1. Stover and Potassium Management in an Upland Rice–Soybean Rotation
on an Indonesian Ultisol
Thomas S Dierolf* and Russ S. Yost
ABSTRACT Potassium and stover management are critical to the
uplands of Sitiung, Indonesia, where the predominantFertilizer and stover management greatly determine the extent of
cropping system is upland rice followed by soybean orK deficiency on weathered, low-K soils in the humid tropics. This
study quantified the effects of K fertilization and stover management peanuts (Colfer, 1991). Rice stover (straw) in this system
on soil properties and crop yields on a Typic Kanhapludult in Indone- is usually burned in piles after threshing. Even if farmers
sia. Cowpea [Vigna unguiculata (L.) Walp. subsp. unguiculata]– rotate the location of the rice burn piles each season,
cowpea–rice (Oryza sativa L.)–soybean [Glycine max (L.) Merr.]– incorporation of burnt stover generally results in an
rice–soybean were grown during a 2-yr period. Fertilizer KCl was
uneven distribution of nutrients, which can hasten nutri-
applied as 70 and 250 kg K haϪ1
to the first crop only, and as a total
ent depletion. The large concentration of cations in one
of 250 and 600 kg K haϪ1
applied to several crops. The effect of stover
area may exceed the soil effective cation exchange ca-removal or return was examined for each K rate. Critical soil K levels
pacity (ECEC) and render the nutrient cations suscepti-were 0.14 cmolc kgϪ1
for the final rice crop and 0.14 and 0.16 cmolc kgϪ1
ble to leaching losses (Ritchey, 1979). The objectives offor the two soybean crops. By returning stover, a single application of
this experiment were to (i) quantify the effects of vary-70 kg K haϪ1
to the first crop was adequate to maintain soil K above
the critical level for all six crops. When stover was removed, a total ing rates and timing of K applied as fertilizer KCl or
of 250 kg K haϪ1
applied during several crops maintained soil K above cattle manure and crop stover return or removal on soil
critical levels. A one-time 250 kg K haϪ1
application to the first crop, base cation status; and (ii) relate the effect on base
however, resulted in yield declines by the fifth crop. A maintenance cation status to crop yields for a 2-yr period during
rate of about 45 kg K haϪ1
per crop was required when stover was which six crops were grown.
removed. Stover removal also hastened soil Mg depletion, and thus
a maintenance rate of about 6 kg Mg haϪ1
cropϪ1
is recommended.
MATERIALS AND METHODS
Site Characteristics
Highly weathered, leached Ultisols and Oxisols are Field experiments were conducted at a site near the village
often used for the expansion of agricultural pro- of Sitiung 1A´ , West Sumatra, Indonesia (102Њ E, 1Њ S), from
1989 to 1991. The rainforest originally covering the site wasduction in Indonesia (Donner, 1987). Although soil
cut and cleared by bulldozer in 1976. After three seasons ofacidity and low soil P are the major fertility constraints
rice, the field was fallowed and became covered with alang-in these soils, K usually becomes limiting to crop growth
alang grass [Imperata cylindrica (L.) P. Beauv.]. In 1989, theunder continuous cultivation (von Uexku¨ll, 1985). This
alang-alang was cut and removed after it was sprayed with a
is primarily because both available and reserve K are
glyphosate herbicide. The surface 15 cm was then plowed and
usually low in these soils (Sparks and Huang, 1985). roots of alang-alang were removed. The pedon at the site
Soil and crop management factors also contribute to was classified as a clayey, kaolinitic, isohyperthermic Typic
the occurrence of K deficiency. For example, in the Kanhapludult by staff at the Center for Soil and Agroclimate
move to intensify crop production on highly weathered Research, Bogor. Selected soil chemical and physical data
soils, K application usually ranks far behind N and P with depth are shown in Table 1. Rain ranges from 2500 to
3000 mm yrϪ1
and averages more than 200 mm per month(Pushparajah, 1985). K uptake by plants is similar to
except from June to September, when monthly rain rangesthat of N, however, and is usually an order of magnitude
from 100 to 200 mm. Total rain during the 24-month experi-greater than that for P (Cooke, 1985). Additionally,
mental period was 6870 mm, of which 5440 mm fell whileremoving crop stover from the field hastens the deple-
crops were growing in the plots.
tion of soil K (Pieri, 1982; Vilela and Ritchey, 1985).
Through a combination of the above factors, the soil
Experimental and Treatment Designeventually becomes depleted of available K. Although
Treatment, fertilization, and planting information areK can be replenished through fertilization, excessive
shown in Table 2. A six-crop rotation of cowpea (cv. local),fertilization can result in leaching losses (Ritchey, 1979;
cowpea (cv. local), rice (cv. Tondano), soybean (cv. Lokon),Gill and Kamprath, 1990) and in losses from the luxury
rice (cv. Laut Tawar), and soybean (cv. Wilis) was grown fromconsumption of K (Brady, 1984) when stover is not re-
May 1989 to May 1991. Treatments consisted of combinationsturned.
of rates and timing of KCl fertilizer application and the return
or removal of crop stover. An additional treatment examined
T.S Dierolf, Jasa Katom, Jalan Kehakiman No. 283, Bukittinggi, West
the replacement value of cattle manure for KCl fertilizer. The
Sumatra, Indonesia 26136; R.S. Yost, Dep. Agronomy and Soil Sci.,
field experiment consisted of nine treatments in a randomizedUniv. of Hawaii, Honolulu, HI 96822. This work was supported by the
complete block design with four replications. Plots were 42 m2
Center for Soil Research and Agroclimate (CSAR), Bogor, Indonesia,
with a 12-m2
harvest area. Calcitic lime was applied at theand the Soil Management Collaborative Research Support Program
(USAID). Received 7 Aug. 1998. *Corresponding author (Katom@
bukittinggi.wasantara.net.id).
Abbreviations: DAP, days after planting; ECEC, effective cation ex-
change capacity.Published in Agron. J. 92:106–114 (2000).
106

2. DIEROLF & YOST: STOVER AND K MANAGEMENT IN A RICE–SOYBEAN ROTATION 107
Table 1. Profile description and selected chemical and physical data (described by CSAR staff).
Exchangeable‡ pH (1:1) Particle size¶
Org.§
Horizon Depth Color Structure† Ca Mg K AlϩH Na ECEC# H2O KCl carbon sand clay BD††
cm cmolc kgϪ1
g kgϪ1
%
A 0–12 10YR 3/4 f/m sbk 2.00 0.25 0.14 4.63 0.03 7.05 4.9 3.9 27.1 8 62
Bt1 12–35 10YR 4/4 f/m sbk 0.35 0.03 0.05 4.01 0.03 4.47 4.9 3.8 7.7 11 66
Bt2 35–72 7.5YR 4/4 f/m sbk 0.35 0.05 0.05 4.54 0.04 5.03 5.1 3.9 3.8 6 73
Bt3 72–97 7.5YR 4/6 f/m sbk 0.30 0.03 0.05 4.88 0.04 5.30 5.1 3.8 1.6 7 65
Bt4 97–143 5YR 5/8 f/m sbk 0.30 0.05 0.07 4.88 0.05 5.35 5.0 3.8 3.4 7 54
Bt5 143–160 5YR 5/8 f/m sbk 0.35 0.07 0.04 4.86 0.04 5.36 5.0 3.8 1.0 7 52
† F, fine; m, medium; sbk, subangular blocky.
‡ Ca, Mg, and K extracted with 1 M NH4OAc; AlϩH extracted with 1 M KCl.
§ Organic carbon by Walkley-Black method.
¶ Pipette method after sonification in sodium hexametaphosphate.
# ECEC, effective cation exchange capacity.
†† BD, bulk density (g cmϪ3
) at mean soil depths: 3.75 cm ϭ 0.91; 15 cm ϭ 0.93; 30 cm ϭ 1.09; 45 cm ϭ 1.05; 60 cm ϭ 1.03; 75 cm ϭ 1.00; 90 cm ϭ 1.01.
beginning of the experiment to reduce acid saturation to 25%. biomass was harvested prematurely to allow for timely plant-
ing of the subsequent rice crop. Rice and soybean plants wereAn additional 2 t haϪ1
of calcitic lime was applied before crop
6 (soybean) to reduce the possibility that acidity would limit cut near ground level, sun-dried, and hand-threshed to harvest
grain. Returned stover was cut into 10-cm pieces and returnedsoybean growth. All lime and fertilizer applications were in-
corporated into the surface 15 cm. Plant spacing was 25 by to the appropriate plots. Stover and grain subsamples were
taken for chemical analyses and moisture content determina-25 cm for the first five crops and 20 by 25 cm for the final
crop. Legumes were thinned to 2 plants hillϪ1
at 30 days after tion. Grain yields are reported at 140 g kgϪ1
moisture for
all crops. Stover weights are reported on an oven-dry basis.planting (DAP). The first soybean crop was inoculated with
rhizobium. Urea was side-dressed to the rice crops (Table 2) Harvest fraction samples were dry-ashed at 550ЊC and nutri-
ents were determined by inductively coupled plasma spec-and 10 kg haϪ1
of ZnSO4 and 15 kg haϪ1
of borate were applied
as supplements to crop 5 (rice). Each crop was hand-weeded trometry.
Potassium and Mg uptake in grain and stover were calcu-at least twice and pesticides were applied several times per
crop to control insects and rice blast disease (Pyricularia lated by multiplying the subsample cation concentration by
the respective biomass. Because tissue cation concentrationsoryzae).
Crop 1 (cowpea) was harvested by collecting the entire are based on oven-dried samples, grain biomass was multiplied
by 0.90 to adjust from 140 g kgϪ1
moisture to an average oven-pod and separating the seeds after air-drying. Pods were not
included in the grain weight, but were used to calculate nutri- dry weight of 40 g kgϪ1
. Nutrient removal for the stover-
returned treatments was calculated by summing the uptakeent removal when pods were not returned to the plot. No
grain was harvested for crop 2 (cowpea) because the entire in grain (and pod for the first cowpea crop) for all crops and
Table 2. Treatment codes, fertilization rates, planting and harvesting dates for cowpea (Cp), rice (Rc), and soybean (Sy). An S in the
code indicates that the treatment was split over several crops. Rainfall is shown for period between planting and harvesting of each crop.
Crop
Stover
1 Cp 2 Cp 3 Rc 4 Sy 5 Rc 6 Sy Total returned?
kg K haϪ1
Treatment code†
70ϩ 70 0 0 0 0 0 70 yes
250ϩ 250 0 0 0 0 0 250 yes
250/Sϩ 70 0 45 45 45 45 250 yes
600/Sϩ 120 0 120 120 120 120 600 yes
70Ϫ 70 0 0 0 0 0 70 no
215/SϪ 70‡ 0 37 31 38 39 215 no
250Ϫ 250 0 0 0 0 0 250 no
250/SϪ 70 0 45 45 45 45 250 no
600/SϪ 120 0 120 120 120 120 600 no
kg haϪ1
Blanket treatments of fertilizer and lime
Lime (calcite) § 0 0 0 0 2000¶
P (TSP) 80 0 20 20 20 30 170
N (urea)†† 25 0 90 25 100 0 240
Mg (MgSO4) 16 0 0 0 14 16 46
Planting time May 1989 Aug. 1989 Oct. 1989 Mar. 1990 Oct. 1990 Feb. 1991
Harvest date July 1989 Oct. 1989 Feb. 1990 June 1990 Jan. 1991 May 1991
Rainfall (cm) 51 41 135 97 145 75 544
† Plus sign indicates stover was returned; minus sign indicates stover was removed.
‡ 70 kg K haϪ1
applied as KCl fertilizer to crop 1, and 20 kg K haϪ1
applied as KCl fertilizer to crops 3, 4, 5, and 6. Remainder applied as 2 t haϪ1
air-
dried cattle manure to crops 3 (17 kg K haϪ1
), 5 (18 kg K haϪ1
) and 6 (19 kg K haϪ1
) and as 1 t haϪ1
to crop 4 (11 kg K haϪ1
).
§ Calcitic lime was applied to attain 25% acid saturation; rates ranged from 1.3 to 2.7 t haϪ1
.
¶ A maintenance application of calcitic lime was added at same rate to all plots.
# The total amount of lime applied varied with the initial lime requirement, and ranged from 3.3 to 4.7 t haϪ1
.
†† Applied all at planting for crops 1 and 4; 1/3 at planting and 2/3 at 60 days after planting (DAP); for crop 3; and 1/4 at planting, 1/4 at 21 DAP, and
1/2 at 50 DAP for crop 5.

3. 108 AGRONOMY JOURNAL, VOL. 92, JANUARY–FEBRUARY 2000
Table 3. Grain (140 g kgϪ1
moisture content) and oven-dried sto- yields for soybean (crops 4 and 6) and rice (crop 5) were based
ver yields for cowpea (Cp), rice (Rc), and soybean (Sy). on the maximum treatment yield for the respective crop. Soil
samples used to determine relative yield-exchangeable K rela-Crop
Treatment tionships were taken prior to fertilization for crops 4 (soybean)
code 1 Cp† 2 Cp‡ 3 Rc 4 Sy 5 Rc 6 Sy Total
and 5 (rice), and after harvest for crop 6 (soybean).
kg haϪ1
To more accurately reflect soil K during crop growth, mea-
sured soil K was corrected to include fertilizer K for crops 4Grain yields
and 5, and K taken up in the grain and stover for crop 6.70ϩ 1431 4107 724 3263 1136 10 661
250ϩ 1323 4181 682 3321 1109 10 616 Exchangeable soil K values were corrected for crops 4 (soy-
250/Sϩ 1331 3946 800 3812 1100 10 989 bean) and 5 (rice) by the equation
600/Sϩ 1330 4515 719 3864 1149 11 577
70Ϫ 1451 3299 361 1076 104 6 291 KSoil ϭ KTest ϩ (KFert/␣/BD␳) [1]215/SϪ 1452 3553 643 2913 957 9 518
250Ϫ 1226 3895 598 2655 267 8 641 and for crop 6 (soybean) by the equation
250/SϪ 1507 3910 582 3132 993 10 124
600/SϪ 1396 3835 545 3619 1005 10 400
KSoil ϭ KTest ϩ (KCrop/␣/BD) [2]
LSD (0.05) ns§ ns 178.6 599.1 256.1 1 206.3
C.V. (%) 13.3 11.8 19.5 13.4 20.2 8.4 where KSoil is exchangeable soil K after fertilization
Stover yields (cmolc kgϪ1
), KTest is exchangeable soil K determined from the
70ϩ 1515 915 4628 1282 4291 1652 14 283 soil sample (cmolc kgϪ1
), KFert is the amount of fertilizer applied
250ϩ 2250 991 4762 1234 4652 1732 15 621 to either crop 4 or 5 (kg haϪ1
), ␣ ϭ 585 kg K haϪ1
in the
250/Sϩ 1815 951 4620 1341 5192 1684 15 603
surface 15 cm of soil per cmolc K kgϪ1
, BD ϭ 0.92 g cmϪ3
,600/Sϩ 1756 1035 4842 1191 5278 1686 15 788
the soil bulk density in the surface 15 cm, ␳ ϭ 0.90, the propor-70Ϫ 1567 758 4141 891 2138 541 10 036
215/SϪ 1696 802 4221 1136 3979 1481 13 315 tion of applied K recoverable in this soil by 1 M NH4OAc
250Ϫ 2448 1111 4459 1011 3541 807 13 377 extraction (Dierolf, 1992), and KCrop is the amount of K taken
250/SϪ 1467 982 4422 1116 4284 1521 13 792
up in grain and stover for crop 6.600/SϪ 1819 935 4481 1006 4851 1582 14 674
LSD (0.05) 442.4 ns ns 211.1 661.7 272.3 1 188.3
C.V. (%) 16.7 18.5 7.6 12.8 10.7 13.2 5.8 Statistical Analyses
† Pod yields, which were removed for all treatments, are not included. An analysis of variance was used to test treatment effects
‡ Crop 2 (cowpea) was harvested before maturity and the entire harvested on crop yield and nutrient uptake for each harvest date, and
portion was included in stover yield.
on exchangeable K and Mg for each soil sampling date. An§ ns ϭ not significant.
analysis of variance was conducted to test treatment and time
effects on the change in soil organic carbon. In all cases, an
the stover for crop 6 (soybean). Nutrient removal for the
F-protected LSD (P ϭ 0.05) was calculated when the analysis
stover-removed treatments was considered equivalent to nu-
of variance indicated a significant variable effect (P Ͻ 0.05).
trient uptake, because all aboveground biomass was removed. Relative yield-exchangeable K relationships were estimated
Nutrient analyses for crop 3 (rice) were not available and using several models, with the most suitable being a linear
were estimated by multiplying the average nutrient concentra- response and plateau model,
tions in crop 5 (rice) grain (2.4 g K kgϪ1
and 1 g Mg kgϪ1
)
and stover (12.5 g K kgϪ1
and 1 g Mg kgϪ1
) with the grain Y ϭ ␤1 ϩ ␤2(KCritical Ϫ X) [3]
and stover yields of crop 3 (rice).
where Y is the relative grain yield (%), ␤1 is the plateau yield,Grain discoloration from predominantly Helminthospor-
␤2 is the change in relative yield with respect to extractableium sp. (J. Klap and J. Castano, 1991, personal communica-
K when extractable K was less than the critical K level, KCriticaltion) was measured in rice (crop 3) at 90 DAP by randomly
is the critical soil K level (cmolc kgϪ1
), and X is extractableselecting 15 panicles per plot. Individual grains were visually
soil K (cmolc kgϪ1
) (Systat, 1992). A similar model was usedinspected and considered discolored if Ͼ10% of the grain was
to describe the relationship between grain discoloration anddarkened. The percentage of grains not discolored was related
exchangeable K, except that Y was grains not discolored (%).
to extractable K determined from soil samples taken 35 d
after grain sampling. These were the soil samples, of those
available, that most closely represented the soil K status at RESULTS AND DISCUSSION
panicle sampling for grain discoloration.
Grain Yields
Critical Soil Potassium Levels Potassium fertilization responses were affected by
stover management (Table 3). When stover was re-The critical soil K level is defined as the exchangeable soil
turned, there were no grain yield responses to K fertil-K level above which no further grain yield increase is expected.
The critical soil K for a crop was determined from the relation- ization for any of the six crops (with one exception in
ship between relative yield and exchangeable soil K for the crop 5). However, when stover was removed, responses
respective crop. Soil samples were taken from the surface to K fertilization began to appear by crop 4 (soybean) at
15 cm for crop 1 at 0 months (before fertilization), for crop the lowest K rate (70Ϫ). Relative yield for this treatment
3 at 5 months, for crop 4 at 9 months, for crop 5 at 16 months, continued to decrease for the remainder of the experi-
and for crop 6 at 24 months (after harvest). Soils were sampled
ment. When 250 kg K haϪ1
(250Ϫ) was applied as aprior to fertilization to avoid the high variablity in measured
single dose before planting the first crop, relative yieldexchangeable soil K, which our previous experience had indi-
did not decrease until crop 5 (rice), and the decreasecated was the result of sampling immediately after fertilization
continued through crop 6 (soybean). In comparison,on these low-K soils. Cations from the soil samples were ex-
yields did not decline when the 250 kg K haϪ1
wastracted with 1 M NH4OAc. Extracts were then analyzed for K,
Mg, and Ca by atomic absorption spectrophotometry. Relative applied over several crops (250/SϪ). For any given K

4. DIEROLF & YOST: STOVER AND K MANAGEMENT IN A RICE–SOYBEAN ROTATION 109
Fig. 1. Relationship of exchangeable soil K in the surface 15 cm and relative grain yield for (a) rice, crop 5; and (b) soybean, crops 4 and 6.
treatment, stover removal generally resulted in lower Higher critical levels for rice (0.16 to 0.20 cmolc kgϪ1
)
than for soybean (0.10 to 0.18 cmolc kgϪ1
) were reportedyields, starting with crop 4 (soybean) (Table 3).
in an Indonesian Ultisol (Gill, 1988). Gill also found
that values of critical K levels for soybean increasedCritical Soil Potassium Levels
with higher lime rates and critical levels for both upland
Critical soil K levels were 0.14 cmolc kgϪ1
for rice rice and soybean crops were affected by the soil extrac-
(crop 5) (Fig. 1a) and 0.14 and 0.16 cmolc kgϪ1
for the tion method used. Based mostly on the work by Gill
two soybean crops (Fig. 1b). No critical soil K levels (1988), Wade et al. (1988) recommended critical K levels
were established for cowpea because there was no grain of 0.20 for upland rice and 0.16 for soybean in the Sitiung
yield response to soil K levels in the first crop and area. In contrast, Cox and Uribe (1992a, 1992b) reported
the second crop was harvested before maturity. Ritchey higher critical K levels for soybean (0.19) than for rice
(1979) reported that critical K levels for food crops and cowpea (0.10). The authors noted, however, that
ranged from 0.13 to 0.20 cmolc kgϪ1
for several Ultisols the coefficients of determination for the linear response
and Oxisols in South America and Puerto Rico. A criti- and plateau models for rice (r2
ϭ 0.41) and cowpea (r2
ϭ
cal level of 0.13 cmolc K kgϪ1
was recommended for 0.21) were low in comparison to soybean (r2
ϭ 0.95).
cereal crops in the Brazilian humid savannas (cerrados) Rice panicles in crop 3 exhibited grain discoloration
caused predominantly by Helminthosporium sp. Figure(Vilela and Ritchey, 1985).

5. 110 AGRONOMY JOURNAL, VOL. 92, JANUARY–FEBRUARY 2000
Fig. 2. Relationship of exchangeable soil K in the surface 15 cm and percentage of grains rated as not discolored from fungal infection
(predominantly Helminthosporium sp.) at 90 DAP for rice (crop 3). Soil samples were taken 35 d after grain discoloration was rated.
2 shows the relationship between exchangeable soil K was planted (Fig. 3a). Although this did not have a
and the percent of grains not discolored that was mea- significant effect on yield between treatments 70ϩ and
sured at 90 DAP. The critical K level was 0.135 cmolc 70Ϫ (Table 3), the 70Ϫ treatment had the highest rate
kgϪ1
. Field observations indicated that the severity of of grain discoloration (data not shown). Even with a
discoloration increased up to harvest (109 DAP), with one-time application of 250 kg K haϪ1
and stover re-
treatment 70Ϫ most adversely affected. These results moval, the 250Ϫ treatment could not maintain the soil
confirm other reports of the positive effect of K in reduc- K above the critical level beyond crop 4 (soybean). This
ing grain discoloration by Helminthosporium sp. (Huber is reflected in the lower yield for 250Ϫ as compared to
and Arny, 1985). 250ϩ for crop 5 (Table 3).
The split-application of 250 kg K haϪ1
(250/SϪ) main-
tained soil K above the critical level even when stoverStover Management Effects on Soil Potassium
was removed (Fig. 3a). (As mentioned earlier, soil sam-Figures 3a and 3b show the effect of stover manage-
ples for crops 1, 3, 4, and 5 were taken prior to fertiliza-ment on soil K levels in the surface 15 cm. The soil K
tion. Thus, the exchangeable soil K values shown in Fig.levels in Fig. 3a and 3b underestimated the soil K status
3a and 3b do not include the fertilizer that was appliedat planting for each crop, since soil sampling was com-
to that crop. For example, for the 250/SϪ treatment,pleted before basal fertilization and planting of the sub-
45 kg K haϪ1
was applied to crop 5 and increased thesequent crop (indicated by arrows). For example, the
measured prefertilization soil K by approximatelysoil K levels at planting for both the 250/SϪ and 250/Sϩ
0.08 cmolc kgϪ1
.) In contrast, soil K decreased to belowtreatments would be about 0.08 cmolc K kgϪ1
greater
the critical level by crop 5 for the one-time application(to account for the addition of 45 kg K haϪ1
) than those
of 250 kg K haϪ1
when stover was removed (250Ϫ).shown in Fig. 3a and 3b for crops 3, 4, and 5. Soil samples
Thus, the yield decline that occurred in the one-timeat 5, 9, and 24 months were taken within 1 to 2 wk after
application of 250 kg K haϪ1
(250Ϫ), but did not occurharvest. The short time interval between the return of
when the same amount of fertilizer was split over severalstover and soil sampling probably did not allow suffi-
treatments (250/SϪ), is related to the maintenance ofcient time for all of the K to be released from the
soil K above the critical soil K level.decomposing stover. This would result in an additional
These results are similar to those reported by Coxunderestimation of soil K for the residue-returned
and Uribe (1992b) on a Peruvian Ultisol recently clearedtreatments.
of an 18-yr-old secondary forest. Returning stover main-Returning the stover always resulted in the mainte-
tained soil K above the critical level of 0.10 cmolc K kgϪ1
nance of soil K above the critical level. Thus, the lack
for 12 crops of rice and cowpea. Removing stover re-of grain yield response to K fertilization when stover
quired a maintenance rate of 40 kg K haϪ1
to maintainwas returned (Table 3) was a direct result of the soil K
the soil K above the critical level. Wade et al. (1988)for each treatment remaining above the critical K levels
recommended slightly higher rates of 60 kg K haϪ1
for(Fig. 3b). (Note that the drop below the critical level in
rice and 40 kg K haϪ1
for soybean when stover wasthe last crop for the 70ϩ treatment is because stover
removed from Ultisols in Sitiung, Indonesia.from the final harvest was not returned to the soil at
Removing the stover in the 600/SϪ treatment resultedthe termination of the experiment. Therefore, the quan-
in significantly lower yields for crop 4 as compared withtity of K in the stover is not reflected in the soil analyses.)
the one-time application of 70 kg K haϪ1
and stoverStover removal for the 70Ϫ treatment decreased soil
K to below the critical level by the time crop 3 (rice) return treatment (70ϩ). The exchangeable soil K was

6. DIEROLF & YOST: STOVER AND K MANAGEMENT IN A RICE–SOYBEAN ROTATION 111
Fig. 3. Change in exchangeable K in the surface 15 cm with time as affected by K fertilization and stover (a) removal or (b) return. Rice and
soybean critical K levels were taken from Fig. 1. The soil K levels in both figures underestimate the soil K status at planting for each crop,
because soil sampling was conducted before basal fertilization and planting of the subsequent crop (indicated by arrows). Soil samples at 5,
9, and 24 months were taken before adequate decomposition of the previous crop’s stover, thus resulting in additional underestimation of
soil K for the residue returned treatments. Error bars in Fig. 3b indicate the LSDs (P ϭ 0.05) for all treatment means at a given sampling
time and apply to both parts of the figure.
not the limiting factor in this case (Fig. 3a), and indicates
Table 4. Change in soil organic carbon after six crops were grownthat stover removal may have detrimental effects on
during 24 months.soybean grain yield that can not be corrected by K
Monthreplenishment with fertilizer KCl.
Treatment code 0 24
Stover Management Effects on Soil Magnesium g kgϪ1
70ϩ 26.9 24.5Yields for the stover-removed treatments in crop 4 250ϩ 27.8 24.8
(soybean) were, in general, lower than for the stover- 250/Sϩ 27.5 25.4
600/Sϩ 27.1 27.0returned treatments (Table 3). Extractable soil K was
70Ϫ 24.5 24.0
above the critical K level for soybean except when stover 215/SϪ 26.9 24.6
was removed for the 70Ϫ treatment (Fig. 3a). A yield 250Ϫ 28.2 24.7
250/SϪ 27.9 25.2response to increased soil K may not explain all of the
600/SϪ 26.4 24.0
yield increase due to stover return, since after applying
LSD (0.05) CV%
the maintenance rate of 45 kg K haϪ1
to the 250/SϪ Treatment NS 7.2
Date 0.7 5.8treatment, the available soil K should be higher than
Treatment ϫ Date NS 5.8
that for the 70ϩ treatment (Fig. 3a and 3b). However,

7. 112 AGRONOMY JOURNAL, VOL. 92, JANUARY–FEBRUARY 2000
Fig. 4. Change in exchangeable soil Mg in the surface 15 cm with time as affected by stover return (250/Sϩ) or removal (250/SϪ) for the split-
application of 250 kg K haϪ1
over several crops. Vertical bars indicate one SE of the mean.
grain yield for crop 4 was higher for the 70ϩ treatment between stover treatments for the 600/Sϩ and 600/SϪ
than for the 250/SϪ treatment (Table 3). treatments (Table 3). Although yield differences due to
A possible explanation for the relatively lower yields stover management for the 250/Sϩ and 250/SϪ treat-
of the stover-removed treatments in crop 4 (soybean) ments were still evident for crop 5, they were no longer
may be the reduction in exchangeable soil Mg. Figure significantly different for crop 6 (Table 3).
4 shows the change in soil Mg for the 250/S treatments.
Magnesium was not applied after the initial application Cattle Manure as Substitute for Potassium
for crop 1 to crop 5 (Table 2). The gradual decrease in
Chloride Fertilizer
soil Mg up to crop 5 was hastened by stover removal
One ton of air-dried cattle manure contained about(Fig. 4). The soil Mg values for both treatments were
9 to 10 kg K haϪ1
, in addition to the 150 kg K haϪ1
thatnear the critical limits of 0.24 cmolc Mg kgϪ1
for soybean
was applied as KCl fertilizer, and resulted in a totaland 0.21 cmolc Mg kgϪ1
for rice recommended by Wade
application of 215 kg K haϪ1
(215/SϪ) for the experi-et al. (1988). After Mg was reapplied to crops 5 (rice)
and 6 (soybean), there were no significant differences ment (Table 2). The amount and timing of K in the
Table 5. Uptake and removal of K and Mg in grain (G) (includes pod for cowpea) and stover (S). Nutrient analyses were not obtained
for crop 3 (rice) and were estimated to calculate nutrient uptake (see text). The total uptake in biomass and total removal (in biomass
not returned to the treatment) includes estimates from crop 3 (rice) and are shown as ‘six–crop estimate’. Stover for the sixth and
final crop was not returned for any treatment and was considered removed.
1 Cowpea 2 Cowpea 4 Soybean 5 Rice 6 Soybean Total uptake
Treatment Total removal
code G S S G S G S G S Five crops Six–crop estimate (Six–crop estimate)
kg haϪ1
K
70ϩ 28 31 12 14 31 7 47 19 24 213 280 101
250ϩ 28 52 14 13 31 6 78 20 31 273 341 107
250/Sϩ 24 25 16 14 29 8 88 19 26 249 315 101
600/Sϩ 25 37 14 13 29 9 100 21 33 281 352 111
70Ϫ 32 25 7 6 17 2 9 1 2 102 161 161
215/SϪ 27 26 10 13 19 6 41 15 13 172 232 232
250Ϫ 25 63 19 11 24 6 23 3 7 182 246 246
250/SϪ 33 24 12 11 16 7 42 16 16 175 239 239
600/SϪ 32 40 15 10 20 8 88 18 28 259 323 323
LSD (0.05) NS 14.2 6.1 3.5 4.7 1.8 19.5 4.5 5.9 29.9
CV (%) 19.2 27.2 31.4 20.7 13.2 18.5 23.3 20.9 20.0 9.6
MG
70ϩ 4 4 3 2 4 3 6 3 7 36 45 22
250ϩ 4 4 2 2 4 2 4 3 6 30 38 20
250/Sϩ 4 3 3 2 5 3 4 3 6 32 40 20
600/Sϩ 3 4 3 2 4 4 3 3 5 29 38 20
70Ϫ 4 4 2 1 4 1 4 0 3 22 29 29
215/SϪ 3 4 2 2 4 3 5 2 6 31 39 39
250Ϫ 3 5 3 2 3 3 4 1 3 26 34 34
250/SϪ 4 3 2 2 3 3 4 2 6 29 37 37
600/SϪ 4 4 3 2 3 3 3 3 5 31 35 35
LSD (0.05) NS NS NS 0.5 1.1 1.0 1.2 0.6 1.4 4.5
CV (%) 14.1 25.3 28.0 21.9 19.9 25.9 19.5 21.7 18.3 10.5
† Pod yields, which were removed for all treatments, are not included.
‡ Crop 2 (cowpea) was harvested before maturity and the entire harvested portion was included in stover yield.
§ NS ϭ not significant.

8. DIEROLF & YOST: STOVER AND K MANAGEMENT IN A RICE–SOYBEAN ROTATION 113
cattle manure treatment (215/SϪ) was most similar to Recommendations for Potassium, Magnesium,
and Stover Managementthe 250/SϪ treatment (Table 2). Grain yields in Table
3 were not significantly different between these two Our results suggest that stover should be returned to
treatments. Exchangeable soil K for the two treatments the field when possible. Returning crop stover after
was also very similar throughout the experiment (Fig. harvest can substantially reduce the need for K fertilizer
3). These results indicate that 1 t air-dried cattle manure inputs in a rice–soybean cropping system. When the
haϪ1
could effectively replace about 18 kg KCl fertil- stover is returned, a one-time application of 70 kg K haϪ1
izer haϪ1
. (70ϩ) can sustain up to six sequential plantings of cow-
pea, soybean, and rice over a period of 24 months. When
Stover Management Effects on Soil
stover is removed, however, an additional 35 to
Organic Carbon
45 kg K haϪ1
per crop is required to sustain soil K above
We also considered the effect of stover management the critical levels for rice and soybean. In systems where
on organic carbon content, and some of the indirect stover is removed because of its usefulness for other
effects of organic carbon content on soil fertility. There purposes, applying smaller rates of K to each crop rather
were no treatment differences after 24 months, although than a large single amount to one crop can reduce luxury
there was an overall decrease (LSD0.05 ϭ 0.7) in organic K uptake. Cattle manure can be used to replace some
carbon between the beginning and the end of the experi- of the KCl fertilizer at a rate of about 1 ton haϪ1
for
ment (Table 4). every 18 kg haϪ1
of KCl fertilizer (9.5 kg K haϪ1
).
Stover removal also hastened the depletion of soil Mg;
Cation Removal in Biomass about 6 kg Mg haϪ1
is required to replace Mg removed in
the grain and stover of each crop. These recommenda-The effect of soil cation loss through crop biomass
tions also apply to systems where stover is not removed(grain and stover) removal is shown in Table 5. Calcium
from the field but may be gathered and not equallyloss through biomass removal is not discussed because
redistributed across the field.of the relatively low percentage of crop uptake of Ca
compared with that applied (Dierolf, 1992). Average K
depletion through biomass removal was 40 kg K haϪ1
REFERENCES
when both stover and grain were removed and Brady, N.C. 1984. The nature and properties of soils. 9th ed. Macmil-
18 kg K haϪ1
when only grain was removed. The replen- lan, New York.
ishment rate when stover was removed is similar to the Colfer, C.J.P. 1991. Toward sustainable agriculture in the humid trop-
ics. Soil Manage. Collaborative Support Progr. TropSoils Bull. 91-maintenance rate of 45 kg K haϪ1
that was applied to
02. North Carolina State Univ., Raleigh, NC.crops 3 to 6 in the 250/Sϩ and 250/SϪ treatments.
Cooke, G.W. 1985. Potassium in the agricultural systems of the humidMagnesium lost through biomass removal would have
tropics. p. 21–28. In Potassium in agricultural systems of the humid
reduced Mg in the surface 15 cm by about 0.12 cmolc tropics. Int. Potash Inst., Bern, Switzerland.
Mg kgϪ1
even when stover was returned, and by about Cox, F.R., and E. Uribe. 1992a. Potassium in two humid tropical
0.21 cmolc kgϪ1
when stover was removed. Considering Ultisols under a corn and soybean cropping system: I. Management.
Agron. J. 84:480–484.that initial Mg levels were about 0.25 cmolc kgϪ1
, it seems
Cox, F.R., and E. Uribe. 1992b. Management and dynamics of potas-that Mg can be rapidly depleted through stover removal.
sium in a humid tropical Ultisol under a rice–cowpea rotation.When stover was removed, the one-time application
Agron. J. 84:655–660.
of 250 kg K haϪ1
(250Ϫ) resulted in higher K removal Dierolf, T.S. 1992. Cation movement and management in an Indone-
in the biomass for crops 1 and 2 (actual plant uptake sian Ultisol. Ph.D. diss. Univ. of Hawaii, Honolulu (Diss. Abstr.
data were not available for crop 3), but the split-applica- 93-12189).
tion (250/SϪ) resulted in higher K removal for crops 4, Donner, W. 1987. Land use and environment in Indonesia. Univ. of
Hawaii Press, Honolulu.5, and 6. This trend is similar to the relative soil K
Gill, D.W. 1988. Response of upland crops to potassium at three levelslevels for the two treatments shown in Fig. 3. However,
of aluminum saturation in the humid tropics of West Sumatra.
because soil K in both treatments remained above the
Ph.D. diss. North Carolina State Univ., Raleigh (Diss. Abstr. 88-
critical level during the first three crops, there was no 21759).
grain yield increase from the one-time application (Ta- Gill, D.W., and E.J. Kamprath. 1990. Potassium uptake and recovery
ble 3). Thus, a partial explanation for the depletion of by an upland rice–soybean rotation on an Oxisol. Agron. J. 82:
329–333.soil K below the critical level in the 250Ϫ treatment is
Huber, D.M., and D.C. Arny. 1985. Interactions of potassium withthe luxury uptake and removal of K in the biomass of
plant disease. p. 467–488. In R.D. Munson (ed.) Potassium in agri-early crops. Another example of luxury uptake is that
culture. ASA, CSSA, and SSSA, Madison, WI.
the 600/SϪ treatment removed 84 kg K haϪ1
more in Pieri, C. 1982. Potassium fertilization of pennisetum millet and its
the biomass than did the 250/SϪ treatment (Table 5), effect on the fertility of a sandy soil in Senegal. Potash Rev. Subj.
but there was no significant yield difference (Table 3) 27, Suite 101, No. 4. Int. Potash Inst., Bern, Switzerland.
between the two treatments. Thus, there is probably no Pushparajah, E. 1985. Trends in the use of potassium fertilizers in the
humid tropics. p. 345–361. In Potassium in agricultural systems ofreason to maintain soil K much higher than the critical
the humid tropics. Int. Potash Inst., Bern, Switzerland.soil K level because the uptake of additional K not
Ritchey, K.D. (ed.) 1979. Potassium fertility in Oxisols and Ultisolsneeded for yield increase is susceptible to removal from
of the humid tropics. Cornell Int. Agric. Bull. 37. Cornell Univ.,
the soil if the biomass is not returned. The incidence of Ithaca, NY.
grain discoloration in rice also disappeared at about 0.13 Sparks, D.L., and P.M. Huang. 1985. Physical chemistry of soil potas-
cmolc K kgϪ1
, which is near the critical soil K level for sium. p. 201–276. In R.D. Munson (ed.) Potassium in agriculture.
ASA, CSSA, and SSSA, Madison, WI.rice grain yields.

9. 114 AGRONOMY JOURNAL, VOL. 92, JANUARY–FEBRUARY 2000
Systat, 1992. Systat Inc. Evanston, IL. agricultural systems of the humid tropics. Int. Potash Inst., Bern,
Vilela, L., and K.D. Ritchey. 1985. Potassium in intensive cropping Switzerland.
systems on highly weathered soils. p. 1155–1175. In R.D. Munson Wade, M.K., D.W. Gill, H. Subagjo, M. Sudjadi, and P.A. Sanchez.
(ed.) Potassium in agriculture. ASA, CSSA, and SSSA, Madi-
1988. Overcoming soil fertility constraints in a transmigration areason, WI.
of Indonesia. TropSoils Bulletin 88-01. Soil Manage. Collaborativevon Uexku¨ll, H.R. 1985. Improvement and maintenance of soil fertility
Support Progr., North Carolina State Univ., Raleigh.in tropical upland farming systems. p. 233–250. In Potassium in
SMALL GRAINS
A Model Analysis of Yield Differences among Recombinant Inbred Lines in Barley
Xinyou Yin,* Martin J. Kropff, Jan Goudriaan, and Piet Stam
ABSTRACT in fact declined in recent years (CIMMYT, 1995). Sup-
port from other disciplines is therefore becoming moreCrop models can support plant breeding if they can predict differ-
important (Bindraban, 1997).ences in performance of different genotypes. In this study, the ability
of a crop model to explain yield differences among genotypes in a Donald (1968) introduced the concept of designing
recombinant inbred line (RIL) population of two-row barley (Hordeum ideotypes with enhanced yield potential based on a com-
vulgare L.) was explored. Yield and model-input traits of 94 RILs and position of favorable traits. Ideotype breeding gives
their parents, ‘Prisma’ and ‘Apex’, were measured in field experiments higher priority to sets of individual traits than selection
conducted in Wageningen, Netherlands, in 1996 at low and in 1997
for yield per se. While Donald (1968) focused mainly
at high N levels. The major gene, denso, with the dwarfing allele
on morphological traits, ideotype breeding was broad-from Prisma, was segregating in this population. Short denso RILs
ened by including physiological traits. For example, theoutyielded tall types in both years, and this yield advantage was
new plant type for rice (Oryza sativa L.), which is ex-stronger in 1997, largely because the tall genotypes lodged. A crop
pected to break through the apparent yield potentialmodel based on existing routines for biomass production explained
barrier of current cultivars, was defined largely on theonly 26 to 38% of the yield variation among genotypes. The model,
using input traits measured from the 1997 data, did not accurately basis of studies of physiological yield-component traits
predict growth of genotypes in 1996 because some traits varied with (Peng et al., 1994). The need for physiological studies
plant N status, which the model did not account for. Model analysis in plant breeding results at least partly from the fact
in the high-N environment showed that of the seven model-input traits that yield is a very complex trait controlled by numerous
examined, only lodging score, preflowering duration, and fraction of
interacting genes. Physiological studies provide a way
biomass partitioned to spikes had a significant effect on yield. When
to dissect complex traits into simpler components thatthese three traits were used while fixing others at their across-genotype
might be under separate genetic control.means, the model explained 65% of yield variation. To allow effective
A promising approach to analyzing yield is the use ofuse of crop modeling in breeding, the ability of crop models to explain
ecophysiologically based crop growth models (Loomis etyield differences among genotypes has to be improved.
al., 1979; Boote et al., 1996). These models have been
developed by integrating knowledge across disciplines
and are increasingly applied in problem-oriented research
To cope with growing demand for food, new culti-
(Boote et al., 1996). One application is the explanationvars with increased yield potential are required.
of differences in yield potential of genotypes on theThrough intensive selection, largely by selecting for
basis of individual physiological characteristics and theyield per se on an empirical basis, breeders have been
use of this knowledge to evaluate and design new plantsuccessful in increasing crop yield potential. By compar-
types (Loomis et al., 1979; Kropff et al., 1995). In mecha-ing eight wheat (Triticum aestivum L.) cultivars released
nistic crop models, the genotypic expression of geneticbetween 1962 and 1988, Sayre et al. (1997) reported a
characteristics of plants is simulated for a specific envi-linear progress in yield potential, with a rate of progress
ronment. Physiological parameters in crop models, oragainst year of release of 67 kg haϪ1
yrϪ1
. However,
model-input traits, are often referred to as genetic coeffi-further progress has been increasingly difficult and has
cients, indicating that these parameters are mainly under
genetic control (Stam, 1998). Using models, many stud-X. Yin, Plant Research International, Wageningen Univ. and Re-
ies (e.g., Penning de Vries, 1991; Boote and Tollenaar,search Centre, P.O. Box 14, 6700 AA Wageningen, Netherlands; M.J.
Kropff and Jan Goudriaan, Laboratory of Theoretical Production 1994; Kropff et al., 1995) examined opportunities to
Ecology, Wageningen Univ. and Research Centre, P.O. Box 430, increase yield potential. Others have explored better
6700 AK Wageningen, Netherlands; P. Stam, Lab. of Plant Breeding,
Wageningen Univ. and Research Centre, P.O. Box 386, 6700 AJ
Abbreviations: DAE, days after emergence; DS, development stage;Wageningen, Netherlands. Contribution of The C.T. de Wit Graduate
FPleaf, fraction of shoot biomass partitioned to leaves; FPspike, fractionSchool for Production Ecology of Wageningen Univ. Funded by The
of shoot biomass partitioned to spikes; LAI, leaf area index; LNC, leafNetherlands Organization for Scientific Research. Received 1 Feb.
nitrogen content; Pre-F, preflowering duration; Post-F, postflowering1999. *Corresponding author (X.yin@plant.wag-ur.nl).
duration; RIL, recombinant inbred line; SLA, specific leaf area; SYP-
BL, simulator of yield potential for barley.Published in Agron. J. 92:114–120 (2000).