PROTOCOL: Agronomic biofortification strategies to increase grain zinc concentrations for improved nutritional quality of wheat,maize and rice: a systematic review

Zinc deficiency and related malnutrition

Zinc deficiency in soils and related malnutrition in humans pose serious threats both to cereal production and human health worldwide (Alloway, 2009; Cakmak et al., 2010; Zou et al., 2012; Velu et al., 2014). With two thirds of the world's agriculture land being marginal or severely zinc deficient and the current genetic erosion in wheat, maize and rice (FAO, 2013; Zou et al., 2012), there is a strong concern whether agronomic strategies for zinc biofortification would effectively enhance zinc concentrations for bioavailability to consumers post-processing. The major cereals are inherently low in zinc concentrations, thus fortified foods are inherently low in zinc concentrations. Therefore, a judicious use of zinc fertilisers (i.e., appropriate method, rate and timing) is likely to be the primary target not only for correcting soil deficiencies but also increasing zinc bioavailability globally (Alloway, 2009; FAO, 2013; Velu et al., 2014). A judicious use of zinc fertiliser has the potential for improving zinc status of crops up to fourfold from current baseline, and thus, producing crops with zinc concentrations that meet human requirement for dietary intakes. Improving zinc concentration in crops via agronomic strategies (i.e., soil and foliar fertilisation) lowers the risk of malnutrition and chronic diseases especially in children due to inadequate zinc intake and it limits the need for improving zinc nutrition via supplementation of commercial fortification (Alloway, 2009; Velu et al., 2014). However, there is a considerable contention among policy makers and analysts regarding the effectiveness of the soil and foliar application method for zinc biofortification and the crops for which it may or may not work (Cakmak, 2008).

Soil and foliar fertilisation

Soil and foliar application of fertilisers are two common agronomic strategies used to correct nutrient deficiencies in the soil and address plant demand for a specific nutrient. In the recent years, there has been an increasing interest and research about how these strategies could be used to increase zinc concentrations in the edible parts of crops, particularly wheat, maize and rice (Cakmak, 2008), in a way that best fits farmers’ interests and farm economics. This increasing interest results from the large investments that have been made for production of nutraceutical preparations and development of functional foods to address the occurrence of zinc deficiencies in the human diet. Generally, soil strategy involves application of granular Zn fertiliser, usually zinc sulphate (ZnSO₄ ^. 7H₂O), before sowing at rates based on soil test results of the soil properties prior to the intervention. Foliar strategy involves spraying a zinc (ZnSO₄ ^. 7H₂O) solution on the leaves one or multiple times during growth.

How soil and foliar zinc fertilisation might work

For decades, the use of zinc fertilisers in agriculture has been justified economically for improving plant nutrition by increasing soil availability of nutrients for plant uptake and increasing productivity by increasing yields (Manzeke et al., 2012; 2014). Application of zinc fertilisers via soil or foliar approach is an agronomic short-term strategy that affects mainly mineral nutrition and yield of crops. However, the effects of soil and foliar fertilisation on zinc concentrations beyond basic plant nutrition and productivity remain not fully understood.

When zinc is effectively loaded into the grains by continuous uptake from the soil or by remobilisation from vegetative tissues, there is an increase in the grain concentrations and thus increases the potential for bioavailability post-processing. However, zinc may not be loaded effectively into the grains depending on the ability of the crop to continuously take it up from the soil or remobilise from leaves and stems as well as the timing and rate of fertiliser application (Palmgren et al., 2008; Olsen and Palmgren, 2014). Therefore, several factors within farming systems limit implementation and effectiveness of soil and foliar strategies to increase grain zinc concentrations as listed below. •

Crop and nutrient management: crop genetics and efficiency, timing and rate of fertilisation;

The outcomes and effects flowing from soil and foliar zinc fertilisation strategies to increase grain zinc concentrations suggest that a policy-relevant systematic review of the effects of soil and foliar fertilisation on zinc biofortification requires examination of direct effects on: •

whole grain (including grain fractions) zinc concentrations;

Why it is important to do the review

While commercial zinc fortification has played a key role in addressing zinc malnutrition by producing zinc fortified foods, empirical evidence now show that agronomic zinc biofortification could be an attractive strategy to mitigate zinc deficiencies in the soil and related malnutrition in humans (Cakmak et al., 2008). Hence, there is a scope to identify not only zinc fertilisation methods that increase grain zinc concentrations and grain yield and minimise negative impacts on the environment and crop performance but also that could improve zinc bioavailability post-processing.

This viewpoint prompted research about the effects of soil and foliar zinc fertilisation on zinc biofortification particularly in countries severely affected by zinc deficiency and malnutrition. Recently, several primary studies attempted to study the effects of soil and foliar zinc fertilisation on grain zinc concentrations and bioavailability of wheat, maize and rice (Cakmak et al., 2010; Phattarakul et al., 2012; Wang et al., 2015; Zou et al., 2012; Zhang et al., 2012; Li et al., 2015). The evidence suggests that foliar strategy is more effective than soil strategy in increasing grain zinc concentration with inverse relation to grain yield (Phattarakul et al., 2012; Ram et al., 2016). The evidence also shows that great proportion of grain Zn in wheat and rice comes from remobilisation from leaves before grain formation and from continuous uptake, respectively, suggesting that soil application would be more effective for zinc biofortification of rice and foliar application would be more effective for zinc biofortification of wheat (Waters and Sankara, 2011; Phattarakul et al., 2012).

Ashong et al. (2012) reviewed the positives and negatives of rice biofortification with zinc and other minerals and vitamins on micronutrient status and health-related outcomes suggesting that industrial zinc fortification of rice could aid in the designing and implementation of appropriate food fortification. Brnic et al. (2016) studied zinc absorption from hydroponically and industrially zinc-fortified rice varieties by consumers fed with the same total zinc content meal, yet the study was inconclusive which strategy is more effective in increasing zinc bioavailability to consumers. Thus, evidence regarding the direct human health impacts of zinc biofortification is somewhat equivocal, largely because of low precision. Estimates based on the upper and lower confidence intervals of pooled effects from meta-analysis range from highly effective to no substantive effect (Ashong et al. 2012). Nevertheless, biofortification remains a mainstream development activity with annual spending millions of dollars. Identifying efficiencies in biofortification therefore remains an important development and food security objective.

Although there has been extensive primary research on zinc biofortification through soil and foliar application of zinc fertilisers, to our knowledge, no systematic review has been published using systematic data collection, critical appraisal and statistical synthesis using network meta-analysis. The only existing evidence synthesis undertook a cost effectiveness analysis of the potential of zinc-enriched fertilisers to alleviate human dietary zinc deficiency in sub-Saharan countries (Joy et al. 2015). Whilst useful, this synthesis did not adopt the standard methodologies of systematic review to minimise bias. The synthesis was not guided by a protocol, searches and inclusion criteria were not specified, critical appraisal was not undertaken, effect modifiers and publication bias were not investigated and the reporting does not conform to the standards of the Methodological Expectations of Campbell Collaboration Intervention Reviews (MECCIR). Another existing systematic review focused on quantifying the effect of zinc deficiency on human health in relation to the incidence and related mortality risk of diseases such as diarrhoea, pneumonia and malaria, particularly among children in developing countries (Caulfield and Black, 2004). Nonetheless, this review did not assess the evidence regarding biofortification strategies to mitigate the prevalence of inadequate zinc intakes. In addition, none of the syntheses used network meta-analysis to compare multiple treatment effects and explore inconsistency. We are unaware of any network meta-analysis in agricultural research to date, despite its widespread use in other domains and clear application with multiple treatments regularly used in single study sites. Therefore, the present review is important because it could inform policy makers and food industries with respect to the use of inherent high zinc genotypes of the three major cereal crops for the development of functional foods worldwide. This review will evaluate the effects of zinc fertilisation in terms of increased grain zinc concentrations with potential for increased bioavailability to the consumer post-processing, and this review will also inform policy makers and farmers in relation to adoption of a zinc fertilisation method for zinc biofortification of the major cereals that could optimise utilisation efficiency, farm economics and environmental sustainability.

Type of studies

This review will be limited to studies examining the effect of application of zinc fertilisers on grain zinc concentration and yield of wheat, maize and rice will be included in the review.

Population

The study population will be wheat, maize and rice genotypes as determined by each individual primary study under zinc fertilisation. Studies focusing on wheat, maize and rice under a fertilisation regime other than zinc will not be included in the review, likewise studies examining the effect of soil and foliar application of zinc fertilisers on the yield and zinc concentration of crops other than wheat, maize and rice will not be included in the review. Although such studies might be useful to understanding zinc biofortification mechanisms and processes, they are not relevant to the target intervention and population of the present review.

Intervention

Intervention studies will be limited to those examining the effects of timing, rate and solute concentration of zinc fertiliser, through soil, foliar, and soil plus foliar application, on grain zinc concentration and grain yield of wheat, maize and rice will be included in the review. Soil, foliar and soil plus foliar applications are assumed to be comparable with high exchangeability and are assumed to be jointly randomizeable. This is a strong assumption which will be tested by comparison with results from pairwise meta-analysis and meta-regressions to explore inconsistency within pairwise treatments.

Outcomes

Outcomes will be used to determine study eligibility. Grain zinc concentrations will be considered primary outcomes in this review. Secondary outcomes will be limited to grain yield. Grain zinc concentrations are usually measured in milligrams per kilogram (mg kg^-1) and grain yield in tonnes per hectare (t ha^-1). This review will not include outcomes relating to micronutrient status and human health. The meta-analyses will be used to parameterise a downstream systems model which will make explicit links between the systematic review outcomes and the quality of life and behaviour change factors of interest to policy makers. Heuristic discussion of these relationships will be included when considering the strength of evidence (indirectness) in the summary of findings.

Study designs

This review will include replicated field trials. Such studies will be included because of the practical implications of their findings on the decision-making for a generalised adoption or recommendation of a given fertilisation regime by farmers and policy makers. The review will include original primary research from all countries. Relevant studies written in a language other than English will be considered eligible for inclusion.

Search strategy for identification of studies

Description of methods used in research on agronomic zinc biofortification

Field research on agronomic zinc biofortification is well established with little or no variation in methodological quality. The little variability is in relation to fertiliser recommendations which follow guidelines established by individual countries based on soil test results. Glasshouse studies are also well established but with significant variation in methodological quality. These glasshouse studies are stronger in exploring and explaining mechanisms under controlled conditions but weaker in addressing fertiliser recommendations and extension programmes. Outcomes from glasshouse studies do not reflect conditions from the real world.

Example of studies which are eligible for inclusion in the review are provided in Table (?). Example would be excluded are presented in Table (?).

Criteria for determination of independent findings

Only independent effect sizes will be included in any single meta-analysis. If pooling across studies, only a single effect estimate for that study will be taken. The following decision criteria to determine independent findings will be used: •

a ‘summary effect size’ will be calculated using weighted random effects meta-analysis to explore heterogeneity between studies attributable to effect modifiers and estimate the treatment effect obtained from several potentially heterogeneous sources of evidence (Jansen et al., 2008);

Dealing with missing data

Where measure of dispersion (i.e. standard error, standard deviation, confidence intervals) is not reported, primary study authors will be contacted to obtain this data, where possible. Failing this, standard deviation of the outcome variable will be imputed using a bootstrapping procedure based on runif in the R statistical software environment.

Treatment of qualitative research

No qualitative data is planned to be included in this review.

Method of synthesis

Random effects network meta-analysis will be undertaken for each crop species separately. Effect sizes and 95 per cent confidence intervals will be presented using forest plots. Pairwise meta-analyses will be conducted to further explore the sources of heterogeneity where appropriate. Pairwise inconsistency will be assessed using the GRADE framework and heterogeneity quantified using I². If excessive heterogeneity is detected in the data, then a narrative synthesis will instead be conducted guided by the data extraction form in terms of the ways in which studies may be grouped and summarised. The narrative analysis will follow guidance laid out in the ESRC Narrative Synthesis Guidance document (Popay et al, 2006). We will not explore inconsistency within the network itself because the models will not converge with the high number of covariates included in the exploration of heterogeneity. Instead we will perform pairwise meta-regressions using AIC/DIC to avoid overfitting. Moderator analysis will be conducted according to the moderator variables defined above, and if sufficient studies permit we will conduct analysis of publication bias using conventional methods such as funnel plots.

The main findings of the review will be set out in summary of findings (SoF) tables incorporating primary outcomes only, to explain the significance of the findings. The outcome for each comparison will be listed with estimates of relative effects contributing data for those outcomes.