A long-term study was made of the carrying capacities (K) of UK grasslands restored to support the endangered butterfly Phengaris arion. This iconic species is the focus of major restoration programmes and provides a blueprint for the conservation of other threatened insects. P. arion larvae are phytophagous during their early instars before switching to become obligate social parasites of Myrmica ant colonies for their main growth. A mechanistic model, incorporating life-table measurements of natality and mortalities, was used to predict the value of K on 19 restorations. Predictions were compared with actual butterfly densities measured for up to 33 consecutive generations per site over 43 years. The model suggested that individual sites vary by up to a hundred-fold in the density of butterflies each site would support. Observed densities closely correlated with model predictions, with the most productive site supporting 129-times more butterflies per square metre than the least productive one. Of seven life-table parameters modelled, the three that explained observed butterfly densities best describe the distribution and fitness of its primary host-ant species, Myrmica sabuleti. Population densities correlated most closely with the density of host-ant nests. Annual fluctuations in P. arion numbers attributable to the weather were one to two orders of magnitude lower than the differences between site carrying capacity densities. Even extreme events, such as drought, caused perturbations one order of magnitude smaller than inter-site carrying capacity variations.

Synthesis and application. Observed population dynamics of P. arion conform with theories of a definable ceiling for insect numbers, controlled by density-dependence and limited by a larval resource that is typically more specialised, and rarer, within sites than was once perceived. At a practical level, the model provides a useful tool for determining which future sites should be prioritised for the persistence of this endangered species and for new restorations. More generally, the results support two concepts: (i) once suitable management has been implemented, effects of intrinsic site characteristics on numbers greatly exceed inter-annual fluctuations; and (ii) understanding how the quality of larval habitat varies between sites is paramount when restoring the size and resilience of populations of declining invertebrates.