Power-law productivity of highly biodiverse agroecosystems supports land recovery and climate resilience

The study addresses how to transform food production systems to simultaneously improve productivity, biodiversity, and climate resilience, especially for smallholder-dominated agriculture. Global assessments warn of accelerating biodiversity loss and ecosystem deterioration, with agriculture contributing substantially to greenhouse gas emissions and biogeochemical disruption. However, coarse global scenarios and economic models often miss ground-truth complexities and the social-ecological contexts of small farms. Practical research gaps identified include: (P1) doubling incomes and productivity of small-scale producers; (P2) making production more environmentally sound and resilient to climate shocks; and (P3) meeting smallholder needs with local-context data. To bridge agronomy and ecology, the paper investigates synecological farming (synecoculture)—high-diversity, low-input mixed polycultures that rely on self-organization of plant communities and aim at ecological rather than physiological optima. Research questions are: (Q1) how community dynamics of biodiversity and productivity self-organize and respond to climatic variability, independent of social confounders; (Q2) whether productivity exhibits specific dynamical properties beyond aggregate means (e.g., links between fluctuations and resilience); and (Q3) whether common principles emerge across species compositions, soils, and climates.

Empirical ecology has shown positive diversity–productivity relationships and power-law spatial patterns in natural vegetation, but these insights are underutilized in crop production. Global scenario studies (e.g., on land use, afforestation) provide important orientations but often neglect key ecological and socio-technical constraints (e.g., nutrient cycles, albedo trade-offs, feasibility of converting natural grasslands) and lack applicability to smallholder contexts. Big-data-driven cohort approaches also miss the temporal resolution and community-level complexity pertinent to highly diverse agroecosystems. Meta-analyses highlight practical gaps for smallholders (P1–P3). The paper situates synecoculture within community ecology and open complex systems science, proposing an integrated model of physiological and ecological optima (IMPEO) to conceptualize overyielding and symbiotic gains in mixed communities, especially under marginal environmental conditions.

Design: Out of 60+ implementations, three representative small-scale sites were analyzed: Field A (Oiso, Japan; temperate; 420 m²; non-harvested vegetation monitoring), Field B (Ise, Japan; temperate; 1,000 m²; four years of production records), and Field C (Mahadaga, Burkina Faso; tropical semi-arid; 500 m²; three years of production records). All fields followed synecoculture protocols: no tillage, no synthetic or organic fertilizers, no agrochemicals or ground covers, minimal machinery, and only seeds/seedlings and necessary water introduced. Field setups: Field A (2010–2011): randomly mixed communities of 52 edible species plus spontaneous vegetation; no harvest or watering; surface cover measured. Field B (2010–2014): strategic mixed association of 133 edible species; harvesting and occasional watering; products sold (including a delivery-box scheme). Field C (2015–2018): strategic mixed association of 150 edible species; harvesting and watering; sales in local market; five alternative farming methods run in parallel for on-site comparison. Data collection and analyses: (1) Vegetation surface patterns (Field A): Low-ground cover of each species measured using a two-step visual analog scale (extended Braun–Blanquet) on 80 sections of 2 m², observed 22 times across 2011. Species categorized as introduced crops vs spontaneous, and edible vs non-edible. Inverse cumulative distributions of species-wise surface coverage fitted against an integrated Box–Cox model spanning exponential (competition-dominated) to power-law (symbiosis-influenced) forms; goodness-of-fit estimated via bcPower() and nls() in R. (2) Productivity distributions (Fields B and C): Species-wise daily sales recorded; probability densities estimated via density() in R and fitted on log–log scale to Pareto distributions (thresholds: >1,000 JPY for B, >1,000 XOF for C; density >1e−7). To compare synecoculture’s frequent, asynchronous harvests with conventional annual reports, a regularized productivity metric r_i = m s_i (daily species-wise sale multiplied by yearly number of harvest events m) was used so that mean regularized productivity equals annual gross sales. Costs plotted as negative offsets to compare annual profits and costs. (3) Benchmarks and comparisons: Field B compared to official Japanese open-field market gardening statistics (2010–2014), scaled to 1,000 m². Field C compared to five simultaneously implemented alternative methods (system of rice intensification with trees, conservation agriculture, permaculture, bio-intensive market gardening, traditional market gardening; all scaled to 500 m²) and to FAO conventional estimates for Burkina Faso. Profitability defined as gross sales minus costs. (4) Biomass estimation: Defined a community-based relative biomass ratio (BR) to avoid biases of land equivalent ratio (LER). BR := (Σ X_i) / (Σ Y_j), using sales-weighted biomass per unit price; for diverse sets V/W ≈ price-rate R, enabling BR estimation from sales and R. Field B used R ≈ 1.5; Field C used R ≈ 2.0 (first two years) and ≈1.0 (third year). Non-harvest ecosystem biomass (e.g., trees, spontaneous non-edibles, residues, roots) excluded from BR. (5) Climate–biodiversity correlations: For Field A, daily 1-km gridded agro-meteorological data (temperature, humidity, radiation, precipitation) for the 30 days prior to each observation were obtained from NARO. For Field B, similar 30-day windows were used. For Field C, daily 19.2-km CFS Reanalysis data (19 variables including air temperature, specific humidity, soil moisture at multiple depths, radiation fluxes, latent/sensible heat fluxes, precipitation, potential evaporation) were aggregated over 14-day intervals. Linear regressions tested correlations between plant/product diversity and means vs variances of meteorological parameters. (6) IMPEO simulation: Conceptual simulations illustrated overyielding dynamics for centrally vs marginally competent species under environmental gradients, showing competitive losses and symbiotic gains, and transitions of total overyielding with departures from physiological optima. LER components depicted on transformed scale LER' = log(log(LER)+1).

• Power-law community structure: In Field A, the inverse cumulative distribution of species-wise surface coverage more closely followed a power-law than an exponential distribution, indicating dominant symbiotic interactions beyond simple resource competition. Species-wise surface probability densities within 2 m² sections also followed power laws. Spontaneous species exhibited stronger symbiotic signatures and more positive diversity responses to climate variability than introduced crops.
• Power-law productivity: In Fields B (Ise, Japan) and C (Mahadaga, Burkina Faso), species-wise daily sales exhibited long-tailed, power-law distributions, diverging from normal distributions typical of monoculture-based annual reports. This pattern held despite differing climates and crop portfolios.
• Ecosystem function gains under no-input practice: Across sites, on-site observations indicated multiple ecosystem improvements with succession: higher yields, complex food webs enabling pest regulation, porous soil structure, increased humus and soil organic matter, improved water retention and permeability, and activated soil microbiota.
• Productivity and profitability increases: • Field B (1,000 m²; 4 years; 78 products): Average profitability rose 2.35- to 3.87-fold over conventional databases (all scales and <0.5 ha), corresponding to an estimated 0.981- to 1.16-fold increase in harvest biomass (via BR). • Field C (500 m²; 3 years; 37 products): Compared with median conventional market gardening, profitability rose 88.0-fold (202/54.4), corresponding to an estimated 33.8-fold biomass increase across two 18-month periods spanning a change in market accessibility (pre/post Nov 2016). Under high market access: 121-fold profitability, 37.8-fold biomass; under low market access: 55.0-fold profitability, 29.9-fold biomass. On-site comparison vs five alternative sustainable methods showed a 258-fold profitability increase and an estimated 12.4-fold harvest biomass increase.
• Regime shift reversal in semi-arid environment: Satellite imagery and field observations indicated pre-existing spotted vegetation patterns suggesting imminent desertification around Field C. Introducing 150 edible species (including 40 staples) re-established a lush, year-round productive ecosystem, with positive spillover on neighboring plots and development of mature successional layers (annuals, perennials, shrubs, vines, light-demanding and shade-tolerant trees).
• Climate resilience via adaptive diversification: Across Fields A–C, plant/product diversity was significantly positively correlated with the variance (fluctuation) of meteorological parameters over relevant growth windows (14–30 days), often with no or negative correlations with parameter means. Seasonality was weaker in fluctuation than in mean components. Results support adaptive diversification of crop portfolios in response to climatic variability, aligning with biodiversity maintenance mechanisms.
• Management-relevant statistics: Fitted Pareto exponents were in ranges with finite analytical means (α>1), yet annual sales deviated widely (e.g., annual gross sales 56–141% of average in Field C; 27–214% in Field B). Arithmetic means were unstable indicators, whereas cumulative cost–benefit ratios converged to superior performance, consistent with power-law productivity and the stability of harmonic means.
• General principle and recommendations: Findings support three recommendations: (R1) high-biodiversity, low-input mixed associations generate symbiotic gains reflected in power-law biomass distributions; (R2) characterize productivity via diversity levels, medians/quartiles, and cost–benefit ratios rather than arithmetic means, interpreting temporal fluctuations as adaptive diversification; (R3) positive biodiversity responses to climatic variability are expected across differing crop portfolios, climates, and soils, meriting broader verification. Proposed solutions include improving market access and literacy (S1), expanding access to plant genetic resources and benefit-sharing (S2), and education/certification and data-driven marketing to match diverse supplies with demand (S3).

The study demonstrates that when crop production is organized as highly diverse, low-input mixed communities, ecological self-organization yields power-law structures in both spatial vegetation patterns and productivity distributions. This challenges monoculture-based optimization, where Gaussian assumptions and arithmetic means guide management. For Q1, the results show that community dynamics—biodiversity and productivity—self-organize in response to climatic variability, with diversity increasing alongside environmental fluctuations irrespective of mean conditions. For Q2, productivity fluctuations are not noise around a mean but embody adaptive diversification that enhances resilience; arithmetic means are poor management indicators under heavy-tailed distributions, whereas medians, quartiles, and cumulative cost–benefit metrics are more robust, and harmonic-mean stability is theoretically compatible with power-law behavior. For Q3, similar power-law productivity and positive biodiversity–variance correlations across temperate and semi-arid sites with different species and soils indicate a transferable principle: ecological-optimum production can deliver symbiotic gains, especially in marginal environments. Practically, synecoculture reduced inputs while increasing ecosystem functions, reversed early-warning patterns of desertification in the Sahel site, and outperformed conventional and alternative methods in profitability and estimated biomass. Scaling such systems for smallholders could shift agricultural externalities from negative to positive for biodiversity, contribute to land recovery, and support climate adaptation by expanding viable crop portfolios. Policy and implementation should prioritize the R1–R3 framework and S1–S3 actions to align markets, genetic resources, and training with ecological-optimum production.

This work establishes an agroecological rationale for highly biodiverse, low-input mixed polycultures (synecoculture) that leverage ecological optima and self-organization. Across temperate Japan and semi-arid Burkina Faso, primary production and sales follow power laws, biodiversity increases with climatic variability, and systems exhibit enhanced ecosystem functions with minimal inputs. Productivity and profitability gains—particularly the dramatic improvements and regime-shift reversal in the Sahel—indicate that ecological optimization can overcome the traditional trade-off between productivity and biodiversity. The study recommends shifting management metrics from arithmetic means toward medians/quartiles and cost–benefit ratios, and advancing supportive measures: market access and literacy, expanded plant genetic resources with fair benefit-sharing, and education/certification with data-driven marketing. Future research should include broader cohort studies and multi-site validations to test the generality of biodiversity–variance responses and refine management under differing socio-ecological contexts, while integrating supportive information technologies and distribution networks to scale synecoculture for smallholders globally.

• Scope and generalizability: The study focuses on three in-depth case sites selected from >60 implementations, using an in-depth profiling strategy rather than large cohort designs; broader validation across regions, soils, and market conditions is needed.
• Proxy measures and market effects: Productivity and biomass estimation rely on sales data and price-rate adjustments (R) to derive relative biomass ratios (BR), and market accessibility changed during the Field C study period, which may influence profitability comparisons.
• Data resolution and design: Meteorological datasets differ in spatial resolution (1 km for Field A/B vs ~19.2 km for Field C) and variables; experiments were not randomized controlled trials.
• Biomass accounting: BR estimates exclude persistent ecosystem biomass (e.g., trees, spontaneous non-edibles, residues, roots), potentially underestimating total biomass and ecosystem carbon accrual.
• Comparative benchmarks: Conventional and alternative method datasets come from official statistics or contemporaneous trials with differing aggregation (daily vs monthly vs annual), which may introduce comparability constraints despite regularization methods.