Increasing crop rotational diversity can enhance cereal yields

The study addresses whether and how increasing crop rotational diversity (CRD)—the number and functional variety of crops rotated over time—affects cereal grain yields under varying nitrogen (N) fertilization and over long time scales. Modern commodity cropping often relies on short rotations or monocultures, contributing to environmental degradation and requiring high external inputs. Prior knowledge suggests that diverse rotations can enhance soil fertility, microbial biomass, water holding capacity, and nutrient availability, reducing pests, weeds, and diseases, yet the extent and temporal dynamics of yield responses across cereals and fertilization regimes remain unclear. The authors hypothesize that diversifying rotations increases rainfed grain yields within a few years of implementation, with benefits growing over decades due to accumulating ecosystem functions; effects would be stronger with more crop functional groups, and CRD benefits would be greater under lower external N inputs.

Long-term grassland experiments show positive diversity–productivity relationships driven by species selection and niche complementarity, but their applicability to arable systems is uncertain due to differences in plant life histories, management intensity, and spatial versus temporal diversification. Existing agricultural studies often compare only monocultures against a single diverse rotation, are site-specific, short-term, or focus on a single indicator crop, limiting generalization. A recent meta-analysis reported higher yields from CRD under low N, especially when legumes were included, but lacked resolution on gradations of diversity and temporal development. Diverse rotations can maintain soil fertility and nutrient use efficiency, improving soil carbon, microbial biomass, and plant N uptake, and reduce pest, disease, and weed pressures. However, arable CRD experiments often confound species and functional traits chosen for agronomic value. There is a need to disentangle species diversity from functional richness and to assess which functional groups (e.g., legumes, broadleaves, leys) drive yield benefits across crops and over time under contrasting N inputs.

Data: 27,460 annual rainfed grain yield observations from 32 long-term experiments (≥10 years; up to 63 years) across Europe (winter- and spring-sown small grain cereals) and North America (maize), totaling 957 site-years. Experiments included at least two rotation treatments with different crop rotational diversity (CRD) levels and similar or comparable management (tillage, fertilization, pest control) within site. Yield data were generally annual; two sites had biennial sampling. Multiple CRD contrasts per site were used when available. CRD metrics: (1) Species diversity (SD), calculated as the inverse Simpson’s index in time (SD = 1/Σp_i^2), where p_i is the proportion of years a species appears in the rotation; SD = 1 for monoculture; higher values reflect greater richness and evenness (e.g., a 4-year maize–maize–soy–winter wheat rotation yields SD = 2.7). (2) Functional richness (FR), the count (1–4) of functional groups present in the rotation: annual cereals, annual legumes, annual broadleaves, and ley (biennial/perennial grass and/or legume, including ≥2 years alfalfa). Nitrogen fertilization: External N input rates were classified as low (below local recommended N rate for the indicator crop) or high (at or above recommendations), based on site-specific sources. Across sites: 12 high-N, 6 low-N, and 14 with both low and high N inputs; inorganic fertilization in 15 sites, organic in 9, and both in 8. Preprocessing and modeling: For each indicator crop within each site, yields were mean-centered by subtracting the long-term within-site average across all CRD and N treatments. Indicator crops were analyzed separately (spring small grain cereals, winter small grain cereals, maize). Statistical analysis: Gaussian mixed-effects models (lme4 in R 4.1.0) related mean-centered yield to CRD (either continuous SD with polynomial terms and interactions, or categorical FR), time since experiment start (years; with polynomial terms where supported), and fertilization level (low/high), including interactions (CRD × time; CRD × fertilization) as supported by model selection (AIC). Final models differed slightly by crop (Eqs. 1–5 in paper). Random effects included site-level grouping by management group (dummy variable for management combinations within site) nested within site, and calendar year as a categorical random effect to account for interannual variation and technological change. Residual diagnostics (DHARMa) indicated acceptable fit. Effect estimates and predictions were generated with emmeans and ggeffects. Functional group presence analysis: An additional model (Eq. 6) assessed the effect of including specific functional groups (binary indicators for annual legume, annual broadleaf, ley) on mean-centered yields, without time interactions. Mapping and visualization: Geographic context provided via rnaturalearth; plots via ggplot2. Data/code availability: Mean-centered yields, metadata, rotation information, and R code are archived at the Swedish National Data Service (https://doi.org/10.5878/8af1-0q60).

• Overall effect of CRD: Grain yields increased with higher crop species diversity for maize and both small grain cereal groups, with responses modulated by N input and time. Benefits generally increased over decades.
• Magnitude of long-term benefits (vs monoculture at year 0, low N): At year 35, maximum yield gains occurred at intermediate-to-high species diversity: spring small grain cereals +0.94 t/ha (95% CI 0.74–1.13) at SD ≈ 3.9; winter small grain cereals +1.32 t/ha (0.101–1.62) at SD ≈ 3.91; maize +4.19 t/ha (3.60–4.78) at SD ≈ 4.57. Winter small grain cereals exhibited a decline in gains beyond SD ≈ 3.9; spring small grains showed a slight decline only at the highest CRD mostly after 35 years; maize showed only a slight, transient decline near the maximum SD tested and only at low N (disappearing by year 35).
• Early vs long-term: After 5 years, yield benefits at low N already emerged: spring +0.36 t/ha (0.16–0.55) at SD ≈ 4.03; winter +0.62 t/ha (0.33–0.91) at SD ≈ 3.54; maize +2.26 t/ha (1.81–2.71) at SD ≈ 3.72. From years 5 to 35, additional gains accrued: spring +0.58 t/ha (0.38–0.77), winter +0.68 t/ha (0.37–0.98), maize +1.70 t/ha (1.18–2.22). Over the same period, monocultures increased little (spring +0.08 t/ha; winter +0.08 t/ha; maize +0.58 t/ha).
• Nitrogen interaction: CRD effects were stronger at low N, especially for maize. At year 35, the slope (t/ha per SD unit) for maize was +1.00 (0.74–1.26) at low N vs +0.61 (0.40–0.82) at high N. Winter small grains: +0.48 (0.40–0.56) low N vs +0.41 (0.34–0.48) high N. Spring small grains: +0.29 (0.24–0.34) low N vs +0.19 (0.14–0.23) high N.
• Functional richness (FR): Greatest benefits generally occurred with 2–3 functional groups. For small grains, three functional groups gave increasing benefits over time; adding a fourth group at low N reduced the incremental benefit for spring cereals but still outperformed monoculture at year 35. For maize, two functional groups led to the highest long-term benefit.
• Specific functional groups: Benefits were driven by inclusion of key groups rather than diversity per se. Winter small grains benefited from annual broadleaf and legume crops; spring small grains benefited from legumes and annual broadleaves (ley had no impact); maize benefited from annual legumes and ley (annual broadleaves did not benefit maize).
• CRD vs N fertilization (Fig. 3): In early years, increasing N to monocultures produced larger yield gains than diversification at low N for small grains. Over time, diversified low-N rotations approached or exceeded high-N monocultures. Combining diversification with high N produced the greatest yields. In maize, differences among strategies were small by year 5; diversified systems surpassed high-N monocultures by 20–35 years depending on CRD level.
• Model performance (from figures): Marginal/conditional R² (species diversity models) approximately: spring small grains 0.154/0.719; maize 0.128/0.771; winter small grains 0.295/0.775, indicating substantial site/year random-effect variance but consistent fixed-effect signals of CRD and fertilization.

Findings support the hypothesis that increasing crop rotational diversity enhances cereal yields and that benefits build over decades, especially under low external N inputs. Stronger responses at low N, particularly in maize, suggest nutrient-mediated mechanisms such as improved nitrogen supply from soil organic matter, increased nitrogen use efficiency, and biological N fixation via legumes. Inclusion of specific functional groups (legumes, broadleaves, leys) explains much of the benefit, consistent with species selection effects and niche complementarity. Additional mechanisms likely include reduced pest, disease, and weed pressures, enhanced soil microbial activity and soil structure, greater soil carbon and water retention, and complementary rooting architectures improving resource capture. For small grain cereals, a hump-shaped yield response to species diversity indicates diminishing or negative returns at very high species counts, possibly due to functional redundancy or shared pest/pathogen pressures among similar crops. In practice, moderate species diversity with high functional richness may maximize yield benefits. Over time, diversified rotations at low N can approach yields of high-N monocultures, offering a pathway to reduce fertilizer dependence and environmental impacts. However, combining CRD with adequate N generally maximizes yields, which can reduce land needed for a target production level. Pre-crop effects (e.g., maize often following legumes versus cereals preceding small grains) likely contributed to differential responses but could not be formally tested due to unbalanced designs.

Across 32 long-term experiments and nearly three decades on average, increasing crop rotational diversity, measured as species diversity and functional richness, consistently enhanced maize and small grain cereal yields, with benefits strengthening over time and under lower nitrogen inputs. Functional composition matters: including legumes, and for maize also ley phases, was particularly beneficial. Intermediate levels of species diversity often maximized small grain yields, while maize showed large gains even at higher diversity levels tested. These results indicate that strategically increasing functional richness in rotations can sustain or enhance yields while reducing dependence on external N inputs, offering environmental co-benefits such as lower greenhouse gas emissions and reduced nitrogen pollution. Future research should disentangle species versus functional effects with fully crossed designs, clarify temporal sequences and pre-crop effects, quantify mechanistic pathways (soil biology, nutrient cycling, pest dynamics) over decadal scales, and tailor management (e.g., crop protection) to rotation treatments to capture full CRD benefits. Policy and extension efforts are needed to incentivize and support adoption given the multi-year timeframe for realizing maximal benefits.

• Model explanatory power was limited (low marginal R²) due to large geographic, management, and environmental variability; results demonstrate general trends rather than precise predictions.
• Rotation designs were not fully crossed for species diversity and functional richness, constraining inference on their independent effects and optimal combinations.
• Unbalanced datasets prevented formal testing of pre-crop (preceding crop) effects, which likely influence yield responses.
• In roughly half of experiments, crop protection and some management practices were standardized across CRD treatments, likely underestimating potential CRD benefits from reduced pest/weed pressures.
• Benefits at very high diversity declined for small grains; generalizability depends on local crop options, markets, and management contexts.
• Yield benefits accrue over decades; short-term evaluations may understate CRD advantages.