International Journal of Biometeorology

, Volume 47, Issue 4, pp 188–192

Spatial trends in the sighting dates of British butterflies


    • Centre for Ecology and Hydrology, Monks Wood, Abbots Ripton, Huntingdon, Cambridgeshire, PE28 2LS, UK
    • University of Southampton, Bassett Crescent East, Southampton, Hampshire SO16 7PX, UK
  • J. Asher
    • Butterfly Conservation, Manor Yard, East Lulworth, Wareham, Dorset, BH20 5QP, UK
Original Article

DOI: 10.1007/s00484-003-0170-6

Cite this article as:
Roy, D.B. & Asher, J. Int J Biometeorol (2003) 47: 188. doi:10.1007/s00484-003-0170-6


A strong relationship between appearance dates and temperature has been demonstrated over two decades for most British butterflies. Given this relationship over time, this paper tests whether comparable spatial trends in timing are also apparent. A major survey of British butterflies is used to calculate mean sighting dates of adults across the country, and these are compared with geographic patterns in temperature. With the use of regression techniques, we calculated latitudinal (south–north) and longitudinal (east–west) gradients in sighting date and temperature. The majority of butterflies appear later in the east of Britain where temperatures are lower during summer, but not the rest of the year. Most butterflies are also seen later in the cooler north of the country, by upto 3–4 days/100 km. However, no geographical relationship between temperature and timing of appearance was detected for over a third of the species analysed, suggesting their populations may be adapted to their local climates. We suggest possible mechanisms for this and discuss the implications of such adaptation for the ability of butterfly species to respond to rapid climate warming.


PhenologyClimate changeButterfly populationsButterfly censusTemperature


Recent evidence suggests that responses to global warming are consistent across a range of taxonomic groups, organisational levels and throughout all major biomes (e.g. Hughes 2000; Walther et al. 2002). Phenological changes have proved particularly sensitive (Peñuelas and Filella 2001) and well studied, with numerous aspects of plant and animal life cycles showing marked trends with warmer temperatures across Europe and North America (e.g. Bradley et al. 1999; Menzel and Fabian 1999).

Most phenological studies report climate-related changes over time in events such as bird migration (Sparks 1999) and egg laying (Crick and Sparks 1999), plant growth and flowering (Abu-asab et al. 2001) and insect life cycles (Zhou et al. 1995). Similarly, on average, the appearance of British butterflies has advanced by 2–10 days/1 °C increase in temperature over a 23-year period (Roy and Sparks 2000).

Fewer studies have examined spatial trends in phenological events but, given the often-reported strong relationship between timing and temperature, it is expected that events such as the appearance of adult butterflies will occur later in the north of their range than in the warmer south. However, previously published data for the butterfly Pyronia tithonus (Brakefield 1987; Pollard 1991) show no clear trend in the timing of the flight period with latitude over its British range. Conversely, anecdotal evidence suggests that other butterfly species fly later in the north, as expected (Warren 1992). The aim of this paper is to quantify spatial trends in butterfly phenology using data from a major survey of their distribution in Britain, to determine the extent to which appearance is synchronized across regions.

Materials and methods

Latitudinal and longitudinal trends in temperatures across Britain were calculated using 10-km × 10-km climate summaries available from the UK Climate Impacts Programme (Hulme and Jenkins 1998). Monthly, seasonal and annual mean temperatures were calculated for each 100-km × 100-km area for which butterfly sighting data were available. Trends in temperature were calculated by multiple linear regressions with easting and northing as explanatory variables. The temperature data used are the means of values for 1960 to 1990; more recent, comparable climate data were not available. The spatial trends in temperatures across the country are likely to be applicable to recent years, and these data have been widely used in analyses of the response of British butterflies to climate change (e.g. Hill et al. 1999; Warren et al. 2001).

The timing of butterfly sightings across Britain was derived from distributional records (1.5 million records) collected over a 5-year period, 1995–1999 (Asher et al. 2001). The mean sighting date per 100-km × 100-km grid square was calculated to reduce local effects of variation in recording intensity. Recording continued throughout periods where adult butterflies were active and the mean sighting date is taken to represent an average flight-period time. Species with more than one generation per year were excluded to overcome the difficulty in separating generations for multivoltine species. Non-resident species such as Vanessa atalanta and Cynthia cardui were also excluded as their appearance within Britain is dependent on conditions in breeding areas further south. Twenty-nine remaining species were suitable for analysis. Trends over space in mean sighting date were examined using linear regression with easting and northing as separate explanatory variables, but models were weighted by the number of sightings per grid square to accommodate variation in recording intensity and density.


Unsurprisingly there are clear east–west and north–south gradients in temperature across Britain (Table 1) reflecting global latitudinal gradients from the equator to the pole. Annual, monthly and seasonal temperatures are constantly warmer by approximately 0.4 °C/100 km from the north to the south. Longitudinal gradients in temperature are not consistent throughout the year, however. The west is warmer than the east in the winter and spring months, by up to 0.42 °C/100 km, but the gradient is reversed during the summer months, with a less marked gradient of 0.23 °C/100 km in July.
Table 1.

Spatial trends (1960–1990) in temperature. TEAST and TNORTH are the regression coefficients (SE) from a multiple linear regression of east–west and south–north effects (per 100 km) on monthly, seasonal and annual temperature. Seasons are defined as 3-month periods with winter as November, December, January. n = 47








–0.42 (0.08)***

–0.33 (0.04)***



–0.32 (0.08)***

–0.31 (0.04)***



–0.16 (0.06)*

–0.31 (0.03)***



–0.07 (0.03)

–0.32 (0.03)***



0.06 (0.04)

–0.34 (0.02)***



0.16 (0.03)***

–0.36 (0.02)***



0.23 (0.04)***

–0.43 (0.02)***



0.22 (0.04)***

–0.43 (0.02)***



0.13 (0.06)*

–0.43 (0.03)***



–0.29 (0.08)**

–0.41 (0.04)***



–0.42 (0.08)***

–0.36 (0.05)***





–0.39 (0.04)***

–0.34 (0.04)***



–0.06 (0.05)

–0.33 (0.03)***



0.21 (0.04)***

–0.41 (0.02)***



–0.07 (0.07)

–0.42 (0.04)***



–0.08 (0.05)

–0.37 (0.03)***


*P < 0.05, **P < 0.01, ***P < 0.001

There is a significant trend towards later sighting dates in the north for over a third of the butterflies analysed (Table 2) reflecting the north–south temperature gradient in Britain (Table 1). Habitat specialists with a northern (e.g. Limentis camilla, Argynnis aglaja) or southern range margin in Britain (e.g. Coenonympha tullia, Aricia artaxerxes), as well as wider-ranging species (Anthocharis cardamines, Fig. 1a; Thymelicus sylvestris) are seen as adults earlier in the southern than in the northern parts of their range. These species also have a range of life histories, overwintering as eggs, caterpillars or chrysales. A further 11 species had a positive, but non-significant, trend towards earlier sighting dates in the south. Callophrys rubi was the only species with a significantly negative relationship between timing and distance north (Table 2). This species has an extended flight period, and late mean flight date, in the most southerly parts of its range (Asher et al. 2001). Ten species have a negative relationship between northing and mean sighting date and there is little variation in the mean sighting date of P. tithonus (Fig. 1b) from the south to north parts of its range, as noted by previous authors (Brakefield 1987; Pollard 1991).
Table 2.

Spatial trends (1995–1999) for mean date (Julian day) of butterfly sightings per 100-km × 100-km grid cell of the British Ordinance Survey grid. n is the number of grid cells analysed; BEAST and BNORTH are the regression coefficients (SE) from a multiple linear regression of east–west and south–north effects (per 100 km) on mean sighting date


Common name





Thymelicus sylvestris

Small skipper


–0.94 (0.41)*

2.34 (0.38)***


Thymelicus lineola

Essex skipper


–0.60 (0.79)

0.82 (0.74)


Ochlodes venata

Large skipper


2.13 (0.54)***

–0.42 (0.45)


Erynnis tages

Dingy skipper


1.26 (1.94)

1.42 (1.30)


Pyrgus malvae

Grizzled skipper


–0.01 (1.05)

0.36 (1.11)


Anthocharis cardamines

Orange tip


0.54 (0.33)

2.15 (0.20)***


Callophrys rubi

Green hairstreak


–2.84 (0.83)**

–2.07 (0.49)***


Lysandra coridon

Chalkhill blue


–1.59 (0.77)

–2.10 (0.91)


Limenitis camilla

White admiral


–1.21 (0.39)**

1.11 (0.47)*


Clossiana selene

Small pearl-bordered fritillary


0.97 (1.24)

2.19 (0.46)***


Clossiana euphrosyne

Pearl-bordered fritillary


0.79 (1.21)

3.75 (0.55)***


Argynnis aglaja

Dark green fritillary


1.62 (0.88)

1.22 (0.42)**


Argynnis paphia

Silver-washed fritillary


–2.80 (0.77)**

0.88 (0.80)


Melanargia galathea

Marbled white


0.39 (0.76)

–0.16 (0.73)


Hipparchia semele



2.02 (0.58)**

–0.32 (0.32)


Pyronia tithonus



–0.80 (0.29)*

0.16 (0.31)


Maniola jurtina

Meadow brown


–0.70 (0.46)

–0.08 (0.25)


Aphantopus hyperantus



0.70 (0.39)

0.36 (0.26)


Apatura iris

Purple emporer


–2.78 (0.96)

–1.00 (1.15)


Aricia artaxerxes

Northern Brown argus


1.70 (1.85)

3.65 (0.94)**


Coenonympha tullia

Large heath


2.18 (1.32)

2.64 (0.62)***


Erebia aethiops

Scotch argus


–0.07 (0.97)

1.15 (0.56)


Eurodryas aurinia

Marsh fritillary


–6.19 (2.31)*

–1.50 (1.26)


Hamearis lucina

Duke of Burgundy


–0.94 (1.09)

1.51 (0.83)


Hesperia comma

Silver-spotted skipper


0.22 (1.09)

–1.55 (2.83)


Plebejus argus

Silver-studded blue


1.52 (1.13)

–1.64 (1.57)


Strymonidia w-album

White letter hairstreak


–1.46 (1.00)

2.55 (0.76)**


Thecla betulae

Brown hairstreak


0.27 (2.30)

0.32 (3.11)


Quercusia quercus

Purple hairstreak


–2.25 (0.60)**

2.30 (0.46)***


Number of significantly negative relationships




Number of significantly positive relationships




*P < 0.05, **P < 0.01, ***P < 0.001

Fig. 1a, b.

Geographic patterns in the mean dates (Julian day) of butterfly sightings for (a) Anthocharis cardamines and (b) Pyronia tithonus within a 100-km × 100-km grid square of the British Ordinance Survey grid. Symbols, in decreasing size mean dates: a 206–210, 210–212, 212–214, 214–216, 216–219; b 125–130, 130–135, 135–140, 140–145, 145–151

Mean sighting dates are earlier in the east for most species (Table 2), and significantly so for almost a quarter of the species. The reverse pattern, with significantly earlier sighting dates in western Britain, is found in only two species, Ochlodes venata and Hipparchia semele. Temperatures are warmer in eastern Britain during summer, but not during the rest of the year (Table 1) and we may expect this to affect the timing of appearance differently for spring- and summer-emerging species. However, the directions of east–west relationships do not relate to the overall timing of the flight period. For example, spring-emerging species such as Anthocharis cardamines and Pyrgus malvae do not appear significantly earlier in the west even though autumn and spring temperatures are warmer in this part of the country during important periods for the development of immature stages for these species.


As expected, the flight period of most butterflies is earlier in the warmer south than in the cooler north of Britain. This relationship is predicted by the year-to-year correspondense between temperature and timing of butterfly appearance reported for most British species (Roy and Sparks 2000), and the tendency for timing to be later at high, and cooler, altitudes (Gutiérrez and Menéndez 1998). However, for a number of species, the mean sighting dates recorded appear to be synchronized across latitude, supporting the findings of Brakefield (1987) and Pollard (1991) for P. tithonus; the latitudinal gradient in temperature is not mirrored by butterfly phenology. This finding raises important questions, such as how is the synchrony of flight periods achieved? And what are the implications for butterfly populations under climate warming?

Synchrony in appearance across temperature gradients may provide evidence for local adaptation of butterfly populations to regional climates, and possible mechanisms include behavioural, morphological, physiological and developmental characteristics.

Lepidopteran larvae can attain temperatures 5–20 °C above the ambient temperature by adjusting posture and orientation, exploiting thermal heterogeneity within the environment, and minimizing convective heat losses (Weiss et al. 1988). Larvae can also exhibit positive phototaxy within the host canopy, which tends to put them in high-radiation microsites. These behaviours may be well developed in cooler parts of a species' range. Butterfly populations are also more localised at the northern range margin (Asher et al. 2001) as the suitable habitat patches available are typically smaller, more isolated and short-lived (Bourn and Thomas 2002; Thomas et al. 1999) than those at its core. Thomas (1993) argues that many species of insect in the Palaearctic have been living a few hundred kilometers north of their "natural" climatic limits during recent centuries or millennia because of an ability to exploit unnaturally warm microclimates generated within semi-natural biotopes by traditional forms of agriculture and silviculture. In contrast, the same species occupy different and broader niches near their centres of range, particularly the mid–late seral stages of ecosystems (Thomas 1991, 1993). Such exploitation of warmer microclimates in the north may allow comparable larval development times, and result in the emergence dates of adults being synchronised between populations in different parts of a species' range.

Compensation for cooler temperatures may also be achieved by reduced size in the north of the range. Ayres and Scriber (1994) report smaller larvae and adults in Alaskan and Michigan populations of P. canadensis, and a similar size cline has been suggested for a number of butterflies in Sweden, individuals measured from museum collections being smaller in the north of the country as a result of a shorter growing season and therefore reduced development time (Nylin and Svard 1991). However, butterflies may compensate for limited development time in seasonal environments by accelerated growth rates. Several species of fish exhibit faster increases in growth with temperature in northern than in southern populations (Conover and Present 1990; Schultz et al. 1996) and similar results have been documented for ectothermic organisms (Conover and Schultz 1995; Nylin and Gotthard 1998). The adaptive explanation for such a counter-gradient in growth rate is that temperatures favourable for growth and development occur during a shorter period in northern areas, yet high growth rates are associated with fitness costs (Conover and Present 1990). An analysis of the phenology of four species of butterfly in Sweden demonstrated that the growth rate of these species is indeed finely tuned to the season, but also that larvae can both hibernate and aestivate when extra time is available (Wickman et al. 1990).

If local adaptation to temperature occurs widely, as suggested here, this has implications for the conservation of butterflies by introduction from one locality to another where extinction has occurred. Butterflies moved from a cooler to a warmer locality may emerge earlier in the season with possible consequences for survival.

Similarly, a locally adapted butterfly may be unable to cope with the rapid climate warming predicted for much of Europe. In both cases, there may be loss of synchrony with important resources, such as larval foodplants becoming available in the right condition, or key nectar resources. Although detailed study has shown species to respond to climate change at similar rates in the same system (Buse and Good 1996), loss of synchrony can have detrimental effects on populations (Visser et al. 1998). Conversely, climate warming may be beneficial to prey species if their predators and/or pathogens respond differently. Evidence suggests that mobile, wide-ranging butterflies are already expanding in range and increasing in abundance within Britain as a result of climate warming, yet sedentary habitat-specialist species, e.g. those more likely to be locally adapted, are becoming increasingly restricted owing to habitat loss and degredation (Roy et al. 2001; Warren et al. 2001). It is unclear which mechanism(s) allow synchronised emergence of butterflies in Britain, and different strategies may be operating in different species, yet this intriguing phenomenon deserves further attention to enable more accurate predictions of the future response of butterflies to climate warming.


We are indebted to the dedicated recorders who submitted records to the Butterfly for the New Millenium project and made this work possible. We thank Tim Sparks for helpful discussions and Arnold van Vliet and the European phenology network (EPN) for the opportunity to present this work. D.R. received funding from the CEH Integrating fund (round 7).

Copyright information

© ISB 2003