Skip to content
Field Notes

The research side of permaculture.studio.

Agriculture & Soil6 min read

What a 62-Paper Meta-Analysis Actually Shows About Nitrogen and Phosphorus Use Efficiency

A single meta-analysis of 62 papers reports nitrogen use efficiency up 20.6% and phosphorus use efficiency up 18.9%, with high heterogeneity and a corrected phosphorus estimate of 16.4%.

What the research shows

  • Heterogeneity was high (I2 = 74.3% for nitrogen, 71.8% for phosphorus), and trim-and-fill adjustment reduced the pooled phosphorus estimate from 18.9% to 16.4% after Egger's test indicated funnel-plot asymmetry (p = 0.041). Meta-analysis
  • The pooled figures combine agronomic efficiency, apparent recovery efficiency, partial factor productivity and partial nutrient balance, which the authors state are not numerically interchangeable; separate analyses by metric gave consistent directions. Meta-analysis
  • Microdosing, deep placement, starter-band phosphorus and integrated organic-inorganic management showed the largest effects, while conservation tillage showed the smallest. Meta-analysis
  • In a phosphorus-limited plantation study, leaf phosphorus rose from 0.253 in monocultures to 0.337 in two-species and 0.467 in three-species mixtures, but the authors caution that the three-species result rested on six plots. Model or simulationThis comes from a model, not from measurement.
  • A meta-analysis of 62 papers and 1,148 paired samples found nitrogen use efficiency rose 20.6% and phosphorus use efficiency 18.9% under improved management compared with conventional practice. Meta-analysispartially supported
  • A modelling study of Chinese livestock production estimated that integrated spatial planning could reduce harmful nitrogen exposure for 90% of the population, but required reallocating 14% of production and cutting livestock numbers by one-third against a 2050 baseline. Unclassifiedpartially supportedThe study design could not be determined automatically.

Each of these was checked against the abstracts of the papers it cites. See the receipts.

What a 62-Paper Meta-Analysis Actually Shows About Nitrogen and Phosphorus Use Efficiency

Only one of the sources I was given addresses the question directly: a systematic review and meta-analysis of management practices and nutrient use efficiency [2]. Compared with conventional practice, nitrogen use efficiency (NUE) rose by 20.6% (95% CI 17.4–23.9) and phosphorus use efficiency (PUE) by 18.9% (95% CI 15.2–22.7).

That is one study, and its own numbers say to hold the result loosely. Heterogeneity is high, a publication-bias correction lowered the phosphorus figure to 16.4%, and the pooled estimates mix four metrics that are not interchangeable [2]. The authors screened 4,908 papers, kept 62, and extracted 1,148 paired samples for a random-effects meta-analysis using log response ratios [2].

Start with the heterogeneity. I2 was 74.3% for nitrogen and 71.8% for phosphorus [2]. Roughly three-quarters of the variation between study results is not explained by the practice being tested. When heterogeneity is that high, a pooled average describes a scattered cloud, not a reliable prediction for any particular field.

Then the publication bias check. For phosphorus, Egger's regression test flagged funnel-plot asymmetry (p = 0.041), and a trim-and-fill analysis pulled the pooled PUE estimate down from 18.9% to 16.4% [2]. The nitrogen estimate did not trip the same test, but the phosphorus number is the one that moved when the analysis corrected for missing small studies. A 2.5-point drop is not fatal. It is a reminder that the published literature on this topic is not a random sample of everything that was tried.

There is a third problem, and the authors name it themselves. The pooled NUE and PUE figures combine four different metrics: agronomic efficiency, apparent recovery efficiency, partial factor productivity, and partial nutrient balance [2]. These are not numerically interchangeable. A 20% gain in partial factor productivity is not the same thing as a 20% gain in apparent recovery. The authors ran separate analyses by metric and report that the direction of effect held across all of them [2]. That helps. It does not turn the headline number into a single quantity.

What the practices actually do

The practices with the largest effects were microdosing (28.4%), deep placement (26.1%), starter-band placement of phosphorus (24.8%), and integrated organic–inorganic nutrient management (24.3%). Conservation tillage had the smallest effect, at 9.7% [2].

The mechanisms differ, and it is worth separating them.

Microdosing applies a small quantity of fertiliser close to the seed or plant rather than broadcasting it across the field. The nutrient goes where roots can reach it while the total applied stays low, which reduces the fraction lost to volatilisation, leaching, or fixation before the plant gets to it.

Deep placement does something similar in flooded systems. Placing fertiliser below the soil surface reduces the ammonia that escapes to the air and the phosphorus that leaves with surface water.

Starter-band phosphorus places a concentrated band of P near the root zone at planting. Phosphorus is immobile in soil, so a broadcast application can sit in the top centimetres unavailable to roots. A band intercepts the root as it grows.

Integrated organic–inorganic management combines mineral fertiliser with manure or compost. Organic matter improves soil structure and water holding, and it can slow nutrient release so that supply better matches crop demand.

Conservation tillage, by contrast, changes soil physical conditions and biology but does not by itself change where or when nutrients are placed. That is a plausible reason for its smaller effect on efficiency, though the review does not test that explanation directly.

What the numbers do and do not support

The review reports that efficiency gains were larger on sandy, acidic, and infertile soils, and that efficiency declined rapidly once nitrogen application rates exceeded 120 kg per hectare [2]. It also reports that soil texture, soil organic carbon content, rainfall, and application rate were more important moderators than the type of practice used [2].

That last finding deserves more attention than it usually gets. If where you farm and how much you apply matter more than which technique you pick, then a list of best practices ranked by average effect size is the wrong output. The authors say as much: the estimates are context-dependent and should not be read as a ranking of universally best practices [2].

The review pays particular attention to smallholder farming systems in Zimbabwe, including granitic sandveld soils and the fertility gradients between homefields and outfields [2]. That is a specific context: sandy, acidic, nutrient-poor, with strongly differentiated fields around a homestead. The direction of the findings may well travel. The magnitudes should not be assumed to.

Adjacent evidence, and why it does not transfer

Two other sources touch on nutrient questions, and both are worth naming precisely because they are easy to over-read.

A modelling study of livestock production in China found that single-purpose spatial planning, targeting ammonia emissions alone or instream nitrogen concentrations alone, produces trade-offs that leave more than half the population exposed to at least one other nitrogen problem. An integrated approach, the authors estimate, could reduce harmful nitrogen exposure for 90% of the population while maintaining food security, but it requires reallocating 14% of livestock production, cutting livestock numbers by one-third relative to a 2050 baseline, and deploying advanced mitigation technologies [8]. This is a model, not an observed outcome. It describes what the authors' scenarios produce under their assumptions.

A study of tree species mixing in phosphorus-limited plantations on the Yellow River floodplain found leaf phosphorus rising from 0.253 in monocultures to 0.337 in two-species mixtures and 0.467 in three-species mixtures [9]. That is a large relative increase, and the authors flag the caveat themselves: the three-species result rests on six plots with limited species combinations and should be interpreted with caution [9]. It also measures leaf tissue phosphorus concentration, not phosphorus use efficiency, and it concerns trees in an alluvial plantation rather than crops in a rotation.

A review of nanobiochar reports crop yield improvements of 15–40% and improved nutrient retention [10]. It is a review, and its own conclusion is that scalability, long-term environmental impacts, and regulatory questions remain unresolved [10]. Treat the yield range as a summary of what has been reported, not as an expectation for a farm.

Where the evidence runs out

The direct evidence for the central question is one meta-analysis [2]. Its pooled estimates carry high heterogeneity, a publication-bias correction that moved the phosphorus figure, and a composite metric that mixes four non-interchangeable measures. The authors' own framing, context-dependent and not a universal ranking, is the correct one to carry forward.

Several things are simply not in the sources I was given. There is no long-term field trial data here on whether these efficiency gains persist over decades, or whether they hold soil carbon and biology steady while they do. There is no cost or labour analysis, though the review names cost, labour, and farmers' resource capacity as determinants of what is actually adoptable [2]. There is no evidence in these sources on how any of these practices perform under a permaculture design, in polycultures, perennial systems, or systems with no external inputs, because none of the studies examined that.

There is also a measurement problem worth flagging. A field study of ridge–furrow plastic mulching found that ridges and furrows within the same system had different moisture responses for all three greenhouse gases measured, with furrows showing a significant negative moisture response for nitrous oxide and ridges a positive but non-significant one [11]. The authors conclude that the two positions should not be treated as interchangeable observation units [11]. If adjacent strips within one field behave differently, then averaging nutrient measurements across a field, which is what most of the studies in the meta-analysis will have done, carries its own uncertainty. That uncertainty is not quantified in the review.

The honest summary: the practices with the strongest reported effects are placement and timing techniques that put nutrients where roots can use them. The average effect is around a fifth. The variation around that average is large, the context matters more than the technique, and the whole picture rests on one synthesis of a literature that is probably not complete.

References

  1. [1]Asymmetric Au(I)/Au(III) Catalysis with Chiral bis(P'̂'N) Ligands Deriving from Tröger’s BasesVivek W. Bhoyare, Nidal Saleh, Stéphane Grass, Charles-Louis Serpentini, Sonia Mallet-Ladeira, Gennaro Pescitelli · 2026 · Journal of the American Chemical Societydoi:10.1021/jacs.6c13732
  2. [2]Improving nitrogen and phosphorus use efficiency in agriculture: a critical review of management practicesCosmas Parwada, Philip T. Masere · 2026 · Frontiers in Soil Sciencedoi:10.3389/fsoil.2026.1978463
  3. [3]Assessing the adaptation between recreational fisheries development and aquaculture efficiency in multifunctional freshwater fisheries systems: Evidence from ChinaXianglan Hu, Jingzhou Liu · 2026 · Israeli Journal of Aquaculture - Bamidgehdoi:10.46989/001c.172992
  4. [4]Integrative multivariate analysis of Pisum sativum L. genotypes using agro-morphological traits under contrasting semi-arid highland agro-ecosystemsMohammed Mareai, Nagi Zaid, Tawfiq Al-Omri · 2026 · Scientific Reportsdoi:10.1038/s41598-026-79441-x
  5. [5]Molecular mapping of sheath blight resistance QTLs from the wild rice Oryza rufipogon accession CR100438Safoora Javed, Ankita Babbar, Kamal Thakur, Sheezana Rasool, Kulveer Singh Dhillon, Kishor Kumar · 2026 · Molecular Biology Reportsdoi:10.1007/s11033-026-13101-1
  6. [6]Alpine wetland degradation reshapes the balance between hydrolytic and reductive soil nitrogen cycling enzyme potentials on the Qinghai–Tibetan PlateauWenhua Chang, Weiwei Ma, Jianfang Wang, Liangcui Song, Jianan Du, Wanpeng He · 2026 · SOILdoi:10.5194/soil-12-1121-2026
  7. [7]Toward climate-resilient date palm agriculture in the United Arab Emirates: a strategic triangulated knowledge mappingSajid Safeer, Giuseppe Natale Basile, Sajid Maqsood, Cataldo Pulvento · 2026 · Agriculture & Food Securitydoi:10.1186/s40066-026-00652-6
  8. [8]Can the Spatial Planning of Livestock Production Tackle Future Nitrogen Pollution in China?Xi Chen, Zhaohai Bai, Maryna Strokal, Michelle T. H. van Vliet, Ling Liu, Xiangwen Fan · 2026 · Environmental Science & Technologydoi:10.1021/acs.est.6c11284
  9. [9]Tree species mixing shifts community-level leaf economics trait composition and functional diversity in phosphorus-limited plantationsMingjing Wei, Xiaorui Qin, Wenjie Zhu, Changxiao Lyu, Shouchao Yu, Chuanqiang Yang · 2026 · Frontiers in Forests and Global Changedoi:10.3389/ffgc.2026.2005754
  10. [10]Nanobiochar derived from organic waste: a sustainable solution for agriculture, environment, and circular bioeconomyA. Tanveer, S. Pradhan, H. Muhammad, G. McKay · 2026 · International Journal of Environmental Science and Technologydoi:10.1007/s13762-026-07487-5
  11. [11]Contrasting moisture responses of greenhouse-gas fluxes between ridge and furrow positions in a ridge–furrow plastic-mulching systemMinmin Zhang, Jianing Wang, Jiaying Li, Jianxiong Zhou, Xiangyu Xu, Shujun Zhao · 2026 · Agricultural Water Managementdoi:10.1016/j.agwat.2026.110927
  12. [12]A “cradle-to-grave” life-cycle assessment of biokerosene production through chemical looping gasification for use in jet engine: Comparing second and third generation biomassesDaofeng Mei, Alberto Navajas, Francisco García-Labiano, Alberto Abad · 2027 · Energy Conversion and Managementdoi:10.1016/j.enconman.2026.122375
  13. [13]Differential effects of neutral and alkaline salt stress on wheat photosystem II photochemistryGuangrong Cai, Yanchuan Gong, Xiaotong Zeng, Qingyao Wang, Liyuan Hu, Xin Wang · 2026 · Frontiers in Plant Sciencedoi:10.3389/fpls.2026.1966363
  14. [14]β-1,3-Glucan, Precision Biotic, and Bacillus subtilis as Antibiotic Growth Promoter Alternatives: Effects on Broiler Performance, Gut Morphology, and Serum Biochemistry2026 · Pakistan Veterinary Journaldoi:10.29261/pakvetj/2026.233

All 14 references resolved against Crossref on Nov 7, 2026. Full verification record.

Get the findings, not the noise

A short email when something worth knowing clears the verification gate. No takes. Unsubscribe in one click.