Invasive mammal eradications are commonplace in island conservation. However, post-eradication monitoring beyond the confirmation of target species removal is rarer. Seabirds are ecosystem engineers on islands and are negatively affected by invasive mammals. Following an invasive mammal eradication, the recovery of seabird populations can be necessary for wider ecosystem recovery. Seabirds fertilise islands with isotopically heavy nitrogen, which means nitrogen stable isotope analysis (δ15N) could provide a useful means for assessing corresponding change in ecosystem function. We quantified decadal changes in δ15N on eight temperate New Zealand islands subject in pairs to distinct mammal invasion and seabird restoration histories: invaded, never-invaded, invader-eradicated and undergoing active seabird restoration. First, we investigated long-term changes in δ15N values on individual islands. Second, we used a space for time analysis to determine if δ15N levels on islands from which invaders had been removed eventually recovered to values typical of never-invaded islands. On each island soil, plants (Coprosma repens, C. robust and Myrsine australis) and spiders (Porrhothelidae) were sampled in 2006/07 and 2022 allowing δ15N change on individual islands over 16 years to be assessed. Combined, the samples from invader-eradicated islands provided a 7 – 32 year post-eradication dataset. Change in δ15N was only detected on one island across the study period, following the unexpected recolonisation of seabirds to an invaded island. Invader-eradicated islands generally had higher δ15N values than invaded islands however, they were still lower than never-invaded islands and there was no trend in δ15N with time since eradication. This, and the measurable increase in δ15N following seabird recolonisation on one island, may suggest that δ15N change occurs rapidly following invader-eradication, but then slows, with δ15N values staying relatively constant in the time period studied here. Isotope and seabird population studies need to be coupled to ascertain if plateauing in δ15N reflects a slowing of seabird population growth and subsequent basal nutrient input, or if the baseline nutrients are entering the ecosystem but then not propagating up the food web.

Invasive mammal eradications are widely used for managing island ecosystems. However, tracking the outcomes of such large-scale, whole ecosystem projects is challenging and costly, and monitoring all components of an ecosystem is near impossible. Instead, indicators of ecosystem change may provide more practical and integrated measures of ecosystem response to eradications. As high-order marine predators, seabirds subsidise island ecosystems with nutrients isotopically enriched in nitrogen. Invasive mammals have caused a global decline of seabirds on islands, reducing this nutrient subsidisation. Following eradications, nitrogen stable isotope analysis may provide a useful and resource-efficient indicator of ecosystem functional change on eradicated islands. However, isotope ratios are affected by a myriad of factors, with potential sources of variation being introduced by spatial and temporal variation in sampling, and within and between different taxa and ecosystem components. To correctly attribute isotopic change to post-eradication ecosystem function change, these confounding variables need to be understood. To address this need, we analysed stable isotopes of nitrogen in soil, plant, spider, and seabird guano samples collected at different distances from seabird colonies and at different stages of the short-tailed shearwater breeding cycle on six island sites around south-eastern Tasmania, Australia. Across these cool, temperate islands we detected no temporal variability in δ15N throughout the breeding season. However, there was notable spatial variability in δ15N values. The effects of seabird-derived nutrient subsidisation were highly localised with high δ15N values found inside seabird colonies and then rapidly decreasing from the colony boundary. Higher δ15N values also occurred in areas of higher burrow density within a colony. Variability in δ15N values also existed both within and between ecosystem components. Our results highlight the importance of context dependency when using ecological indicators and have important implications for the design, implementation and interpretation of studies employing stable isotopes as indicators for ecosystem change. We provide recommendations for designing future stable isotope studies on seabird islands.