Estimating food production in an urban landscape

There is increasing interest in urban food production for reasons of food security, environmental sustainability, social and health benefits. In developed nations urban food growing is largely informal and localised, in gardens, allotments and public spaces, but we know little about the magnitude of this production. Here we couple own-grown crop yield data with garden and allotment areal surveys and urban fruit tree occurrence to provide one of the first estimates for current and potential food production in a UK urban setting. Current production is estimated to be sufficient to supply the urban population with fruit and vegetables for about 30 days per year, while the most optimistic model results suggest that existing land cultivated for food could supply over half of the annual demand. Our findings provide a baseline for current production whilst highlighting the potential for change under the scaling up of cultivation on existing land.


Supplementary Methods 2: Detailed sampling methodology for public fruit trees
A 500 m mesh grid was placed across the entirety of each urban area, and each 500  500 m grid square (hereafter 'tile') was classified into one of 25 urban form classes according to (a) percentage building cover (five categories), and (b) percentage cover of vegetation over 0.5 m tall (five categories). Five representatives of each urban form class were randomly selected, where available, yielding a total of 112 survey tiles. Within 78 survey tiles, a greenspace 'fragment' covering all land uses, public and private, was selected using a stratified random approach based on urban form. Within each fragment, the species/size composition and abundance of all trees were estimated using a combination of transects and complete surveys across the whole site, or sub-areas selected using a stratified random approach, to represent the range of land uses present.
Separately, in all 112 tiles, a 1 km (not straight line) transect was walked along public rights of way, sampling as wide a variety of the land uses present as possible (given access constraints). Due to these constraints, 9 transects were between 200-947 m in length and 9 were between 1052-1218 m in length. All other transects were 1km in length, +/-50 m. As in the fragment sites, the composition and abundance of upright tree species (excluding hedges and shrubs) were recorded in a 10 m radius around each of eight equally spaced stopping points on the transect. The centre of the survey area was always on public routes, but the surveys extended into both public and private adjacent land.
Both sets of samples (fragments and transect stopping points) were classified according to the landuse in which they occurred, and used to separately estimate the proportional composition of studied fruit tree species for areas of tree cover within each land-use type across the study area. The two data sets yielded broadly similar estimates of fruit tree species proportional abundance, although the transect samples had slightly lower estimates of overall fruit tree abundance (0.67% of all recorded transect trees were studied fruit tree species, compared to 0.72% for fragment trees). We thus combined the proportional abundance of study species for both datasets by mean value into a single dataset of proportional tree occurrence by land-use type.
Study species accounted for 10% of total surveyed species (by number of species) producing parts that are edible with minimal preparation, but are believed to be the most well-known, recognisable and frequently managed for food crops. They are therefore expected to have a disproportionately high importance and recognisability in foraging circumstances, so production estimates for nongarden fruit trees were calculated for these species. Some other species producing edible components had high occurrence but were excluded from analysis on the basis of being deemed less relevant to potential foraging as a direct food source (e.g. sloe Prunus spinose; wild cherry Prunus avium; and hazel Corylus avellana).

Abstract for Clark, Rachel (2014), "Investigating land use on Sheffield's allotments: Potential for food production and provision other services", Research Thesis, University of Sheffield, Department of Animal and Plant Sciences 18
This study uses a multidisciplinary approach to explore the land use on 38 allotment plots in Sheffield, involving mapping plots to quantify the proportion cultivated and questionnaires to obtain details of management practices and motivations of plot holders. Cultivated land occupied on average 27% of the total plot area, ranging widely from 6% to 67%, driven predominantly by the tenancy length under the current plot holder, and decreasing with more people sharing the plot. Produce grown within plots was diverse, the most prevalent being crops such as potatoes, onions, leeks and green beans which occupied a relatively large proportion of the growing area. Evidence for substantial meal provision emphasises the need for more accurate quantification of the extent of food production on allotments. Plot holders reported the most important aspects of allotment management to be spending time outdoors and associated wellbeing benefits, and along with evidence for pro-environmental management practices this highlights the potential for involvement of allotment gardeners in bottom-up conservation schemes. This study highlights avenues for further investigation into allotments for food production and conservation of urban biodiversity, and emphasises the need for increased support from local authorities to fully utilise allotments as multifunctional green infrastructure.  19 , and mean relative proportions of cultivated crops 18 (mean of that used in Clark 18 and Edmondson et al. 19