Phenology affects the abiotic and biotic conditions that an organism encounters and, consequently, its fitness. For populations of high-latitude species, spring phenology often occurs earlier in warmer years and regions. Here we apply a novel approach, a comparison of slope of phenology on temperature over space versus over time, to identify the relative roles of plasticity and local adaptation in generating spatial phenological variation in three interacting species, a butterfly, Anthocharis cardamines, and its two host plants, Cardamine pratensis and Alliaria petiolata. All three species overlap in the time window over which mean temperatures best predict variation in phenology, and we find little evidence that a day length requirement causes the sensitive time window to be delayed as latitude increases. The focal species all show pronounced temperature-mediated phenological plasticity of similar magnitude. While we find no evidence for local adaptation in the flowering times of the plants, geographic variation in the phenology of the butterfly is consistent with countergradient local adaptation. The butterfly's phenology appears to be better predicted by temperature than it is by the flowering times of either host plant, and we find no evidence that coevolution has generated geographic variation in adaptive phenological plasticity.

The effect of spring temperature on spring phenology is well understood in a wide range of taxa. However, studies on how winter conditions may affect spring phenology are underrepresented. Previous work on Anthocharis cardamines (orange tip butterfly) has shown population‐specific reaction norms of spring development in relation to spring temperature and a speeding up of post‐winter development with longer winter durations. In this experiment, we examined the effects of a greater and ecologically relevant range of winter durations on post‐winter pupal development of A. cardamines of two populations from the United Kingdom and two from Sweden. By analyzing pupal weight loss and metabolic rate, we were able to separate the overall post‐winter pupal development into diapause duration and post‐diapause development. We found differences in the duration of cold needed to break diapause among populations, with the southern UK population requiring a shorter duration than the other populations. We also found that the overall post‐winter pupal development time, following removal from winter cold, was negatively related to cold duration, through a combined effect of cold duration on diapause duration and on post‐diapause development time. Longer cold durations also lead to higher population synchrony in hatching. For current winter durations in the field, the A. cardamines population of southern UK could have a reduced development rate and lower synchrony in emergence because of short winters. With future climate change, this might become an issue also for other populations. Differences in winter conditions in the field among these four populations are large enough to have driven local adaptation of characteristics controlling spring phenology in response to winter duration. The observed phenology of these populations depends on a combination of winter and spring temperatures; thus, both must be taken into account for accurate predictions of phenology.

The temporal aspects of life cycle characteristics, such as diapause development, are under strong selection in seasonal environments. Finetuning of the life cycle may be particularly important to match the phenology of potential mates and resources as well as for optimizing abiotic conditions at eclosion. Here, we experimentally study the spring phenology of the orange tip butterfly, Anthocharis cardamines, by analysing post-winter pupal development in three populations along a latitudinal cline in each of Sweden and the United Kingdom. These countries differ substantially in their seasonal temperature profile. By repeatedly recording pupal weights, we established that post-winter development has two separate phases, with a more rapid weight loss in the second phase than in the first, likely corresponding to a ramping up of the rate of development. Variation in the duration of the first phase contributed more strongly than the second phase to the differences in phenology between the localities and sexes. We found that insects from Sweden had a faster overall rate of development than those from the United Kingdom, which is consistent with countergradient variation, as Sweden is colder during the spring than the United Kingdom. Similar trends were not observed at the within-country scale, however. A cogradient pattern was found within Sweden, with populations from the north developing more slowly, and there was no clear latitudinal trend within the United Kingdom. In all localities, males developed faster than females. Our results point to the importance of variation in the progression of post-winter development for spring phenology.

Understanding and predicting phenology has become more important with ongoing climate change and has brought about great research efforts in the recent decades. The majority of studies examining spring phenology of insects have focussed on the effects of spring temperatures alone. Here we use citizen-collected observation data to show that winter cold duration, in addition to spring temperature, can affect the spring emergence of butterflies. Using spatial mixed models, we disentangle the effects of climate variables and reveal impacts of both spring and winter conditions for five butterfly species that overwinter as pupae across the UK, with data from 1976 to 2013 and one butterfly species in Sweden, with data from 2001 to 2013. Warmer springs lead to earlier emergence in all species and milder winters lead to statistically significant delays in three of the five investigated species. We also find that the delaying effect of winter warmth has become more pronounced in the last decade, during which time winter durations have become shorter. For one of the studied species, Anthocharis cardamines (orange tip butterfly), we also make use of parameters determined from previous experiments on pupal development to model the spring phenology. Using daily temperatures in the UK and Sweden, we show that recent variation in spring temperature corresponds to 10-15 day changes in emergence time over UK and Sweden, whereas variation in winter duration corresponds to 20 days variation in the south of the UK versus only 3 days in the south of Sweden. In summary, we show that short winters delay phenology. The effect is most prominent in areas with particularly mild winters, emphasising the importance of winter for the response of ectothermic animals to climate change. With climate change, these effects may become even stronger and apply also at higher latitudes.

An insect species that shows variation in host species association across its geographical range may do so either because of local adaptation in host plant preference of the insect or through environmentally or genetically induced differences in the plants, causing variation in host plant suitability between regions. In the present study, we experimentally investigate the host plant preference of Anthocharis cardamines (orange tip butterfly) in two populations from the UK and two from Sweden. Previous reports indicate that A. cardamines larvae are found on different host plant species in different regions of the UK, and some variation has been reported in Sweden. Host plant choice trials showed that females prefer to oviposit on plants in an earlier phenological stage, as well as on larger plants. When controlling for plant phenological stage and size, the host species had no statistically significant effect on the choice of the females. Moreover, there were no differences in host plant species preference among the four butterfly populations. Based on our experiment, the oviposition choice by A. cardamines mainly depends on the phenological stage and the size of the host plant. This finding supports the idea that the geographical patterns of host-plant association of A. cardamines in the UK and Sweden are consequences of the phenology and availability of the local hosts, rather than regional genetic differences in the host species preference of the butterfly.