Abstract. Stenothermal habitat specialists such as monophagous insects may be threatened by climatechange induced range shifts, if resource requirements along migration routes or future habitats are not met. Little is known about altitudinal shifts of inconspicuous, less mobile species relative to latitudinal range shifts in prominent, mobile organisms such as butterflies. Here, we address three questions: What are the resource requirements-habitat area, quality and fragmentation-for two specialist herbivores? How do habitat characteristics change with altitude? Do species resource requirements suggest an increased vulnerability towards temporal or spatial bottlenecks under climate change?We mapped the spatial distribution of a riparian shrub, Veronica stricta (Plantaginaceae), at local (300 m elevational gradient, n ¼ 252 patches) and regional scales (600 m gradient; n ¼ 102) in the Tongariro National Park, New Zealand. Patch occupancy and patch-level population dynamics of Trioza obscura (Hemiptera: Triozidae) and an undescribed gall midge (Diptera: Cecidomyiidae) were recorded in 2010 and 2011. Habitat suitability and population dynamics were predicted with generalized linear models, applying information theoretic model averaging.Habitat area had the largest positive influence on insect presence and survivorship, the latter benefitting from larger insect populations. Habitat quality showed a species-dependent effect on occurrence (T. obscura: negative effect of plant shade; cecidomyiid: positive effect of host inflorescences) and survival (T. obscura: positive response to leaf-nitrogen and host size; cecidomyiid: positive response to inflorescence abundance). Only cecidomyiid colonization had a weak negative response to habitat fragmentation. Altitude contributed positively to patch suitability and survivorship of the cecidomyiid, but decreased survivorship of T. obscura.Altitude-related changes in the landscape matrix, such as the transition from forested to subalpine vegetation, positively affected habitat area and fragmentation, which in turn offset a steady decrease in host plant size (i.e., patch quality). Our study revealed that habitat area, quality and fragmentation do not always follow simple linear trends along elevational gradients. We recommend that species distribution models should consider these complex patterns when predicting range shifts. We suggest that species vulnerability to climate change depends on whether resource requirements allow for upward migration and suitable habitats are available in future distribution ranges.