Providing evidence to improve practice

Action: Use crop rotation Soil Fertility

Key messages

Biodiversity: Four randomized, replicated trials from Canada, Portugal and Zambia measured the effect of including legumes in crop rotations and found the number of microbes and diversity of different soil animals increased.

Erosion: One randomized, replicated trial from Canada found that including forage crops in crop rotations reduced rainwater runoff and soil loss, and one replicated trial from Syria showed that including legumes in rotation increased water infiltration (movement of water into the soil).

Soil organic carbon: Four studies from Australia, Canada, and Denmark (including two controlled replicated trials and one replicated site comparison study), found increased soil organic carbon under crop rotation, particularly when some legumes were included.

Soil organic matter: Three of five replicated trials from Canada, Portugal and Syria (one also randomized, one also controlled and randomized), and one trial from the Philippines found increased soil organic matter, particularly when legumes were included in the rotation. One study found lower soil organic matter levels when longer crop rotations were used. One randomized, replicated study found no effect on soil particle size.

Soils covered: Clay, clay-loam, fine clay, loam, loam/silt loam, sandy clay, sandy loam, silty clay, silty loam.


Supporting evidence from individual studies


A replicated, randomized, controlled experiment in 1988 on a loam/clay loam soil in Saskatchewan, Canada (Campbell et al. 1992) found that crop rotations including grain crops and alfalfa Medicago sativa decreased soil organic matter in fallow and grain crop phases, but increased soil organic nitrogen in 4- (37.7 kg/ha/yr) and 6-year (43.9 kg/ha/yr) barley rotations in comparison to continuous wheat (31.1 kg/ha/yr). Microbial biomass was increased by including alfalfa in the rotation (by 26.6 mg C/kg and 10 mg N/kg), as was carbon release into the soil (by 6 mg C/kg). The experiment was part of a long term crop rotation study started in 1964, and over the course of the trial included wheat Triticum aestivum (replaced with canola Brassica campestris), barley Hordeum vulgare (replaced with oats Avena sativa) and alfalfa. There were 10 rotations of 2-5 crop types for four or six years, replicated four times in plots of 7.3 x 30.4 m. There were two, three, four, and six year rotations. Crop management followed recommended field practice. Soil samples were taken to 15 cm depth. Organic carbon and nitrogen, carbon release, and microbial biomass were measured.



A randomized, replicated, controlled experiment in 1991-1992 on a silty loam in Ontario, Canada (Rasiah & Kay 1995) found that including forage crops in crop rotations and minimizing tillage reduced rainwater runoff by 70% and 27%, and soil loss by 87% and 63%, respectively, compared to continuous cropping of maize Zea mays. Treatments included alfalfa Medicago sativa or bromegrass Bromus inermis followed by maize, and a continuously cropped, conventionally tilled maize control. Forage crops were grown for either two, four or six years prior to the reintroduction of maize. There were four replicates. A rainfall simulator was used to simulate rain events at 16 mm/h in 1 m2 subplots within each treatment. Runoff and soil lost from plots were collected manually. The results did not distinguish between forage crops.



A controlled, replicated experiment in 1995-1996 on clay soil in Saskatchewan, Canada (Campbell et al. 1999) found soil organic carbon was 30% higher in the fertilized continuous wheat Triticum aestivum after 39 years, compared to the beginning of the experiment. Soil organic carbon was also 41 % higher under the fallow-wheat-wheat-hay-hay-hay (brome Bromus inermis-alfalfa Medicago sativa) compared to fallow-wheat rotation after 39 years. However soil organic carbon did not differ between rotations over the two years sampled. A long term rotation study was established in 1957, and treatments included: continuous wheat Triticum aestivum unfertilized, continuous wheat fertilized, fallow-wheat unfertilized, fallow-wheat-wheat straw retained, fallow-wheat-wheat straw harvested, green manure (sweet clover Melilotus officinalis or black lentil Lens culinaris)-wheat-wheat unfertilized, and fallow-wheat-wheat-hay-hay-hay unfertilized. Fallow treatments were treated regularly with Roundup, Banvel and Rustler to prevent weeds. There were four replicates. Plot size was not specified. Soils were sampled in 1995-1996, depth not specified. Effects of rotation on microbial biomass were reported but not clear.



A randomized, replicated experiment, established in 1992 on loam/silt-loam soil at Fort Vermilion, Canada (Lupwayi et al. 1999) found higher soil microbial biomass in rotations with legume crops (red clover Trifolium pratense: 593.99 mg/kg soil, field pea Pisum sativum: 448.40 mg/kg soil) compared to fields left fallow (322.68 mg/kg soil) or cropped continuously with wheat Triticum aestivum (432.25 mg/kg soil). The trial treatments were zero tillage and conventional tillage (3-4 mechanical cultivations/year), combined with four different crop rotations: wheat-field peas, wheat-red clover, wheat-summer fallow, or continuous wheat. The trial included three replicate plots of each treatment combination, and 10 soil samples were taken from each plot during wheat cropping and mixed before analysis.



An experiment in 1997 on clay soils in New South Wales, Australia (Blair & Crocker 2000) found that crop rotation decreased soil organic carbon across all crop rotations (in two soil types) up to 71% compared to un-cropped controls. Including legumes like clover Trifolium subterraneum and lucerne Medicago sativa in rotations increased soil organic carbon levels by 41% and 32% respectively compared to wheat Triticum aestivum and long fallow controls, and 25% more than when grain legumes were included. Stability of soil aggregates was higher in continuous wheat than in rotations including lucerne, clover, and snail medic Medicago scutellata. A long-term rotation started in 1966 included six rotation treatments with three phases arranged in a 6 x 6 m plot (size/replication not specified). (1): lucerne followed by wheat. (2): lucerne or sorghum Sorghum bicolor on three plots, and a chickpea Cicer arietinum-wheat rotation, a wheat-long fallow rotation and continuous wheat on the remaining three. (3): cowpea Vigna unguiculata, clover or snail medic on three plots, and the same wheat rotations as for (2). Soil porosity was measured and soil samples were taken.



An experiment in 1994-1995 on silty clay in the Philippines (Witt et al. 2000) found 33-41% higher available carbon and nitrogen under the rotation compared to continuous rice Oryza sativa, resulting in 11-12% less carbon and 5-12% less nitrogen stored in the soil compared to under continuous rice. There were two crop systems: continuous rice Oryza sativa, and a maize Zea mays-rice rotation. Maize was grown in the dry season, and rice in the dry. Within each 12 x 25 m cropping system were four 12 x 8 m nitrogen treatments: control (no nitrogen fertilizer), low (30 kg N/ha), medium (40 kg N/ha) and high application (50 kg N/ha). Within these were two 6 x 8 m sub-treatments: early (63 days before rice seedling transplanting) or late crop residue (rice or maize) incorporation (14 days before transplanting). Soils were sampled to 15 cm depth.



A long-term replicated, controlled experiment between 1959 and 1994 on clay loam soil in Ontario, Canada (Gregorich et al. 2001) found that cultivating maize Zea mays in crop rotations increased soil carbon (20 Mg/ha more carbon than in maize monoculture) and increased maize yield by 30% in fertilized plots, and 360% in unfertilized plots, compared to maize monoculture.  In 1959, 12 plots (76.2 x 12.2 m) were established comprising three cropping treatments were maize in monoculture, bluegrass Poa pratensis in monoculture, and a maize-oat Avena sativa-alfalfa Medicago sativa-alfalfa rotation. The 12 plots comprised six replicate plots with fertilizer (16.8 kg N, 29.3 kg P and 27.4 kg K/ha) and six without. Three soil cores were taken from each plot to measure soil density and carbon.



A randomized, replicated experiment in 2003 on a sandy-loam soil in Quebec, Canada (Jiao et al. 2006) found similar sized soil aggregates in continuous maize Zea mays (1.89 mm) and in a soybean Glycine max-maize rotation (1.90 mm). There were four replicates of two tillage systems: conventional; and no-tillage. Within these were continuous maize, soybean -maize, and maize-soybean rotations (in 20 x 24 m plots). Within these were four fertilizer treatments: inorganic fertilizers, composted cattle manure, and the two mixed together (tested in 20 x 6 m plot sections). Soil samples (to 10 cm depth) were taken after crop harvest from the maize phase in October 2003. The size of soil aggregates was measured using a wet-sieving procedure. Soil carbon, nitrogen, and phosphorus were measured using finely ground soil samples.



A replicated experiment from 1983 to 1995 on fine clay soil in Syria (Masri & Ryan, 2006) found that soil organic matter increased in medic Medicago sativa and vetch Vicia sativa legume-cereal rotations (12.5-13.8 g/kg), compared to continuous wheat Triticum aestivum and wheat-fallow (10.9-11 g/kg). Higher levels of water filtered into the soil and hydraulic conductivity (see background section) was higher in legume rotations (16.2-21.8 cm/h and 8.7-12.4 cm/h) compared to continuous wheat and wheat-fallow (13.9-14.4 cm/h and 6.2-7.4 cm/h). Cropping sequences included: (1) durum wheat (var durum) grown in rotation with lentil Lens culinaris, chickpea Cicer arietinum, medic and vetch and watermelon Citrullus vulgaris; (2) continuous wheat; (3) wheat with a clean-tilled fallow. Each crop treatment was 36 x 150 m. There were seven replicates. Within these plots were secondary grazing treatments and tertiary nitrogen fertilizer treatments, but no results were presented. Soils were sampled annually prior to planting to 20 cm depth.



A replicated site comparison study in 2001-2003 on a sandy-loam in Denmark (Schjønning et al, 2007) found that soil organic carbon was higher under crop rotation (2.14 g/100g) relative to continuous cereal at Foulum (2.01 g/100 g). At Flakkebjerg, organic carbon was highest in the cereal plus manure treatment (1.06 g/100 g) compared to continuous cereal (0.91g/100 g). Soil porosity at Flakkebjerg was much higher under crop rotation (0.120 m3m-3) compared to continuous cereal (0.094 m3m-3), and cereal with manure (0.090 m3m-3). There were three 4-year crop rotations at two sites: cereal (oats Avena sativa, barley Hordeum vulgare, lupin Lupinus angustifolius, wheat Triticum aestivum) no manure; cereal plus manure; cereal-grass Lolium perenne-clover Trifolium repens and Trifolium pratense rotation without manure. Part of each plot was compacted by a medium-sized tractor. There were two replicates of 216 m2 plots at Foulum, and 169 m2 plots at Flakkebjerg. Soils were sampled to 13 cm depth at Foulum and 10 cm depth at Flakkebjerg in the wheat plots, from compacted and uncompacted plots.



A replicated experiment from 1989 to 1997 on a clay soil in northern Syria (Ryan et al. 2008) found that a wheat Triticum aestivum-fallow rotation had the lowest level of soil organic matter (235 t/ha) while a wheat-medic Medicago spp. (no species specified) rotation had the highest level (290 t/ha). The other rotations, listed according to the level of soil organic matter they maintained and starting with the lowest, were: wheat-melon Citrullus vulgaris (235 t/ha), continuous wheat (246 t/ha), wheat-lentil Lens culinaris (249 t/ha), wheat-chickpea Cicer arietinum (257 t/ha), and wheat-vetch Vicia sativa (266 t/ ha). Rotations of wheat with fallow, wheat with other crops and a continuous wheat control were replicated three times in plots of 36 x 120 m. The wheat rotation treatment included wheat with lentil, chickpea, vetch, pasture medic or watermelon. Within each rotation were four smaller 36 x 30 m sub-plots with 0, 30, 60 or 90 kg N/ha applied. Within these were 12 x 30 m grazing treatments: no grazing, medium and heavy grazing. Soil organic matter, phosphorus, nitrogen and nitrates were measured each cropping season.



A controlled, randomized, replicated study in 1999-2007 on sandy-clay soil in Zambia (Sileshi et al. 2008) found higher soil animal diversity and improved maize Zea mays yield in crop rotations including legumes (3.7 orders and 4 t/ha, respectively) compared to a continuously cropped maize control (2.9 orders and 2.7 t/ha). There were no differences in overall soil animal abundance and crop yield between single- and two-species legume treatments, but the abundance of earthworms and millipedes were higher in pure stands of pigeon pea Cajanus cajan (4.3 earthworms, 2.4 millipedes) and earthworms in sesbania Sesbania sesban (1 earthworm), compared to continuously-cropped maize (<1 earthworm, <0.5 millipedes per plot). Two fallow and cropping cycles and one control treatment (continuous maize with two applications of 200 kg/ha NPK fertilizer) were established on 10 × 10 m plots. The treatments in the two cropping cycles were: pure stands of sesbania, tephrosia Tephrosia vogelii, or pigeon pea; 1:1 mixtures of sesbania/pigeon pea, and sesbania/tephrosia. The fallow consisted of native legume and grass species. During the cropping phase, fertilizer was only applied to continuously cropped maize. There were three replicates.



A replicated experiment in 2001-2006 on loam soil in Saskatchewan, Canada (Malhi et al. 2009) found that nitrate-N was lower in arable rotations (82 kg N/ha) and grain and forage crop rotations (60 kg N/ha) than in in low diversity crop rotation (92.3 kg N/ha). Three treatments were replicated four times: an arable rotation comprising wheat Triticum aestivum-oilseed (mustard Brassica juncea or canola Brassica napus), a grain-forage rotation comprising barley Hordeum vulgare-perennial forage crop (sweet clover Melilotus officinalis, pea Pisum sativum, flax Linum usitatissimum or alfalfa Medicago sativa), and  a grain-perennial forage crop comprising mustard-wheat-barley-alfalfa-hay. Fallowing (and green manuring) followed each crop stage for one season, creating six cropping phases over each six-year rotation period. Each treatment plot measured 40 x 12.8 m. Each year, two soil samples were taken during each crop phase, and a third taken in 2006. Nitrate-N, carbon, nitrogen, phosphorus, and yield were measured.



A randomized, replicated experiment in 2004-2005 on sandy soil in Portugal (de Varennes & Torres, 2011) found greater enzyme activity after the lupin Lupinus spp. (2.2μmol nitrophenol/g/h) than oat Avena sativa crop (1.7 μmol nitrophenol/g/h). A higher number of fungal colonies (0.004 colonies/g soil), higher soil organic carbon levels (6.75 g C/kg) and extractable phosphorus (63.25 mg P/kg) were found following oat compared to lupin (0.003 colonies/g soil, 6.15 g C/kg, and 60.25 mg P/kg respectively). There were two treatments in 400 m2 plots: conventional tillage (two passes with a disc plough to 15cm), and no-till (left undisturbed). Each tillage treatment was divided into six 5 x 10 m plots. Three were sown with lupin and three with oat. Soils were sampled to 15 cm depth six times each year during the experiment.


Referenced papers

Please cite as:

Key, G., Whitfield, M., Dicks, L.V., Sutherland, W.J. & Bardgett, R.D. (2017) Enhancing Soil Fertility. Pages 383-404 in: W.J. Sutherland, L.V. Dicks, N. Ockendon & R.K. Smith (eds) What Works in Conservation 2017. Open Book Publishers, Cambridge, UK.