Hostname: page-component-745bb68f8f-l4dxg Total loading time: 0 Render date: 2025-02-11T01:50:19.627Z Has data issue: false hasContentIssue false

Terrestrial arthropod abundance and phenology in the Canadian Arctic: modelling resource availability for Arctic-nesting insectivorous birds

Published online by Cambridge University Press:  12 February 2013

Elise Bolduc*
Affiliation:
Département de biologie & Centre d’études nordiques, Université du Québec à Rimouski, 300 Allée des Ursulines, Rimouski, Quebec, Canada G5L 3A1
Nicolas Casajus
Affiliation:
Département de biologie & Centre d’études nordiques, Université du Québec à Rimouski, 300 Allée des Ursulines, Rimouski, Quebec, Canada G5L 3A1
Pierre Legagneux
Affiliation:
Département de biologie & Centre d’études nordiques, Université du Québec à Rimouski, 300 Allée des Ursulines, Rimouski, Quebec, Canada G5L 3A1
Laura McKinnon
Affiliation:
Département de biologie & Centre d’études nordiques, Université du Québec à Rimouski, 300 Allée des Ursulines, Rimouski, Quebec, Canada G5L 3A1
H. Grant Gilchrist
Affiliation:
National Wildlife Research Centre, Carleton University, 1125 Colonel By Drive (Raven Road), Ottawa, Ontario, Canada KJA OH3
Maria Leung
Affiliation:
Wild Tracks Ecological Consulting, 39 Harbottle Road, Whitehorse, Yukon, Canada Y1A 5T2
R.I. Guy Morrison
Affiliation:
National Wildlife Research Centre, Carleton University, 1125 Colonel By Drive (Raven Road), Ottawa, Ontario, Canada KJA OH3
Don Reid
Affiliation:
Wildlife Conservation Society Canada, PO Box 31127, Whitehorse, Yukon, Canada Y1A 5T2
Paul A. Smith
Affiliation:
Smith and Associates Ecological Research Ltd, 772 – 7th Conc. South, Pakenham, Ontario, Canada K0A 2X0
Christopher M. Buddle
Affiliation:
Department of Natural Resource Sciences, McGill University 21, 111 Lakeshore Road, Ste-Anne-de-Bellevue, Quebec, Canada H9X 3V9
Joël Bêty
Affiliation:
Département de biologie & Centre d’études nordiques, Université du Québec à Rimouski, 300 Allée des Ursulines, Rimouski, Quebec, Canada G5L 3A1
*
1Corresponding author (e-mail: elise_bolduc@uqar.ca).
Rights & Permissions [Opens in a new window]

Abstract

Arctic arthropods are essential prey for many vertebrates, including birds, but arthropod populations and phenology are susceptible to climate change. The objective of this research was to model the relationship between seasonal changes in arthropod abundance and weather variables using data from a collaborative pan-Canadian (Southampton, Herschel, Bylot, and Ellesmere Islands) study on terrestrial arthropods. Arthropods were captured with passive traps that provided a combined measure of abundance and activity (a proxy for arthropod availability to foraging birds). We found that 70% of the deviance in daily arthropod availability was explained by three temperature covariates: mean daily temperature, thaw degree-day, and thaw degree-day2. Models had an adjusted R2 of 0.29–0.95 with an average among sites and arthropod families of 0.67. This indicates a moderate to strong fit to the raw data. The models for arthropod families with synchronous emergence, such as Tipulidae (Diptera), had a better fit (average adjusted R2 of 0.80) than less synchronous taxa, such as Araneae (R2 = 0.60). Arthropod abundance was typically higher in wet than in mesic habitats. Our models will serve as tools for researchers who want to correlate insectivorous bird breeding data to arthropod availability in the Canadian Arctic.

Résumé

Dans la toundra arctique, les arthropodes constituent des proies essentielles pour de nombreux vertébrés dont les oiseaux. Cependant, les populations d'arthropodes et leur phénologie sont susceptibles de subir des modifications face aux changements climatiques. Notre étude utilise des données sur les arthropodes terrestres provenant d'une initiative pancanadienne (Îles Southampton, Herschel, Bylot et Ellesmere), afin de modéliser la relation entre les changements saisonniers d'abondance d'arthropodes et les variables environnementales. Des pièges fournissant une mesure combinée de l'abondance et de l'activité des arthropodes ont été utilisés afin d'obtenir un indice de la disponibilité des arthropodes pour les oiseaux. Dans nos modèles, trois covariables liées à la température (température, degrés-jours et degrés-jours2) expliquent 70% de la déviance. Selon les sites et les familles d'arthropodes modélisés, les R2 ajustés des modèles ont variés de 0.29–0.95 (moyenne de 0.67). Les modèles pour les familles d'arthropodes ayant une émergence synchronisée, comme les Tipulidae (Diptera), avaient de meilleurs R2 ajustés (0.80 en moyenne) comparativement aux groupes dont la disponibilité est plus répartie dans le temps, comme les araignées (0.60). L'abondance d'arthropodes était généralement plus grande dans les milieux humides que dans les milieux plus secs. Nos modèles pourront servir d'outil aux chercheurs qui désireraient corréler leurs données sur la reproduction des insectivores avec des données sur la disponibilité d'arthropodes dans l'Arctique Canadien.

Type
Behaviour & Ecology
Copyright
Copyright © Entomological Society of Canada 2013

Introduction

Climate change is a significant environmental disturbance that will alter the distribution and abundance of species (Thomas et al. Reference Thomas, Cameron, Green, Bakkenes, Beaumont and Collingham2004; Berteaux et al. Reference Berteaux, Humphries, Krebs, Lima, McAdam and Pettorelli2006; Post et al. Reference Post, Forchhammer, Bret-Harte, Callaghan, Christensen and Elberling2009). Changes in climate are not only associated with temperature changes but also with environmental stochasticity that directly affects life cycles of animals (Saether Reference Saether1997; Thomas et al. Reference Thomas, Cameron, Green, Bakkenes, Beaumont and Collingham2004; Jenouvrier et al. Reference Jenouvrier, Caswell, Barbraud, Holland, Stroeve and Weimerskirch2009). As ectotherms, arthropods are particularly sensitive to climate variation since they are biochemically, physiologically, and behaviourally dependent on temperature (Huey and Berrigan Reference Huey and Berrigan2001; Frazier et al. Reference Frazier, Huey and Berrigan2006) and their abundance is primarily driven by temperature (Deutsch et al. Reference Deutsch, Tewksbury, Huey, Sheldon, Ghalambor and Haak2008; Tulp and Schekkerman Reference Tulp and Schekkerman2008). Effects of climate change on arthropods may be most acute and significant in the Arctic since this region's climate is warming at a disproportionate rate relative to the rest of the planet (Arctic Climate Impact Assessment 2004) and rapid changes in arthropod phenology, abundance or species assemblage are expected (Deutsch et al. Reference Deutsch, Tewksbury, Huey, Sheldon, Ghalambor and Haak2008).

Arctic food webs are relatively simple (Elton Reference Elton1927; Gauthier et al. Reference Gauthier, Berteaux, Bêty, Tarroux, Therrien and McKinnon2012) and many trophic interactions in the Arctic are linked to arthropods (Hodkinson and Coulson Reference Hodkinson and Coulson2004). Millions of insectivorous birds breed in the Arctic (Committee for Holarctic Shorebird Monitoring 2004) and rely on terrestrial arthropods for their survival and reproduction (Pearce-Higgins and Yalden Reference Pearce-Higgins and Yalden2004; Schekkerman et al. Reference Schekkerman, Tulp, Calf and de Leeuw2004). On the Arctic tundra, surface-active arthropods are abundant only for a short period of time varying from a few days to a few weeks every year (MacLean and Pitelka Reference MacLean and Pitelka1971; Hodkinson et al. Reference Hodkinson, Coulson, Webb, Block, Strathdee and Bale1996; Schekkerman et al. Reference Schekkerman, Tulp, Calf and de Leeuw2004; Tulp Reference Tulp2007). In Arctic-nesting shorebirds, chick growth rates appear to be influenced strongly by the availability of arthropods (Tulp and Schekkerman Reference Tulp and Schekkerman2001; Schekkerman et al. Reference Schekkerman, Tulp, Piersma and Visser2003; Schekkerman et al. Reference Schekkerman, Tulp, Calf and de Leeuw2004; McKinnon et al. Reference McKinnon, Picotin, Bolduc, Juillet and Bêty2012). The timing and duration of this period of high arthropod availability, however, can change rapidly because of global warming as recorded over the last 10–30 years in Siberia, Russia, and eastern Greenland (Høye et al. Reference Høye, Post, Meltofte, Schmidt and Forchhammer2007; Tulp and Schekkerman Reference Tulp and Schekkerman2008).

The relationship between climate change and the ecology of species is often assessed via correlations within long-term data sets including climate, primary producers, and consumers (Both and Visser Reference Both and Visser2001; Root et al. Reference Root, Price, Hall, Schneider, Rosenzweig and Pounds2003; Dickey et al. Reference Dickey, Gauthier and Cadieux2008; Visser Reference Visser2008). Although long-term data sets on insectivorous birds exist, seasonal changes in Arctic arthropod abundance are poorly documented (but see Tulp and Schekkerman Reference Tulp and Schekkerman2008) and this remains an obstacle in determining the effect of climate change on insectivorous bird populations.

Since climate (for which long-term data exist and are freely available; Hijmans et al. Reference Hijmans, Cameron, Parra, Jones and Jarvis2005) is directly related to the abundance of surface-active arthropods (Danks Reference Danks1981; Hodkinson et al. Reference Hodkinson, Webb, Bale, Block, Coulson and Strathdee1998; Bale et al. Reference Bale, Masters, Hodkinson, Awmack, Bezemer and Brown2002; Tulp and Schekkerman Reference Tulp and Schekkerman2008), predicting or hindcasting Arctic arthropod abundance requires the selection of relevant climatic covariates to model arthropod abundance. This is the objective of our research. Here we report the results of a collaborative pan-Canadian study of surface-active Arctic arthropods and provide predictive models of daily arthropod availability for four sites that differ in terms of their climate and arthropod communities.

Methods

Arthropod sampling

Arthropods were sampled from June to August using a rectangular pitfall trap (38 cm × 5 cm and 7 cm deep). Above the pitfall trap, a 40 cm × 40 cm mesh screen was set vertically. Above the screen, a plastic cone funnelled flying insects into a collecting jar (Fig. 1). Traps were placed with the mesh perpendicular to prevailing winds, and their design was similar to traps used by Schekkerman et al. (Reference Schekkerman, Tulp, Piersma and Visser2003). These passive traps provided a combined measure of abundance and activity levels of arthropods, and so a proxy for arthropod availability to foraging birds, and data from such methods have been correlated to chick growth rate (Schekkerman et al. Reference Schekkerman, Tulp, Piersma and Visser2003; McKinnon et al. Reference McKinnon, Picotin, Bolduc, Juillet and Bêty2012). Traps were used on four different Arctic islands across the Canadian Arctic: Southampton (63°59′N, 81°40′W; mean summer temperature = 7.1 °C) from 2006 to 2008, Herschel (69°35′N, 138°55′W; 10.6 °C) in 2007 and 2008, Bylot (73°8′N, 79°58′W; 5.8 °C) from 2005 to 2008, and Ellesmere (Alert) (82°29′N, 62°21′W; 3.8 °C) in 2007 and 2008. At each site, five traps located 20 m apart from each other on a straight line were set in both dry upland (mesic) or low wetland (wet) tundra, the main foraging habitats for the dominant insectivorous bird species (passerines and shorebirds) during their brood-rearing period. Site-specific habitat descriptions are available in Smith et al. (Reference Smith, Gilchrist and Smith2007) (Southampton), Ale et al. (Reference Ale, Morris, Dupuch and Moore2011) (Herschel), Gauthier et al. (Reference Gauthier, Berteaux, Bêty, Tarroux, Therrien and McKinnon2011) (Bylot), and Morrison et al. (Reference Morrison, Davidson and Piersma2005) (Ellesmere). Traps were emptied at approximately two-day intervals, and arthropods were stored in ethanol (70%) until sorting and identification in the laboratory. Insects were sorted into families, and spiders were grouped together. Springtails and mites were not included in the analyses because of their very low contribution to total arthropod biomass. Butterflies and bumblebees were also excluded because few individuals were collected due to the design of the traps and because these few heavy specimens had a strong influence on daily variation in biomass. Moreover, adults of these taxa are not key prey for shorebirds or passerines. Sorting and identification was conducted on a subsample of three to five traps for each habitat and site. A standardised daily index of arthropod availability (mg/trap) was calculated by dividing the total arthropod biomass (dry mass) by the number of traps sorted and by the number of days between sampling event. To transform arthropod counts into dry mass, we used length to dry mass equations derived from our samples (Picotin Reference Picotin2008) or from the literature (Rogers et al. Reference Rogers, Buschbom and Watson1977; Sage Reference Sage1982; Sample et al. Reference Sample, Cooper, Greer and Withmore1993; Hodar Reference Hodar1996). For some of the dominant groups, we dried and weighed specimens and calculated a mean individual dry mass (Picotin Reference Picotin2008). When individual variation in size was high, individuals were grouped within size categories and mean dry mass was obtained for each category. A list of equations is provided in the supplementary materials of McKinnon et al. (Reference McKinnon, Picotin, Bolduc, Juillet and Bêty2012).

Fig. 1 Arthropod trap in the field.

Climate data

Mean hourly weather data (temperature in °C, precipitation in mm, relative humidity in %, radiation in W/m2, and wind speed in km/h) from the closest automated weather stations were used to build predictive models of daily arthropod availability. Stations were located ∼0.5 km (Southampton), 0.5–3.0 km (Herschel), 1 km (Bylot), and 2 km (Ellesmere) away from the trapping sites. Radiation data were unavailable for Ellesmere and Herschel Island.

Statistical analyses

Climatic variables known to influence the phenology of emergence, activity patterns, and/or abundance of arthropods (Wigglesworth Reference Wigglesworth1972; Strathdee et al. Reference Strathdee, Bale, Block, Coulson, Hodkinson and Webb1993; Whittaker and Tribe Reference Whittaker and Tribe1998; Roy et al. Reference Roy, Rothery, Moss, Pollard and Thomas2001; Goulson et al. Reference Goulson, Derwent, Hanley, Dunn and Abolins2005; Høye and Forchhammer Reference Høye and Forchhammer2008; Tulp and Schekkerman Reference Tulp and Schekkerman2008) were used to construct models of arthropod availability: daily temperature (T), relative humidity (H), precipitation (Rain), wind speed (Wi), thaw degree-days (D), and solar incidental radiation (R, log-transformed to improve normality). Correlations among meteorological variables ranged between 0.02 and 0.52 except between T and H (Pearson correlation = 0.70). High multicolinearity was coming from one study site: Ellesmere (Pearson correlation = 0.92). For this particular site, H and T were not entered simultaneously in the model. All variables were averaged over the number of days between trap checks except for precipitation and thaw degree-days. The sum of precipitation between trapping sessions was used (95% of the trapping sessions lasted two days, 4% lasted one or three days and 1% lasted four days), and thaw degree-days represented the accumulated mean daily temperatures between the first spring day above 0 °C and the day of sampling (sub-zero temperatures being treated as zeros). The quadratic form of thaw degree-days (D 2) was included in the models in order to represent the curvilinear pattern of seasonal arthropod availability. We also included the interaction between T and D as a variable because (i) insect development is proportional to accumulation in degree-days (Wagner et al. Reference Wagner, Olson and Willers1991; Gullan and Cranston Reference Gullan and Cranston2005) and (ii) temperature can affect insects differently depending on their developmental stage (Gullan and Cranston Reference Gullan and Cranston2005).

We used generalised linear mixed models with a Poisson distribution (McCullagh and Nelder Reference McCullagh and Nelder1989) to analyse the relationships between weather variables and arthropod dry mass. To account for repeated data across years, year was treated as a random factor in the analyses. Since the purpose of our models was not to test hypotheses but rather to select the best predictors of arthropod availability, model selection was based on adjusted R 2. When more than one model had the same adjusted R 2 (<1% variation), the model with the least number of parameters was preferred. For each of the four study sites, we created separate models for all of the dominant families. A family was considered dominant for a given site if its dry mass accounted for >10% of the total arthropod dry mass excluding spiders for at least one year (spiders had a high biomass contribution in early season and that could interfere with the contribution of other groups during the peak in arthropod abundance). Spiders were, by default, considered a dominant group at all sites. A threshold of 10% was set in order to select a limited number of dominant families while still accounting for most of the dry mass encountered. Dominant groups accounted for between 78% and 91% of total dry mass by site. Separate models were created for wet and mesic habitats.

For each model, the associated deviance for each selected variable (based on our model selection) was calculated. We then calculated the average deviance for each covariate based on the 37 models (each site/habitat/family) presented in Table 2. Proportions of deviance were calculated based on the partial R 2 and variables that were not included in a model were set to 0% except when data were unavailable (radiation in Ellesmere and Herschel).

Using the models described above, we calculated the estimated dry mass for each study site (sum of the predictions for the dominant groups). We then compared these predictions to the actual dry mass measured for all families (dominant and no-dominant altogether).

In order to validate the models, we developed a cross-validation technique to assess the fit of our models on an independent data set. For this sake, we used data from Bylot Island, the only site for which we had a relatively large data set available (i.e., up to four years). We constructed models based on three years of data (except for Araneae and Ichneumonidae, where only three and two years were available) and confronted the predicted values from these models to the independent data of the fourth year. We repeated this procedure four times (for each three-year combination) for Chironomidae (Diptera), Carabidae (Coleoptera), Muscidae (Diptera), and Tipulidae (Diptera) and three or two times for Araneae and Ichneumonidae (Hymenoptera). We then calculated the adjusted R 2 and generated a figure for each year and taxon for the wet habitat. Such approach is ideal to assess the reliability of a predictive model (Efron and Tibshirani Reference Efron and Tibshirani1993).

We also performed a Linear Mixed Model to investigate both effects of habitat and study sites on arthropod availability (all families combined). In this model, we linked arthropod availability to site, habitat, and the interaction of both. We put days and years as random factors (days nested in year) in order to account for repeatability.

Results

Descriptive results

A total of 300 days of sampling over up to four years at four different sites yielded 342 451 arthropods identified to the family level, except for spiders. In total, representatives of 92 families were found. The most common groups were Araneae, Carabidae, Chironomidae, Tipulidae, Muscidae (Diptera), Ichneumonidae, Mycetophilidae (Diptera), and Scatophagidae (Diptera) (Table 1). Diversity (expressed as the number of different families) was higher at Herschel (n = 81) than at Southampton, Bylot or Ellesmere (n = 37, 38, 26, respectively). At each site, only a few families (6–10) contributed significantly (more than 1%) to the total dry mass. There was considerable inter-annual variation in seasonal trends of arthropod availability in terms of timing, duration, and magnitude of peaks in total dry mass, both within and among sites (Fig. 2). Although variable among sites, arthropod availability was generally highest in early July (Fig. 2). On Bylot Island, for example, peaks were short-lived each year (two to seven days) and usually occurred within the same 17-day period (28 June to 14 July).

Table 1 Total dry mass of the different families averaged over the number of years of sampling.

Only families representing more than 1% of total biomass for at least one year are presented in the table but all families are included in the totals.

Fig. 2 Predicted and observed total daily arthropod availability in wet habitat in four Canadian Arctic sites.

Modelling results

Our models had a relatively strong fit to the raw data for most arthropod families (adjusted R 2 up to 0.95; Table 2 and Fig. 3) across all sites. Seasonal change in daily availability of arthropods was determined primarily by the following environmental variables: cumulative temperatures above 0 °C (thaw degree-days) and its quadratic form (mean % deviance explained = 48.5% ± 4.7 SE), and mean daily temperature (mean % deviance explained = 23.2% ± 4.6 SE). The other climatic variables (mean daily wind speed, mean daily relative humidity, total daily precipitation, mean daily incidental radiation, and the interaction between daily temperature and thaw degree-day) each accounted for <10% of the deviance explained (Fig. 4).

Table 2 Parameter estimates for the generalized linear mixed models of weather variables on availability of the dominant arthropod groups (dry mass of arthropod expressed in mg/trap/day).

Coefficients in bold were significant and where coefficients are missing, the corresponding variable was absent from the model selected. Numbers in parentheses represent the standard errors associated with each coefficient. Certain variables were not measured (NA) at some sites. Weather variables are as follows: D is the accumulation of thaw degree-days (10−1 °C), D 2 is the square of the accumulation in thaw degree-days (10−4 °C2), T is average daily temperature (10−2 °C), Wi is average daily wind speed (10−1 km/h), H is average daily relative humidity (10−1 %), R is average daily incidental radiation (10−2 log(W/m2)), Rain is total daily precipitation between two trapping sessions (10−1 mm), and T × D is the interaction between T and D (10−3).

Fig. 3 Predicted and observed arthropod availability split by sites, year, and habitat for the most abundant families.

Fig. 3 continued

Fig. 3 continued

Fig. 4 Average (+SE) of the % of deviance explained by the different weather variables from models generated for each family and year. Weather variables are as follows: D is the accumulation of thaw degree-days, D 2 is the square of the accumulation in thaw degree-days, T is average daily temperature, Wi is average daily wind speed, H is average daily relative humidity, R is average daily incidental radiation, Rain is total daily precipitation between two trapping sessions, and T × D is the interaction between T and D.

Predicted and observed data are presented for each site, year, habitat, and family in Figure 3. Based on these models, we calculated an estimated dry mass for each study site (sum of the predictions for the dominant groups), which explained well the total dry mass of arthropods measured for all dominant and nondominant arthropods family pooled (R 2 varied from 0.30 to 0.95 depending on the site or year considered Fig. 2).

In the cross-validation, models generally performed slightly less than models with full data but overall the predictive power of the validation models was still good. Average adjusted R 2 of the validation models was 0.48 ± 0.042 SE and 0.35 ± 0.05 SE for the wet and mesic habitats, respectively. In all the validation models, the date of the peak of abundance could be assessed with great accuracy (Fig. 5 on wet habitat).

Fig. 5 Cross-validation models based on data from Bylot Island. We constructed models from all the data available minus one year. The observed values from the year left out (indicated on the left of the graphs) were then contrasted with the predictions from the cross-validation model.

Arthropod availability differed according to habitat type. The mean dry mass was 63.37 ± 6.03 SE mg per trap per day and 40.84 ± 3.14 SE in wet and mesic habitats, respectively, all sites combined (Linear Mixed Models with days nested in year: F 1,341 = 12.69, P < 0.001). The interaction between habitat and study site was not significant (F 3,335 = 0.25, P = 0.86) revealing that similar patterns of availability occurred in both habitats within a study site. The abundance also greatly differed according to study site (F 3,339 = 34.28, P < 0.001).

Discussion

The objective of our study was to generate predictive models of daily arthropod availability in the Canadian Arctic. Based on data collected at four different sites that varied in terms of their climate and arthropod communities, we found part of the variation to be dependent on climatic variables measured daily such as precipitation or wind. The other part of the variation was explained by daily temperature and a larger time scale measure of weather (thaw degree-day). Indeed, with mean daily temperature, thaw degree-day, and its quadratic form, more than 70% of the daily variation in arthropod availability was explained by our models. This is a substantial portion of the deviance, emphasising the overarching importance of both temperature and variables measured over larger time scales in determining seasonal change in arthropod availability. Our research results complement work conducted in other Arctic regions such as the Taymir Peninsula in Siberia, Russia, where cumulative degree-days was a better predictor of the number of arthropods caught than the combination of date and temperature (Schekkerman et al. Reference Schekkerman, Tulp, Calf and de Leeuw2004). Together with our results, this highlights the importance of including temperature, cumulative temperature, and its quadratic form in studies aimed at forecasting arthropod abundance rather than focusing mostly on daily temperature variation as is often the case (Deutsch et al. Reference Deutsch, Tewksbury, Huey, Sheldon, Ghalambor and Haak2008).

As ectotherms, arthropods are highly sensitive to climate variation (Danks Reference Danks1981; Hodkinson et al. Reference Hodkinson, Webb, Bale, Block, Coulson and Strathdee1998; Bale et al. Reference Bale, Masters, Hodkinson, Awmack, Bezemer and Brown2002). Population growth rates and development of many arthropod species is linked to temperature (Huey and Berrigan Reference Huey and Berrigan2001; Frazier et al. Reference Frazier, Huey and Berrigan2006). As expected, we found higher dry mass (and diversity) of arthropods for sites with a warmer summer. However, predicting the effect of global warming also requires further information on possible lagged effects (like density dependence and previous summer temperature).

There was considerable variation in seasonal trends of arthropod availability in terms of timing, duration, and magnitude of peaks in total biomass both between sites and between habitats within sites. Arthropod availability was higher in wet habitats than in mesic habitats and this was consistent for all our study sites; we uncovered no significant interaction between habitat and site. Interestingly, for a specific site, climatic variables had similar effects on the abundance and phenology of a particular family for both habitats. Knowing the habitat characteristic of a site and its temperature thus appear to be an important covariate to model Arctic arthropod availability.

Across the sites, different patterns of abundance and phenology were evident among the different arthropod families that were sampled. For example, Tipulidae (Diptera; crane flies) showed very synchronised peaks of abundance, whereas Araneae (spiders) were present throughout the study period. This is likely related to the natural history of the different families. In certain species of crane flies, for example, all adults that emerge in a given summer overwintered in the prepupal stage (MacLean Reference MacLean1973). As a result, there is no feeding requirement before their emergence as adults and a more synchronised emergence is therefore possible. Studies of crane flies in the Arctic, including ours (Fig. 3), have revealed well-defined peaks of abundance (MacLean and Pitelka Reference MacLean and Pitelka1971; Tulp and Schekkerman Reference Tulp and Schekkerman2008) supporting the concept of a synchronised emergence. As a consequence, our climatic models for crane flies had a good fit to the data. This life-history strategy has also been observed in a number of Arctic Chironomidae (nonbiting midges) species (Danks and Oliver Reference Danks and Oliver1972). However, the timing of emergence of nonbiting midges is directly linked to the temperature of the pond in which the prepupal stage resides (Danks and Oliver Reference Danks and Oliver1972). Since ponds can warm up at different rates (based on depth, for example), emergence can be synchronised within ponds rather than across broad spatial scales. In our observations, there is more than one peak of nonbiting midges in any given year and the peaks are not as well defined as for crane flies (see Fig. 3). Numerous peaks may also be due to several species emerging at different times considering that nonbiting midges contributes a high proportion of species to the tundra insect fauna (MacLean and Pitelka Reference MacLean and Pitelka1971). In spiders, we did not observe a synchronised peak of abundance (see Fig. 3). They are present throughout the season and usually are active at the very beginning of snow melt (Meltofte and Høye Reference Meltofte and Høye2007; this study).

These contrasting patterns of arthropod phenology may be important in their role as food for birds. Both spiders and crane flies are important to successful reproduction of insectivorous birds but they likely play different roles. Spiders are active at the very beginning of snow melt and are present throughout the season (Meltofte and Høye Reference Meltofte and Høye2007; this study). It has been found that the abundance of wolf spider Pardosa glacialis (Thorell, 1872) (Araneae: Lycosidae) was probably the only variable influencing the timing of reproduction of jaegers Stercorarius longicaudus (Vieillot, 1819) (Aves: Stercoriidae) in Greenland (Meltofte and Høye Reference Meltofte and Høye2007). Arthropods that are present early in the season may be especially important for the long-distant migrants that arrive at their Arctic breeding grounds with little to no stored energy and must rely upon early emerging arthropods in order to rebuild fat reserves and produce eggs (Danks Reference Danks1971; Klaassen et al. Reference Klaassen, Lindstrom, Meltofte and Piersma2001; Meltofte et al. Reference Meltofte, Høye and Schmidt2008). On the other hand, arthropods such as crane flies, which exhibit peaks later in the season may be more important for the growth and survival of offspring (Pearce-Higgins and Yalden Reference Pearce-Higgins and Yalden2004).

Arthropods with an availability that is limited in time (synchronised emergence) can provide a great source of food for chicks if hatching is synchronised with emergence. But achieving this synchrony can be challenging, especially in the context of climate change (Both and Visser Reference Both and Visser2001; Thomas et al. Reference Thomas, Blondel, Perret, Lambrechts and Speakman2001). For example, studies have now indicated that an asynchrony between hatch of shorebird chicks and peaks in crane flies can reduce chick growth rates (McKinnon et al. Reference McKinnon, Picotin, Bolduc, Juillet and Bêty2012) and even lead to potential population declines in some shorebird species (Pearce-Higgins et al. Reference Pearce-Higgins, Yalden and Whittingham2005). However, abundance of arthropods rather than asynchrony may better predict population decline in different shorebirds species (Pearce-Higgins et al. Reference Pearce-Higgins, Yalden, Dougall and Beale2009; Pearce-Higgins Reference Pearce-Higgins2010).

Changes in climate are likely to induce changes in the patterns of arthropod availability with the potential to affect several trophic levels within the tundra food web. The unique data we have collected as part of this pan-Canadian effort, and the models we have tested will help us to forecast and/or hindcast arthropod availability over time, so that we can gain greater insight into the potential effects of changing arthropod availability for Arctic insectivores. Future research should continue to refine our understanding of seasonal variation in arthropod availability and attempt to study variation at lower taxonomic levels such as genus and, if possible, species. However, with over 2000 species of arthropods in arctic North America (Danks Reference Danks1992), this could prove to a be a very interesting, though rather challenging task.

Acknowledgements

Funding and logistic assistance for the collection of specimens was provided by grants from the Canadian federal government's International Polar Year program, the Natural Science and Engineering Research Council of Canada, and the Canadian Ministry of Environment to the Arctic Wildlife Observatories Linking Vulnerable EcoSystems International Polar Year project. Our study was also made possible due to funding by the Fonds Québecois de recherche sur la nature et les technologies (FQRNT), Wildlife Conservation Society Canada, the Garfield Weston Award for Northern Research, and the Northern Scientific Training Program. Logistical support was provided by Polar Continental Shelf Project and Parks Canada. Special thanks to Ludovic Jolicoeur, Benoit Laliberté, Émilie D'astous, Pierre-Yves l'Hérault, Maryse Cloutier, Jean-Rémi Julien, Scott Gilbert, Georges-Olivier Cimon, William Halliday, and Alice Kenney for field work. Thanks also to Valérie Delisle-Gagnon, Sarah Loboda, Hirondelle Varady-Szabo, Annie Hibbert, Samuel Pinna, Maude Picotin, Ludovic Jolicoeur, and Karine Richer for identification of arthropods in the laboratory. Thank you to Alain Caron, Nicolas Lecomte, Arnaud Tarroux, Eeva Soininen, the members of the Bêty-Berteaux laboratory, and two anonymous referees for their helpful comments.

References

Ale, S.B., Morris, D.W., Dupuch, A., Moore, D.E. 2011. Habitat selection and the scale of ghostly coexistence among Arctic rodents. Oikos, 120: 11911200 . doi:10.1111/j.1600-0706.2010.18933.x.CrossRefGoogle Scholar
Arctic Climate Impact Assessment. 2004. Impacts of warming climate: Arctic Climate Impact Assessment. Cambridge University Press, Cambridge, United Kingdom.Google Scholar
Bale, J.S., Masters, G.J., Hodkinson, I.D., Awmack, C., Bezemer, T.M., Brown, V.K., et al. 2002. Herbivory in global climate change research: direct effects of rising temperature on insect herbivores. Global Change Biology, 8: 116 . doi:10.1046/j.1365-2486.2002.00451.x.CrossRefGoogle Scholar
Berteaux, D., Humphries, M.M., Krebs, C.J., Lima, M., McAdam, A.G., Pettorelli, N., et al. 2006. Constraints to projecting the effects of climate change on mammals. Climate Research, 32: 151158.CrossRefGoogle Scholar
Both, C.Visser, M.E. 2001. Adjustment to climate change is constrained by arrival date in a long-distance migrant bird. Nature, 411: 296298.CrossRefGoogle Scholar
Committee for Holarctic Shorebird Monitoring. 2004. Monitoring Arctic-nesting shorebirds: an international vision for the future. Wader Study Group Bulletin, 103: 25.Google Scholar
Danks, H.V. 1971. A note on the early season food of arctic migrants. Canadian Field-Naturalist, 85: 7172.Google Scholar
Danks, H.V. 1981. Arctic arthropods: a review of systematics and ecology with particular reference to the North American fauna. Entomological Society of Canada, Ottawa, Canada.Google Scholar
Danks, H.V. 1992. Arctic insects as indicators of environmental change. Arctic, 45: 159166.CrossRefGoogle Scholar
Danks, H.V.Oliver, D.R. 1972. Seasonal emergence of some high Arctic Chironomidae (Diptera). The Canadian Entomologist, 104: 661686.CrossRefGoogle Scholar
Deutsch, C.A., Tewksbury, J.J., Huey, R.B., Sheldon, K.S., Ghalambor, C.K., Haak, D.C., et al. 2008. Impacts of climate warming on terrestrial ectotherms across latitude. Proceedings of the National Academy of Sciences, 105: 66686672 . doi:10.1073/pnas.0709472105.CrossRefGoogle ScholarPubMed
Dickey, M.-H., Gauthier, G., Cadieux, M.-C. 2008. Climatic effects on the breeding phenology and reproductive success of an arctic-nesting goose species. Global Change Biology, 14: 19731985 . doi:10.1111/j.1365-2486.2008.01622.x.CrossRefGoogle Scholar
Efron, B.Tibshirani, R.J. 1993. An introduction to the bootstrap. In Monographs on statistics and applied probability. Edited by B. Raton. Chapman & Hall, London, United Kingdom. Pp. 413425.Google Scholar
Elton, C.S. 1927. Animal ecology. Macmillan Company, New York, United States of America.Google Scholar
Frazier, M.R., Huey, R.B., Berrigan, D. 2006. Thermodynamics constrains the evolution of insect population growth rates: “Warmer is better”. American Naturalist, 168: 512520 . doi:10.1086/506977.CrossRefGoogle Scholar
Gauthier, G., Berteaux, D., Bêty, J., Tarroux, A., Therrien, J.-F., McKinnon, L., et al. 2011. The tundra food web of Bylot Island in a changing climate and the role of exchanges between ecosystems. Ecoscience, 18: 223235 . doi:10.2980/18-3-3453.CrossRefGoogle Scholar
Gauthier, G., Berteaux, D., Bêty, J., Tarroux, A., Therrien, J.F., McKinnon, L., et al. 2012. The Arctic tundra food web in a changing climate and the role of exchanges between ecosystems. Ecoscience, 18: 223235.CrossRefGoogle Scholar
Goulson, D., Derwent, L.C., Hanley, M.E., Dunn, D.W., Abolins, S.R. 2005. Predicting calyptrate fly populations from the weather, and probable consequences of climate change. Journal of Applied Ecology, 42: 795804.CrossRefGoogle Scholar
Gullan, P.J.Cranston, P.S. 2005. The insects: an outline of entomology. Blackwell Publishing, Oxford, United Kingdom.Google Scholar
Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A. 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology, 25: 19651978 . doi:10.1002/joc.1276.CrossRefGoogle Scholar
Hodar, J.A. 1996. The use of regression equations for estimation of arthropod biomass in ecological studies. Acta Oecologica, 17: 421433.Google Scholar
Hodkinson, I.D.Coulson, S.J. 2004. Are high Arctic terrestrial food chains really that simple? The Bear Island food web revisited. Oikos, 106: 427431.CrossRefGoogle Scholar
Hodkinson, I.D., Coulson, S.J., Webb, N.R., Block, W., Strathdee, A.T., Bale, J.S., et al. 1996. Temperature and the biomass of flying midges (Diptera: Chironomidae) in the high Arctic. Oikos, 75: 241248.CrossRefGoogle Scholar
Hodkinson, I.D., Webb, N.R., Bale, J.S., Block, W., Coulson, S.J., Strathdee, A.T. 1998. Global change and Arctic ecosystems: conclusions and predictions from experiments with terrestrial invertebrates on spitsbergen. Arctic and Alpine Research, 30: 306313.CrossRefGoogle Scholar
Høye, T.T.Forchhammer, M.C. 2008. The influence of weather conditions on the activity of high-arctic arthropods inferred from long-term observations [online]. BMC Ecology, 8. Available from http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2390509/pdf/1472-6785-8-8.pdf [accessed 27 December 2012].CrossRefGoogle Scholar
Høye, T.T., Post, E., Meltofte, H., Schmidt, N.M., Forchhammer, M.C. 2007. Rapid advancement of spring in the High Arctic. Current Biology, 17: R449R451.CrossRefGoogle ScholarPubMed
Huey, R.B.Berrigan, D. 2001. Temperature, demography, and ectotherm fitness. The American Naturalist, 158: 204210.CrossRefGoogle ScholarPubMed
Jenouvrier, S., Caswell, H., Barbraud, C., Holland, M., Stroeve, J., Weimerskirch, H. 2009. Demographic models and IPCC climate projections predict the decline of an emperor penguin population. Proceedings of the National Academy of Sciences of the United States of America, 106: 18441847.CrossRefGoogle ScholarPubMed
Klaassen, M., Lindstrom, A., Meltofte, H., Piersma, T. 2001. Ornithology – Arctic waders are not capital breeders. Nature, 413: 794794 . doi:10.1038/35101654.CrossRefGoogle Scholar
MacLean, S.F. 1973. Life cycle and growth energetics of the Arctic Crane fly Pedicia hannai antenatta. Oikos, 24: 436443.CrossRefGoogle Scholar
MacLean, S.F.Pitelka, F.A. 1971. Seasonal patterns of abundance of tundra arthropods near Barrow. Arctic, 24: 1940.CrossRefGoogle Scholar
McCullagh, P.Nelder, J.A. 1989. Generalized linear models. Chapman and Hall, London, United Kingdom.CrossRefGoogle Scholar
McKinnon, L., Picotin, M., Bolduc, E., Juillet, C., Bêty, J. 2012. Timing of breeding, peak food availability, and effects of mismatch on chick growth in birds nesting in the High Arctic. Canadian Journal of Zoology, 90: 961971 . doi:10.1139/z2012-064.CrossRefGoogle Scholar
Meltofte, H.Høye, T.T. 2007. Reproductive response to fluctuating lemming density and climate of the long-tailed Skua Stercorarius longicaudus at Zackenberg, Northeast Greenland, 1996–2006. Dansk Orn Foren Tidsskr, 101: 109119.Google Scholar
Meltofte, H., Høye, T.T., Schmidt, N.M. 2008. Effects of food availability, snow and predation on breeding performance of waders at Zackenberg. In High-Arctic ecosystem dynamics in a changing climate. Edited by H. Meltofte, T.R. Christensen, B. Elberling, M.C. Forchammer, and M. Rasch. Elsevier Academic Press Inc., San Diego, California, United States of America. Pp. 325341.CrossRefGoogle Scholar
Morrison, R.I.G., Davidson, N.C., Piersma, T. 2005. Transformations at high latitudes: why do red knots bring body stores to the breeding grounds? Condor, 107: 449457 . doi:10.1650/7614.CrossRefGoogle Scholar
Pearce-Higgins, J.W. 2010. Using diet to assess the sensitivity of northern and upland birds to climate change. Climate Research, 45: 119130 . doi:10.3354/cr00920.CrossRefGoogle Scholar
Pearce-Higgins, J.W.Yalden, D.W. 2004. Habitat selection, diet, arthropod availability and growth of a moorland wader: the ecology of European Golden Plover Pluvialis apricaria chicks. Ibis, 146: 335346.CrossRefGoogle Scholar
Pearce-Higgins, J.W., Yalden, D.W., Dougall, T., Beale, C. 2009. Does climate change explain the decline of a trans-Saharan Afro-Palaearctic migrant? Oecologia, 159: 649659 . doi:10.1007/s00442-008-1242-4.CrossRefGoogle ScholarPubMed
Pearce-Higgins, J.W., Yalden, D.W., Whittingham, M.J. 2005. Warmer springs advance the breeding phenology of golden plovers Pluvialis apricaria and their prey (Tipulidae). Oecologia, 143: 470476.CrossRefGoogle ScholarPubMed
Picotin, M. 2008. Variation climatique, abondance d'arthropodes et phénologie de la reproduction chez deux espèces de limicoles nichant dans le haut ArctiqueMaster. Université du Québec à Rimouski, Rimouski, Québec, Canada.Google Scholar
Post, E., Forchhammer, M.C., Bret-Harte, M.S., Callaghan, T.V., Christensen, T.R., Elberling, B., et al. 2009. Ecological dynamics across the Arctic associated with recent climate change. Science, 325: 13551358 . doi:10.1126/science.1173113.CrossRefGoogle ScholarPubMed
Rogers, L.E., Buschbom, R.L., Watson, C.R. 1977. Length–weight relationships of shrub-steppe invertebrates. Annals of the Entomological Society of America, 70: 5153.CrossRefGoogle Scholar
Root, T.L., Price, J.T., Hall, K.R., Schneider, S.H., Rosenzweig, C., Pounds, J.A. 2003. Fingerprints of global warming on wild animals and plants. Nature, 421: 5760.CrossRefGoogle ScholarPubMed
Roy, D.B., Rothery, P., Moss, D., Pollard, E., Thomas, J.A. 2001. Butterfly numbers and weather: predicting historical trends in abundance and the future effects of climate change. Journal of Animal Ecology, 70: 201217.CrossRefGoogle Scholar
Saether, B.-E. 1997. Environmental stochasticity and population dynamics of large herbivores: a search for mechanisms. Trends in Ecology and Evolution, 12: 143149.CrossRefGoogle ScholarPubMed
Sage, R.D. 1982. Wet and dry-weight estimates of insects and spiders based on length. American Midland Naturalist, 108: 407411.CrossRefGoogle Scholar
Sample, B.E., Cooper, R.J., Greer, R.D., Withmore, R.C. 1993. Estimation of insect biomass by length and width. American Midland Naturalist, 129: 234240.CrossRefGoogle Scholar
Schekkerman, H., Tulp, I., Calf, K.M., de Leeuw, J.J. 2004. Studies on breeding shorebirds at Medusa Bay, Taimyr, in summer 2002 [online]. In Alterra report 922. Alterra, Wageningen, The Netherlands. Available from http://edepot.wur.nl/23058 [accessed 30 December 2012].Google Scholar
Schekkerman, H., Tulp, I., Piersma, T., Visser, G.H. 2003. Mechanisms promoting higher growth rate in Arctic than in temperate shorebirds. Oecologia, 134: 332342.CrossRefGoogle ScholarPubMed
Smith, P.A., Gilchrist, G.H., Smith, J.N.M. 2007. Effects of nest habitat, food, and parental behavior on shorebird nest success. The Condor, 109: 1531 . doi:10.1650/0010-5422(2007)109[15:eonhfa]2.0.co;2.CrossRefGoogle Scholar
Strathdee, A.T., Bale, J.S., Block, W.C., Coulson, S.J., Hodkinson, I.D., Webb, N.R. 1993. Effects of temperature elevation on a field population of Acyrthosiphon svalbardicum (Hemiptera: Aphididae) on Spitsbergen. Oecologia, 96: 457465.CrossRefGoogle ScholarPubMed
Thomas, D.W., Blondel, J., Perret, P., Lambrechts, M.M., Speakman, J.R. 2001. Energetic and fitness costs of mismatching resource supply and demand in seasonally breeding birds. Science, 291: 25982600.CrossRefGoogle ScholarPubMed
Thomas, C.D., Cameron, A., Green, R.E., Bakkenes, M., Beaumont, L.J., Collingham, Y.C., et al. 2004. Extinction risk from climate change. Nature, 427: 145148 . doi:10.1038/nature02121.CrossRefGoogle ScholarPubMed
Tulp, I. 2007. The Arctic pulse, timing of breeding in long-distance migrant shorebirds [online]. Ph.D. thesis, University of Groningen. Available from http://www.waddenacademie.nl/fileadmin/inhoud/pdf/06-wadweten/Proefschriften/Thesis_ITulp_verkl.pdf [accessed 30 December 2012].Google Scholar
Tulp, I.Schekkerman, H. 2001. Studies on breeding shorebirds at Medusa Bay, Taimyr, in summer 2001 [online]. In Alterra report 451. Alterra, Wageningen, The Netherlands. Available from http://edepot.wur.nl/21801 [accessed 30 December 2012].Google Scholar
Tulp, I.Schekkerman, H. 2008. Has prey availability for Arctic birds advanced with climate change? Hindcasting the abundance of tundra arthropods using weather and seasonal variation. Arctic, 61: 4860.CrossRefGoogle Scholar
Visser, M.E. 2008. Keeping up with a warming world; assessing the rate of adaptation to climate change. Proceedings of the Royal Society B – Biological Sciences, 275: 649659 . doi:10.1098/rspb.2007.0997.CrossRefGoogle ScholarPubMed
Wagner, T.L., Olson, R.L., Willers, J.L. 1991. Modeling arthropod development time. Journal of Agricultural Entomology, 8: 251270.Google Scholar
Whittaker, J.B.Tribe, N.P. 1998. Predicting numbers of an insect (Neophilaenus lineatus: Homoptera) in a changing climate. Journal of Animal Ecology, 67: 987991.CrossRefGoogle Scholar
Wigglesworth, V.B. 1972. The principles of insect physiology. Chapman and Hall, London, United Kingdom.CrossRefGoogle Scholar
Figure 0

Fig. 1 Arthropod trap in the field.

Figure 1

Table 1 Total dry mass of the different families averaged over the number of years of sampling.

Figure 2

Fig. 2 Predicted and observed total daily arthropod availability in wet habitat in four Canadian Arctic sites.

Figure 3

Table 2 Parameter estimates for the generalized linear mixed models of weather variables on availability of the dominant arthropod groups (dry mass of arthropod expressed in mg/trap/day).

Figure 4

Fig. 3 Predicted and observed arthropod availability split by sites, year, and habitat for the most abundant families.

Figure 5

Fig. 3 continued

Figure 6

Fig. 3 continued

Figure 7

Fig. 4 Average (+SE) of the % of deviance explained by the different weather variables from models generated for each family and year. Weather variables are as follows: D is the accumulation of thaw degree-days, D2 is the square of the accumulation in thaw degree-days, T is average daily temperature, Wi is average daily wind speed, H is average daily relative humidity, R is average daily incidental radiation, Rain is total daily precipitation between two trapping sessions, and T × D is the interaction between T and D.

Figure 8

Fig. 5 Cross-validation models based on data from Bylot Island. We constructed models from all the data available minus one year. The observed values from the year left out (indicated on the left of the graphs) were then contrasted with the predictions from the cross-validation model.