UC Davis

Changes in crop species diversity can affect agroecosystem function. However, most crop diversity studies insufficiently account for the influence of scale on spatial crop diversity, and its relation to temporal diversity has not been explored. Moreover, crop diversity might be limited by environmental constraints and market demand for specific crops, which needs to be considered when assessing opportunities for diversification. This dissertation developed and applied new approaches to gaining a quantitative understanding of diversity patterns and processes, allowing for improved comparison between regions and countries. It includes an analysis of the scale dependency of crop species diversity and its relation with temporal diversity using high-resolution crop-specific land-cover data for the conterminous US. It also shows the magnitude of environmental and demand-side constraints to crop diversity globally. For that purpose, a theoretical framework of hierarchical levels of crop species diversity is presented, in which potential, attainable, and current diversity levels are compared to compute diversity gaps.

We found that spatial diversity monotonically increases with the size of the observational unit, and the strongest association between spatial and temporal diversity is observed when measured in areas comparable to farm sizes. In larger areas, the association weakens because of the increasing diversity among farms. At the national level, the diversity among farms is usually higher than the diversity within them, which needs to be considered when inferring diversity effects with national-level data.

Environmental limits to crop diversity are higher in temperate and continental areas than in tropical and coastal regions. Crop diversity is also constrained by a high demand for a few crop species, which results in an attainable diversity that is much lower than the potential. Nevertheless, there are large gaps between current and attainable diversity levels in most croplands. These gaps are particularly large in the Americas, where croplands are dominated by a few major annual crops (maize, soybean, wheat) mostly grown on fields with a very low temporal diversity. In contrast, diversity gaps are relatively small in Europe and East Asia. Changes in food demand favoring minor crops could positively impact spatial and temporal crop species diversity by increasing the attainable diversity. But given current consumption patterns, the most effective strategy to increase crop diversity in areas with high diversity gaps might be to expand the area of a major crop adapted to that specific environment, but that is not widely planted.

Securing adequate soil fertility is also critical for diversification, especially in the tropics, where low soil pH is one of the main limiting factors of potential crop diversity, and soil acidity remains a key management challenge for smallholder farmers. Liming can boost the productivity of acid soils, but the lime rate required to achieve this is unknown for many tropical regions where food production increases are urgently needed. Therefore, lime requirement models based on readily available soil data could be very useful in these places. However, the great variety of lime requirement models available in the literature introduces much uncertainty. We evaluated current lime requirement models for acid tropical soils and introduced a new model based on acidity saturation using data from four soil incubation studies and 31 soil types. Foundational models based on acidity or base saturation are reasonably accurate (r ≥ 0.9), but later attempts to improve these models were unsuccessful. The new model, in contrast, has more precision than all earlier models across a wide range of acid tropical soils from different regions. Moreover, lime requirement estimates largely depend on the target soil chemical property of the model. For instance, many more African soils would require liming based on base saturation models than acidity saturation models, regardless of the accuracy. The new acidity saturation model can effectively estimate the lime rate required to address aluminum toxicity. This model could be incorporated into more comprehensive models once lime rates needed for other acidity problems are well established.