Sunflower Yield Risk Narrows Only When Seed and Fertilizer Improve Together
Domestic sunflower production in Tanzania sits at the center of a larger food-system problem: the country relies heavily on imported edible oil, yet its own growers still work with low-input systems that leave yields exposed to wide swings. A new simulation study using nationally representative household data shows just how uneven those outcomes can be. In the most common production setup, local seed with no fertilizer, more than half of simulated yields fell below 1.0 t ha−1, while only a small fraction rose above 2.0 t ha−1.
The pattern is blunt. Most sunflower farmers are still operating in a low-yield, high-uncertainty zone.
Researchers used a Multivariate Empirical model inside a Monte Carlo simulation framework to examine six management combinations based on seed type and fertilizer use: local seed or improved seed, each paired with no fertilizer, organic fertilizer, or inorganic fertilizer. The analysis drew on household data from Tanzania’s 2019/20 National Sample Census of Agriculture and generated 500 yield outcomes for each farming system using Latin Hypercube sampling. Instead of relying on a single average yield, the approach mapped the distribution of likely outcomes and the probability that production would fall below or exceed defined thresholds.
That matters in a crop system shaped by rainfall volatility, nutrient stress, and uneven access to inputs. In Tanzania’s sunflower regions, especially the semi-arid central corridor, yield is not just a function of what is planted. It also reflects how much water the crop receives, whether the soil can supply essential nutrients, and whether the chosen seed can convert limited resources into grain. When those pieces do not line up, the result is not simply a lower harvest; it is a wider spread of possible harvests, which complicates planning for farmers, traders, and policymakers alike.
Local Seed Dominates, but Low Returns Follow Closely Behind
Nearly 87% of sampled sunflower-producing households used local seed without fertilizer, a striking reminder of how entrenched low-input cultivation remains. In that group, the simulated yield profile was poor and tightly clustered around low returns. The probability of yields below 1.0 t ha−1 reached 54%, while the chance of exceeding 2.0 t ha−1 was only 3%.
The study’s threshold framing gives the result real agronomic force. A low average yield can sometimes hide a long tail of better-performing farms, but here the dominant system was not only low-yielding; it was also highly unlikely to produce a breakout harvest. For a crop meant to support both household income and national edible oil supply, that combination leaves little room for commercial surplus or resilience against seasonal shocks.
Adding inorganic fertilizer to local seed improved the outlook, though only modestly. The probability of surpassing 2.0 t ha−1 increased to 7%, showing that nutrient input alone can lift the upper end of the yield distribution, but not transform the system entirely. Organic fertilizer with local seed did not produce the same degree of improvement, reinforcing the idea that nutrient source, timing, and crop response all interact with the underlying genetics of the seed.
Improved Seed Helps Most When Nutrient Supply Keeps Pace
The strongest simulated performance came from the combination of improved seed and inorganic fertilizer. That system produced the highest mean yield, 1.14 t ha−1, and the highest probability of exceeding 2.0 t ha−1, at 16%. Yet this result needs to be read carefully. Only 11 observed households represented that system, which leaves the estimate vulnerable to sampling uncertainty.
Even so, the direction of the signal is biologically plausible. Improved seed carries greater yield potential, but that potential is only expressed when the crop is not constrained by missing nutrients. Sunflower, like other oilseed crops, responds to soil fertility because nutrient availability influences vegetative growth, flowering, seed filling, and final grain weight. Inorganic fertilizer can relieve nitrogen and phosphorus limitations common in Tanzanian soils, while improved varieties may be better able to exploit the added resources. The simulated gains therefore make sense not as a miracle of seed alone, but as a match between genetics and input support.
Improved seed with organic fertilizer performed comparatively poorly, which suggests that organic inputs alone may not be enough to unlock the yield potential assumed by improved varieties under these conditions. Organic fertilizer can enhance soil structure and moisture retention, but its nutrient release is slower and less predictable than mineral fertilizer. In a semi-arid production setting where rainfall variability already narrows the margin for error, that slower release may leave the crop undernourished at critical growth stages.
Production Risk Matters as Much as Mean Yield
The study’s main contribution is not simply ranking management systems by average output. It shows that different seed-fertilizer combinations produce distinct yield distributions, and those distributions carry practical consequences. Farmers do not live on averages; they farm under uncertainty, and risk exposure influences whether they can invest, borrow, store grain, or commit to larger area planting the next season.
For Tanzania, the implications extend beyond one crop cycle. Domestic demand for edible oil continues to outpace production, and that gap leaves the country dependent on imports. Raising sunflower productivity could help narrow the deficit, but the pathway has to be realistic for smallholders who face input costs, limited extension support, and variable weather. A strategy that increases yield only on paper, or only in a few well-resourced farms, will not shift the sector at scale.
The simulation framework used here offers a way to think about that problem more clearly. By preserving empirical yield variation rather than collapsing it into a single average, the analysis captures how likely different outcomes are under each farming system. That is especially valuable in regions where climate variability can quickly move production from marginal to poor, or from adequate to profitable, within a single season.
Sample Size Limits the Strongest Claim, but the Direction Is Consistent
The central caution sits in plain view: the most favorable system, improved seed with inorganic fertilizer, rested on only 11 observed households. Such a small base limits confidence in the exact magnitude of the estimated advantage. Larger samples, multi-year data, and explicit climatic variables are needed before anyone treats the 16% probability of exceeding 2.0 t ha−1 as a stable national estimate.
Even with that limitation, the study points toward a practical agronomic message. Better seed and better nutrient management appear to move sunflower production into a higher-yield zone, and they also reduce the odds that farms stay trapped in chronically low-output states. The next step is not to assume those gains will generalize automatically, but to test whether they hold across seasons, regions, and rainfall conditions that are increasingly unstable.
The unanswered question is how much of the observed yield spread is driven by management and how much by environment. Until multi-year evidence is available, the safest interpretation is narrower: in Tanzania’s sunflower sector, local seed without fertilizer leaves most farms in a low-yield regime, while improved seed paired with inorganic fertilizer offers the strongest simulated upside, but from a very small observed base. The DOI for the study is 10.1371/journal.pone.0363039.
- • PLOS ONE DOI 10.1371/journal.pone.0363039: https://doi.org/10.1371/journal.pone.0363039