SEARCH

SEARCH BY CITATION

Keywords:

  • altitude;
  • biotic interactions;
  • climate change;
  • phenology;
  • range shifts

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • 1
    The ranges of many species have expanded in cool regions but contracted at warm margins in response to recent climate warming, but the mechanisms behind such changes remain unclear. Particular debate concerns the roles of direct climatic limitation vs. the effects of interacting species in explaining the location of low latitude or low elevation range margins.
  • 2
    The mountains of the Sierra de Guadarrama (central Spain) include both cool and warm range margins for the black-veined white butterfly, Aporia crataegi, which has disappeared from low elevations since the 1970s without colonizing the highest elevations.
  • 3
    We found that the current upper elevation limit to A. crataegi's distribution coincided closely with that of its host plants, but that the species was absent from elevations below 900 m, even where host plants were present. The density of A. crataegi per host plant increased with elevation, but overall abundance of the species declined at high elevations where host plants were rare.
  • 4
    The flight period of A. crataegi was later at higher elevations, meaning that butterflies in higher populations flew at hotter times of year; nevertheless, daytime temperatures for the month of peak flight decreased by 6·2 °C per 1 km increase in elevation.
  • 5
    At higher elevations A. crataegi eggs were laid on the south side of host plants (expected to correspond to hotter microclimates), whereas at lower sites the (cooler) north side of plants was selected. Field transplant experiments showed that egg survival increased with elevation.
  • 6
    Climatic limitation is the most likely explanation for the low elevation range margin of A. crataegi, whereas the absence of host plants from high elevations sets the upper limit. This contrasts with the frequent assumption that biotic interactions typically determine warm range margins, and thermal limitation cool margins.
  • 7
    Studies that have modelled distribution changes in response to climate change may have underestimated declines for many specialist species, because range contractions will be exacerbated by mismatch between the future distribution of suitable climate space and the availability of resources such as host plants.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The ranges of many species have expanded at high latitudes and elevations but contracted at their warm margins in response to recent climate change (Walther et al. 2002; Thomas, Franco & Hill 2006). Changes to the distribution of ‘habitable climate space’ may lead to extinctions if future ranges are too small or isolated from current ranges, and consequently the impact of climate change on biodiversity is of increasing concern (Thomas et al. 2004). The mechanisms that determine geographical range margins remain poorly understood, limiting our ability to predict future species distributions and to model the effects of climate change on biodiversity (Pearson & Dawson 2003; Ibáñez et al. 2006).

Climate can limit distributions directly by influencing survival and fecundity, or indirectly through its effects on interacting species, including food sources, natural enemies and competitors (Gaston 2003). The prevailing view that climate limits distributions directly at cool, upper latitude range margins is supported by considerable empirical evidence (see references in Gaston 2003). An increasing number of examples show how climate warming has allowed range expansions at high latitudes by reducing mortality (Crozier 2004; Walther, Berger & Sykes 2005) or increasing fecundity (Davies et al. 2006). In contrast, there is little evidence for direct climate limitation at warm margins, where biotic interactions are believed to be more important (MacArthur 1972; Brown, Stevens & Kaufman 1996; Parmesan et al. 2005). Considering that the first symptoms of biodiversity loss are expected at lower elevational and latitudinal boundaries, empirical evidence for the mechanisms that limit species distributions in these areas is of particular importance for modelling future species ranges and for adapting biodiversity conservation to climate change (Hampe & Petit 2005).

Phytophagous insects and their host plants are useful model systems for testing the effects of climate and biotic interactions on species distributions (Hodkinson 1999; Bale et al. 2002), and have provided some of the first evidence for the climatic mechanisms behind population extinctions at warm range margins (Parmesan 1996, 2005; McLaughlin et al. 2002). The progressive restriction of phytophagous insects to warmer microhabitats is a well-documented explanation for the locations of cool range boundaries (Thomas 1993; Thomas et al. 1999), but evidence for the reverse pattern at warm margins is lacking. In this study we determine the relative roles of climate and larval host plants in determining distribution and abundance for an oligophagous butterfly across a naturally occurring thermal gradient incorporating both warm (low elevation) and cool (upper elevation) range limits. We first determine changes to the elevational range of the black-veined white Aporia crataegi L. after 30 years of climate warming, and test for differences between the current range of the species and its larval host plants. We then investigate how climatic effects on the phenology, habitat associations and survival of A. crataegi could cause its elevational range to be narrower than that of its host plant. In contrast to prevailing explanations for range margins, we find stronger evidence for direct climate limitation at the warm margin of the species distribution and for host plant limitation at the cool margin.

Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

study system

The black-veined white A. crataegi L. is a widespread Palearctic butterfly, distributed from north-west Africa and western Europe to east Asia (40–70° N) and Japan (Tolman & Lewington 1997). In Europe, the species has expanded its range in Scandinavia but has also suffered serious declines, including extinctions from the United Kingdom in the 1920s and recently from the Czech Republic and the Netherlands (Asher et al. 2001). At its south-western range margin, in central and southern Spain and in North Africa, A. crataegi is restricted to high elevations in mountains (García-Barros et al. 2004; personal communication, J. Tennent, 2007).

A. crataegi has one annual flight period, and females lay batches of up to c. 100 eggs on the leaves of both wild and cultivated Rosaceae (Emmet & Heath 1989). In particular, the species uses Crataegus spp. (hawthorn) and Prunus spinosa L. (blackthorn). Eggs hatch after approximately 2 weeks, and larvae live gregariously in a silken web from which they emerge to feed. The web is converted into a nest, constructed from leaves and silk, in which larvae over-winter in the second instar. In spring, larvae continue feeding in groups on the host plant before dispersing and pupating alone (Emmet & Heath 1989).

The study location was the Sierra de Guadarrama (central Spain), a 100 × 30 km mountain range located at c. 40°45′ N 4°00′ W, that rises to a maximum elevation of 2428 m from plains of ≥ 700 m to the north and ≥ 535 m to the south (Fig. 1). The main host plants for A. crataegi in the region are C. monogyna Jacq. and P. spinosa, although eggs have also been observed rarely on Rosa spp. (D. Gutiérrez, personal observation). Mean annual temperature in the region increased by 1·3 °C between 1967 and 1973 and 1997–2003 (equivalent to an uphill shift in isotherms of c. 225 m), and over the same time-period the lower elevational limits of 16 butterfly species with herbaceous host plants shifted uphill by an average of 212 m (Wilson et al. 2005).

image

Figure 1. Sample distribution for A. crataegi and its host plants in 1967–73 and 2006. Triangles show 1967–73 sites. Transect sites in 2006 are squares (C. monogyna and/or P. spinosa present) and circles (host plants absent). Filled symbols show sites where A. crataegi was observed, open symbols where absent.

Download figure to PowerPoint

elevational range of a. crataegi

The elevational range of A. crataegi was recorded in 1967–73 and 2004–06 at grassland, scrub and woodland sites that were visited repeatedly to record regional butterfly distributions. Data for 1967–73 (Monserrat 1976) are butterfly counts at 38 sites that were each visited five or more times in total, including at least once during the flight period of A. crataegi (mid-May to late July). In 2006, A. crataegi was counted every 2 weeks at 43 sites, stratified by elevation, of which 20 had also been sampled in 2004 and 2005. From May to August, butterflies were counted on standardized 500 m long × 5 m wide transects (Pollard & Yates 1993). The overall elevational range of sites was 640– 1860 m asl in 1967–73, and 550–2240 m in 2006. A. crataegi was considered to be present in either time-period at locations where two or more individuals were counted, and absent where no individuals were observed. Sites where only one individual was recorded were excluded from analyses, because sampled butterflies might be vagrants, rather than representatives of a local breeding population.

To test the relation between the elevational range of A. crataegi and its larval host species, the abundance of host plants was estimated at each of the 43 transect sites in 2006. The route of the 500 m transect was followed in August–September 2006 (before leaf fall) and the number of plants of C. monogyna, P. spinosa and Rosa spp. that occurred in the 5 m wide butterfly transect was recorded, to give a density of each species per 0·25 ha (500 × 5 m). If any of the plant species was not present in the 5 m wide transect, then the transect was repeated with increasing widths of 10 m, 20 m, and up to a maximum of 50 m width (i.e. 25 m on either side of the recorder): host plant density per 0·25 ha was then estimated based on the increased transect width. Host plant species are considered present at a site if they were found in transects of ≤ 50 m wide. When host plant species were not found in ≤ 50 m wide transects, further searches up to 100 m away from the transect route revealed plants at only one site (one individual of C. monogyna with abundant eggs, larvae and pupal cases of A. crataegi). In this case, we include C. monogyna as ‘present’ at the site (given its evident importance for the local population of A. crataegi), and estimate its density as 0·1 per 0·25 ha (i.e. as if the one plant had been included in a 50 m wide transect).

Universal Transverse Mercator (UTM) coordinates were recorded at least every 100 m along each transect using a handheld Garmin GPS unit. The coordinates were used to plot each transect in ArcView geographic information ststem (GIS) (ESRI 1996), and the average elevation of 100 m grid squares intercepted by the transect was determined using a digital elevation model (NASA/JPL-Caltech 2004). To determine the elevational associations of A. crataegi and its host plants, binary logistic regressions were carried out for presence (1) and absence (0) of each species against elevation (km) and elevation2. To test the validity of results based on logistic regression models, species’ elevational distributions were also fitted to Huisman–Olff–Fresco (HOF) models (Oksanen & Minchin 2002) (see Supplementary material, Appendix S1). For A. crataegi, logistic regression models were also fitted with the additional variables of host plant abundance and its interaction with elevation. For this analysis, host plant abundance included the total transect count of the two main hosts C. monogyna and P. spinosa, with Rosa spp. as a separate variable.

Linear regression was used to test for the effects of elevation on the density of A. crataegi and its host plants at sites where the respective species were present. The dependent variables were log-transformed counts of each species per 500 × 5 m transect. Normal rather than Poisson error structures were used because the data were highly over-dispersed. For A. crataegi, we used the sum of annual transect counts and included elevation, the density of larval host plants and their interaction as independent variables. As a further analysis, we calculated the density of A. crataegi per host plant (i.e. A. crataegi total count divided by the total count of C. monogyna and P. spinosa per 0·25 ha), and regressed this variable against elevation. We tested the validity of results based on the 43 intensive transects in 2006 by repeating analyses using data from the 20 transects that had also been walked every 2 weeks in 2004 and 2005.

elevational trends in phenology

The flight period of A. crataegi was summarized by calculating mean flight date and first appearance date at each transect (each measured as the number of days since 1 April). Mean flight date was calculated as the weighted mean date of transect counts (Stefanescu, Peñuelas & Filella 2003). Linear regressions for flight date were fitted with elevation (km) as an explanatory variable.

To estimate the temperatures experienced by A. crataegi during its flight period (that could affect habitat associations), Hobo dataloggers were located at transect sites in 2006, and the temperature was recorded every hour from the beginning of May onwards. Dataloggers were placed in semishaded locations 1·5 m above the ground, corresponding to typical A. crataegi egg-laying sites on shrubs. Daytime temperature for the entire flight period was calculated as mean temperature (°C) for the 12-hourly intervals of 0700– 1800 h GMT, over the 2 months from 15 May to 14 July (encompassing all A. crataegi observations in 2006). Daytime temperature during the peak flight period was estimated as the mean for the 29 days centred on the mean flight date at each occupied site.

egg-site location

Host plants were searched for eggs and larval webs at sites across and beyond the elevational range of A. crataegi. Twenty-three sites ranging in elevation from 580 m to 1780 m (14 of which corresponded to butterfly transect locations) were visited between 23 June and 12 July 2006. Repeated searching at elevations above 1800 m revealed no C. monogyna or P. spinosa, and no eggs were found on Rosa spp. at any sites. At each location the first 25 C. monogyna and P. spinosa plants encountered were searched for A. crataegi egg batches and larval webs. Where fewer than 25 plants were encountered, all available host plants were searched. The larval web is almost always on the same or an adjacent leaf to the oviposition site, and so is an acceptable measure of egg-site location. The height above the ground (H) and side of the host plant (S) were recorded for each egg batch or larval web. In the field S was recorded as the cardinal and ordinal points (i.e. north, north-east, east, etc.); data were converted to degrees from north for subsequent analyses (so that north = 0, both north-east and north-west = 45, both east and west = 90, etc.). The side of the leaf on which eggs were located (upperside or underside) was recorded for egg batches.

We hypothesized that at higher elevations eggs would be laid in warmer microhabitats: on the south side of plants, closer to the ground, and on the upper surface of leaves. Because egg batches from the same site are not statistically independent, summary values for each site were used in analyses. Dependent variables were mean values of S and H, and the proportion of egg batches on the upperside of the leaf (L). Generalized linear models were constructed for side of host plant (S) and height above ground (H, log-transformed) using elevation as a predictor variable. The proportion of egg batches laid on the leaf upperside (L) was tested using analysis of deviance, including both elevation and elevation2 to test for a possible curvilinear effect of elevation. Because the data were over-dispersed, the model for L was tested using quasi-binomial error structure and F-tests (Crawley 2002). To test whether the inclusion of sites with summary values based on few data had a large effect, analyses were repeated excluding sites where fewer than five egg batches/larval webs were recorded.

egg and larval survivorship

Trends in egg and young larval survivorship were investigated by transplanting egg batches onto host plants at different elevations. Ten female A. crataegi were caught at 1450 m and maintained in captivity outside at 1300 m elevation, where they were provided with semishaded C. monogyna plants for egg-laying. The number of eggs in each batch was recorded. Within 3 days of being laid, egg batches were transplanted to nine sites at approximately 100 m intervals between 900 m and 1700 m, corresponding to the elevational range of A. crataegi, and to a further location at 580 m, below the current low elevation limit. Leaf fragments with egg batches were attached using a drop of UHU® solvent-based glue to the upper surface of C. monogyna leaves, with two batches in shaded microhabitats and two in unshaded microhabitats on each tree. This was repeated for two trees at each site (resulting in eight experimental batches at each elevation) except at 580 m, where only one tree (four egg batches) was used because of a shortage of egg batches. In total, 76 experimental egg batches were established between 22 and 28 June 2006. The location of each egg batch was marked so that it could be relocated and identified.

The experimental sites were visited on 17 or 18 July to record mortality of eggs and young larvae. The criterion for egg batch survival was that living larvae were present. Sites with unhatched egg batches were revisited on 27 July in case larvae had not had time to emerge by the earlier visit. Egg batches were excluded from subsequent analyses if the leaf onto which they had been transplanted could not be recovered.

Survivorship of egg batches in shaded and unshaded treatments was compared using a paired t-test. Generalized linear models with binomial error structure were then used to test the effects of elevation on survivorship. The proportion of surviving batches at each experimental site was the dependent variable, with elevation as the predictor. The analysis was repeated excluding data for the 580 m site, which is outside the current elevational range of A. crataegi.

To determine whether Rosa spp. were potential hosts for A. crataegi in the region, 10 egg batches were collected from C. monogyna in the field on 5 July 2006. Half of these were transplanted onto the leaves of Rosa spp. and the remaining control batches onto C. monogyna. These egg batches were all placed within 100 m of each other with the same elevation (c. 1350 m) and aspect (south). Egg batches were attached to leaves on the south side of trees at approximately 1 m above ground and were revisited on 17 July.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

elevational range of a. crataegi

In 1967–73, A. crataegi was present at 17 sample locations (based on observations of two or more adults) and absent from 19 locations. The lowest elevation population was at 640 m (with a maximum count of 37 in one visit), and the highest at 1800 m (with a maximum of 6). In 2006, A. crataegi was observed at 26 of 43 transect sites; of these, four were excluded from analysis because only one individual was recorded. Sites with two or more adults ranged in elevation from 930 to 1765 m. The elevational range appeared to have shifted uphill from 1967 to 73– 2006, with the loss of populations below 900 m (Fig. 2a).

image

Figure 2. Proportion of occupied sites in 250 m elevation bands for A. crataegi and its host plants. (a) A. crataegi in 1967–73 (black) and 2006 (white); (b) C. monogyna (black) and P. spinosa (white) in the 43 transect sites in 2006. Number of samples per elevation band shown above each bar [in (a), n for 1967–73 and 2004–05 separated by hyphens]. Asterisks show 250 m bands where C. monogyna or P. spinosa were found, but not at transect sites.

Download figure to PowerPoint

The three potential host plants (C. monogyna, P. spinosa and Rosa spp.) were present, respectively, at 25, 17 and 38 of the transect sites. The distributions of the two main host plants extended to lower elevations than that of A. crataegi but were absent from high elevations: transects with C. monogyna ranged from 550 to 1535 m, and of P. spinosa from 840 to 1525 m (Fig. 2b). Additional field searches encountered C. monogyna up to elevations of 1792 m, and P. spinosa as low down as 580 m. Rosa spp. occurred at both higher and lower elevations than A. crataegi (elevational range 590–2040 m; see Supplementary material, Fig. S1).

The probability of occurrence of A. crataegi was not significantly related to elevation in 1967–73. In 2006, probability of occurrence of A. crataegi and of the three plant species peaked at mid-elevations, with significant positive effects of elevation (km) and negative effects of elevation2 in logistic regressions (Table 1). Logistic regressions for A. crataegi occurrence in 2004 and 2005 also showed significant positive effects of elevation and negative effects of elevation2 (results not shown). HOF models showed unimodal relationships of probability of occurrence with elevation for A. crataegi, P. spinosa and Rosa spp., but the best model for C. monogyna was a plateau with a probability of occurrence of > 75% up to 1600 m, which rapidly declined at higher elevations (see Supplementary material, Table S1).

Table 1.  Logistic regression models for the elevational range of Aporia crataegi, its main host plants and Rosa spp. Models for logit (probability of occurrence), showing number of presences (NP) and absences (NA); coefficients for elevation and elevation2; and summary statistics, with R2 based on Nagelkerke (1991)
SpeciesNPNAElevation (km) (coefficient)Elevation2 (coefficient)Intercept–2 LLR2× 2P
  1. Significance for terms in logistic regression models: ***P < 0·001, **P < 0·01, *P < 0·05, +P < 0·1.

A. crataegi221745·25**–16·84**–27·66**25·500·6927·92< 0·001
C. monogyna251818·60* –8·72**–7·81+34·480·5824·00< 0·001
P. spinosa172659·60*–25·87*–32·46*30·170·6427·55< 0·001
Rosa spp.38 534·42***–13·56***–14·97***11·100·7219·82< 0·001

Despite the differences between the distributions of A. crataegi and its host plants, the best-fitting logistic regression model for A. crataegi occurrence in 2006 included a positive significant effect of host plant count (C. monogyna + P. spinosa). The coefficient for host plant count was not significant (P = 0·27) but the removal of the term produced a significant change in log likelihood (P = 0·006) (overall model: logit probability of occurrence = –33·02 (± SE 12·68) + 50·12 elevation (km) (± 18·75) – 17·88 elevation2 (± 6·65) + 0·08 host plant density (± 0·07); –2 log likelihood = 19·94, R2 = 0·77, χ2 = 33·49, P < 0·001).

The mean annual count of A. crataegi at occupied transects ranged from 14·8 to 16·1 in 2004–06, with maxima per site ranging from 38 in 2004–72 in 2006. Linear regressions for A. crataegi counts at occupied sites showed significant effects of elevation (positive) and elevation2 (negative) (Table 2a; Fig. 3a). Counts of host plants (C. monogyna + P. spinosa) in sites where they occurred had a marginally non-significant unimodal relationship with elevation (Table 2a, Fig. 3b). Rosa spp. abundance also increased from low to mid-elevations, before decreasing at high elevations (see Supplementary material, Fig. S1). The density of A. crataegi per host plant increased significantly with elevation in 2004 and 2006 (Fig. 3c), but the relationship was marginally non-significant in 2005 (Table 2b).

Table 2.  Linear regressions for the density of Aporia crataegi, its main host plants and Rosa spp. at occupied sites against elevation, showing coefficients (± 1 SE). Elevation in km above sea level.
SpeciesNElevation(coefficient)Elevation2 (coefficient)InterceptR2FP
  1. Significance for terms in models: ***P < 0·001, **P < 0·01, *P < 0·05,P < 0·1, NSP > 0·1.

(a) Models for log10 counts at occupied sites
 A. crataegi (2004)1011·34 (± 3·71)*–4·10 (± 1·35)*–6·62 (± 2·52)*0·574·70·051
 A. crataegi (2005)11 8·46 (± 3·43)*–3·13 (± 1·20)*–4·54 (± 2·40)0·493·80·070
 A. crataegi (2006)2212·21 (± 3·34)*–4·78 (± 1·26)**–6·54 (± 2·17)**0·457·90·003
 C. monogyna+P. spinosa2710·71 (± 4·83)*–5·18 (± 2·20)*–4·27 (± 2·55)NS0·203·10·065
 Rosa spp.38 8·91 (± 1·70)***–3·44 (± 0·64)***–4·39 (± 1·07)***0·4614·6< 0·001
(b) Models for log 10A. crataegi density per host plant
 Year
 2004 9 3·72 (± 1·46)* –5·16 (± 1·92)*0·486·550·038
 200510 2·86 (± 1·28) –3·99 (± 1·67)*0·384·970·056
 200619 2·90 (± 1·04)* –3·47 (± 1·32)*0·317·840·012
image

Figure 3. Changes in the abundance of A. crataegi and its host plants with elevation. (a) A. crataegi annual abundance. (b) C. monogyna and P. spinosa density per 0·25 ha (c) A. crataegi density per host plant. Regression lines plotted based on the equations for 2006 in Table 2.

Download figure to PowerPoint

elevational trends in phenology

In 2006, the flight period of A. crataegi was later at higher locations (Fig. 4), whether analysed using mean flight or first appearance date. Data from the smaller sample of transects in 2004–05 confirmed this elevational delay, although mean flight date was not significantly related to elevation in 2005 (Table 3). Flight was approximately 30–40 days later per 1 km increase in elevation in 2006, whereas in 2004–05 the delay was 10–30 days per 1 km.

image

Figure 4. The relationship of A. crataegi mean flight date with elevation. Mean flight date (no. days since 1 April) on butterfly transects in 2006 is plotted against elevation (km) (regression equation shown in Table 3).

Download figure to PowerPoint

Table 3.  Linear regressions for Aporia crataegi mean flight date and first appearance date with elevation (km), showing coefficients (± SE). Date measured as number of days since 1 April
 NElevation (km)(coefficient)InterceptR2FP
  1. Significance for terms in models: ***P < 0·001, **P < 0·01, NSP > 0·1.

(a) Mean date
 20041024·6 (± 7·0)**52·5 (± 9·6)**0·6112·60·008
 200511 9·2 (± 6·7)NS64·9 (± 9·2)***0·17 1·9 0·204
 20062233·1 (± 6·6)***30·0 (± 8·7)**0·5625·2< 0·001
(b) Appearance date
 20041028·5 (± 5·3)**39·0 (± 7·3)**0·7828·80·001
 20051116·2 (± 4·6)**43·9 (± 6·3)***0·5812·50·006
 20062242·7 (± 7·1)*** 7·8 (± 9·4)NS0·6436·2< 0·001

Using data from 17 temperature data-loggers at transect sites where A. crataegi and its host plants were recorded, mean daytime temperature (from 0700 to 1800 GMT) over the 2 months of the entire regional flight period in 2006 decreased by 8·9 °C per 1 km increase in elevation (mean temperature = 33·1 °C (± SE 2·7)–8·9 × elevation (km) (± SE 2·1); R2 = 0·54; F1,15 = 17·4, P = 0·001). Mean daytime temperature for the month of peak flight at each site had a less pronounced decline of 6·2 °C per 1 km increase in elevation (mean = 28·8 °C (± SE 2·4)–6·2 × elevation (km) (± SE 1·9); R2 = 0·41; F1,15 = 10·3, P = 0·006), because of the phenological delay with increasing elevation. Data from a smaller number of dataloggers in 2004–05 also showed steeper declines in temperature with elevation for the 2 months encompassing the entire flight period than for the local month of peak flight (results not shown).

egg-site location

In total, 351 host plants were searched, and data were recorded for 236 egg batches and 63 larval webs. Egg batches were not located randomly among host plants: the distribution of egg batches/larval webs among trees differed significantly from that expected under a Poisson distribution (χ2-test: observations grouped ≥ 4 egg batches/larval webs per host plant; χ2 = 237·98, d.f. = 3, P < 0·0001), indicating a clumped distribution of egg batches on relatively few host plants. No egg batches or larval webs were found below 900 m.

For sites with at least five egg batches, the side of the tree on which eggs were laid (degrees difference from north) was positively related to elevation (S° = 104·9 (± SE 18·1) × elevation (km) – 30·6 (± SE 23·9); R2 = 0·72, F1,13 = 33·66, P < 0·001), indicating that eggs were located on the north side of plants at lower elevations, and on the south side of plants at higher elevations (Fig. 5a). Height above ground of egg-sites was not related to elevation. The proportion of eggs laid on the upper surface of leaves (L) showed evidence of a curvilinear relationship with elevation, peaking at middle elevations (Fig. 5b). However, the relationship was not significant in the analysis of sites with five egg batches or more, where only one site below 1100 m was included (because most eggs had already hatched into larvae by the time of egg searches at low elevations).

image

Figure 5. The relationship of egg-site microhabitat with elevation. (a) Side of host plant (°difference from north: higher values represent the south-facing side of trees). (b) Proportion of egg batches on the leaf upperside. Solid symbols show sites with ≥ 5 egg batches or larval webs (trend line shown in a); open symbols show sites with < 5 egg batches or larval webs.

Download figure to PowerPoint

egg and larval survivorship

There was no significant difference in the proportion of egg batches surviving to form larval webs in shaded and unshaded environments (paired t-test, t = 0·20, d.f. = 8, P = 0·847). Data for shaded and unshaded locations were therefore pooled for subsequent analyses. Elevation was a highly significant predictor of the proportion of egg batches surviving to form larval webs at experimental sites (generalized linear model with quasi-binomial error structure to correct for overdispersion: F1,8 = 11·61, P = 0·009; Fig. 6), and also after excluding data from the 580 m site (F1,7 = 8·91, P = 0·020). The parameter estimates for elevation were positive, showing that mortality decreases with elevation.

image

Figure 6. Proportion survival of A. crataegi egg batches as a function of elevation (km). Solid symbols show locations within the range of A. crataegi (n = 8 batches transplanted per site); open symbol shows one location outside the elevational range (n = 4 batches).

Download figure to PowerPoint

None of the eggs transplanted onto Rosa spp. survived as larvae. Living larvae were found at three of five control manipulations on C. monogyna.

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We present evidence for a shift in the elevational range of the butterfly A. crataegi in central Spain between 1967 and 73 and 2006. The butterfly has apparently disappeared from low elevations where its larval host plants remain, while its ability to expand its distribution above its upper elevation limit has been constrained by the absence of host plants at higher elevations. We propose increasing temperatures as a causal factor for the retraction of A. crataegi from low elevations, based on evidence that larval survivorship increases with elevation. Information on population density, phenology and habitat use support the importance of temperature in influencing the population dynamics and elevational range of A. crataegi.

elevational range of a. crataegi

The elevational range of A. crataegi retracted uphill between 1967 and 73 and 2006, with the species disappearing from elevations below 900 m. In 1967–73 the butterfly's range coincided closely with that of its host plants (assuming that host plant distribution did not change markedly between 1967–73 and 2006), with large populations recorded as low down as 640 m and up to 1800 m. By 2006, A. crataegi was absent from sites below 900 m, in spite of the presence of host plants. The loss of A. crataegi from 600 to 900 m elevations almost certainly results from local population extinctions rather than adult dispersal, as the nearest known extant populations of the species are 25 km or further from the lowest elevation locations where it was recorded in 1967–73. In contrast, the upper elevation limit has not increased, and seems to be determined by the distribution of C. monogyna and P. spinosa. The main host plants are rare above 1500 m and were not found above 1792 m despite extensive searching. The few transect locations without host plants where A. crataegi was observed (all of them at 1400 m elevation or above) were within 2·5 km of the nearest known C. monogyna plant, and had annual counts of four or fewer A. crataegi.

At occupied sites, the population density of A. crataegi increased from low to mid-elevations before declining above 1500 m (Fig. 3). The decline in population density at high elevations appears to be related to a concurrent decline in host plant density. There was an indication that the density of A. crataegi relative to that of its host plants increased linearly with elevation (Fig. 3c). One possible explanation for such a pattern is that the proportion of host plants that was suitable (e.g. for egg-laying or larval survival) increased with elevation.

egg and larval survivorship

Field transplant experiments across and below A. crataegi's current range revealed a significant effect of elevation on the survivorship of egg batches and young larvae. Direct temperature effects seem to be the most likely explanation for the trends in A. crataegi survival (for effects of high temperatures on egg and larval survival in other Lepidoptera see Bryant, Thomas & Bale 1997; Alonso 1999). Differences in rates of leaf maturation or senescence at different elevations (e.g. Weiss et al. 1988; Bale et al. 2002) are unlikely to explain differences in mortality of young A. crataegi larvae, as leaves are fully mature at all elevations at dates of egg hatch (June–July). In addition, mortality often occurred before the emergence of larvae, with no evidence of larval webs for nine of the 14 experimental egg batches that did not survive at sites below 1200 m.

Insects may escape their natural enemies at higher elevations or latitudes by entering predator-free space (Randall 1982; Hodkinson 1999), representing another possible explanation for elevational trends in A. crataegi survivorship. We observed no evidence that parasitoids or predators were responsible for the pattern observed, and in most cases the unhatched eggs remained intact on the leaves. Nevertheless, it would be necessary to extend the experimental manipulation (e.g. Crozier 2004; Hill & Hodkinson 1995) to determine the contributions to A. crataegi mortality of direct climatic effects, and of effects mediated by other trophic levels, that potentially reflect the indirect effects of climate. Whichever mechanism is responsible, the effects of the elevational gradient on egg batch and young larval survivorship will have a negative effect on the viability of A. crataegi populations at low elevations, with the potential to limit the distribution of this species.

causes and consequences of phenological variation

Elevation was a significant predictor of both mean and first flight date, consistent with previous studies reporting later emergence dates for insects at higher elevations (Fielding et al. 1999; Crozier 2004; Hodkinson 2005). Elevation affects mean temperature, potentially influencing the date of diapause cessation and rates of larval and pupal development after hibernation. The differences in phenology in 2006 were pronounced, with butterflies flying 30–40 days later per 1 km increase in elevation (Fig. 4).

Climate-driven changes to species phenology could affect survival or fecundity, by shifting key stages of life cycles to cooler or warmer times of the year (Wilson, Davies & Thomas 2007) or by disrupting biotic interactions (Ibáñez et al. 2006). Because A. crataegi butterflies fly later at higher elevations, the temperatures experienced by adults do not decrease with altitude as sharply as might be predicted, all else being equal. However, results from temperature dataloggers at transect sites show that mean daytime temperature during the local peak month of A. crataegi flight (i.e. the temperatures experienced by egg-laying females, eggs and young larvae) still decreased by 6·2 °C per 1 km increase in elevation.

egg-site location

At high elevations eggs were laid on the south side of host plants, corresponding to hotter microclimates, whereas at low elevations they were found on the north side of plants (Fig. 5a), suggesting that egg-laying site may depend on ambient temperature (Davies et al. 2006). Based on our results for survivorship, the pattern could result from selection on female oviposition behaviour for sites whose microclimate favours the development and survival of larvae (Gilbert & Singer 1977). Further experiments are needed to confirm whether egg-site location in A. crataegi is directly related to temperature during oviposition, or whether egg-site choice and larval survival differ among populations from different elevations, representing possible local adaptation.

Further work is also required to determine whether additional climatic variables influence egg-site selection. For example, exposure to ultraviolet-B radiation (UV-B) can increase by c. 20% per 1 km elevation (Blumthaler, Ambach & Ellinger 1997). UV-B can cause tissue damage in insect larvae (Caldwell et al. 1998), and high levels of UV-B are avoided by some insects (Mazza et al. 2002). In A. crataegi, there was an indication that the proportion of egg batches laid on the leaf upperside peaked at middle elevations (Fig. 5b), a pattern that could result if host plant leaves protect eggs from UV-B. At low elevations, exposure both to UV-B and high ambient temperatures might be reduced for eggs on the leaf underside. Reduced ambient temperatures at increasing elevations could then favour the upperside, but at the highest elevations UV-B exposure may be so high that the leaf underside is again favoured. Research is needed to determine whether the effects of UV-B radiation on insect development and survival could limit species’ ability to track climate change, by preventing range shifts towards higher elevations.

implications of climate change for specialist species

Increasing temperatures are likely to drive the warm range margin of A. crataegi away from low elevations, and the future distribution of the species in the region will depend on its ability to colonize high elevation sites. However, suitable host plants might be unlikely to shift their distributions as quickly in response to climate change. Recent research shows that a broadening of host plant use by butterflies can increase rates of range expansion at a landscape scale (Thomas et al. 2001). The ability to exploit alternative host plants such as Rosa spp., which have a much wider elevational range (see Supplementary material, Fig. S1), could greatly increase A. crataegi's ability to colonize high elevations. However, A. crataegi has rarely been observed laying eggs on Rosa spp. in the region, and our experimental transplants showed no evidence that larvae are able to develop on these plants. Unless A. crataegi can broaden its host plant use at higher elevations, it seems inevitable that the species will experience an increasingly restricted elevational range as the climate warms.

Bioclimate envelope models have been used to predict future species distributions, based either on the geographical range of suitable climate space or its overlap with currently suitable conditions (e.g. Pearson & Dawson 2003; Thomas et al. 2004). Our research emphasizes how, for specialist species such as many phytophagous insects, observed retractions in potential or realized distributions are likely to be much greater than previously estimated, because of reduced overlap between host species and suitable climate (see Andrew & Hughes 2004, 2005). Realistic models of future species ranges should include the effects both of species’ physiological requirements and climatic associations, as well as those of their key interacting species (Hodkinson 1999; Araújo & Guisan 2006) or habitats (Franco et al. 2006).

Our work suggests that temperature is the most likely factor determining the low elevation limit to A. crataegi's distribution, whereas interacting species are responsible for its upper elevation margin. This result contrasts with the prevailing theory that species interactions determine warm limits to species distributions, while direct climatic limitation imposes the cool boundaries to species ranges (MacArthur 1972; Brown et al. 1996; Parmesan et al. 2005). The clearest change to the distribution of A. crataegi was the loss of populations from elevations of 600–900 m, despite the continuing presence of larval host plants. This distribution change is unlikely to reflect the effects of adult resources, since during A. crataegi's June flight period the density of flowers does not increase with elevation, and many nectar sources are available in the habitats sampled below 900 m (S. B. Díez, unpublished data). Instead, we show that egg and young larval survival increases with elevation, with > 75% estimated mortality below the current lower elevation margin. Increased mortality at low elevations because of increased temperatures in the region since 1967–73 could represent a mechanism behind the range contraction of A. crataegi, but further experimental work is needed to rule out other competing explanations. For example, egg transplant experiments that exclude predators or parasitoids could be particularly informative (e.g. Crozier 2004), as well as translocation experiments using the progeny of females from lower elevation populations, to test for a potential role of adaptation across the elevational range.

Predictions indicate further warming of up to 4 °C by 2100 (IPCC 2007), corresponding to a c. 650 m shift in the elevational distribution of isotherms. Considering that A. crataegi is currently restricted to a 900 m band in the Sierra de Guadarrama and may have little capacity to move to higher elevations, a considerable contraction of its current range seems inevitable. If the distribution patterns revealed by this study also apply to other areas in southern Europe and North Africa where the species is restricted to mountains (García Barros et al. 2004; J. Tennent, personal communication), then the future of A. crataegi, and indeed other plant and animal species with ranges limited in similar ways, may not be as secure as its current widespread status implies.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank V. J. Monserrat for access to historical data, and J. Bridle and J. Harcourt for assistance in the field. Funding to D.G., R.J.W. and J.G. was provided by the Ministerio de Educación y Ciencia (grant reference CGL2005-06820/BOS), Spain. R.M.M. was supported by an MSc studentship from the Biotechnology and Biological Sciences Research Council (BBSRC) and by a travel grant from St Peter's College, University of Oxford. O.T.L. is a Royal Society University Research Fellow and was also supported by the Ernest Cook Research Fellowship, Somerville College, University of Oxford. Access and research permits were provided by Comunidad de Madrid, Parque Regional de la Cuenca Alta de Manzanares, Parque Natural de Peñalara, and Parque Regional del Curso Medio del Río Guadarrama.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Alonso, C. (1999) Variation in herbivory by Yponomeuta mahalabella on its only host plant Prunus mahaleb along an elevational gradient. Ecological Entomology, 24, 371379.
  • Andrew, N.R. & Hughes, L. (2004) Species diversity and structure of phytophagous beetle assemblages along a latitudinal gradient: predicting the potential impacts of climate change. Ecological Entomology, 29, 527542.
  • Andrew, N.R. & Hughes, L. (2005) Diversity and assemblage structure of phytophagous Hemiptera along a latitudinal gradient: predicting the potential impacts of climate change. Global Ecology and Biogeography, 14, 249262.
  • Araújo, M.B. & Guisan, A. (2006) Five (or so) challenges for species distribution modelling. Journal of Biogeography, 33, 16771688.
  • Asher, J., Warren, M., Fox, R., Harding, P., Jeffcoate, G. & Jeffcoate, S. (2001) The Millennium Atlas of Butterflies in Britain and Ireland. Oxford University Press, Oxford.
  • Bale, J.S., Masters, G.J., Hodkinson, I.D., Awmack, C., Bezemer, T.M., Brown, V.K., Butterfield, J., Buse, A., Coulson, J.C., Farrar, J., Good, J.E.G., Harrington, R., Hartley, S., Jones, T.H., Lindroth, R.L., Press, M.C., Symrnioudis, I., Watt, A.D. & Whittaker, J.B. (2002) Herbivory in global climate change research: direct effects of rising temperatures on insect herbivores. Global Change Biology, 8, 116.
  • Blumthaler, M., Ambach, W. & Ellinger, R. (1997) Increase in solar UV radiation with altitude. Journal of Photochemistry and Photobiology, 39, 130134.
  • Brown, J.H., Stevens, G.C. & Kaufman, D.M. (1996) The geographic range: size, shape, boundaries and internal structure. Annual Review of Ecology and Systematics, 27, 597623.
  • Bryant, S.R., Thomas, C.D. & Bale, J.S. (1997) Nettle-feeding nymphalid butterflies: temperature, development and distribution. Ecological Entomology, 22, 390398.
  • Caldwell, M.M., Bjorn, L.O., Bornman, J.F., Flint, S.D., Kulandaivelu, G., Teramura, A.H. & Tevini, M. (1998) Effects of solar ultraviolet radiation on terrestrial ecosystems. Journal of Photochemistry and Phytobiology, Series B, Biology, 46, 4052.
  • Crawley, M.J. (2002) Statistical Computing – an Introduction to Data Analysis Using S-Plus. John Wiley & Sons, Chichester.
  • Crozier, L.G. (2004) Field transplants reveal summer constraints on a butterfly range expansion. Oecologia, 141, 148157.
  • Davies, Z.G., Wilson, R.J., Coles, S. & Thomas, C.D. (2006) Changing habitat associations of a thermally constrained species, the silver-spotted skipper butterfly, in response to climate warming. Journal of Animal Ecology, 75, 247256.
  • Emmet, A.M. & Heath, J. (1989) The Moths and Butterflies of Great Britain and Ireland, vol. 7, Part 1. Harley Books, Colchester.
  • Environmental Systems Research Institute (ESRI) (1996) Arc View Spatial Analyst. Advanced Spatial Analysis Using Raster and Vector Data. Environmental Systems Research Institute, Redlands, CA.
  • Fielding, C.A., Whittaker, J.B., Butterfield, J.E.L. & Coulson, J.C. (1999) Predicting responses to climate change: the effect of altitude on the phenology of the spittlebug Neophilaenus lineatus. Functional Ecology, 12 (Suppl. 1), 6573.
  • Franco, A.M.A., Hill, J.K., Kitschke, C., Collingham, Y.C., Roy, D.B., Fox, R., Huntley, B. & Thomas, C.D. (2006) Impacts of climate warming and habitat loss on extinctions at species’ low-latitude range boundaries. Global Change Biology, 12, 15451553.
  • García-Barros, E., Munguira, M.L., Martín Cano, J., Romo Benito, H., Garcia-Pereira, P. & Maravalhas, E.S. (2004) Atlas of the Butterflies of the Iberian Peninsula and Balearic Islands (Lepidoptera: Papilionoidea & Hesperioidea). Sociedad Entomológica Aragonesa, Zaragoza, Spain.
  • Gaston, K.J. (2003) The Structure and Dynamics of Geographical Ranges. Oxford University Press, Oxford.
  • Gilbert, L.E. & Singer, M.C. (1977) Butterfly ecology. Annual Review of Ecology and Systematics, 6, 365397.
  • Hampe, A. & Petit, R.J. (2005) Conserving biodiversity under climate change: the rear edge matters. Ecology Letters, 8, 461467.
  • Hill, J.K. & Hodkinson, I.D. (1995) Effects of temperature on phenological synchrony and altitudinal distribution of jumping plant-lice (Hemiptera: Psylloidea) on dwarf willow (Salix lapponum) in Norway. Ecological Entomology, 20, 237244.
  • Hodkinson, I.D. (1999) Species response to global environmental change or why ecophysiological models are important: a reply to Davis et al. Journal of Animal Ecology, 68, 12591262.
  • Hodkinson, I.D. (2005) Terrestrial insects along elevation gradients: species and community responses to altitude. Biological Reviews, 80, 489513.
  • Ibáñez, I., Clark, J.S., Dietze, M.C., Feeley, K., Hersh, M., LaDeau, S., McBride, A., Welch, N.E. & Wolosin, N.S. (2006) Predicting biodiversity change: outside the climate envelope, beyond the species–area curve. Ecology, 87, 18961906.
  • Intergovernmental Panel on Climate Change (IPCC) (2007) Climate Change 2007: the Physical Science Basis. Summary for Policymakers. UNEP Intergovernmental Panel on Climate Change, Geneva, Switzerland.
  • MacArthur, R.H. (1972) Geographyraphical Ecology: Patterns in the Distribution of Species. Princeton University Press, Princeton, NJ.
  • Mazza, C.A., Izaguirre, M.M., Zavala, J., Scopel, A.L. & Ballaré, C.L. (2002) Insect perception of ambient ultraviolet-B radiation. Ecology Letters, 5, 722726.
  • McLaughlin, J.F., Hellmann, J.J., Boggs, C.L. & Ehrlich, P.R. (2002) Climate change hastens population extinctions. Proceedings of the National Academy of Sciences USA, 99, 60706974.
  • Monserrat, V.J. (1976) La distribución ecológica de las mariposas diurnas del Guadarrama. Thesis, Universidad Complutense de Madrid, Madrid, Spain.
  • Nagelkerke, N.J.D. (1991) A note on a general definition of the coefficient of determination. Biometrika, 78, 691692.
  • NASA/JPL-Caltech (2004) Shuttle Radar Topography Mission. Available at: http://www2.jpl.nasa.gov/srtm/ (acessed 9 June 2007).
  • Oksanen, J. & Minchin, P.R. (2002) Continuum theory revisited: what shape are species responses along ecological gradients? Ecological Modelling, 157, 119129.
  • Parmesan, C. (1996) Climate and species range. Nature, 382, 765766.
  • Parmesan, C. (2005) Detection at multiple levels: Euphydryas editha and climate change. Climate Change and Biodiversity (eds T.E.Lovejoy & L.Hannah), pp. 5660. Yale University Press, New Haven CT/London.
  • Parmesan, C., Gaines, S., Gonzalez, L., Kaufman, D.M., Kingsolver, J., Peterson, A.T. & Sagarin, R. (2005) Empirical perspectives on species borders: from traditional biogeography to global change. Oikos, 108, 5875.
  • Pearson, R.G. & Dawson, T.P. (2003) Predicting the impacts of climate change on the distribution of species: are bioclimatic envelope models useful? Global Ecology and Biogeography, 12, 361371.
  • Pollard, E. & Yates, T.J. (1993) Monitoring Butterflies for Ecology and Conservation. .Chapman & Hall, London.
  • Randall, M.G.M. (1982) The dynamics of an insect population throughout its altitudinal distribution: Coleophora alticolella (Lepidoptera) in northern England. Journal of Animal Ecology, 51, 9931016.
  • Stefanescu, C., Peñuelas, J. & Filella, I. (2003) Effects of climate change on the phenology of butterflies in the northwest Mediterranean Basin. Global Change Biology, 9, 14941506.
  • Thomas, J.A. (1993) Holocene climate changes and warm man-made refugia may explain why a sixth of British butterflies possess unnatural early-successional habitats. Ecography, 16, 278284.
  • Thomas, C.D., Bodsworth, E.J., Wilson, R.J., Simmons, A.D., Davies, Z.G., Musche, M. & Conradt, L. (2001) Ecological and evolutionary processes at expanding range margins. Nature, 411, 577581.
  • Thomas, C.D., Cameron, A., Green, R.E., Bakkenes, M., Beaumont, L.J., Collingham, Y.C., Erasmus, B.F.N., Ferreira de Siqueira, M., Grainger, A., Hannah, L., Hughes, L., Huntley, B., Van Jaarsveld, A.S., Midgley, G.F., Miles, L., Ortega-Huerta, M.A., Peterson, A.T., Phillips, O.L. & Williams, S.E. (2004) Extinction risk from climate change. Nature, 427, 145148.
  • Thomas, C.D., Franco, A.M.A. & Hill, J.K. (2006) Range retractions and extinctions in the face of climate warming. Trends in Ecology and Evolution, 21, 415416.
  • Thomas, J.A., Rose, R.J., Clarke, R.T., Thomas, C.D. & Webb, N.R. (1999) Intraspecific variation in habitat availability among ectothermic animals near their climatic limits and their centres of range. Functional Ecology, 13, 5564.
  • Tolman, T. & Lewington, R. (1997) Butterflies of Britain and Europe. HarperCollins, London.
  • Walther, G.-R., Berger, S. & Sykes, M.T. (2005) An ecological ‘footprint’ of climate change. Proceedings of the Royal Society, Series B, 272, 14271432.
  • Walther, G.-R., Post, E., Convey, P., Menzel, A., Parmesan, C., Beebee, T.J.C., Fromentin, J.-M., Hoegh-Guildberg, O. & Bairlein, F. (2002) Ecological responses to recent climate change. Nature, 416, 389395.
  • Weiss, S.B., Murphy, D.D. & White, R.R. (1988) Sun, slope and butterflies: topographic determinants of habitat quality for Euphydryas editha bayensis. Ecology, 69, 14861496.
  • Wilson, R.J., Davies, Z.G. & Thomas, C.D. (2007) Insects and climate change: processes, patterns and implications for conservation. Insect Conservation Biology (eds A.J.A.Stewart, T.R.New & O.T.Lewis). CABI Publishing, Wallingford, UK.
  • Wilson, R.J., Gutiérrez, D., Gutiérrez, J., Martínez, D., Agudo, R. & Monserrat, V.J. (2005) Changes to the elevational limits and extent of species ranges associated with climate change. Ecology Letters, 8, 11381146.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The following supplementary material is available for this article.

Appendix S1. Methods: Huisman–Olff–Fresco (HOF) models.

Table S1. HOF models for the elevational range of A. crataegi and its host plants.

Fig. S1. The elevational range and abundance of Rosa spp. (a) Proportion of occupied sites against elevation (km). (b) Rosa abundance against site elevation.

This material is available as part of the online article from: http://www.blackwell-synergy.com/doi/full/10.1111/j.1365-2656.2007.01303.x

(This link will take you to the article abstract).

Please note: Blackwell Publishing is not responsible for the content or functionality of any supplementary material supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.

FilenameFormatSizeDescription
JANE_1303_sm_Supmat.doc50KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.