9

cacao E. poepp~na Humus Roots 22 54 19 5 20 65 11 4

Thus the basic exchangeable elements follow the order Ca > K > Mg. Ca and Mg accumulation was noticeably higher in the shade trees than in the cacao trees. However, highest reserves of these elements are found in the soil, although only their exchangeable forms have been analysed. Percentage values are 85% and 88% for Ca and 78% and 85% for Mg respectively. Thus soil exhaustion is not as high as with potassium.

The dynamics of these elements is characterized by little exportation with harvest (4-5 kg Ca.ha-~.a - t and 4 kg Mg.ha-l.a -I) and residues (124.8 and 163.8kg Ca.ha-~.a-~; 50.3 and 53.7kg Mg.ha'-t.a-~). These elements are therefore more stable than potassium.

References

1. Alpizar L e t al (1986) Sistemas agroforestales de caf6 (Coffea arabica) con laurel (Cordia alliodora) y caf~ con por6 (Erythrina poeppigiana) en Turrialba, Costa Rica. I Biomasa y reservas nutritivas. Turrialba 35: 233-242, 1985

2. Alpizar L e t al (1986) Modelling agroforestry systems of cacao (Theobroma cacao) with Cordia alliodora and Erythrina poeppigiana in Costa Rica. 1 Inventory of organic matter and nutrients. Agroforestry Systems 4:231-257

3. Aranguren J e t al (1982) Cycles of nitrogen of tropical perennial crops under shade. II Cacao. Plant and Soil 67:259-269

4. Boyer J (1973) Cycles de la materie organique et des element mineraux dans une cacaoyere camerounaise. Cafe, Cacao, The 17, 2-23

5. Enriquez G (1979) Ensayo Central de Cultivos Perennes en comparaci6n con algunos

62

anuales. In CATIE: Taller de sistemas agroforestales en Am6rica Latina. 1979. Actas. Editado por G. de las Salas. Turrialba, Costa Rica, 226 p

7. Fassbender HW (1968) Phosphate retention and its chemical forms under laboratory conditions for 14 Costa Rican soils. Agrochimica 12:512-521

8. Fassbender HW (1984) Bases edafol6gicas de los sistemas de producci6n agroforestales. CATIE. Serie de materiales de ensefianza El. CATIE, Turrialba, Costa Rica, 191 p

9. Fassbender HW (1985) Ciclos da materia organica e dos nutrientes em agroecosistemas c o n cacauerios. In Seminar on Nutrient Research and low input agriculture for the

tropics, Anales, editados por P. Cabala, CEPLAC, Itabuna, Brasil, pp 231-257 10. Fassbender HW (1985) Sisternas agroforestales de caf6 (Coffea arabica) con laurel (Cordia

alliodora), caf6 con por6 (Erythrina poeppigiana) en Turrialba, Costa Rica. III Modelos de la materia org~.nica y elementos nutritivos. Turrialba 35:403-413

11. Fassbender HW (1987) Nutrient cycling in agroforestry systems of coffee with shade trees in the Central Experiment of CATIE. In Beer J, Heuveldop J and Fassbender HW: Seminar on Advances in Agroforestry Research, CATIE, Turrialba, Costa Ric a, (In press)

12. Hendry CD et al (1984) Precipitation chemistry at Turrialba, Costa Rica. Water Resour- ces Research. (In press)

13. Heuveldop J et al (1985) Sistemas agroforestales de caf6 (Coffea arabica) con laurel (Cordia alliodora) y de caf6 con por6 (Erythrina poeppigiana) en Turrialba, Costa Rica. II Producci6n agricola, maderable y de residuos vegetales. Turrialba 35:347-355

14. He uveldopJetal(1987)Modellingagroforestrysystemsofcacao(Theobromacacao)with laurel (Cordia alliodora) and por6 (Erythrina poeppigiana) in Costa Rica. II Cacao and wood production, litter production and decomposition. Agroforestry systems. 6:37-48

15. Jordan J (1985) Nutrient cycling in tropical forest ecosystems. J Wiley and sons. New York, 190 p

16. Martinez M and Blasco M (1972) Metabolismo en t6rminos de CO2 en ins suelos cacaoteros de Turrialba, C.R. Turrialba 22:415-419

17. Nair PKR (1984) Soils aspects of agroforestry ICRAF, Nairobi, Kenya, 164 p 18. Santana MBM and Cabala P (1982) Dynamics of nitrogen in a shaded cacao plantation.

Plant and Soil 67:271-281 19. Santana MBM and Cabala P (1984) Reciclagem de nutrientes em una plantacao de cacau

sombreada com Erythrina. 9th. International Cacao Research Conference Togo, 205-212