Global warming is changing the dynamics of Arctic host–parasite systems

Abstract

Global climate change is altering the ecology of infectious agents and driving the emergence of disease in people, domestic animals, and wildlife. We present a novel, empirically based, predictive model for the impact of climate warming on development rates and availability of an important parasitic nematode of muskoxen in the Canadian Arctic, a region that is particularly vulnerable to climate change. Using this model, we show that warming in the Arctic may have already radically altered the transmission dynamics of this parasite, escalating infection pressure for muskoxen, and that this trend is expected to continue. This work establishes a foundation for understanding responses to climate change of other host–parasite systems, in the Arctic and globally.

1. Introduction

(a) Arctic climate change and infectious diseases

Recent anthropogenic climate change has altered physical and biological systems globally (McCarthy et al. 2001; Hassol 2004). Shifting patterns of abundance and distribution of pathogens, including metazoan and protozoan parasites, and the emergence of infectious disease in people, livestock and wildlife, are among the most important impacts of climate change (Epstein 1997; Daszak 2000; Kovats et al. 2001; McCarthy et al. 2001; Parmesan & Yohe 2003; Root et al. 2003). In the Arctic, where the effects of global climate change are already profoundly evident, dramatic alterations in host–parasite interactions are anticipated (Dobson et al. 2003; Hoberg et al. 2003; Hassol 2004; Kutz et al. 2004). Arctic species, including ungulates and their pathogens and the invertebrate vectors, have evolved under severe seasonal and environmental constraints (Strathdee & Bale 1998; Hoberg 2005), and the life history patterns of these species can be dramatically altered by even minor climatic perturbations (Graham et al. 1996). Thus, the Arctic serves as a sentinel, where studies to detect, understand and predict the responses of high latitude host–parasite systems to changing temperature can provide considerable insight into biotic implications of warming on a global scale. Empirical data necessary to evaluate the impacts of climate change on infectious diseases are, however, rare, and the few descriptive models available have focused on tropical or temperate ecosystems (Harvell et al. 2002).

(b) A model for the impacts of climate change on the ecology of arctic parasites

To explore the impacts of climate change on host–parasite systems in the Arctic, we investigated the ecology of an important protostrongylid lung-dwelling nematode, Umingmakstrongylus pallikuukensis, in muskoxen from the Canadian Arctic (Kutz et al. 2001a, 2002). Prevalence of infection approaches 100% in the endemic region, and the parasite can cause respiratory compromise and predispose muskoxen to predation (Kutz et al. 2001a). Development from first-stage larvae (L1), shed in the faeces of the muskox, to third-stage larvae (L3), infective to the definitive host, occurs in a gastropod intermediate host. Infection follows ingestion of a gastropod containing L3, or L3 emerged from a gastropod (Kutz et al. 2001a). The slug Deroceras laeve, common in the region endemic for U. pallikuukensis, is an important intermediate host for this and other protostrongylids (Samuel et al. 1985; Kutz et al. 2001b).

Previously, we determined that rates of development of U. pallikuukensis from L1 to L3 in D. laeve on the arctic tundra can be predicted from a simple degree-day mathematical model (Kutz et al. 2002). This model incorporates hourly soil-surface temperatures, the threshold temperature (8.5 °C), the thermal constant (167 degree days (DD), the amount of heating above threshold required for development to L3 as determined in laboratory studies), and a maximum temperature set at 21 °C to account for microhabitat selection by slugs (Kutz et al. 2001b, 2002). In the present study, we apply this model to examine temporal and quantitative patterns of parasite development in the past, and in a future of climate warming. We assumed that the availability, survival and immunity of slugs, survival of L1 and L3, and all other climate parameters other than temperature remained constant. In this conservative model, these and other potentially relevant variables cannot be easily estimated in the absence of empirical data from this or other related host–parasite systems.

2. Material and methods

(a) Historical patterns of larval development

To estimate historical rates of larval development we used hourly air temperatures from the airport at Kugluktuk, Nunavut (67°50′N, 115°06′W; Environment Canada Meteorological Service) for the period 1978–2003 (hourly temperatures were not available for years prior to 1978). DD accumulated above the threshold of 8.5 °C were calculated as per Kutz et al. (2002) for each year from 1st May to 30th September. Hourly soil-surface temperatures, which are consistent predictors of development for U. pallikuukensis in slugs, are typically warmer than air temperatures in the early summer (up to 5 °C depending on year, month and time), but approximate air temperatures later in the summer (Kutz et al. 2002). Unfortunately, historical soil-surface temperatures were not available for this or any nearby site and, therefore, our predictions based on air temperatures may underestimate development rates in the early summer (for example, in 1998 from 10th June to 1st July, almost four times as many DD were accumulated based on surface (74DD) versus air (19DD) temperatures, but over the following three weeks only 1.2 times as many DD were accumulated at the surface (138DD) versus air (116DD)).

(b) Predicted effects of climate warming on patterns of larval availability

(i) Calculation of accumulated degree days for warming scenarios

In the region where U. pallikuukensis is endemic, a climate warming scenario projects increases in mean temperature of ≥2 °C by the 2020s, ≥4 °C by the 2050s and ≥6 °C (to a maximum of 8 °C) by the 2080s (Canadian Centre for Climate Modelling and Analysis Global Coupled Model 2, A21 Economic Regional Focus Simulation 1 http://www.cics.uvic.ca/scenarios/). We used the mean hourly air temperatures from 1978 to 2000 as a baseline and increased these hourly values by temperature increments of 1 °C to represent climate warming (note that although scenarios for warming indicate minimum temperatures will increase more than maximum temperatures, for the sake of simplicity this was not incorporated into our model). We then calculated the accumulated DD for each increase in temperature starting on 29th May. This date was used because this was the first day of the year on which DD were accumulated under the most extreme warming scenario of 8 °C.

(ii) Calculation of numerical availability of L3 for warming scenarios

We modelled L3 availability on a weekly basis throughout the summer under two different larval survival scenarios: (i) No L3 mortality—L3 were available in slugs or as emerged L3 and had 100% survival to October (note: emergence of L3 from slugs is a common phenomenon for this parasite and up to 100% of the larvae can emerge in a summer (Kutz et al. 2000)) and (ii) L3 mortality—L3 did not emerge and died when slugs died.

We assumed, for the purposes of this exercise, that cohorts of 10 slugs were infected on a weekly basis from 29th May to the end of September. Thus, cohort 1 was infected on 29th May and was not re-infected throughout the summer, cohort 2 was infected on 5th June and was not re-infected, and so forth. We also assumed that the infection dose and establishment rates among the different cohorts were the same. The weekly survival rates in our model were derived from a Cox regression of the average survival rates of infected slugs from Arctic field trials that took place from June through September in 1997 (Kutz et al. 2002). We used 13.6 as a constant for the number of L3/slug; this was based on the average number of L3/slug observed in experimentally infected slugs in these field trials (Kutz et al. 2002).

Under scenario (i) no L3 mortality, for each temperature increment we multiplied the slug survival rate for the date that the first L3 was available by 13.6 (L3/slug). The L3 present on that date were assumed to remain available until October even if slugs died, as the L3 would emerge and still be capable of completing the cycle. Under scenario (ii) L3 mortality, we assumed that L3 did not emerge from the slugs and were not available to muskoxen when slugs died. Therefore, calculations incorporated the weekly rates for slug survival (L3 availability in each cohort declined with time as slugs died). Under both scenarios, the L3 availability for a given date is the sum of L3 available from all cohorts on that date.

3. Results

(a) Historical patterns of larval development

Our model calculations for rates of parasite development from 1978 to 2003 indicate considerable inter-annual variability in the accumulated DD (figure 1). Only in some years were there sufficient DD (167) for development to L3 (a 1 year development cycle), and such years were more common from 1991 to 2003 (12 of 13), than from 1978 to 1990 (5 of 13; figure 1). When sufficient DD were accumulated in a single year for development to L3, the window for transmission of L3 from slugs ranged from 6 (1995) to 72 days (1996; table 1). In years when there were insufficient DD for development to L3, larvae would have had to over-winter in gastropods and resume development to L3 the following year (a 2 year development cycle).

Figure 1.

Annual accumulated degree days predicted near Kugluktuk, Nunavut for 1978–2003. Note that development to L3 (as indicated by accumulating at least 167DD) would be predicted in only 5 years from 1978 to 1990, but in 12 years from 1991 to 2003.

Table 1.

Predicted availability of L3 near Kugluktuk, Nunavut from 1978 to 2003.

year	date of first L3a	last date for slug infectionb	last day for transmission by ingestion of infected slugsc	transmission period (days)d	degree days accumulated
1978	incompletee	n/af	29-Sep	0	71
1979	incomplete	n/a	27-Sep	0	133
1980	incomplete	n/a	16-Sep	0	140
1981	16-Aug	10-Jul	2-Oct	47	196
1982	16-Aug	23-Jun	21-Sep	36	174
1983	incomplete	n/a	3-Oct	0	163
1984	17-Aug	20-Jun	28-Sep	42	189
1985	incomplete	n/a	1-Oct	0	111
1986	incomplete	n/a	28-Sep	0	115
1987	incomplete	n/a	3-Oct	0	90
1988	15-Aug	8-Jul	8-Oct	54	237
1989	20-Jul	25-Jul	16-Sep	58	369
1990	incomplete	n/a	28-Sep	0	160
1991	16-Aug	19-Jun	21-Sep	36	180
1992	11-Aug	7-Jul	18-Sep	38	209
1993	19-Aug	24-Jun	23-Sep	35	180
1994	20-Jul	20-Jul	25-Sep	67	337
1995	21-Sep	6-Jun	27-Sep	6	175
1996	16-Jul	6-Jul	26-Sep	72	297
1997	4-Aug	17-Jul	30-Sep	57	232
1998	26-Jul	24-Jul	5-Oct	71	328
1999	incomplete	n/a	27-Sep	0	130
2000	1-Aug	18-Jul	22-Sep	52	253
2001	1-Sep	13-Jul	13-Oct	42	195
2002	6-Sep	9-Jun	28-Sep	22	173
2003	27-Aug	5-Jul	11-Oct	45	198

^aBased on infection of gastropods on 1st May.

^bLast date that larvae infecting slugs could mature to L3 within that summer.

^cLast date before 5 consecutive days with below freezing average daily temperatures.

^dThis denotes the transmission period for L3 in slugs and assumes that slugs become unavailable when temperatures go below freezing for 5 consecutive days; emerged L3 may still be transmitted beyond this period. (Last day for transmission−date of first L3.)

^eIndicates incomplete development—larvae did not complete development to L3 within a single summer.

^fL3 could not develop within a single summer, thus there is no last date for infection.

(b) Predicted effects of climate warming on patterns of larval availability

Climate warming of as little as 1 °C above the 1978–2000 average shifted larval development from a 2 year cycle to a 1 year cycle (figure 2). Warming also expanded the period for transmission of L3 in slugs to muskoxen (defined as the time from when the first L3 became available in slugs to the last day before 5 consecutive days with average temperatures below 0 °C, when slugs are assumed to move deeper into the soil and begin hibernation) from 44 days (1 °C increase) to 105 days (8 °C increase; table 2). This is due to L3 becoming available 42 days earlier, and to a delay of 19 days in anticipated slug hibernation (table 2). Concomitantly, warming also allowed larval development to begin later in the summer and still be completed within the year (table 2). The accelerated development rates and expanded window for development and transmission also resulted in a considerable increase in numbers of infective L3 available to muskoxen in both scenario (i) no L3 mortality and scenario (ii) L3 mortality (figure 3a,b). Actual L3 availability is predicted to be within the range represented by the two scenarios in figure 3a,b.

Figure 2.

Predicted patterns of accumulated degree days with 0–8 °C increases above the 1978–2000 average, and assuming slugs are infected with L1 on 29th May. Note that as little as 1 °C increase results in accumulation of sufficient degree days for development to L3 in single summer.

Table 2.

Predicted availability of L3 near Kugluktuk, Nunavut with climate warming from 1 to 8 °C.

warming scenario (°C increase over 1978–2000 mean)	date for first L3a	last date for slug infectionb	last day of transmission by slug ingestionc	transmission periodd (days)	degree days accumulated
0	incomplete	n/a	n/a	n/a	141
1	15-Aug	30-Jun	28-Sep	44	181
2	03-Aug	11-Jul	29-Sep	57	236
3	26-Jul	20-Jul	03-Oct	69	305
4	20-Jul	26-Jul	11-Oct	83	384
5	16-Jul	02-Aug	14-Oct	90	466
6	11-Jul	07-Aug	15-Oct	96	567
7	07-Jul	12-Aug	16-Oct	101	670
8	04-Jul	16-Aug	17-Oct	105	777

^aBased on infection of gastropods on 29th May.

^bLast date that larvae infecting slugs could mature to L3 within that summer.

^cLast date before 5 consecutive days with below freezing temperatures.

^dThis denotes the transmission period for L3 in slugs and assumes that slugs become unavailable when temperatures go below freezing for 5 consecutive days; emerged L3 may still be transmitted beyond this period.

Figure 3.

Predicted L3 availability (L3 index) in a single season under climate warming scenarios of 1, 2, 4 and 8 °C for assumptions (a) L3 are available in slugs or as emerged L3 and have 100% survival to October and (b) L3 do not emerge and die when slugs die.

4. Discussion

Directional climate change has altered, and will continue to alter, distribution and phenology of a variety of taxa, and is predicted to have considerable impacts on the seasonal patterns of development and transmission of many pathogens in the Arctic and globally (Harvell et al. 2002; Dobson et al. 2003; Parmesan & Yohe 2003; Kutz et al. 2004; Hoberg 2005). Our retrospective analysis for U. pallikuukensis suggests that, as a result of climate warming, L3 development has already shifted from a 2 year cycle to a predominantly 1 year cycle (figure 1). This nonlinear shift may have great ecological significance. For the 2 year cycles characteristic of 1978–1990, high over-winter mortality of slugs and developing larvae (Kutz et al. 2002) would have resulted in only a few L3 developing in the second year, and low-infection pressure for muskoxen. Unusually warm years during this period, such as 1988 and 1989, may have lead to a ‘pulse’ of L3 developing in a single summer, resulting in increased abundance of L3, a shift in temporal availability, and, possibly, disease outbreaks. Such a prediction is consistent with a linkage between unusually warm summer temperatures and periodic outbreaks of clinical disease for a related elaphostrongyline, Elaphostrongylus rangiferi, in reindeer in Finmark, Norway (Handeland & Slettbakk 1994). For the affected muskox population, a decline of 50% was observed from 1988 to 1994 (Fournier & Gunn 1998). Although this corresponded with the initial discovery of U. pallikuukensis in muskoxen, because of the remote nature of this population and the inherent costs and logistical difficulties associated with investigations in the Arctic, the role of the parasite in this decline was not determined.

From 1990 to 2003, the predicted pattern for larval development of U. pallikuukensis was one of multiple, consecutive, 1 year cycles. Based on our analyses of scenarios for climate warming, these 1 year cycles will be typical of the future (figure 2). These will extend the window of availability of L3 in slugs, augment numbers of L3 available (figure 3a,b), and could lead to increased intensity of infection in muskoxen, with adverse effects on their fecundity and survival. This is consistent with overall climate warming predictions for amplification of parasite populations through increased rates for development, reduction in generation times, and broadened seasonal windows for transmission (Hoberg et al. 2001).

The impacts of climate change on host–parasite interactions may differ depending on location within the parasite's ‘climate envelope’ (Sutherst 2001). For example, at the warmer southern and western extents of its current range, U. pallikuukensis is well established, with a prevalence approaching 100%. Muskoxen in these areas of ‘endemic stability’ (Sutherst 2004) may tolerate moderately increased infection pressure resulting from warming. However, extremely high levels of L3 and an expanded temporal window of L3 availability, together with other stressors including habitat perturbation, interspecific competition, or severe weather events associated with climate change, may induce parasitic disease (Gulland 1992; Sutherst 2004). In contrast, at the northeastern edge of the range of U. pallikuukensis, where cooler temperatures are currently a limiting factor, we might expect episodic range expansion (Kutz et al. 2002). In these cooler areas, we predict a pattern of primarily 2 year developmental cycles, with pulses of rapid larval development and accumulation associated with occasional warmer years, similar to the historical pattern (1978–1990). The resulting acute increase in infection pressure during warmer years may have significant implications for these muskox populations that have had little or no previous exposure (low-herd immunity), and may result in outbreaks of overt disease. Such disease outbreaks occurred in all age classes of caribou following the introduction of E. rangiferi to a naive population in Newfoundland (Ball et al. 2001).

Ecological disturbances associated with climate change may alter other biotic parameters that affect the epidemiology of U. pallikuukensis. For example, availability, survival and immunity of intermediate hosts, and survival of free-living larval stages are not incorporated into our model, but are important determinants of parasite persistence. D. laeve, an important intermediate host for U. pallikuukensis, has high phenotypic plasticity and survives in ecologically diverse habitats under a wide range of climatic conditions (Pilsbry 1948; Rollo & Shibata 1991). It will likely thrive in a changing climate, particularly in the north, where warmer and wetter conditions are anticipated (McCarthy et al. 2001). Importantly, larval stages in gastropods are buffered from the external environment because of microhabitat selection by these intermediate hosts, and consequently, survival and development rates of larvae are less likely to be impacted by climate variability and stochasticity (Saunders et al. 2002) than parasites with direct life cycles. Conversely, the free-living L1 and emerged L3 may be adversely affected by extreme temperatures, desiccation, ultraviolet radiation, and frequency of freeze–thaw cycles, all possible outcomes of climate change at high latitudes (Forrester & Senger 1963; Shostak & Samuel 1984; Hassol 2004). Climate-driven habitat perturbations are also predicted to influence habitat quantity and quality, nutrition, behaviour, immune function and patterns of geographic distribution and abundance of definitive hosts, sympatry with other ungulates, and the occurrence and emergence of an array of other parasites and pathogens (Hoberg et al. 2002, 2005; Hassol 2004). The final outcome for the health and persistence of muskox populations will depend on the interactions among these numerous climate-linked factors.

Our model has been validated by field studies on P. odocoilei, a phylogenetically distinct protostrongylid of Dall's sheep in a sub-arctic alpine habitat, and results indicate that possible effects of climate change in that system include parasite range expansion and amplification in endemic regions (Jenkins et al. in press). Our model is also directly applicable to other protostrongylid parasites of wild ungulates. For example, threshold temperatures and thermal constants have been determined for E. rangiferi and Protostrongylus stilesi (parasites of Rangifer and Ovis, respectively; Halvorsen & Skorping 1982; Samson & Holmes 1985) and could be incorporated into the model to develop testable hypotheses for current and future distribution and transmission patterns of these parasites. Application of the U. pallikuukensis model to these and other host–parasite systems will generate insights into transmission patterns across a wide range of species, and will generate hypotheses exploring responses to climate change.

Climate change will alter biotic and abiotic conditions and dissolve ecological barriers, redrawing maps of current distribution of parasites and their hosts (Dobson et al. 2003; Hoberg 2005). The U. pallikuukensis model is a simple and conservative model that identifies and quantifies the effects of climate on critical life history stages of a nematode parasite, and explores responses of host–parasite assemblages to climate change. Such models, founded on empirical data, serve as powerful predictive frameworks for tracking of seasonal, annual and long-term changes in parasitic infections from local to global scales, and provide the foundation for developing more complex quantitative and comparative models for the epidemiology of pathogens in a changing climate.