Butterflies (Lepidoptera) are potentially good indicators of climate change because they are short-lived, highly visible, and can respond rapidly to changes in habitat conditions. Range shifts of butterflies associated with regional warming have been documented (Crozier Reference Crozier2004; Parmesan Reference Parmesan2006) and modelled (Hellman Reference Hellman2001). Such studies of butterflies are also appropriate for gauging the effects of climatic change in Arctic Canada.

Documentation of butterflies is largely limited to reconnaissance data, but does show that butterfly fauna of the northwest Canadian tundra is particularly diverse (Layberry et al. Reference Layberry, Hall and Lafontaine1998). The location of our research, Herschel Island (Qikiqtaryuk), Yukon, lies within this region. Our study assesses: (i) whether butterfly species have changed their distributions in this portion of the tundra; (ii) the relative abundance and habitat affinity of butterflies; and (iii) factors affecting the interannual variability in abundance.

Field work took place on Herschel Island (69°35′20′′N, 139°05′20′′W), a land mass of ∼116 km² situated 3 km north of mainland Yukon in the Beaufort Sea. From 2007 to 2009, we collected new butterfly records on Herschel Island, comprised of voucher specimens deposited at the Royal British Columbia Museum, Victoria, British Columbia, Canada (ENT-010-000637 to ENT-010-000909) and photographs of species that were not collected. The coordinates and dominant plant species were noted for each site where a specimen was collected. These were cross-referenced to vegetation types in Smith et al. (Reference Smith, Kennedy, Hargrave and McKenna1989) and used to assign habitat types to each voucher specimen. Concentrated sampling effort was limited to 20 June 2007–6 July 2007, 3 July 2008–13 July 2008, and 4 July 2009 to 13 July 2009, but field crews recorded occasional observations from early June through September.

Historic butterfly records for Herschel Island were compiled from a database detailing butterflies collected from the Yukon (Guppy Reference Guppy2010). This database is publicly available from Environment Canada.

We sourced temperature data from the National Climate Data and Information Archive of Environment Canada (www.weatheroffice.gc.ca). Very few weather records are available for Herschel Island prior to 1994. Thus, Komakuk Beach (69°35′41′′N, 140°10′25′′W), located 40 km west of Herschel Island, was used as a proxy for comparing historical temperatures to recent temperatures. Daily weather data are available for Komakuk Beach beginning July 1958. Temperatures at Komakuk closely approximate those on Herschel Island (Burn Reference Burn2012). Snow persistence was measured on Herschel Island from mid-May to mid-June of 2008 and 2009 by estimating snow cover from a fixed vantage point every one to two days.

On 7 July 2008 and 5 July 2009, all butterflies encountered along a 14-m wide transect on the east side of Herschel Island were counted. The same 2.2 km transect was followed both years, but in 2009 this was extended an additional 3 km through a similar composition of vegetation types to increase the sampling effort. Six groups were confidently discerned by colour pattern and size. These were yellow sulphurs (Colias hecla Lefèvbre (Lepidoptera: Pieridae) or Colias canadensis Ferris (Lepidoptera: Pieridae)), green sulphurs (Colias nastes Boisduval (Lepidoptera: Pieridae) or Colias tyche (de Böber) (Lepidoptera: Pieridae)), blues (Lycaeides idas (Linnaeus) (Lepidoptera: Lycaenidae)), fritillaries (Boloria Moore species (Lepidoptera: Nymphalidae)), alpines (Erebia Dalman species (Lepidoptera: Nymphalidae)), and arctics (Oeneis Hübner species (Lepidoptera: Nymphalidae)). Encounter rates estimate the relative abundance of the butterfly groups.

Previous butterfly records from Herschel Island consisted of 15 species (Table 1), with collection dates spanning 1916–1983 (Guppy Reference Guppy2010). We found a total of 19 species of butterflies on Herschel Island from 2007 to 2009 (Table 1). New records are documented by voucher specimens, except Papilio machaon Linnaeus (Lepidoptera: Papilionidae) and Nymphalis l-album (Esper) (Lepidoptera: Nymphalidae), which are substantiated by photographs. Six of these were new species records for Herschel Island (Table 1). Except for Lycaena phlaeas (Linnaeus) (Lepidoptera: Lycaenidae), which has been well documented in Canada's Arctic archipelago (Layberry et al. Reference Layberry, Hall and Lafontaine1998), Herschel Island is the northernmost location for these six species. The northward latitudinal extension was greatest for N. l-album, which was previously recorded about 300 km south of Herschel Island. Nymphalis antiopa (Linnaeus) (Lepidoptera: Nymphalidae) and C. canadensis had previously been recorded 30–100 km further south, in the British Mountains inland from the Yukon north coast. Papilio machaon and Pontia occidentalis (Reakirk) (Lepidoptera: Pieridae) were historically in close proximity to Herschel Island, having been recorded along the nearby Arctic coast of mainland Yukon.

Table 1 List of butterfly species for Herschel Island, Yukon, Canada indicating which species were newly recorded and which were not re-found.

Note: Percent of total individuals of each species in each of nine vegetation types (Smith et al. Reference Smith, Kennedy, Hargrave and McKenna1989) using data from voucher specimens collected from 2007 to 2009.

The greater number of days conducive to butterfly flight could explain the new northern records of butterfly species. The percent of days during which temperatures were adequately warm for butterfly flight (maximum temperatures exceeding 12 °C for June and July) was higher from 2007 to 2009 than 1958–1983 (χ² = 4.01, df = 1, P < 0.05). This difference was less evident when considering maximum daily temperatures >16 °C (Fig. 1). Daily temperatures in June were significantly warmer for 2007–2009 ( $$$\bar{x}$$$ = 5.0 °C) than for 1959–1983 ( $$$\bar{x}$$$ = 3.4 °C; t = 4.16, df = 114, P < 0.001). July temperatures were also significantly warmer for 2007–2009 ( $$$\bar{x}$$$ = 8.7 °C) than for 1959–1983 ( $$$\bar{x}$$$ = 7.4 °C; t = 3.57, df = 126, P < 0.001).

Fig. 1 The percent of total days during which the maximum temperature exceeded the colour-coded temperature at Komakuk Beach, Yukon, Canada. Note that data were missing for June 1962 and July 1966.

We did not find two species previously collected on Herschel Island: C. palaeno Linnaeus (Lepidoptera: Pieridae) and Erebia rossii Curtis (Lepidoptera: Nymphalidae). These had been collected in an area that we also covered in the 2007–2009 surveys, so they may actually be absent or rare on Herschel Island now. Alternatively, it is possible that C. palaeno had yet to emerge by early July when the 2007–2009 surveys were underway. The previous records for C. palaeno were dated 1 and 4 August 1953, during a collection period beginning on 3 July, suggesting a late emergence of this species (Guppy Reference Guppy2010).

The six newly found species could conceivably reproduce on Herschel Island, judging by the distribution of host plants (Table 2). Known larval food plants for L. phlaeas, N. antiopa, N. l-album, and P. machaon grow on Herschel Island (Smith et al. Reference Smith, Kennedy, Hargrave and McKenna1989). Legumes (Fabaceae) are the suspected larval food plants for C. canadensis (Layberry et al. Reference Layberry, Hall and Lafontaine1998) and are also common on Herschel Island. The known food plants of P. occidentalis have not been documented on Herschel Island or the adjacent mainland, but the closely related Thlaspi arcticum Porsild (Brassicaceae) has, along with a variety of other mustards (Smith et al. Reference Smith, Kennedy, Hargrave and McKenna1989, Cody 2000). The vegetation types associated with each species found on Herschel Island from 2007 to 2009 are given in Table 1.

Table 2 A list of host plants associated with the six newly found butterfly species on Herschel Island.

The new records of N. antiopa and N. l-album are of particular interest because they overwinter as adults. The sightings of N. antiopa were consistently near shrubby willows (Salix Linnaeus (Salicaceae) species) along creeks, both in early and late summer, suggesting that they are breeding and possibly overwintering on Herschel Island. Nymphalis l-album is characterised as a woodland species, overwintering above ground in structures such as tree cavities (Layberry et al. Reference Layberry, Hall and Lafontaine1998). With only one sighting on Herschel Island and unsuitable overwintering sites, it is likely that this species was a vagrant to the Arctic coast in 2008, the same year that an unusually large number were sighted elsewhere in the Yukon (S. Cannings, Environment Canada, Whitehorse, Yukon, Canada, personal communication).

Within 2007–2009, we observed interannual variability in butterfly abundance and we related differences in abundance to weather variables, similar to historical records versus recent records, but at a smaller time scale. Not only were butterflies more abundant in 2008, but their phenology was more advanced (Fig. 2). The lower encounter rate of alpines and higher encounter rate of sulphurs in 2008 indicate that the alpines were at the tail end of their flight season, whereas the sulphurs were peaking in numbers. The species composition of the fritillaries also indicates that butterfly emergence was later in 2009 when Boloria improba (Butler) (Lepidoptera: Nymphalidae), an early emerging species, dominated the fritillaries found. In contrast, Boloria chariclea (Schneider) (Lepidoptera: Nymphalidae), a later emerging species, was the most common of the fritillaries in 2008. A noteworthy aside is that only Boloria polaris (Boisduval) (Lepidoptera: Nymphalidae) appeared to have a strongly synchronous two-year life cycle, with markedly few in flight in 2008 compared with 2007 and 2009.

Fig. 2 The encounter rate of different groupings of butterflies observed along a transect run in 2008 and 2009 on eastern Herschel Island, Yukon, Canada. See text for description on how species were grouped.

The differences in snow persistence (Fig. 3) and the number of hours of warmer air temperature among years (Fig. 4) probably explain much of the difference in phenology and hence relative abundance and species composition at a set point in summer. Persistent snow cover can delay adult emergence, especially for those species emerging in moist habitats characteristic of topographical depressions. These include the Erebia species and Oeneis species, whose larvae feed on sedges.

Fig. 3 Snow persistence in spring 2008 and 2009 on Herschel Island, Yukon, Canada.

Fig. 4 The number of hours per day above the indicated colour-coded temperatures, on Herschel Island, Yukon, Canada.

The development of butterflies into adults can be accelerated by warmer temperatures (Scott Reference Scott1986). This may have occurred in the latter half of June 2008 when air temperatures commonly exceeded 12 °C (Fig. 4). By the time sampling of butterflies began in July, many had already emerged and were flying for many days of their short adult lives. In contrast, temperatures in June and early July of 2009 rarely exceeded 12 °C. Not only had fewer emerged, but it is likely that there were fewer flying as air temperature is positively correlated with flight activity (Pollard and Yates Reference Pollard and Yates1993) and it was not possible to find a day in 2009 with equivalent warmth as 2008. Temperature during counts ranged from 13 °C to 16 °C in 2008, whereas they ranged from 9 °C to 11 °C in 2009. Butterflies were rarely seen below 8 °C, but it is possible that butterflies adapted to northern latitudes can prolong their adult lifespan by remaining inactive when weather is too cold for flying, feeding, and finding mates. An inadvertent discovery of cold tolerance was made when specimens were placed in a −10 °C freezer to kill for mounting. Individuals of Colias, Oeneis, and Erebia survived up to five days, indicating that adults are adapted to sub-zero conditions in summer.

Northward range shifts of butterflies have been well documented where long-term monitoring programmes are in place (Parmesan et al. Reference Parmesan, Ryrholm, Stefanescu, Hill, Thomas and Descimon1999; Warren et al. Reference Warren, Hill, Thomas, Asher, Fox and Huntley2001; Hill et al. Reference Hill, Thomas, Fox, Telfer, Willis and Asher2002). Our observations also suggest that butterfly species are moving northwards in low Arctic Canada. Since soils and vegetation were first mapped on Herschel Island (Smith et al. Reference Smith, Kennedy, Hargrave and McKenna1989), more prolific growth has taken hold where ground was previously sparsely vegetated (Kennedy et al. Reference Kennedy, Smith and Cooley2001), and willows have proliferated (Myers-Smith et al. Reference Myers-Smith, Hik, Kennedy, Cooley, Johnstone and Kenney2011). These changes, coupled with the fact that Herschel Island is the northernmost tundra region in Yukon, may continue to influence the relative abundance of species there.

Annual systematic collection of butterfly observations akin to the British Butterfly Monitoring Scheme (Pollard and Yates Reference Pollard and Yates1993) does not exist in northern Canada. Instead, projected changes in distribution and abundance have been modelled based on presence/absence data collected from multiple studies (Kerr Reference Kerr2001; Peterson et al. Reference Peterson, Martinez-Meyer, Gonzalez-Salazar and Hall2004; Kharouba et al. Reference Kharouba, Algar and Kerr2009). It is essential that such models be reevaluated periodically, as climate change in the Arctic is more rapid than the global average (Zhang Reference Zhang2005; Kittel et al. Reference Kittel, Baker, Higgins and Haney2010), affects numerous ecosystem processes (Arctic Monitoring and Assessment Programme 2011), and its influence may not be linear. Although the current increased vegetation cover and growth may provide enhanced habitat for some butterflies, these changes may not be a persistent benefit of climate change. Earlier snowmelt, melting permafrost, and warmer summers are making the Arctic tundra drier and more vulnerable to fire, which has been a very rare natural disturbance (Hu et al. Reference Hu, Higuera, Walsh, Chapman, Duffy and Brubaker2010). Fire and limited moisture would significantly reduce the distribution and vigour of many butterfly food plants, sometimes for considerable periods, thereby altering the spatial array of suitable habitats.