Published March, 2003
WHEAT Influence of Diverse Cropping Sequences on Durum Wheat Yield and Protein in the Semiarid Northern Great Plains Y. T. Gan,* P. R. Miller, B. G. McConkey, R. P. Zentner, F. C. Stevenson, and C. L. McDonald ABSTRACT
(Typic Borolls) soil zones of the Canadian prairies (Gan and Noble, 2000). The area planted to lentil in Saskatchewan increased from 300 000 ha in 1995 to 670 000 ha in 2000 (Anonymous, 2001). The inclusion of these crops as alternatives to cereals allows producers to become less reliant on summer fallow and monoculture cropping systems. Expanded production of these alternative crops provides producers the opportunity to grow cereal crops on different types of stubble. The rotational benefits derived from these opportunities are not well documented in this region. Types of crops grown in previous years may impact the soils differently, affecting the amounts of residual soil water and nutrients available for subsequent plant growth. Arranging crops in an appropriate sequence allows them to use the available resources more efficiently and improves soil productivity at a system’s level. Zentner et al. (2001) reported that spring wheat GCPC was higher in 11 of 18 yr when grown on lentil stubble compared with wheat stubble at Swift Current, SK, Canada. Residual NO3–N below 60-cm soil depth following the pulse crop was the main factor contributing to the increased GCPC and occasionally grain yield of the subsequent wheat crop (Campbell et al., 1992). Similar responses of cereal grain yield and/or GCPC to previous annual pulses have been observed in both semiarid and subhumid environments (Wright, 1990a, 1990b; Miller et al., 2001). Other studies have shown that the narrow C/N ratio of pulse residues enhances soil N availability (Beckie and Brandt, 1997; Beckie et al., 1997). The accumulation of crop residues with frequent inclusion of pulse crops in a rotation is shown to improve the biochemical and physical properties of the soil by increasing labile organic matter (Biederbeck et al., 1994). The semiarid northern Great Plains is one of the major durum wheat production areas in the world (Clarke et al., 1998). The largest proportion of Canadian durum wheat is grown in the semiarid Brown (Aridic Haploborolls) and Dark Brown (Typic Borolls) soil zones of the prairies. However, there is little information available regarding the most suitable crop sequences for durum wheat production under no-till, dryland cropping systems. The objective of this study was to determine the effects of crop type and cropping sequences from the previous 2 yr on the yield and quality of durum wheat in the semiarid northern Great Plains.
Crops grown in previous years impact the amounts of residual soil water and nutrients available for subsequent plant growth. Appropriate sequences allow efficient use of the available soil resources by the crop to increase yields at a system’s level. This study was conducted to determine whether the grain yield and grain crude protein concentration (GCPC) of durum wheat (Triticum turgidum L.) were related to crops grown in the previous 2 yr. Durum was grown following pulses [chickpea (Cicer arietinum L.), lentil (Lens culinaris Medik.), and dry pea (Pisum sativum L.)], oilseed [mustard (Brassica juncea L.) or canola (B. napus L.)], and spring wheat (Triticum aestivum L.) in southwest Saskatchewan from 1996 to 2000. Durum increased grain yields by 7% and GCPC by 11% when grown after pulse crops rather than after spring wheat. Durum after oilseeds increased grain yield by 5% and GCPC by 6%. Pulse and oilseed crops grown for the previous 2 yr increased durum grain yield 15% and GCPC 18% compared with continuous wheat systems. Fall residual soil NO3–N and available soil water accounted for 3 to 28% of the increased durum yield in two of five site-years, whereas those two factors accounted for 12 to 24% of the increased GCPC in three of five site-years. Durum grain yield was negatively related to GCPC. The relationship was stronger when durum was preceded by oilseeds compared with pulses. Broadleaf crops in no-till cropping systems provide significant rotational benefits to durum wheat in the semiarid northern Great Plains.
C
ropping systems in the semiarid northern Great Plains traditionally have been dominated by dryland production of cereal grain crops, namely spring wheat and durum wheat. In recent years, low prices for cereal grains and rising input costs, coupled with changes in grain transportation policies and government support programs, have provided strong incentives for producers to seek alternative production opportunities. Furthermore, increasing emphasis on soil and environmental quality and economic innovation is stimulating changes in cropping systems. Consequently, producers in this region have been extending and diversifying their crop rotations by including broadleaf crops such as chickpea, dry pea, lentil, canola, and mustard. The area planted to chickpea in Saskatchewan increased from 4000 ha in 1995 to 280 000 ha in 2000 (Anonymous, 2001). Over 90% of this production area is concentrated in the semiarid Brown (Aridic Haploborolls) and Dark Brown
Y.T. Gan, B.G. McConkey, R.P. Zentner, and C.L. McDonald, Semiarid Prairie Agric. Res. Cent., Agric. and Agri-Food Can., Box 1030, Swift Current, SK, S9H 3X2, Canada; P.R. Miller, Dep. of Land Resour. and Environ. Sci., Montana State Univ., Bozeman, MT 597173120; and F.C. Stevenson, Ag-Stat. Serv., Rogers Rd., Saskatoon, SK, S7N 3T6, Canada. Received 02 Oct. 2001. *Corresponding author (
[email protected]).
Abbreviations: GCPC, grain crude protein concentration; PASW, plant available soil water.
Published in Agron. J. 95:245–252 (2003).
245
246
AGRONOMY JOURNAL, VOL. 95, MARCH–APRIL 2003
MATERIALS AND METHODS Site and Year Field experiments were conducted from 1996 to 2000 at two sites in southwestern Saskatchewan. The first site was on an Orthic Brown Chernozem (Aridic Haploborolls) with loam to silt loam texture and a saturated-paste pH of 6.5 in the 0- to 15-cm depth (Ayers et al., 1985). This site was at the Agriculture and Agri-Food Canada Semiarid Prairie Agricultural Research Centre near Swift Current (50⬚12⬘ N, 107⬚24⬘ W). The second site was on Rego Brown Chernozem (Vertic Cryoborolls) with heavy clay texture and a saturated-paste pH of 6.8 in the 0- to 15-cm depth (Ayers et al., 1985) in a farmer’s field near Stewart Valley (50⬚12⬘ N, 107⬚24⬘ W).
Experimental Design and Plot Management Three pulse crops (chickpea, lentil, and dry pea), one oilseed crop (oriental mustard), and one cereal crop (hard red spring wheat) were planted on tilled, fallow soil in the first year. In the following year, spring wheat, an oilseed (mustard or canola), and a pulse (lentil or dry pea) crop were notill–seeded on soil in each of the five previous crop stubbles. In the third year, ‘Kyle’ durum wheat was no-till–planted on soil in standing stubbles of all 15 combinations of previous crop types. At each site, the 3-yr cropping sequences were duplicated for three cycles, staggered 1 yr apart. The first cycle of the crop sequences initiated in 1996 and completed in 1998, the second cycle began in 1997 and completed in 1999, and the third cycle began in 1998 and completed in 2000. Before the initiation of each cycle, soil samples were collected from the experimental sites by taking two 30-mm-diam. cores to a depth of 120 cm, the rooting depth for cereal and most field crops grown in this semiarid region (Angadi et al., 2001). Each core of the soil was divided into 0- to 15-, 15- to 30-, 30- to 60-, 60- to 90-, and 90- to 120-cm increments that were bulked by soil depth. The soil samples were analyzed for NO3–N and P using the methods described by Hamm et al. (1970). Soil water was gravimetrically determined. Values for soil bulk density were obtained from a previous study at Swift Current (Campbell et al., 1983) and were measured by grid sampling at the Stewart Valley site. These bulk densities were used to express water content on a volumetric basis (Table 1). Plant available soil water (PASW) was determined using the method described by Ratliff et al. (1983) and Cutforth et al. (1991). With that method, PASW is expressed as the field-measured values subtracted by a lower limit of water contents obtained for a specific soil. In our study, the lower limit was 130 mm at Swift Current and 348 mm at Stewart Valley. Ratliff et al. (1983) demonstrated that field-measured lower limits of soil water were preferable to laboratory-measured lower limits. Cutforth et al. (1991) found that fieldmeasured lower limits did not vary markedly at a given envi-
ronment and that the available soil water measured using the lower limits could be used to better compare different crop species. In the first year of the crop sequence, the five crops were grown in a randomized complete block design with three replications. Plot size was 16 by 4.5 m. All crops were grown using the recommended agronomic practices in regard to seeding date and depth, plant density, pest control, and fertilizer application (Anonymous, 2000). Mustard and spring wheat were fertilized using ammonium nitrate to supply 70 kg ha⫺1 total available N (i.e., residual soil N in a 120-cm depth plus fertilizer N), according to preplanting soil tests. All crops received 4.5 to 7.5 kg P ha⫺1 as monoammonium phosphate placed with the seed. The pulse crops received 5 to 8 kg ha⫺1 of an appropriate Rhizobium spp. inoculant (a peat granular form). Crops were individually harvested after they reached maturity. Uncut crop stubble was left standing. Crop residues cut by the combine were chopped and spread evenly in the field with a combineattached chopper. In mid-September, glyphosate [isopropylamine salt of N-(phosphonomethyl)glycine] was sprayed for weed control on all plots at a rate of 200 g a.e. ha⫺1. In Year 2, the recropped oilseed (canola or mustard) and spring wheat were fertilized to supply 60 to 70 kg ha⫺1 of total available N, based on previous fall soil test results (in a 120-cm depth). Nitrogen credits from pulse stubble were taken into account in the fertilizer calculations using equations provided by Saskatchewan Soil Testing Laboratory (B. Green, personal communication, 1995):
N credit (kg ha⫺1) ⫽ 0.005 ⫻ grain yield [dry pea] N credit (kg ha⫺1) ⫽ 0.004 ⫻ grain yield [lentil, chickpea] As a result, canola, mustard, and spring wheat grown on pea stubble received an average of 20 kg N ha⫺1 less fertilizer than when grown on spring wheat stubble and 10 to 15 kg N ha⫺1 less than when grown on lentil and chickpea stubble. All plots received 7.5 kg P ha⫺1 as monoammonium phosphate placed with the seed. Crops were separately harvested after they reached maturity. Crop stubble was handled similar to the first year as was postharvest weed control. In October, soil samples were collected from each plot (to a 120-cm depth) using the same methods previously described. Residual soil NO3–N and PASW were measured from those soil samples (Table 2) using the method of Hamm et al. (1970) for residual NO3–N and the methods of Ratliff et al. (1983) and Cutforth et al. (1991) for PASW. In the third year, durum wheat was planted on the 15 combinations of previous crop stubbles. Seed was treated with Vitaflo (Gustafson, Canada) at 2.6 g kg⫺1 seed and planted 5 cm deep at a rate of 250 viable seeds m⫺2. Row spacing was 20
Table 1. Soil nutrients and available water measured before the initiation of the study at Swift Current and Stewart Valley, SK, Canada. Swift Current Variable Date of sampling Soil NO3–N†, kg ha⫺1 in 0–60 cm depth in 60–120 cm depth Soil P†, kg ha⫺1 in 0–60 cm depth in 60–120 cm depth Available soil water‡, mm
Stewart Valley
1996
1997
1998
1996
1997
1998
8 May
28 May
22 Apr.
9 May
29 May
14 May
44.5 22.9
30.2 24.9
92.2 39.1
33.3 25.9
22.1 12
53.8 17.8
– – 236
21.7 13.6 247
31.1 31.1 127
– – 132
6.1 5.2 166
17.5 12.3 128
† Nitrate N and P were measured in samples collected to a 120-cm soil depth before seeding, using the methods described by Hamm et al. (1970). ‡ Residual available soil water (in the 120-cm soil depth) was determined using the methods described by Ratliff et al. (1983) and Cutforth et al. (1991).
247
GAN ET AL.: DURUM WHEAT IN DIVERSE CROPPING SYSTEMS
Table 2. Postharvest residual soil NO3–N and plant available soil water (PASW) in a 120-cm depth measured in October from the various crop sequence plots in which durum wheat was grown in the following spring at Swift Current and Stewart Valley, SK, Canada. Residual soil NO3–N† Swift Current Crop sequence Cereal–cereal Cereal–oilseed Cereal–pulse Mustard–cereal Mustard–oilseed Mustard–pulse Pulse–cereal Pulse–oilseed Pulse–pulse LSD0.05 P value
1997 18.6 15.9 24.9 18.7 23.6 20.6 23.5 26.0 40.6 3.6 0.01
1998 17.5 17.4 31.4 22.2 29.6 30.0 36.5 28.4 46.3 10.4 ⬍0.01
PASW‡
Stewart Valley 1999 kg 11.0 12.3 17.6 9.4 10.8 16.0 12.5 15.3 21.5 12.5 0.76
1997
Swift Current
Stewart Valley
1998
1999
1997
1998
1999
20.0 20.7 28.8 27.5 30.2 50.0 48.7 25.6 46.2 17.5 ⬍0.01
8.4 9.8 10.5 7.1 12.4 18.1 10.1 19.1 24.2 9.3 ⬍0.01
32.4 46.5 51.5 35.1 42.1 68.2 42.2 46.6 54.4 37.9 0.27
38.9 25.5 35.9 24.9 19.6 28.1 40.4 27.3 45.8 27.8 0.25
22.1 24.4 41.1 5.3 0.0 32.1 5.6 22.6 14.0 25.3 0.01
ha⫺1 20.0 33.0 34.4 15.7 48.5 49.2 24.8 40.8 77.5 18.6 ⬍0.01
1997
1998
1999
94.4 96.8 115.4 104.9 95.7 124.4 113.3 90.6 124.0 52.3 ⬍0.01
5.1 54.7 48.5 20.7 74.8 37.8 4.0 46.7 41.5 51.2 ⬍0.01
mm 56.6 80.7 128.8 43.8 64.1 105.4 36.7 39.0 110.9 44.5 ⬍0.01
† Soil N was measured using the methods described by Hamm et al. (1970). ‡ Plant available soil water was determined using the methods described by Ratliff et al. (1983) and Cutforth et al. (1991).
cm. Fertilizer N as ammonium nitrate was placed between the rows at the rate of 45 to 62 kg N ha⫺1 (based on the fall soil test in the cereal–cereal sequence plots), along with 7.5 kg P ha⫺1 and 6 to 7 kg S ha⫺1. The same rates of N, P, and S were applied to all 15 combinations of previous crop stubble to determine residual soil N contributions from the different crop sequences. Weeds were controlled with a 200 g a.e. ha⫺1 application of glyphosate before seeding. When needed, weeds in the standing crop were controlled with appropriately labeled herbicides. Plant height at maturity was measured in each plot, and the center eight rows (19.2 m2) were harvested with a plot combine. The grain samples were air-dried, cleaned, and weighed. Test weight and kernel weight were determined and the grain yield reported on a dry weight basis. Grain N concentration was measured using the standard micro-Kjeldahl method and was multiplied by 5.7 to convert to GCPC. Crude protein yield was calculated by multiplying the grain yield by GCPC.
Statistical Analysis We analyzed the data using the PROC MIXED procedure of SAS (Littel et al., 1996), with blocks as a random effect and site-years (where applicable) and treatments as fixed effects. The analysis was conducted separately for each site-year when significant site-year ⫻ treatment interactions occurred. Data from the Stewart Valley site in 2000 were excluded from analysis due to seeding errors. We conducted a separate analysis for the Year-1 crop type effect, Year-2 crop type effect, and Year-1 ⫻ Year-2 crop sequence effect. Three pulse crops (pea, lentil, and chickpea) grown in Year 1 had similar effects on the subsequent durum wheat (Table 3). Therefore, effects of the three pulses were averaged when comparing with the effects of oilseed and cereal crop stubble. Treatment effects were considered significant at P ⱕ 0.05. We used a covariance analysis, similar to that used by Stevenson and van Kessel (1996), to explain the differences in grain yield and GCPC that were due to variations of PASW and residual soil NO3–N. We determined correlations between cumulative residual soil NO3–N by depth and durum wheat grain yield and GCPC and found that total residual soil NO3–N from the 120-cm depth provided the best correlations to the yield-related variables (results not presented). Therefore, we only presented the results of covariance analysis in which the residual soil NO3–N from the 120-cm depth was used. An estimate of the average deviation of all crop sequences relative to the cereal–cereal–durum crop sequence was derived from this analysis. This estimate of the crop sequence effect was
compared with that derived with the covariables out of the analysis of variance model to calculate the percentage of crop sequence effect explained by the two soil-related variables. The percentage estimates were determined using the ratio of covariable out and covariable in, with the percentage being none when the ratio was ⬎1. The final component of the statistical analysis confirmed the effect of crop sequences on the relationship between durum wheat grain yield and GCPC. In this analysis, we excluded grain yield data ⬎3200 kg ha⫺1 because data greater than this level clearly were not in the continuous portion of the association. The quadratic relationship between durum wheat grain yield and GCPC was not significant (data not presented).
RESULTS AND DISCUSSION The 3-yr crop sequence study was duplicated for three cycles, with the first from 1996 to 1998, the second from 1997 to 1999, and the third from 1998 to 2000. In each of the cycles, durum wheat was grown as a Year-3 crop (i.e., in 1998 for the first cycle, in 1999 for the second cycle, and in 2000 for the third cycle). Growing season precipitation in the durum-grown years was generally average to above average (Table 4). In 1998, total soil water (to a 120-cm depth) at spring seeding time was 20% lower than long-term averages though the growing season precipitation was near the 40-yr average. In contrast, in 1999 and 2000, the growing season rainfall was 50 mm (26%) more than the 40-yr average due to greater-than-average precipitation in May and July, with total soil water at spring seeding time being close to or Table 3. The ANOVA values of contrast, showing the effect of crop types grown 2 yr before durum wheat on the grain yield and protein of durum wheat at Swift Current (SC) and Stewart Valley (SV), SK, Canada. Contrast
1998 SC 1999 SC 2000 SC 1998 SV 1999 SV Mean P value
Grain yield Among pulses Pulse vs. oilseeds Pulse vs. cereal Protein concentration Among pulses Pulse vs. oilseeds Pulse vs. cereal
0.09 0.25 0.26
0.26 0.35 ⬍0.01
0.42 0.17 0.01
0.05 0.94 0.88
0.09 0.94 0.51 0.2 ⬍0.01 ⬍0.01
0.01 0.86 0.96
0.90 0.01 ⬍0.01
0.83 0.06 0.1
0.16 0.14 0.02
0.85 0.1 0.55 0.02 0.05 ⬍0.01
248
AGRONOMY JOURNAL, VOL. 95, MARCH–APRIL 2003
Table 4. Growing season precipitation and temperatures at Swift Current and Stewart Valley, SK, Canada, and total soil water of different soil depths measured at spring seeding time at Swift Current. Swift Current 1998 May June July August Total
38 91 37 35 201
May June July August Mean
12.6 14.0 20.1 20.9 16.9
Soil depth, cm 0–30 30–60 60–120
1998 56(⫺19)§ 94(⫺21) 162(⫺23)
1999 94 86 60 17 257 9.9 14.1 16.4 18.9 14.8 1999 75(9) 121(2) 208(0)
Stewart Valley† 2000
40 yr†
Precipitation, mm 65 45 54 71 127 48 13 40 259 204 Air temperatures, ⴗC 10.9 10.9 13.8 15.6 19.1 18.2 18.4 18 15.6 15.6 Soil water, mm 2000 34 yr‡ 78(13) 69 ⫾ 5 125(5) 119 ⫾ 24 206(⫺1) 209 ⫾ 34
1998
1999
2000
30 83 37 54 204
77 78 38 18 211
44 51 108 15 218
16.9 15.4 21.2 22.3 18.9
10.7 15.1 17.2 18.8 15.5
11.1 13.8 19.6 18.6 15.8
Not measured
† Means of 1961–2000 for the Swift Current site. Long-term weather data are not available for the Stewart Valley site. ‡ Means of 1967–2000. (Soil water at spring seeding time was not measured for the Stewart Valley site.) § Data in parentheses are percentage changes from the 34-yr mean.
slightly above the long-term averages. The patterns and amounts of growing season precipitation were similar between the two sites though the long-term weather data at Stewart Valley were not available on the farmer’s field.
Grain Yield and Yield Components Grain yields of durum wheat ranged from 1430 kg ha⫺1 at Swift Current in 1998 to 4700 kg ha⫺1 at Stewart Valley in 1999 (Table 5). Limited preseeding soil water and the low rainfall in the early period of the growing season in 1998 (Table 4) reduced durum vegetative growth at Swift Current. Crop height at anthesis was 98 cm in 1998, 15% lower than canopy measured in 1999 and 2000. Higher-than-normal temperatures in the
late part of the 1998 growing season caused durum wheat to have significantly lower (29–35%) kernel weight and lower (7–9%) test weight than those measured in the other study years. A similar soil water deficit at Stewart Valley in 1998 was the likely cause of the reduced yield at that site. Crops grown 2 yr before durum wheat (i.e., Year-1 crop effect shown in Table 5) influenced grain yields of durum wheat in three of five site-years, with the yield of durum wheat grown on mustard and pulse stubble averaging 6 to 8% higher than when grown on spring wheat stubble (P ⬍ 0.01). A similar response was observed for durum wheat grown on the previous year’s stubble (i.e., Year-2 crop effect shown in Table 5), with durum wheat grown on oilseed and pulse stubble yield-
Table 5. Mean grain yield of durum wheat responses to different crop sequences at Swift Current (SC) and Stewart Valley (SV), SK, Canada. Crop type
1998 SC
1998 SV
1999 SC kg
Year-1 crop sequence effects Cereal Oilseed Pulse LSD0.05 P value Year-2 crop sequence effects Cereal Oilseed Pulse LSD0.05 P value Year-1 ⫻ Year-2 crop sequence effects Cereal–cereal Cereal–oilseed Cereal–pulse Oilseed–cereal Oilseed–oilseed Oilseed–pulse Pulse–cereal Pulse–oilseed Pulse–pulse LSD0.05 P value
1999 SV
2000 SC
Mean
ha⫺1
1430 1430 1520 NS† 0.54
1810 1790 1800 NS 0.98
2660 2830 2900 181 0.01
3910 4420 4370 248 ⬍0.01
2280 2390 2500 190 0.04
2420 2570 2620 43 ⬍0.01
1330 1580 1530 183 0.02
1850 1730 1830 NS 0.28
2940 2650 2920 143 ⬍0.01
4040 4350 4480 222 ⬍0.01
2320 2640 2350 142 ⬍0.01
2500 2590 2620 47 ⬍0.01
1240 1480 1560 1300 1530 1440 1370 1630 1540 NS 0.06
1830 1930 1680 2000 1590 1790 1810 1710 1890 NS 0.17
2590 2610 2760 2900 2750 2820 3070 2630 3010 221 ⬍0.01
3530 3980 4220 4240 4330 4700 4150 4470 4490 330 ⬍0.01
2000 2550 2280 2380 2460 2330 2410 2720 2380 238 ⬍0.01
2240 2510 2500 2560 2540 2620 2560 2630 2660 157 ⬍0.01
† NS ⫽ no significant differences among treatments within the group.
249
GAN ET AL.: DURUM WHEAT IN DIVERSE CROPPING SYSTEMS
Table 6. Mean yield components and crude protein yields of durum wheat responses to different crop sequences; data are means across five site-years in Saskatchewan. Crop type
Spike density spikes
Year-1 crop sequence effects Cereal Oilseed Pulse LSD0.05 P value Year-2 crop sequence effects Cereal Oilseed Pulse LSD0.05 P value Year-1 ⫻ Year-2 crop sequence effects Cereal–cereal Cereal–oilseed Cereal–pulse Oilseed–cereal Oilseed–oilseed Oilseed–pulse Pulse–cereal Pulse–oilseed Pulse–pulse LSD0.05 P value
m ⫺2
Kernel weight mg
kernel⫺1
Test weight kg
hL⫺1
Protein yield kg ha⫺1
315 323 329 NS† 0.13
35.4 35.6 35.5 NS 0.8
80.2 80.2 79.7 NS 0.1
289 318 333 14.2 ⬍0.01
322 320 333 NS 0.1
35.1 35.2 36.2 0.5 ⬍0.01
79.8 79.9 80.1 NS 0.49
293 326 345 11.7 ⬍0.01
307 305 334 318 314 338 328 328 331 NS 0.09
34.5 35.6 35.9 35 35.5 36.3 35.2 35 36.2 1.1 0.01
79.9 80.2 80.5 80.2 80.1 80.5 79.6 79.7 79.9 NS 0.19
251 298 317 293 321 341 306 337 356 20.3 ⬍0.01
† NS ⫽ no significant differences among treatments within the group.
ing 4 to 5% higher than on spring wheat stubble (P ⬍ 0.01). The Year-1 ⫻ Year-2 crop sequence effects were significant (P ⬍ 0.01) in four of six site-years (Table 5). On average, the durum wheat grain yield was 13% lower when the crop was preceded by 2 yr of continuous spring wheat compared with broadleaf crops (Table 5). No significant yield differences were observed among the various pulse–oilseed, oilseed–pulse, or pulse–cereal alternated crop sequences. Durum wheat protein yields responded to crop sequences similarly to the way grain yield responded (Table 6). Examination of yield components revealed that crops grown 2 yr before durum wheat did not influence any yield components of durum wheat (Table 6). Crops grown the year immediately before durum wheat influenced durum wheat kernel weight (P ⬍ 0.01), with durum wheat grown on pulse stubble having the highest kernel weight. The 2-yr crop combinations (i.e., Year-1 ⫻ Year-2 crop types) did not influence yield components in general, but durum wheat following 2 yr of continuous spring wheat had the lowest kernel weight. Studies conducted in other regions of the world produced similar rotational benefits of pulses in cerealbased dryland cropping systems (Welty et al., 1988; Silsbury, 1990; Stevenson and van Kessel, 1996). Beckie and Brandt (1997) and Beckie et al. (1997) found that enhanced residual soil NO3–N was one of the primary contributors to increased grain yields of cereals following a pulse crop. Gan et al. (2000) reported that PASW measured at spring seeding time was 10% greater in dry pea and lentil stubble than in wheat stubble. Residual soil water in a 60-cm depth did not differ among crops, whereas large differences in residual soil water in the 60- to 120-cm depth existed among crop species. These authors believed that conserved soil water, primarily below 60-cm soil depth, contributed to the increased grain yields of cereals following a shallower-
rooted crop such as lentil or dry pea in semiarid dryland regions. In the present study, we measured residual soil NO3–N and PASW in the previous fall (Table 2). Covariance analyses revealed that soil residual NO3–N and PASW combinations accounted for up to 28% of the durum wheat yield variation in two of five site-years and none for the rest of the site-years (Table 7). The remainder of the yield variation could not be explained with the soil-related measurements. The poor relationship between durum wheat grain yield and the soilrelated variables in this study was probably due to soil sampling that was conducted in the previous fall. Changes in PASW from fall through winter to spring is expected along with a weak response of spring crops to available soil water measured in previous fall. Additionally, the majority of soil water, particularly in Vertic soils, moves through macropores and is not uniformly distributed throughout the soil profile (Heuvelman and McInnes, 1997; Lin et al., 1998). Inadequate soil sampling could generate inaccurate estimates of soil water conserved in the profile. We did not assess diseases in this study, but Fernandez et al. (1998) observed in an adjacent field trial at Swift Current that the severity of leaf-spotting diseases was higher in wheat after wheat than in wheat after lentil.
Grain Crude Protein Concentration Growing conditions strongly influenced durum wheat GCPC, with the GCPC being highest in 1998 (Table 8) when the grain yield was the lowest (Table 5). The crop grown 2 yr before durum wheat affected durum wheat GCPC in two of five site-years (Table 8). The GCPC of durum wheat grown on spring wheat stubble (2 yr prior) averaged 6% lower than when grown on pulse stubble. The three pulses (pea, lentil, and chickpea) had similar rotational effects on durum wheat GCPC
250
AGRONOMY JOURNAL, VOL. 95, MARCH–APRIL 2003
Table 7. The effect of postharvest residual available soil water and residual soil N (120-cm depth) on grain yield and grain crude protein concentration (GCPC) of durum wheat grown in the following spring at Swift Current (SC) and Stewart Valley (SV) in Saskatchewan. 1998 SC Grain yield Crop sequence effect† Covariables out, kg ha⫺1 Covariables in, kg ha⫺1 Percentage explained, %‡ Effect of covariable Soil water effect, kg ha⫺1§ Soil NO3–N effect, kg ha⫺1¶ GCPC Crop sequence effect† Covariables out, g kg⫺1 Covariables in, g kg⫺1 Percentage explained, %‡ Effect of covariable Soil water effect, g kg⫺1§ Soil NO3–N effect, g kg⫺1¶
1998 SV
247 178 28
1999 SC
227 228 None
5.69 ⫺3.90
437 426 3 ⫺3.15 ⫺0.26
3.24 1.98
7.8 12 None
24.7 21.7 12
⫺0.36 0.29
⫺0.03 0.22
9.2 8.8 None ⫺0.01 0.11
1999 SV
2000 SC
⫺31 ⫺59 None
799 801 None
0.11 1.24
⫺0.24 0.1
31.6 27.8 12
9.5 7.2 24
⫺0.10 0.28
⫺0.03 0.18
† The average difference for grain yield or GCPC between spring wheat–spring wheat–durum wheat cropping sequence and all other cropping sequences. ‡ The percentage of the crop sequence effect explained by the covariables. (The percentage was determined using the ratio of covariable out and covariable in. None is used when the ratio ⬎ 1.) § Estimated changes in grain yield (kg ha⫺1) or protein concentration (g kg⫺1) for each unit increase of plant available soil water measured in the 120-cm depth in the previous fall. ¶ Estimated changes in grain yield (kg ha⫺1) or protein concentration (g kg⫺1) for each unit increase of residual soil N measured in the 120-cm depth in the previous fall.
(Table 3), and with a few exceptions, they did not differ from that of mustard or canola. Crops grown the year immediately before durum wheat affected durum wheat GCPC at all sites except Stewart Valley in 1999 (Table 8). Averaged over the five site-years, the GCPC of durum wheat grown on mustard or canola and pulse stubbles from the previous year was 10 and 15% higher, respectively, than when grown on spring wheat stubble (P ⬍ 0.01). Year-1 ⫻ Year-2 crop combinations significantly affected durum wheat GCPC in four of five siteyears (Table 8), and the GCPC values at Swift Current in 1998 were not statistically different among treatments. The durum wheat GCPC averaged 16 to 19% higher when the crop was grown in 2 yr of continuous pulses or pulse–oilseed alternated crop sequences than when grown in 2 yr of continuous spring wheat.
Other researchers observed similar rotational effects of pulse crops on cereal GCPC (Evans et al., 1991; Beckie et al., 1997; Miller et al., 2001). For example, Evans et al. (1991) reported that wheat grown after pea in the semiarid southeastern Australia produced an average of 12% more grain N than did wheat after wheat. Miller et al. (2001) reported 8% GCPC increase in wheat after pulse and 5% GCPC increase in wheat after an oilseed compared with wheat after wheat. These studies, however, only determined the effect of crop types grown immediately before a cereal. The increased protein concentration in durum wheat following pulse–pulse, oilseed–pulse, or pulse–oilseed cropping sequences was partially due to increases in the symbiotically fixed N contained in the pulse crop residues and the gradual release of mineralizable N as
Table 8. Mean grain crude protein concentration of durum wheat responses to different crop sequences at Swift Current (SC) and Stewart Valley (SV) in Saskatchewan. Crop type
1998 SC
1998 SV
1999 SC
1999 SV
2000 SC
Mean
g kg⫺1 Year-1 crop sequence effect Cereal Oilseed Pulse LSD0.05 P value Year-2 crop sequence effects Cereal Oilseed Pulse LSD0.05 P value Year-1 ⫻ Year-2 crop sequence effects Cereal–cereal Cereal–oilseed Cereal–pulse Oilseed–cereal Oilseed–oilseed Oilseed–pulse Pulse–cereal Pulse–oilseed Pulse–pulse LSD0.05 P value
186 187 186 NS† 0.99
178 181 187 NS 0.55
124 136 147 10.6 ⬍0.01
117 127 125 5.3 ⬍0.01
130 130 137 NS 0.22
147 152 156 5.8 ⬍0.01
176 186 198 12 ⬍0.01
158 203 192 8.8 ⬍0.01
129 145 147 9.5 ⬍0.01
124 121 127 NS 0.08
126 127 150 4.8 ⬍0.01
142 156 163 4.4 ⬍0.01
180 185 194 175 184 203 175 187 197 NS 0.18
154 186 194 143 206 195 164 208 190 14.7 ⬍0.01
114 130 128 123 148 137 134 149 157 12.4 ⬍0.01
115 111 125 129 125 127 125 122 127 8.7 ⬍0.01
124 121 146 118 122 148 129 130 151 8.3 ⬍0.01
137 147 158 138 157 162 145 159 165 8.5 ⬍0.01
† NS ⫽ no significant differences among treatments within the group.
GAN ET AL.: DURUM WHEAT IN DIVERSE CROPPING SYSTEMS
crop residues decomposed during the growing season. Campbell et al. (1992), in a long-term wheat–lentil rotation study at Swift Current, observed that there was a cumulative enhancement of the N-supplying power of the soil after lentil due to the pulse residual contribution. The lentil–wheat rotation resulted in a gradual reduction in fertilizer N requirements of the mixed cropping system compared with a wheat-based monoculture. Campbell et al. (1993) also found that there was less deepleached NO3–N associated with the wheat–lentil rotation due to better synchrony of N uptake from the lentil residue decomposition compared with well-fertilized continuous wheat. Heenan et al. (1994) demonstrated that addition of N fertilizer increases cereal protein yields in a continuous cereal rotation, but the protein yield could not be elevated to the same levels as those obtained in pulse–cereal rotations. In the present study, covariance analyses revealed that postharvest residual soil N plus PASW accounted for 12 to 24% of the GCPC variation in three of five site-years and none in the two remaining site-years. The poor correlation between durum wheat GCPC and the soil-related variables was probably due to our soil sampling, which was conducted in the previous fall. We did not measure potential changes of soil NO3–N levels during the winter months and the following spring and summer. Durum wheat grown on the no-tilled soil that had mineralizable, high-N crop stubble might benefit from potentially mineralized N over a longer growing period. In addition to residual soil N and N releases from crop residue decomposition, other factors might be attributable to the sizable increase in durum protein. Biederbeck et al. (2000), in a long-term cereal–lentil rotation study at Swift Current, observed that microbial activity in the rhizosphere and rhizoplane of wheat grown after lentil increased significantly compared with those in monoculture wheat. However, it is unknown whether the increased microbial activity causes an increased GCPC for durum wheat. Further research involving more detailed measurements of microbial activity in durum wheat rhizosphere and rhizoplane is needed to elucidate this effect.
Yield–Protein Relationship There was a negative relationship between grain yield and GCPC in durum wheat (Fig. 1). As grain yield increased from 1500 to 3200 kg ha⫺1, the GCPC decreased from 190 to 130 g kg⫺1, equivalent to protein content on a dry matter basis from 18 to 10%. In 1998, durum wheat produced half as much as the grain yields produced in 1999 and 2000 (Table 5) due to lowerthan-normal moisture in the earlier growing season and greater-than-normal temperatures in the latter part of the growing season (Table 4). In the same year, the durum wheat GCPC was the greatest (Table 8). Cropping sequences strongly influenced the relationship between durum wheat grain yield and GCPC (Fig. 1). At the yield level of 1700 kg ha⫺1, durum grown after a pulse or an oilseed crop produced 15% higher GCPC than durum following spring wheat. As yields increased beyond 1700 kg ha⫺1, there was a tendency for durum
251
Fig. 1. Relationship between grain yield and crude protein concentration of durum wheat grown in different crop sequences in Saskatchewan from 1998 to 2000.
wheat GCPC to decline more sharply when the crop was preceded by an oilseed crop rather than a pulse crop. Coefficients of the regression equations were statistically significant between preceding pulses and oilseed (P ⬍ 0.05). Preceding spring wheat had the lowest intercept value, whereas its slope did not differ from that of the pulses, indicating that at any given yield level, the GCPC of durum wheat grown after a cereal will be 15% lower than when grown after a pulse. In cases where the overall grain yields exceeded 3200 kg ha⫺1, crop sequences had little effect on the association between GCPC and grain yields. This implies that the effects of previous crops on durum wheat GCPC diminish under environmental conditions conductive to higher grain yields. In summary, crops grown immediately before durum wheat influenced the grain yield and GCPC of durum wheat more than crops grown 2 yr before the durum wheat. Continuous cereal systems reduced durum wheat grain yields by 4 to 8% and GCPC by 8 to 16% compared with cropping systems that included an oilseed or a pulse crop 1 or 2 yr before durum wheat. The increased yield of durum wheat preceded by an oilseed or a pulse crop was related to residual soil NO3–N and PASW, but these two factors only accounted for up to 28% of the observed yield variation (Table 7). In years when growing season precipitation was above long-term averages (e.g., 1999 and 2000), the crop sequence effects on durum wheat were more evident (Table 5) than those observed in a dry year (e.g., 1998). Durum wheat GCPC increased with greater residual soil NO3–N and PASW in three of five site-years (Table 7); in these cases, 12 to 24% of the GCPC variation was explained by these two factor combinations. In the present study, potential contribution of residual soil NO3–N to the increased durum wheat grain yield and GCPC may have been underestimated. Sampling soils in the previous fall did not consider potentially mineralizable N from crop residue decomposition during the winter months and the following spring and summer. Nevertheless, significant crop sequence effects existed (Tables 5 and 8), even for studies wherein the residual soil N and PASW were not
252
AGRONOMY JOURNAL, VOL. 95, MARCH–APRIL 2003
attributable to the increased grain yield and GCPC in durum wheat (Table 7). These observations lead us to speculate that, besides residual soil N and PASW measured in the previous fall, other factors such as microbial activity and potential N releases during the postharvest period and the following spring and summer play important roles in boosting grain yields and quality of subsequent cereal crops. Further studies are needed to elucidate these great rotational benefits in the semiarid northern Great Plains. ACKNOWLEDGMENTS We gratefully acknowledge the excellent technical assistance provided by Greg Ford, Lee Poppy, Ray Leshures, Dean James, and Barry Blomert. The project was funded by Agricultural Development Fund of Saskatchewan Agriculture and Food, Saskatchewan Pulse Growers, and the Matching Investment Initiative of Agriculture and Agri-Food Canada. We gratefully thank Dr. Hugh Beckie and Dr. Newton Lupwayi for providing reviews of the manuscript.
REFERENCES Angadi, S., B. McConkey, D. Ulrich, H. Cutforth, P. Miller, M. Entz, S. Brandt, and K. Volkmar. 2001. Developing viable cropping options for the semiarid prairies. Final report. Semiarid Prairie Agric. Res. Cent., Agric. and Agri-Food Can., Swift Current, SK, Canada. Anonymous. 2000. Pulse production manual. Saskatchewan Pulse Growers, Saskatoon, SK, Canada. Anonymous. 2001. 2000 Saskatchewan crop district crop production. StatFacts 10.01, 2001.03.27. Saskatchewan Agric. and Food, Regina, SK, Canada. Ayers, K.W., D.F. Acton, and J.G. Ellis. 1985. The soils of the Swift Current map area 72J Saskatchewan. Sask. Inst. of Pedology Publ. 86. Ext. Div. Publ. 481. Univ. of Saskatchewan, Saskatoon, SK, Canada. Beckie, H.J., and S.A. Brandt. 1997. Nitrogen contribution of field pea in annual cropping systems: I. Nitrogen residual effect. Can. J. Plant Sci. 77:311–322. Beckie, H.J., S.A. Brandt, J.J. Schoenau, C.A. Campbell, J.L. Henry, and H.H. Janzen. 1997. Nitrogen contribution of field pea in annual cropping systems: II. Total nitrogen benefit. Can. J. Plant Sci. 77:323–331. Biederbeck, V.O., H.H. Janzen, C.A. Campbell, and R.P. Zentner. 1994. Labile soil organic matter as influenced by cropping practices in an arid environment. Soil Biol. Biochem. 26:1647–1656. Biederbeck, V.O., N.Z. Lupwayi, K.G. Hanson, W.A. Rice, and R.P. Zentner. 2000. Effect of long-term rotation with lentils on rhizosphere ecology and on endophytic rhizobia in wheat. p. 80. In Abstracts, North Am. Conf. on Symbiotic Nitrogen Fixation, 17th, Laval Univ., Sainte-Foy, QC, Canada. 23–28 July 2000. Laval Univ., Sainte-Foy, QC, Canada. Campbell, C.A., D.W.L. Read, V.O. Biederbeck, and G.E. Winkleman. 1983. The first 12 years of a long-term crop rotation study in southwestern Saskatchewan—nitrate-N distribution in soil and N uptake by the plant. Can. J. Soil Sci. 63:563–578. Campbell, C.A., R.P. Zentner, F. Selles, and O.O. Akinremi. 1993. Nitrate leaching as influenced by fertilization in the brown soil zone. Can. J. Plant Sci. 73:387–397. Campbell, C.A., R.P. Zentner, F. Selles, V.O. Biederbeck, and A.J.
Leyshon. 1992. Comparative effects of grain lentil–wheat and monoculture wheat on crop production, N economy and N fertility in a brown chernozem. Can. J. Plant Sci. 72:1091–1107. Clarke, J.M., B.A. Marchylo, M.I.P. Kovacs, J.S. Noll, T.N. McCaig, and N.K. Howes. 1998. Breeding durum wheat for pasta quality in Canada. Euphytica 100:163–170. Cutforth, H.W., P.G. Jefferson, and C.A. Campbell. 1991. Lower limit of available water for three plant species grown on a mediumtextured soil in southwestern Saskatchewan. Can. J. Plant Sci. 71:247–252. Evans, J., N.A. Fettell, D.R. Coventry, G.E. O’Connor, D.N. Walsgott, J. Mahoney, and E.L. Armstrong. 1991. Wheat response after temperate crop legumes in southeastern Australia. Aust. J. Agric. Res. 42:31–43. Fernandez, M.R., R.P. Zentner, B.G. McConkey, and C.A. Campbell. 1998. Effects of crop rotations and fertilizer management on leaf spotting diseases of spring wheat in southwestern Saskatchewan. Can. J. Plant Sci. 78:489–496. Gan, Y.T., B.G. McConkey, and R.P. Zentner. 2000. Soil water conservation for pea, lentil, and chickpea in the Brown soil zone. p. 9–11. In Proc. Pulse Crop Res. Workshop, 3rd, Winnipeg, MB, Canada. 19–21 Nov. 2000. Gan, Y.T., and G. Noble. 2000. Chickpea in southwestern Saskatchewan–potential and risks. p. 8–9. In Proc. Pulse Crop Res. Workshop, 3rd, Winnipeg, MB, Canada. 19–21 Nov. 2000. Hamm, J.W., F.G. Radford, and E.H. Halstead. 1970. The simultaneous determination of nitrogen, phosphorus and potassium in sodium bicarbonate extracts of soils. p. 65–69 In Advances in automatic analysis. Vol. 2. Industrial Analysis. Proc. Technicon Int. Congr. Futura Publ. Co., Mount Kisco, NY. Heenan, D.P., A.C. Taylor, B.R. Cullis, and W.J. Lill. 1994. Long term effects of rotation, tillage and stubble management on wheat production in southern N.S.W. Aust. J. Agric. Res. 45:93–117. Heuvelman, W.J., and K.J. McInnes. 1997. Spatial variability of water fluxes in soil: A field study. Soil Sci. Soc. Am. J. 61:1037–1041. Lin, H.S., K.J. McInnes, L.P. Wilding, and C.T. Hallmark. 1998. Macroporosity and initial moisture effects on infiltration rates in vertisols and vertic intergrades. Soil Sci. 163:2–8. Littel, R.C., G.A. Milliken, W.W. Stroup, and R.D. Wolfinger. 1996. SAS system for mixed models. SAS Inst., Cary, NC. Miller, P.R., B.G. McConkey, G.W. Clayton, S.A. Brandt, J.A. Staricka, A.M. Johnston, G. Lafond, B.G. Schatz, D.D. Baltensperger, and K. Neill. 2001. Pulse crop adaptation in the northern Great Plains. Agron. J. 94:261–272. Ratliff, L.F., J.T. Ritchie, and D.K. Cassel. 1983. Field-measured limits of soil water availability as related to laboratory-measured properties. Soil Sci. Soc. Am. J. 47:770–775. Silsbury, J.H. 1990. Grain yield of wheat in rotation with pea, vetch or medic grown with three systems of management. Aust. J. Exp. Agric. 30:645–649. Stevenson, F.C., and C. van Kessel. 1996. The nitrogen and nonnitrogen rotation benefits of pea to succeeding crops. Can. J. Plant Sci. 76:735–745. Welty, L.E., L.S. Prestbye, R.E. Engel, R.A. Larson, R.H. Lockerman, R.S. Speilman, J.R. Sims, L.I. Hart, G.D. Kushnak, and A.L. Dubbs. 1988. Nitrogen contribution of annual legumes to subsequent barley production. Appl. Agric. Res. 3:98–104. Wright, A.T. 1990a. Quality effects of pulses on subsequent cereal crops in the northern prairies. Can. J. Plant Sci. 70:1013–1021. Wright, A.T. 1990b. Yield effects of pulses on subsequent cereal crops in the northern prairies. Can. J. Plant Sci. 70:1023–1032. Zentner, R.P., C.A. Campbell, V.O. Biederbeck, P.R. Miller, F. Selles, and M.R. Fernandez. 2001. In search of a sustainable cropping system for the semiarid Canadian prairies. J. Sustainable Agric. 18:117–136.