tropical savanna processes and influence of perennial plants. GOTZ SCHROTH, WOLFGANG ZECH and GONTER HEIMANN 1. Institute of Soil Science and ...
Plant and Soil 147: 1-11, 1992. © 1992 Kluwer Academic Publishers'. Printed in the Netherlands.
PLSO 9517
Mulch decomposition under agroforestry conditions in a sub-humid tropical savanna processes and influence of perennial plants G O T Z S C H R O T H , W O L F G A N G Z E C H and G O N T E R HEIMANN 1
Institute of Soil Science and Soil Geography, University of Bayreuth, P.B. 101251, DW-8580 Bayreuth, Germany and Btometrtctan, Crellestrasse 46, DW-IO00 Berlin 62, Germany 1
•
•
•
Received 21 February 1992. Revised July 1992
Key words': agroforestry, Cajanus cajan, decomposition, Ferric Acrisols, mulching, nutrient release, sub-humid savanna, termites Abstract
For the effective use of mulch materials in tropical agriculture and agroforestry knowledge of the speed of decomposition and nutrient release is of primary importance. The transfer of these informations from one site to another requires comparability of the processes of decomposition and their intensity at the two sites. In a litterbag experiment the decomposition and release of main nutrients from leaves and branches of Cajanus cajan (L.) Millsp. were investigated with regard to the underlying physical and biological processes during an 81 days period. To test the influence of perennial plants on the decomposition process, the study was conducted on an agricultural field in 1.1 m, 6.9 m and 14.9m distance from a tree and hedge band. During the first 11 days leaching was high, especially for N and P (about 50% lost) and K (75-80% lost). After the llth day consumption of the mulch material by the soil fauna was the dominating process of decomposition. During this phase the perennial plants significantly retarded the decomposition of Cajanus branches, but not leaves, probably by their influence on termite activity. Ca release was also retarded in leaves. After about 6-7 weeks, more than 90% of all main nutrients except Ca had been released from the samples. To minimize nutrient losses from nutrient-rich mulch materials, they should be applied repeatedly in small quantities according to the nutrient demand of the crop.
Introduction
The benefits of the application of organic materials as mulch to the soil surface has widely been appreciated in tropical agriculture and agroforestry. They include the reduction of erosion hazards, better infiltration of rain water and less evaporation, lower soil temperatures, supply of organic matter and nutrients, higher biological activity, better root growth and suppression of weeds (Webster and Wilson, 1980). Under tropical smaUholder conditions, mulching materials and especially working capacity for their transport and application on the field are
often in limited supply (Webster and Wilson, 1980). Considering the effort for effective mulching, a maximum of benefits has to be drawn from mulch application to be an acceptable technique for smallholders. Where the primary goal is the substitution of costly mineral fertilizers by mulch, it is necessary to know the release functions of the limiting nutrients to be able to synchronize the periods of maximum supply from the decomposing mulch with those of maximum demand of the crop. Under conditions of high erosion hazard, on the other hand, the organic material should cover the soil when rain-
2
Schroth et al.
fall intensities are highest, thus the speed of dry matter loss is of particular interest. Decomposition and release functions measured at one site may be of limited value at another site, unless the intensity of the decomposition processes, like leaching, catabolism of decomposer organisms and comminution, are known to be comparable between the two sites. The understanding of the processes and their dependance on environmental factors seems thus to be necessary for increasing the efficiency of mulching techniques under varying site conditions. Some factors, like microclimate and soil characteristics, may be influenced by the presence of perennial plants, as, for example, in agroforestry. These influences have to be known to predict mulch decomposition and nutrient release in complex land-use systems like agroforestry. This study was conducted to obtain quantitative informations about the decomposition of mulch from Cajanus cajan (L.) Millsp., the nutrient release from the decomposing material and the processes involved under the conditions of a sub-humid tropical savanna in Central Togo. Litterbags were placed in an agricultural field to measure the rates of decomposition and nutrient release, and different positions to an adjacent tree and hedge band were included in the design to test the influence of the perennial plants on these processes.
Study site and methods
Location and climate The trial was conducted near Kazaboua in the sub-humid savanna of Central Togo (1°5 ' E, 8026 ' N, approx. 300 m above sea level). Mean annual rainfall between 1980 and 1990 was 1134 mm. The rainy season is unimodal and lasts from April to October. The experimental year 1988 was comparatively wet with 1282mm of precipitation. Table 1 shows the rainfall distribution during the experiment. The experimental plot was located on the western slope of a low hill with approx. 3% inclination. The soils are typically Ferric Acrisols according to FAO/Unesco (1988). The topsoil
(0-15 cm) of a nearby soil profile had a texture of loamy sand, a pH in 0.1 M KC1 of 5.6 and was low in organic C (7.7 mg g - l ) , total N (0.54 mg -1 g ) and available P (1.3mg kg 1 extractable with 0.03 N NHaF and 0.025 N HC1). A sandyclayey, usually hardened subsoil restricts root growth at a depth of between 10 and 70 cm. The litterbag experiment was conducted on a field that was 100m long in N-S-direction and 2 4 m wide. At the eastern and western side it was framed by tree and hedge bands which had been planted on the contour lines to serve primarily as erosion barriers and for wood production (Egger, 1986). The trees were 4 years old and 6-8 m high. The dominating species was Cassia siamea Lam. (Cesalpiniaceae). In each of the bands two lines of trees were framed on both sides by hedges of Leucaena leucocephala (Lam.) de Wit. The hedges were about 1.5 m high at the beginning of the trial and were pruned to 80 cm on the 13th of September. On the experimental field maize and peanuts were sown into the smooth soil on the 26th of May after plowing and harrowing. Maize was sown in rows with 160 cm distance between the rows and peanuts were sown in double rows between the maize rows with 40 cm between the rows. Peanuts were harvested on the 8th of September, and maize on the 19th of September.
Preparation of litterbags and deposition in the field Mulch material for the experiment was obtained by pruning hedges of one year old Cajanus cajan (L.) Millsp. at 80cm above soil level. The hedges were 140-160cm high before pruning. The prunings were entirely green although partly lignified. The material was ovendried at 40°C for 20 hours to avoid lengthy air-drying at high atmospheric humidity. The weight ratio of leaves and branches after drying was 5.2:4.8. Nylonmeshbags of 20 × 28 cm, 5 mm mesh size, were filled with 5.2 g of leaves and 4.8 g of branches per bag (40°-dry weight). This was equivalent to an oven dry weight, as determined on subsamples at 80°C, of 4.74g of leaves and 3.13g of branches per bag and corresponded to 1.4 t (dry matter) of mulch per ha when related to the surface covered by the bags.
Mulch decomposition under agroforestry conditions
3
Table 1. Rainfall on the research station at Kazaboua during the seven sampling intervals in 1988 Date
30.6-10.7
Interval (days) Rainfall (mm) Number of events
0-10 171 4
11.7-17.7
18.7-24.7
25.7-31.7
1.8-7.8
8.8-21.8
11-17
18-24
25-31
32-38
39-52
47 4
24 4
20 4
21 3
99 4
22.8-18.9 53-80 230 12
Table 2. Concentration of main nutrients (mg g ~) in the dry matter of prunings of one year old Cajanus cajan, cut at 80 cm height, at the beginning of the experiment (means and standard errors) Leaves Branches
N
P
K
Mg
Ca
33.7 -+ 1.0 9.8 -+0.4
1.83 + 0.10 0.95 +-0.06
11.8 + 0.4 11.2 + (1.8
2.68 -+0.09 1.88 +-0.15
7.55 + 0.21 4.77 +_0.21
On the 30th of June, the litterbags were placed on the soil surface in the field at three distances to the upslope (eastern) tree and hedge band: 1.1 m ("position 1", immediately outside the area covered by the hedge), 6.9 m ("position 7") and 14.9 m ("position 15"). 35 bags were distributed over the field length in each position. They were replaced in lines in the middle between a maize row and a peanut row. Cajanus mulch of the same type as in the litterbags was applied to every second interspace between the samples to avoid artefacts due to an unrealistic concentration of destruents in the bags.
magnesium and calcium were measured by atomic absorption spectrometry. Some of the later samples were too small to be analysed separately and were mixed with other samples of the same position and sampling date before the analysis. Nutrient contents were obtained by multiplying the dry mass of the individual samples by the nutrient concentrations in the pooled samples. This applies to all samples of day 81 and some of the branches of day 53 and of the leaves of days 39 and 53.
Statistical analysis Collection and analysis of mulch samples Five randomly selected litterbags per position were sampled on each of the following days after the exposition in the field: 11, 18, 25, 32, 39, 53 and 81. The samples were washed with tap water followed by distilled water, separated into leaves and branches, dried at 80°C and weighed. The washing process was kept as short as possible to avoid nutrient losses. Carbon data are used instead of dry matter data in this p a p e r to allow for soil particles that were eventually still retained in the samples after the washing. C a r b o n and nitrogen in the samples were measured gas-chromatographically with a C-NAnalyzer. For the other nutrients subsamples were digested in HNO3(conc. ) (Heinrichs et al., 1986). Phosphorus was measured colorimetrically ( v a n a d o m o l y b d a t e - m e t h o d ) , and potassium,
T h e statistical analysis was run separately for each of the measured nutrients in mg per sample and for leaves, branches and total samples, respectively. First, two-sided tests for equality of the three positions were conducted at the 5%level of significance. In case of rejection these were followed by two-sided tests to compare each pair of positions. A conditional version of the Friedman type range test described by M c D o n a l d and T h o m p s o n (1967) was used, with sampling dates as blocks and positions as treatments. Ranks were assigned within each block, introducing average ranks in case of ties. Exact conditional p-values were calculated on the basis of the observed rank sums per position and sampling date. For the two-position comparisons ranks were assigned agaJ.n, ignoring the data of the respective third position.
4
Schroth et al.
Results
bags during the experiment are shown in Figures 1 and 2. The results of the statistical analysis are given in Table 3. The influence of the tree and hedge band on the decomposition process resulted in a significantly slower decrease of branch and total C in position 1 compared to positions 7 and 15. An influence on C loss from the leaves could not be detected. Thus, the leaf data from the three positions were pooled for presentation (Fig. 1). After 11 days, 35-39% of total C and 38-45% of leaf C had been lost from the bags in all positions. 88-96% of leaf C had disappeared after 39 days. The initial decrease of branch C was slightly less rapid with 27-29% within 11 days. C loss from the branches in positions 7 and 15 was almost linear until day 32, when more than 90% of the initial content had disappeared from the bags. From day 11 to 32, C loss from the branches tended to be faster than that from the leaves in positions 7 and 15. Until day 11 the branches exhibited almost the same C loss in all positions. From the 18th day onward the decrease of branch C was slowed down in position 1 compared to positions 7 and 15.
Representativity of the material in the fitterbags and effect on the soil surface The comparison of the material in the litterbags with the unconfined mulch with regard to colour, degree of skeletonization and brittleness indicated an almost identical progression of the decomposition process inside and outside the bags. Only some unconfined leaves that were held away from the soil surface by branches showed a generally drier state and slower decomposition compared with leaves within the bags or outside the bags, but lying directly on the soil surface. Thus, the conditions in the bags during the decomposition process were generally representative of those in a thin mulch layer. The effect of the mulch on the soil surface consisted in a strong increase of surface porosity, accompanied by the concentration of large numbers of plant roots, worms and other soil invertebrates in the protected area directly under the mulch. The effect disappeared after about six weeks, when more than 90% of the mulch had been lost from the litterbags in position 7 and 15 and about 80% in position 1. It did not show any visible differences in duration or intensity under the unconfined mulch and the litterbags.
Nutrient release The original nutrient concentrations in the mulch material are given in Table 2. The release of N, P, K, Mg and Ca from leaves and branches is shown in Figures 1 and 2. The retarding influence of the tree and hedge band
Carbon loss from the litterbags C losses from leaves and branches in the litter-
140 •
C and nutrients except Ca in leaves
----
C N
~~... .....
. ......
p
':,,,v.-
- ....
K
1O 0 E •
80
~
60
"."~.........
100 ~
........... Mg
~ i~
"'.,"~ ......~,, -..
I
l\\ ~ ~
80
-'"i..
0--0 A--A 13--13
T
o
;"~ "'".,
~6 ~
Ca in leaves
120.
120
Position I Position 7 Position 1,5
60.
"6 40. a~
..... -~i
20 0
zo.
,
0
'
10
20
,30
40 Days
50
60
70
80
,
0
0
I
10
,
I
20
,
I
30
,
I
40 Days
,
I",
50
I
60
,
I
70
,
I~ , 80
Fig. 1. Decrease of carbon and main nutrients in the leaves in % of the original content of the litterbags (means and standard errors). For C, N, P, K and Mg the data from the three positions have been pooled due to non-significant differences.
Mulch decomposition under agroforestry conditions 1 2 0T|
C in branches
5
K in branches
/
100111
--
O--O
IX
I
~, 8 0 f N
1
° 6o _~
-l?\
:~ 40
Position 1
Z~--A Position7 I--I--E3 Position 15
,
,30
OoI
40 Days
•c0
40
~
2
~d
o
20
\
-6 .c_
I \~?\~ 10
~ 8°I ~o 6O
50
60
70
80
0
o
20
50
40 Days
50
60
70
80
120-
120
Mg in branches
N in branches 100
lOOt
~
80-
8o
g T
60.
T ~
!/8.__ .
20
T
o
60-
:~'
40-
~
20-
°--~ 6 ~ , , , , ,
10
20
,30
40 Days
50
60
70
80
0
140
10
30
~ ~
60
70
80
1
1oo
8O
50
Ca in branches
P in branches
100
40 Days
-
120-
120
O
, ~ , ~ ~ g ~
20
0
g -6
6O
E
"E O
I
40-
O
b~
"s 20
20-
~ & ~
o
,'~.~,~5..---,,
, , . , , 0
10
20
30
40 Days
50
;__ :~A. 60
70
80
0
I
0
10
,
I
20
'
I
30
,
*1
40
~
I
I
50
60
,
I
70
,
[-] 80
Days
Fig. 2. D e c r e a s e of c a r b o n a n d m a i n n u t r i e n t s in the b r a n c h e s in % of the o r i g i n a l c o n t e n t of the l i t t e r b a g s ( m e a n s and s t a n d a r d errors).
on the release of N, P, Mg and Ca from the branches was highly significant (Table 3). The effect was still detectable in position 7 as compared to position 15. The influence of the tree and hedge band on K release from the branches
and Ca release from the leaves was significant, although no differences between pairs of positions could be assertained. The release of all nutrients except Ca from the leaves was almost identical in the three positions, thus the data
6
Schroth et al.
Table 3.
Results of the tests of significance for the all-treatment comparisons (all) and the comparisons of pairs of decomposition
c u r v e s i n p o s i t i o n s I , 7 a n d 15. F o r l e a f C a n o n e o f t h e t e s t s o f p a i r s o f p o s i t i o n s i n d i c a t e d a s i g n i f i c a n t d i f f e r e n c e Positions
C
N
P
K
Mg
Ca
n.s.
n.s.
n.s.
n.s.
n.s.
*
all
***
***
***
*
***
***
1/7
*
*
*
n.s.
*
*
tested
Leaves all
Branches
1/15
*
*
*
n.s.
*
*
7/15
n.s.
*
*
n.s.
*
*
Total samples all
***
**
**
n.s.
*
***
1/7
*
*
n.s.
-
*
n.s.
1/15
*
*
*
-
*
*
7/15
n.s.
n.s.
n.s.
-
n.s.
n.s.
Symbol meanings:
* = p < 0 . 0 5 , ** = p < 0 . 0 1 , *** = p < 0 . 0 0 1 , n.s. = n o t s i g n i f i c a n t , - = n o t t e s t e d
The release rates of N, P and K from leaves and branches decreased distinctly after the first 11-18 days. However, after abut 6-7 weeks 90% of N and P had been lost from the total samples in position 1 and after 5 weeks in positions 7 and 15. More than 90% of K had already been released after 4 weeks in all positions. Mg disappeared from the total samples in nearly the same rate as C during the whole experiment. The release was faster than that of C from the branches, and slower than that of C from the leaves. Ca exhibited the lowest release rates. Until day 11 the Ca-content had increased absolutely in many samples in all positions. This accumulation was still visible in some samples at day 18. The absolute enrichment in a part of the
from the positions were pooled for presentation (Fig. 1). The differences in nutrient release from the branches did not always yield significant effects for the total samples but were masked by the uniform release rates from the leaves. No differences in carbon and nutrient release were found between positions 7 and 15 for the total samples. Nutrient losses from the mulch samples within the first 11 days amounted to about 50% for N and P and 75-80% for K for the total samples (even 84% for the branches alone). The high initial losses of N and P resulted in an increase of the C/N- and C/P-ratios during the first 11 days for the leaves and 18 days for the branches (Fig.
3). 70.
.... o A .... A [] . . . . []
T
o
n . ~ _ . ~ ,, ":~
60
T
T
1000
Position I Position 7 Position 15
800
t
I
50o 40 I. " " - ' ? ~6 #,,
'
"'" '
30.
"'-,
.--0 . ,;,-~
.9
..:
'"~
~
1 "'m'"
z 20.
._.,....-----~ ~
=
=
600
"5
-5..
400 I
/
c)
.x.
i
!!iiii-° "D
2o0
100
,
0
I 10
~
I 20
~
I 30
F i g . 3. C h a n g e s o f C / N - r a t i o s
,
I 40 Days
,
I 50
,
I 60
'
I
70
,
I 80
~
0
,
I 10
'
I 20
,
I 30
~
I 40 Days
,
I 50
,
I 60
,
I 70
,
I 80
,
and C/P-ratios (means and standard errors) of leaves (solid lines) and branches (dashed lines). For
the leaves the data from the three positions have been pooled.
Mulch decomposition under agroforestry conditions
samples and marked losses of Ca from other samples resulted in comparatively large standard errors for this element. Following the llth day all nutrients were released from the litterbags in about the same ratio
200Day 0 • Days 1 t - 3 2 o Days 3 9 - 5 3
150-
A
"
6
a
and in close proportionality to C loss (Fig. 4). For calculating the regression lines in Figure 4, no mass and nutrient values have been included from the sampling dates where pooled samples had been analyzed. The same regressions were
g 100-
A
20-
e, a
16E
.j.
t_~
7
E ._=
12-
a
8-
~ o o
50-
4:
0
1000
2000
:
3000
f°i
l
.........
4000
1000
2000
Carbon [ m g ]
Carbon
15
4000
5000
[ mg ]
80 t, A
60
Z,
10
A
A
A o',
•
E
oOe|
•
•
00
• •
•
,
,
A
40
L O
E :3 "6
5
J~
(~
e
1000
2000
3000
4000
Carbon [mg ]
o~
•
20 ¸
l
t
J
e
l
1000
'
I
,
2000
~
5000
,
,
I
4000
Carbon [ mg ]
100a A
80-
F E .= o
6040-
El-
0
1000
2000
3000
4000
Carbon [ m g ]
Fig. 4. Relationships between carbon and main nutrients in the total samples. Only values from independently analyzed samples have been included in the graphs. The regressions have been calculated from the data of days 11-32. All regressions are significant at p < 0.001.
8
Schroth et al.
valid for positions 1, 7 and 15. The regressions of C content on the contents of the investigated nutrients were also highly significant ( p < 0.001) for leaves and branches alone.
Discussion
Leaching and Ca translocation
The rapid initial nutrient loss from leaves and branches (Figs. 1 and 2) can be explained by leaching of an easily soluble fraction from the flesh plant material (Swift et al., 1981). The process was about equally pronounced in all of the three positions during the first 11 days. This finding is to be expected under a dominance of leaching, as all of the positions received about the same amount of rainfall. With the exception of Ca, which was obviously not leached to a considerable extent, differences between the positions could only develop after day 11, probably because leaching became less dominant due to lower rainfalls during the subsequent sampling intervals (Table 1) and the loss of the most soluble fraction from the material. An initial "leaching phase" has frequently been observed for mass and nutrient loss from flesh leaves (Babbar and Ewel, 1989; Swift et al., 1981) and also from naturally fallen leaf litter (Maheswaran and Gunatilleke, 1988). The particularly strong leaching of K has also been described by these authors and agrees with the high mobility of K in plants (Marschner, 1986). For this element, the effect of the tree and hedge band has been masked by the intensive leaching and is hardly visible in Figure 2. Rapid release of P, attributable to leaching, during the first three or four weeks of decomposition of leaf litter has been reported by Babbar and Ewel (1989) and Ewel (1976) for some leaf types. In the same experiments, P was released from other leaf types in rates comparable to dry matter loss, which has also been observed by Swift et al. (1981). The variable leachability of P has been explained by Babbar and Ewel (1989) with differences in solubility of P-compounds in the investigated leaf types. The higher leaching rates of N than of C from leaves and branches, which have been found in
this study, differ from results of other authors, who measured initial release rates of N comparable to or slower than dry matter loss from flesh leaves (Babbar and Ewel, 1989; Ewel, 1976; Swift et al., 1981). Babbar and Ewel (1989) concluded, that N is an element of minor leachability. Bernhard-Reversat (1972) reported resuits of four similar decomposition experiments with freshly fallen mixed leaf litter. Initial N release was appreciably higher than dry matter loss in one case, but similar to dry matter loss in three cases. In our Cajanus leaves and branches an unusually high percentage of N was apparently held in easily soluble compounds. The reason might partly lie in the leguminous nature of the decomposing substrate, as none of the experiments cited above has been carried out with pure leguminous materials. Mg was leached from the mulch material to about the same extent as C. This finding is in line with results obtained by Swift et al. (1981), although no general rules seem to exist about the leachability of Mg (Babbar and Ewel, 1989). In contrast, Ca was retained in the plant material during the leaching process and even increased in absolute content in a considerable part of the samples, especially in the leaves of position 1. An increase of Ca content has also been observed by Maheswaran and Gunatilleke (1988), Ewel (1976), Swift (1977) and Swift et al. (1981) during leaf and branch decomposition in temperate and tropical ecosystems. In the latter study the enrichment mainly occurred when animals were excluded from the litterbags. The authors explained this effect with translocation of Ca in fungal hyphae which was more evident when the decomposing material was not simultaneously withdrawn by the litter feeding fauna. This interpretation is in line with our observation of Ca accumulation in some samples and rapid losses of Ca together with C from other samples, which resulted in large standard errors for Ca at day 11 (Figs. 1 and 2). The difference between the positions at day 11 might have partly been a result of higher faunal activity further in the field and partly of a more intensive fungal colonization of the samples near the tree and hedge band. As Ca was not affected by leaching these factors already became evident until the first sampling date. Fungal translocation as an addi-
Mulch decomposition under agroforestry conditions tional factor might explain the significance found in the all-treatment test for leaf Ca (Table 3).
Microbial and faunal decomposition The extent of leaching during the first 11 days was quite different between the nutrients, but independent of the position in the field. Following the eleventh day, all of the nutrients disappeared from the samples in close proportionality to C loss (Fig. 4), and consequently to one another. The position in the field, on the other hand, had an influence on branch decomposition during this second phase (Fig. 2). The processes of decomposition, besides leaching of an easily soluble fraction, are catabolism of decomposer organisms and comminution, mainly by animals feeding on the substrate (Swift et al., 1979). The release of all nutrients in a rather constant ratio suggests that the export of fragments from the litterbags by soil animals was the dominating factor of decomposition. This view is supported by the observed high numbers of invertebrates in and under the mulch layer. K loss always showed the worst relationship to C loss as a reflection of the highest contribution of leaching as a second process. Large numbers of earthworms were observed under the mulch samples mainly at the beginning of the decomposition process and may have contributed significantly to the carbon and nutrient loss, especially from the leaves. During the experiment, the species composition of the "mulch f a u n a " - a t least of the day-active part of it, as no observations were carried out at n i g h t - w a s apparently altered more and more in favour of the termites. Direct observations in the field and feeding marks on collected samples proved that termites were an important factor of mass loss from the lignified branches. In litterbags that had been attacked by termites the woody material was frequently absent or strongly reduced, while bark and leaves were much less affected. Wood (1976) has pointed out, that in savanna regions termites can consume considerable quantities of fresh plant material before it is colonized by saprophytic microorganisms, thus reducing the role of microbes in the primary processes of decomposition. The dominance of the soil fauna in the decom-
9
position process could explain why N and P were still rapidly released from the decomposing material after the leaching phase, although a C/Nratio of 20 and a C/P-ratio of 200, which are normally assumed to be necessary for net N and P release during microbial decomposition (Stevenson, 1986), have never been reached in the branches and only for N in the leaves (Fig. 3). Even the relative enrichment of N and P in the branches after day 18 is not necessarily an indication of microbial activity. As mentioned before, primarily the woody parts were withdrawn from branch samples that had been attacked by termites, while the bark often remained fairly intact. As bark is generally richer in mineral nutrients than wood (Larcher, 1984; Swift et al., 1979), this selective consumption might have significantly contributed to the observed enrichment of N and P in the branch residues. The predominating role of termites in the decomposition of the branches is the most probable explanation for the unusual finding of higher C loss rates from the branches compared to the leaves from day 11 to 32 in positions 7 and 15. Branches have been classified as "low quality resources" for decomposer organisms by Swift and Anderson (1989), with consequently low turnover coefficients compared to leaves. However, any classification of resource quality has to take possible differences in the decomposer community into account. Where termites, and not microorganisms, dominate the decomposition process, woody materials may occasionally exhibit an equal or higher "decomposer-specific" resource quality compared to leafy tissues.
Influence of perennial plants on mulch decomposition The tree and hedge band exerted a significant retarding influence on carbon and nutrient release from the branches in its vicinity in downslope direction, which was still measurable at a distance of 6.9 m, compared to 14.9 m. With the exception of Ca, leaf decomposition was not affected. An initial hypothesis of our experiment had been that microbial decomposition processes might be affected by the trees via their influence
10
Schroth et al.
on the microclimate. If this has been the case, the effect was masked by the rapid withdrawal of the mulch material by the decomposer fauna, otherwise it should have become detectable in the rates of leaf disappearance. According to the identified decomposition processes, it is probable that the influence on branch decomposition was mediated by differences in the feeding activity of termites in the three positions. It was frequently noticed that the termites preferentially attacked mulch samples that had been covered by eroded topsoil during rain storms. This observation agrees with the finding of Abe (cit. in Yoneda et al., 1977) that the speed of the consumption of wood tissue by termites increased with the area of its contact with soil. The downslope vicinity of the tree line was partly protected from surface flow-off, which consequently became more frequent and more intense with increasing distance to this erosional barrier. Thus, the samples in positions 7 and 15 were more often affected by eroded topsoil than the samples in position 1, and were consequently more rapidly invaded by termites. The fauna that fed on the leaves, on the other hand, was apparently not influenced in its activity by the perennial plants.
Consequences for mulching and associated research The existence of a pronounced leaching phase, involving the rapid loss of a large soluble fraction of N, P and K does not correspond with the view of a gradual release of available nutrients from a mulch layer (Webster and Wilson, 1980). This contradiction may be explained by the utilization of a nutrient-rich leguminous mulch material in our experiment, which differed in its decomposition and nutrient release characteristics from other common mulch types like grasses. The following rules hold particularly for mulching in agroforestry, where nutrient-rich leguminous materials are typically used. In order to optimise the utilization of the nutrients contained in the mulch and keep the nutrient cycle as closed as possible, the organic material has to be applied when the nutrients are needed by the crop. To minimize nutrient losses the repeated application of small quantities of mulch to the crop should be preferred to a single
application of a large quantity, expecially if this is done at the beginning of the growing season when the nutrient absorption capacity of the plants is small. The observed gradients of wood decomposition with increasing downslope distance from a line of perennial plants could also be interesting for agroforestry, where perennial plants are more or less intensely mixed with annual crops and where mulching with woody materials is frequently practised. Further research should be undertaken to evaluate the relevance of these observations for the modelling of carbon and nutrient fluxes in tropical agroforestry systems.
Acknowledgements We gratefully acknowledge the technical support by the 'Direction Regional du Developpement Rural', Sokod6 (Togo), especially N Poidy from the research farm at Kazaboua, and the technical and financial support by the 'Projet Integr6 du Developpement Rural' of the 'Deutsche Gesellschaft f/ir Technische Zusammenarbeit' (GTZ) at Sokod6, especially Dr T Zeuner. We thank Kumako Kokouvi and Ibrahim Tchacondo for their careful and committed assistance in the field, A Schroth for the statistical computations, and Dr I K6gel-Knabner for valuable comments on earlier drafts of the manuscript.
References Babbar L I and Ewel J J 1989 Descomposi6n del follaje en diversos ecosistemas sucesionales tropicales. Biotropica 21, 20-29. Bernhard-Reversat F 1972 D6composition de la liti6re de feuiUes en for6t ombrophile de basse C6te d'Ivoire. Oecol. Plant. 7, 279-300. Egger K 1986 Beispiele f/Jr Bodenprobleme und L6sungswege in Afrika. Schr. Reihe Deutsch. Rat. Landespfl. 51, 82-85. Ewel J J 1976 Litter fall and leaf decomposition in a tropical forest succession in eastern Guatemala. J. Ecol. 64, 295307. F A O / U n e s c o 1988 Soil map of the world, revised legend. Rom. Heinrichs H, Brunsack H-J, Loftfield N and K6nig N 1986 Verbessertes Druckaufschlul3system f/Jr biologische und
Mulch decomposition under agroforestry conditions anorganische Materialien. Z. Pftanzenernaehr. Bodenkd. 149, 350-353. Larcher W 1984 Okologie der Pflanzen, 4th ed. Eugen Ulmer, Stuttgart. 403 p. Maheswaran J and Gunatilleke I A U N 1988 Litter decomposition in a lowland rain forest and a deforested area in Sri Lanka. Biotropica 20, 90-99. Marschner H 1986 Mineral Nutrition of Higher Plants. Academic Press, London. 674 p. McDonald B J and Thompson W A 1967 Rank sum multiple comparisons in one- and two-way classifications. Biometrika 54, 487-497. Stevenson F J 1986 Cycles of Soil. Wiley & Sons, New York. 380 p. Swift M J 1977 The roles of fungi and animals in the immobilization and release of nutrient elements from decomposing branch-wood. Ecol. Bull. (Stockholm) 25, 193202. Swift M J, Heal O W and Anderson J M 1979 Decomposition
11
in Terrestrial Ecosystems. Blackwell Scientific Publications, Oxford. 372 p. Swift M J, Russell-Smith A and Perfect T J 1981 Decomposition and mineral-nutrient dynamics of plant litter in a regenerating bush-fallow in sub-humid tropical Nigeria. J. Ecol. 69, 981-995. Swift M J and Anderson J M 1989 Decomposition. In Tropical Rain Forest Ecosystems. Eds. H Lieth and M J H Werger. pp 547-569. Elsevier, Amsterdam. Webster C C and Wilson P N 1980 Agriculture in the Tropics, 2nd ed. Longman, London. 640 p. Wood T G 1976 The role of termites (lsoptera) in decomposition processes. In The Role of Terrestrial and Aquatic Organisms in Decomposition Processes. Eds. J M Anderson and A Macfadyen. pp 145-168. 17th Symp. British Ecol. Soc., Blackwell, Oxford. Yoneda T, Yoda K and Kira T 1977 Accumulation and decomposition of big wood litter in Pasoh Forest, West Malaysia. Jap. J. Ecol. 27, 53-60.
