Biofuels

Cropland footprint modelling was undertaken using LCA. Initially, cropland footprint data for Australian agricultural commodities were obtained from a previous study [31 (link)]. This included Australian-produced livestock products, taking into account the cropland footprints associated with feed rations. Australia is a net exporter of most agricultural commodities and it has been estimated that greater than 90% of available food is Australian-grown [58 (link),59 ]. This data source [31 (link)] also describes the cropland footprints of the main agricultural commodities that are imported because they are not grown in Australia to any significant extent, namely coffee bean, cocoa bean, tea, coconut, hazelnut, hops and oil palm fruit. In brief, these data were developed using a 1.1 km² spatial resolution map of agricultural production including crop yield information and involved the application of three life cycle impact assessment models, described below. Detail is available in the associated reference [31 (link)]. The cropland footprints of individual foods were quantified using conversion factors that translate agricultural commodities into retail products and edible portions, as described previously [49 (link)]. Former cropland areas occupied by food-processing factories, transportation systems and other infrastructure were deemed not materially important and were excluded from the assessment.
Three cropland footprint indicators were quantified for each individual adult daily diet, reflecting different environmental concerns relating to cropland occupation. Firstly, a cropland scarcity footprint (CSF) was quantified. Cropland is a globally finite and scarce natural resource and the occupation of cropland contributes to this scarcity as cropland used for one productive purpose cannot be used for another. However, not all cropland is equally productive. Therefore, CSFs were quantified taking into account productive capability using the net primary productivity of potential biomass at each location, reflecting the natural capability of the land. CSF results were expressed in m² yr-e (equivalent), with cropland of global average productivity as the reference. Secondly, a cropland biodiversity footprint (CBF) was quantified using the biodiversity impact factors of Chaudhary and Brookes [60 (link)]. These impact factors report potential species loss based on 5 taxa in 804 ecoregions of the world. CBF results were expressed as potentially disappeared fraction of species (PDF). Thirdly, a cropland malnutrition footprint (CMF) was quantified using the impact factors of Ridoutt et al. [26 (link)]. These impact factors report potential protein-energy malnutrition impacts considering potential domestic and trade-related food deficits arising from cropland occupation. The factors are highest in countries where protein-energy malnutrition is prevalent and in countries that share a trade relationship with these regions. As is typical in LCA, the factors express the potential impact of production (i.e., occupying the cropland) and not the potential benefits of use (i.e., food consumption). Crop products have the potential to be used in numerous ways: for direct human consumption, for livestock rations, for biofuels or other industrial products. Even when crop products are intended for human consumption, they may be wasted or contribute to energy intakes that exceed a healthy diet. CMF results were expressed in disability-adjusted life years (DALYs). To calculate cropland footprint results, each spatially explicit instance of cropland occupation was multiplied by the spatially relevant impact factor. Cropland footprint results for almost 150 separate food items are presented in the Supplementary Materials, Table S3.

The dependent variable was childhood mortality and this was captured with the question on the survival status of the most recent birth (dead = 1 or alive = 0) in the last 5 years preceding the survey. Under-five mortality was defined as the death of a live-born child before its fifth birthday [27 (link)]. The main predictor was the type of housing materials. This was obtained as aggregate score based on information from roof materials (Improved – cement, roofing sheets, ceramic tiles; Unimproved – natural, no roof, palm leaf, sod, rudimentary, rustic mat, bamboo, cardboard), wall materials (Improved – cement, stone with cement, cement blocks, bricks; Unimproved – natural, no wall, palm/trunks, dirt, rudimentary, bamboo with mud) and floor materials (Improved – cement, ceramic tiles, vinyl asphalt strips, parquet, polished wood, finished; Unimproved – natural, earth, sand, dung, rudimentary, wood planks, palm, bamboo, others). The improved categories assumed a score of 1 while unimproved scored [36 , 38 ]. The overall score (13-point maximum and 0-point minimum) for a woman was disaggregated into three categories: inadequate (<50% of the overall score), moderate (50% ≤ x < 75% of the overall score) and adequate (75% ≤ x ≤ 100%) of the overall score). They are based on quartile classification: 3rd quartile is 75% and 2nd quartile is 50%.
Other independent variables include mother’s age, highest educational level, religion, ethnicity, marital status, place of residence, region, wealth index and media exposure. others are; number of antenatal visits, tetanus injection, gender of a child, size at birth, birth order, preceding birth interval, prenatal care provider, delivery assistance, place of delivery, cooking fuel (Clean – electricity, liquefied petroleum gas, natural gas, biofuel; Unclean/biomass – coal, charcoal, wood, straw/shrubs/grass, agricultural crop, animal dung, kerosene), source of water (Improved - piped into dwelling, public tap, borehole, protected well and spring, rain water and bottle water; Unimproved – other sources not listed as improved sources) and toilet facility (Improved – flush/pour flush to piped sewer system, septic tank or pit latrine, ventilated improved pit latrine, pit latrine with slab, composting toilet; Unimproved – other toilet types not listed as improved). We adapted the groupings of environmental factors documented in the 2013 Nigeria National Demographic Health Survey (NDHS) and the 2010 WHO and UNICEF document on progress on sanitation and drinking water [36 , 39 ]. These variables were used as covariates during multivariate analysis to determine the association between housing materials and childhood mortality.

The PPR (41.7° to 54.7°N latitude, 92.5° to 114.5°W longitude) covers ~820,000 km² of the Great Plains in the United States and Canada and is home to millions of depressional wetlands (16 ). These depressions were formed during the Wisconsin glaciation, approximately 12,000 years ago (17 ). The underlying glacial till has low permeability, allowing depressions to fill and hold water and to ultimately develop into palustrine and lacustrine ecosystems made up of wetlands, ponds, and shallow lakes. We collectively refer to these depressional waterbodies as “wetlands” because the majority of them are less than 1 ha (0.01 km²) in size, less than 1 m deep, and seasonally (nonpermanent) ponded, meeting the definition of a wetland (sensu the Cowardin classification system) (61 ). In the PPR, larger wetlands often are shallow (<2 m) and have similar biogeochemical processes as smaller wetlands (62 ). Wetlands contain methanogenic microbial communities that are adapted to anoxic saturated soils, where they decompose organic material and produce CH₄ (9 (link)). PPR wetlands range from fresh to hypersaline (up to three times more saline than the ocean) (17 ). This salinity is attributable to groundwater transport of dissolved sulfate through the ion-rich glacial till. PPR wetlands with higher sulfate concentrations have suppressed CH₄ emissions (23 ), albeit high dissolved organic carbon concentrations in some PPR wetlands can support CH₄ production in the presence of sulfate (24 (link)). Larger wetlands in topographically lower landscape positions tend to accumulate groundwater-derived solutes, while smaller, shallower wetlands are filled with rainwater and snowmelt (45 ). Vegetation characteristics of PPR wetlands are influenced by water depth and chemistry, typically with concentric zones with open water with submerged aquatic vegetation toward their centers and marshes and meadows with floating and emergent macrophytes such as Typha toward their edges (37 , 44 ).
The climate of the PPR is continental with a north to south temperature gradient ranging from ~2° to 8°C mean annual temperature and a northwest to southeast precipitation gradient ranging from ~400 to 900 mm year⁻¹ (63 , 64 (link)). The PPR is centered on the confluence of tropical Pacific Ocean (El Niño–Southern Oscillation), eastern Pacific Ocean (Pacific Decadal Oscillation), and North Atlantic (Atlantic Multidecadal Oscillation) oscillations (65 ). Synergistic effects from synchronization of these oscillations lead to decadal periods of drought and deluge that influence groundwater levels. Wetland surface water is also extremely sensitive to variability in seasonal and annual precipitation (66 ). Thus, the size and hydroperiod of PPR wetlands are the result of complex interactions between long-term climate and short-term weather. An extreme multiyear drought from 1988 to 1992 led to the drying of many wetlands, with 1991 having the lowest wetland extent (45 ). Following the extreme drought, much of the PPR experienced a shift to a wetter climate. Annual precipitation in 2011 was one of the highest on record, and 2011 had some of the highest numbers of wetlands with ponded water (45 ). Therefore, we use 1991 and 2011 as extreme dry and wet years, respectively, in our modeling.
The PPR is intensively used for agricultural crop and biofuels production, which has led to extensive wetland drainage and upland conversion from prairie grassland to cropland (16 ). Drainage reduces CH₄ emissions and soil organic carbon stocks but increases CO₂ and nitrous oxide (N₂O) emissions (22 (link), 67 ). Upland conversion from grassland to cropland can affect CH₄ emissions indirectly through increased nutrient loading, resulting in increased CH₄ emissions (38 (link)). However, wetlands nested in croplands are also subject to tillage and aeration of soils, thereby lowering CH₄ emissions (22 (link)). Wetland drainage often results in consolidation of water from multiple smaller wetlands into one larger wetland. These larger wetlands are deeper with fewer fluctuations in water levels, as well as greater coverage of invasive emergent hybrid cattail, all of which favors CH₄ production and emissions (37 ). Consolidation of wetlands also targets the smallest wetlands changing the distribution of wetland size classes, albeit the vast majority of wetlands are still <1 ha.

We used our spatially explicit predictors and RF model to make predictions of CH₄ flux for all wetland pixels in the entire PPR for two historical dry and wet periods (1991 and 2011, respectively) to understand the range of natural variation in CH₄ fluxes. We focused on predicting growing season fluxes. Studies of wetland CH₄ flux during winter months have demonstrated the importance of nongrowing season emissions to annual CH₄ budgets (48 (link)–50 ). Ice breakup and its thaw in spring potentially may be hot moments for release of physically trapped and accumulated CH₄ under ice over the winter (51 ). However, our wetlands had low nongrowing season flux rates (median, 0.007 mg of CH₄ m⁻² hour⁻¹) (69 ), as did an inversion study in the same region (82 (link)). Therefore, we assumed that the predicted CH₄ flux rates from the first and last time steps of the frost-free season represented winter rates before and following the frost-free season, respectively. Exclusion of nongrowing season dynamics, especially the physical processes of freeze up and thaw in the shoulder seasons, may have underestimated our cumulative CH₄ flux estimates.
We also developed models for four future climate scenarios to examine how different assumptions about warming and wetland extent may affect CH₄ fluxes by 2100. Our current analysis only focuses on climate change scenarios and not land use scenarios. In a study that compared 11 future land use scenarios in the Northern Great Plains, there was a wide variability in future cropland extent with both gains and losses and an increase in grassland extent for cellulosic biofuels production (40 ).
Wetland extent has been identified as a primary source of uncertainty in CH₄ fluxes (2 , 32 ). Unlike the clear consensus among ESMs in the directionality of future atmospheric temperature warming, there is much less agreement among forecasts of future precipitation in the PPR. Future mean annual precipitation change in the PPR by the end of the 21st century could range from −19 to +33% when compared to 1990 to 2020 normals (55 (link)). Therefore, to simulate wetland extent of future scenarios, we used DSWE and NDVI rasters from 1991 and 2011 to represent bookends of the potential range in wetland extent from dry to wet, respectively, in the PPR. For future temperatures, we used a 13-model ensemble of ESMs (described previously) associated with SSP2-4.5 (moderate warming, ~1.7°C) and SSP5-8.5 (severe warming, ~2.7°C) for the period 2081 to 2100. We combined our dry and wet bookends with the two warming regimes to create four future scenarios: (i) moderate warming and dry, (ii) moderate warming and wet, (iii) severe warming and dry, and (iv) severe warming and wet.
In total, our upscaling efforts covered ~3.8 trillion pixels [2 historical climate conditions × (4 future climate scenarios × 13 ESMs) × 26 time steps per condition or scenario × 1.4 billion pixels in the PPR]. Because predicting CH₄ flux under these scenarios was extremely computationally expensive, we limited our analyses to these two historical conditions and four future scenarios. Additional details on the landscape model are provided in the Supplementary Materials.