Tree and grass quality on the African savannah shows seasonal variation, driving mixed-feeding herbivores to switch between browsing and grazing. During this switch, crop consumption could be an attractive alternative to browsing. We analysed elephant diet variability in the Okavango Delta, Botswana, using faecal stable isotope ratios of carbon (δ^13C) and frequencies of elephant crop consumption, to determine the extent to which crop consumption relates to this potential switch. Although elephants did increase their relative grass consumption in the wet season, browse dominated the annual diet. After February, the proportion of grass in the diet dropped considerably, and continued decreasing through April when farmers reported most crop consumption. Generalized Linear Models revealed that the occurrence of elephant crop consumption increased with the proportion of grass consumed and with decreasing grass quality. The proportion of grass in elephant faeces increased with increasing crop consumption intensity. As crop consumption could also be related to nutrient deficiencies in elephant diet, we calculated the total dietary input of nutrients to reveal potential deficiencies. Elephant diet contained insufficient levels of sodium year-round, and insufficient phosphorus from February to July. As the latter coincides with the timing of crop consumption, we consider our results an indication that phosphorus –and potentially sodium - deficiencies, could play a role in elephant dietary choices, including crop consuming behaviour. Further experimental research is required to show whether supplying elephants with supplementary phosphorus and sodium sources could reduce this micro-nutrient deficiency, and could play a role in reducing elephant crop consumption.

Stable isotope analysis of consumer tissues document patterns of resource use because data are linearly related to isotope compositions of their source(s) (i.e., food, water, etc.). Deviations in parameters estimated for these relationships can arise from variations in consumer tissue–diet spacing (Δ TS ) and the level of isotopic heterogeneity in the source(s). We present a set of simple hypotheses that distinguish between the effects of Δ TS and source isotope heterogeneity. The latter may arise via mixed diets, during tissue turnover, or by isotopic routing of dietary components. We apply these concepts to stable carbon and nitrogen isotope relationships between gut contents and body tissues of large mammal herbivores from mixed C3/C4 South African savannas and test predictions based on the compound- and/or time-specific data archived within each material. Predicted effects of source isotope heterogeneity are readily detected in carbon isotope relationships between materials representing different time periods or comprising bulk versus protein-only diet components. Differences in Δ TS of carbon isotopes across mammal herbivore species with very different feeding niches (and diet isotope compositions) are likely to be small or non-existent in these habitats. Variations in Δ TS estimated for nitrogen isotopes are much greater, leading to inconsistencies that cannot be explained by diet or trophic level effects alone. The effects of source heterogeneity on isotopic relationships generate numerical artefacts that have been misinterpreted as variations in Δ TS . We caution against generalized application of hypotheses based on assumptions of source isotopic homogeneity, even for single diets commonly used in laboratory studies. More careful consideration of how heterogeneity affects consumer–diet relationships is needed for many field and laboratory systems.

The narwhal, Monodon monoceros, is a marine mammal that moves seasonally in Arctic waters. In males, left upper canines grow to between 2 and 3 metres in length as tapered tusks (Figs 1,2). Occasionally, tusks grow on the right side in males, but they are generally short.