Impact of climate change on risk of incursion of Crimean-Congo haemorrhagic fever virus in livestock in Europe through migratory birds

Authors


Paul Gale, Animal Health and Veterinary Laboratories Agency, Weybridge, New Haw, Addlestone, Surrey KT15 3NB, UK. E-mail: paul.gale@ahvla.gsi.gov.uk

Abstract

Aims:  To predict the risk of incursion of Crimean-Congo haemorrhagic fever virus (CCHFV) in livestock in Europe introduced through immature Hyalomma marginatum ticks on migratory birds under current conditions and in the decade 2075–2084 under a climate-change scenario.

Methods and Results:  A spatial risk map of Europe comprising 14 282 grid cells (25 × 25 km) was constructed using three data sources: (i) ranges and abundances of four species of bird which migrate from sub-Saharan Africa to Europe each spring, namely Willow warbler (Phylloscopus trochilus), Northern wheatear (Oenanthe oenanthe), Tree pipit (Anthus trivialis) and Common quail (Coturnix coturnix); (ii) UK Met Office HadRM3 spring temperatures for prediction of moulting success of immature H. marginatum ticks and (iii) livestock densities. On average, the number of grid cells in Europe predicted to have at least one CCHFV incursion in livestock in spring was 1·04 per year for the decade 2005–2014 and 1·03 per year for the decade 2075–2084. In general with the assumed climate-change scenario, the risk increased in northern Europe but decreased in central and southern Europe, although there is considerable local variation in the trends.

Conclusions:  The absolute risk of incursion of CCHFV in livestock through ticks introduced by four abundant species of migratory bird (totalling 120 million individual birds) is very low. Climate change has opposing effects, increasing the success of the moult of the nymphal ticks into adults but decreasing the projected abundance of birds by 34% in this model.

Significance and Impact of the Study:  For Europe, climate change is not predicted to increase the overall risk of incursion of CCHFV in livestock through infected ticks introduced by these four migratory bird species.

Introduction

Crimean-Congo haemorrhagic fever (CCHF) is one of the most widely distributed tick-borne diseases in the world, affecting people in parts of Africa, Asia, Eastern Europe and the Middle East (Ergonul and Whitehouse 2007; WHO 2011). The causative agent, CCHF virus (CCHFV), is geographically the second most widespread of all medically important arboviruses after dengue. It belongs to the genus Nairovirus in the family Bunyaviridae (Ergonul and Whitehouse 2007) and circulates in nature in a tick–vertebrate–tick cycle. CCHFV is transmitted by ticks of the genus Hyalomma, in particular Hyalomma marginatum.

Humans are normally infected either through the bite of an infected tick, or through contact with blood and organs from a host infected with CCHFV during the acute phase of infection. The probability of becoming clinically ill for humans who have been infected with CCHFV is around 20% with case fatality rates of 5% in Turkey and 33% in Kosovo (Ergonul and Whitehouse 2007). Starting in 2002, outbreaks of CCHF have occurred in Turkey with more than 2500 cases in humans and CCHF is endemic in several Balkan countries (Maltezou and Papa 2010). Mild et al. (2010) have reviewed the migration of CCHFV and conclude that future outbreaks may extend to hitherto-naïve areas of Europe, given the fact that the tick vector is present in Western Europe. Therefore, from a European Union perspective, it is important to understand the ecology of the virus and thus identify geographical areas that are suitable for transmission of CCHFV. This will facilitate assessment of how the risk will be altered in the future as a result of changes in climate, land use and socio-economic conditions.

CCHFV infection is generally subclinical in animals and livestock. It is nevertheless important for a number of reasons (Gale et al. 2010). First, domestic ruminants may develop a transmissible viraemia following infection and hence amplify the pathogen in agricultural environments, thus serving as a reservoir for infection of feeding adult ticks (Sang et al. 2011). Second, humans may be infected through those ticks infected by feeding on livestock. Third, the virus may be spread to butchers and abattoir workers through contact with blood or tissues from viraemic livestock during their slaughter and butchering (Tahmasebi et al. 2010).

According to WHO (2011), there is virological and serological evidence of CCHFV in most African countries south of the Sahara in addition to Egypt, and CCHF cases in humans have been reported in South Africa, Democratic Republic of Congo, Kenya, Namibia, Tanzania and Uganda. Many species of almost all wild bird orders that breed in Europe in the summer are migratory and fly south in the autumn to over-winter in sub-Saharan Africa, returning to Europe in the following spring. Hyalomma ticks are two host ticks (Hillyard 1996; Ergonul and Whitehouse 2007). Thus, the larvae moult into nymphs while attached to the bird, lengthening the duration of host attachment (12–26 days) and so enabling the passive transport of the immature Hyalomma ticks by migrating birds over long distances (Hoogstraal 1972; Hillyard 1996). As an example, an adult male Hyalomma marginatum rufipes tick was identified on a horse in the Netherlands during a survey of tick species (Nijhof et al. 2007). As that horse was not imported, Nijhof et al. (2007) speculated that the tick was introduced as a nymph by a migratory bird from Africa. Hyalomma marginatum rufipes is endemic in many regions of Africa and has been recorded on migratory birds in spring in Europe (Molin et al. 2011).

It is well established that the immature tick stages (and not the adult) of H. marginatum infest birds (and small mammals), while adults feed on larger mammals such as livestock and wild boar (Sus scrofa) (Hoogstraal 1972; Hillyard 1996; Molin et al. 2011). This is an important consideration for developing a risk assessment (Gale et al. 2010) because if migratory birds introduce immature ticks but only the adults feed on livestock, then a prerequisite for infection of livestock is that those immature ticks moult into adults before taking their next blood meal on a livestock animal. Certain temperature and humidity conditions are needed for moulting of immature stages of H. marginatum to adults (Estrada-Peña et al. 2011). Thus, climate change will directly affect the efficiency of moulting of immature ticks and is also predicted to affect the distribution of many species of migratory birds in Europe (Huntley et al. 2007).

The objective of this study is to build a spatial model for the risk of incursion of CCHFV in livestock in Europe through the introduction of infected ticks on migratory birds from sub-Saharan Africa at the current time (2005–2014) and in the future (2075–2084), given a predicted climate-change scenario. Four bird species are studied, namely Willow warbler (Phylloscopus trochilus), Northern wheatear (Oenanthe oenanthe), Tree pipit (Anthus trivialis) and Common quail (Coturnix coturnix). The Willow warbler and Tree pipit are two of the most abundant species in northern Europe, and the Willow warbler is the most abundant spring migrant from Africa (BirdLife International 2004). The Common quail is abundant through much of southern and central Europe. Immature Hyalomma ticks have been found on Tree pipits, Willow warblers and Northern wheatears in spring in Egypt, Cyprus, Italy and Greece (Hoogstraal et al. 1964; Kaiser et al. 1974; Molin et al. 2011) and on Northern wheatear in Norway (Hasle et al. 2009) and Great Britain (Martyn 1988).

Methods

The risk pathway is taken from Gale et al. (2010) and involves three layers:

  • Layer 1 – abundance of migratory birds to predict the release of CCHFV through infected Hyalomma nymphs carried on migratory birds;
  • Layer 2 – temperature to predict the moult of nymphs to adult ticks; and
  • Layer 3 – Livestock density to estimate the probability of an adult tick finding a livestock host to feed upon.

For the spatial risk map, the land mass of Europe was divided into 14 282 grid cells of dimensions 25 × 25 km reflecting the resolution of the climate and bird data.

Number of CCHFV-positive Hyalomma marginatum nymphal ticks entering Europe through infested migratory birds

Estimation of the prevalence of CCHFV in Hyalomma marginatum nymphs on birds  CCHFV prevalence, although focally endemic, is highly variable in space and time in sub-Saharan Africa (Wilson et al. 1990; Sang et al. 2011). There are no data on the proportion of H. marginatum nymphs in Africa which are infected with CCHFV. A recent study has reported that of 1144 pools of 8600 Hyalomma spp. ticks taken from different livestock animals in north-eastern Kenya in April/May 2008, 23 pools were infected with CCHFV as determined by RT-PCR (Sang et al. 2011). If just one tick were infected in each pool, then the proportion of ticks infected with CCHFV would be 0·0027. However, more than one tick in each pool may have been infected. The average number of ticks per pool is 7·5, and if all ticks in each of the 23 positive pools were infected, then an estimate of the proportion of ticks infected with CCHFV would be 0·020. Although not specified, those ticks were likely to have been adult ticks and all positive pools were taken from large animals, namely cattle or camels. Immature ticks can be infected through feeding on viraemic, small vertebrate hosts and through transovarial transmission from the adult female tick. Interpretation of data from Gonzalez et al. (1992) suggests that the proportion of infected nymphs from an infected female Hyalomma truncatum tick may be low. Logan et al. (1989) demonstrated that the overall CCHFV infection rate for larval H. truncatum ticks after engorging on viraemic newborn mice was only 4·4%, and in endemic regions, only a proportion of those small mammal hosts on which the larvae feed will be infected and viraemic. Furthermore, not all of the birds will come from CCHF endemic areas in sub-Saharan Africa. Thus, in Senegal, which is a wintering area for migrant birds, Wilson et al. (1990) reported that the spatial pattern of CCHFV transmission varied with prevalence highest in the north, decreasing to nil in the south. Given this information, it is assumed that the probability, pprev, that a nymph on a migratory bird is infected is 10−4 (Table 1).

Table 1.   Summary of parameters for estimation of the probability of incursion of CCHFV in one or more livestock animals in a 25 × 25 km grid cell (pinc)
ParameterDescriptionValue (see text)
  1. CCHFV, Crimean-Congo haemorrhagic fever virus.

pprevProportion of immature Hyalomma marginatum nymphs which are infected with CCHFV10−4
NbirdsTotal number of birds per grid cell per yearLayer 1
NtempNumber of years per decade in which criterion of moulting is met in a each grid cellLayer 2
NTotal livestock numbers per grid cellLayer 3
MeantickAverage number of H. marginatum nymphs per migrant bird0·049
pmonthProportion of birds arriving in Europe in given month (Table 3)0·22, March
0·64, April
0·14, May
pfindProbability of adult H. marginatum tick finding livestock animal in a grid cell in a 30 days questing period (given N = 1)0·000046
pTSTProbability of trans-stadial transmission of CCHFV from nymph to adult during moult0·69
psurvProbability of survival of nymph in the environment (based on Ixodes ricinus)0·1

Selection of migratory bird species for inclusion in the risk assessment  Most migrant bird species infested with Hyalomma ticks on arriving in Europe were ground-feeding species including warblers, shrikes and chats which frequent dry habitats suitable for Hyalomma ticks (Kaiser et al. 1974; Martyn 1988; Molin et al. 2011). For this preliminary assessment, it was agreed that there are too many migratory bird species in Europe for all to be included. Therefore, four species of bird were chosen (Table 2) on the basis that they have high abundance in Europe, migrate from sub-Saharan Africa and have contact with the ground in arid, shrub or grassland-type habitats where H. marginatum ticks may be present.

Table 2.   European populations of the four species of migrant birds studied
SpeciesTotal European population* (Breeding pairs)Total number of individual birds used in model (2 × average number of breeding pairs)% of total birds in study (based on averages)Projected number of individual birds in model in climate-change scenario
  1. *Higher values from EBCC (1997) and BirdLife International (2004).

  2. †Total of averages.

Common quail640 000–1300 000 (Average 970 000)1940 0001·61987 560
Northern wheatear2600 000–3800 000 (Average 3200 000)6400 0005·45044 596
Tree pipit15 000 000–19 000 000 (Average 17 000 000)34 000 00028·726 116 913
Willow warbler27 000 000–49 000 000 (Average 38 000 000)76 000 00064·244 435 762
Total59 170 000†118 340 000 77 584 831

Data on the current abundance and distribution of the four bird species across Europe  Average numbers of breeding pairs were taken from EBCC (1997) or BirdLife International (2004) and multiplied by two to give the number of individual birds (Table 2) for the risk assessment. Simulated present distributions of the four bird species were obtained from Huntley et al. (2007). To convert the presence/absence records on the bird distribution maps of Huntley et al. (2007) into quantitative estimates of bird density (Nbirds) for each species (Layer 1), it was assumed that the total European populations (set out in Table 2) are evenly distributed across all grid cells where the species is present.

Monthly variation in arrival times of migrants over the 3 months of spring in Europe  There are large differences in the temperatures between March and May in Europe, and to synchronise the impact of the temperature criterion for moulting of the nymphal ticks, it is important to accommodate the variation in the arrival times of the individual birds at their breeding sites. The proportions of Willow warblers, Northern wheatears and Tree pipits arriving in Europe during each month were estimated from data provided by Finlayson (1992) on migratory bird passage in southern Spain (Table 3). No data were available for Common quail.

Table 3.   Proportion of spring migrants arriving in March, April and May in Gibraltar (southern Spain)
SpeciesMarchAprilMay
  1. Data from Finlayson (1992).

  2. *The average proportion is applied to all birds in this model.

Willow warbler0·1820·5450·273
Northern wheatear0·2360·6730·091
Tree pipit0·2310·6920·077
Average (pmonth)*0·220·640·14

Estimation of the average number of Hyalomma marginatum ticks per bird Molin et al. (2011) collected 386 ticks from 7453 birds (giving an average of 0·052 ticks per bird) in a study of migratory birds trapped at Capri (Italy) and Antikythera (Greece) between 2 April 2009 and 18 May 2009 (Table 4). Allowing for the finding that 95·5% of those ticks were Hyalomma spp., the average number of Hyalomma spp. ticks per migratory bird (Meantick) is 0·049. This value is used in the risk assessment for all four bird species (Table 1). Tick infestation frequencies are not allocated according to bird species for this risk assessment because only four individual Northern wheatears and no Common quails were trapped in that study (Table 4).

Table 4.   Data for tick detections on selected migratory bird species during spring migration in Italy and Greece in 2009
Bird species*No. birds trappedNo. ticks†No. (%) birds infestedMean no. ticks per infested birdMean no. larvae per infested birdMean no. nymphs per infested bird
  1. Data from Molin et al. (2011).

  2. *No Common quail were trapped by Molin et al. (2011). No ticks were recovered on four Common quail trapped in the spring in Egypt (Hoogstraal et al. 1964).

  3. †95·5% of the 386 ticks found were Hyalomma species, and DNA analysis of 10 Hyalomma ticks revealed nine were Hyalomma marginatum rufipes and one was Hyalomma marginatum marginatum.

Tree pipit208117 (3·4)1·60·11·3
Northern wheatear422 (50)1·00·01·0
Willow warbler46433 (0·7)1·00·30·7
Total (22 species of passerine)7453386200 (2·7)1·90·61·7

Calculation of number of CCHFV-infected Hyalomma marginatum nymphs coming in to each grid square on birds  The total number of CCHFV-infected nymphal ticks (NPosTicks) entering each grid cell per year is calculated as:

image(1)

These are converted to monthly values for March, April and May using the average proportion (pmonth) for each month in Table 3.

Impact of climate change on the abundance and range of migrant birds in Europe  Simulated potential distributions for each of the four species of bird in the late 21st century were taken from Huntley et al. (2007). For the purpose of the work here, it is assumed that the number of birds per grid cell remains constant with climate change, i.e., with the same density as in the current map, given the bird is present in a cell.

Probability of moult of nymphs to adults

Nymph stages out-numbered larvae by over 2 : 1 in studies of H. marginatum rufipes ticks on migratory birds trapped in Italy, Greece and Cyprus in spring (Kaiser et al. 1974; Molin et al. 2011). It is therefore assumed here that all H. marginatum ticks on birds are nymphs and that only one moult needs to be modelled, namely that of nymphs to adults. This is a worst case assumption.

Temperature criterion for moulting of Hyalomma marginatum nymphs to adults  Temperature is the main factor affecting the seasonal pattern of the H. marginatum tick (Estrada-Peña et al. 2011) and is therefore used here as the sole parameter for moulting. For the purpose of the model here, the criterion for moulting of the immature (nymph) tick is a temperature of >8°C for 15 continuous days.

Number of years per decade during which moulting criterion is met in each month.  Predictions for noon air temperature at 1·5 m above ground level each day over the 150-year period between December 1949 and December 2099 were provided from the HadRM3 model (Met Office and Hadley Centre 2008). The medium emissions (SRESA1B) scenario was used. The number of years in a decade (Ntemp) in which the condition for moulting is met (at least once) was calculated for each 25 × 25 km grid cell for the months of March, April and May in the current decade (2005–2014) and with the climate-change scenario in the decade, 2075–2084.

Distribution of livestock in Europe

Data from FAO (2007) for current distributions of cattle, sheep and goats were combined to give a single layer representing the main livestock in Europe susceptible to CCHFV (Wilson et al. 1990). This is shown in Fig. 1.

Figure 1.

 Density for cattle, sheep and goats combined in Europe in 2005 (FAO 2007). (inline image) 0.0000–9.2849; (inline image) 9.2850–24.7197; (inline image) 24.7198–45.8911; (inline image) 45.8912–74.2713; (inline image) 74.2714–113.0512; (inline image) 113.0513–168.5150; (inline image) 168.5151–243.8801; (inline image) 243.8802–351.7977; (inline image) 351.7978–540.4635 and (inline image) 540.4636–1105.0063.

Probability of an adult tick finding a livestock animal

Let pfind be the probability of an adult H. marginatum tick finding one livestock animal (given one livestock animal is present) in a 25 × 25 km grid cell. Hyalomma marginatum actively seeks out host animals, being guided mainly by visual cues. On the basis of data from target perception experiments conducted by Kaltenrieder (1990), it is assumed here that under average field conditions, a Hyalomma tick walks at 1·0 cm s−1 and is able to detect a livestock animal at a distance of ≤10 m. There are 10 000 cells of 10 × 10 m dimensions in a 1 × 1 km area. The probability of a tick detecting a livestock animal in a 1 × 1 km area is represented by the number of 10 × 10 m cells that the tick can walk through during the 30-day questing period used by Estrada-Peña et al. (2011). Thus, overlaying a first-order mortality rate of 0·0095 day−1 (Estrada-Peña et al. 2011) and allowing a 14-day period for cuticle hardening before commencement of questing predicts that a tick questing for 8 h day−1 covers, on average, 577 10 × 10 m cells over the 30-days questing period. This represents a 0·058 fraction of a 1 × 1 km area. The probability of an adult tick finding a livestock animal in a 25 × 25 km (625 km2) grid cell is therefore (0·058/625) 9·2 × 10−5. Large wild mammals such as horses, deer and wild boar would ‘compete’ with the livestock for the adult ticks, such that the risk of tick contact with livestock would be reduced (dilution effect). Although such data have not yet been obtained, a preliminary assumption here is that large wild mammals decrease the probability of an adult H. marginatum tick finding a livestock animal by 50%. The value used for pfind in the spatial model is therefore 4·6 × 10−5 (Table 1). It should be noted that any movement of the one livestock animal does not affect pfind, as the probability of contact is not affected by whether both parties are moving or just one.

Survival of the tick and trans-stadial transmission

The probability of survival of the nymph stage to the adult stage (psurv) of Ixodes ricinus is 0·1 according to Hartemink et al. (2008) and, in the absence of data for H. marginatum, the same value is adopted here (Table 1). Gordon et al. (1993) reported that a pool of H. truncatum nymphs in which 0·85% (16 of 1890) was infected gave a pool of adults in which 0·58% (12 of 2049) was infected, suggesting a trans-stadial transmission efficiency (pTST) of 0·69 (Table 1).

Calculation of the risk of incursion of CCHFV in livestock in Europe

The objective of the equations developed here is to calculate the risk (pinc) of incursion of CCHFV in one or more livestock animals in a given 25 × 25 km grid cell using the data for that cell in each of the three layers. Layer 2 gives the predicted number of years (Ntemp) over a 10-year period when the nymph moulting criterion (temperature >8°C for 15 consecutive days) is met during a particular month (March, April or May). The probability that the condition is met once during that month for a particular year in that decade is therefore:

image(2)

The probability of the CCHFV-infected nymph moulting to an adult and surviving to be an adult and of that adult being CCHFV-positive through trans-stadial transmission is given by

image(3)

The probability (pfind) of the adult tick finding one livestock animal (given one livestock animal is present) is 4·6 × 10−5. The probability, p2, of the adult tick finding a livestock animal, given N livestock hosts (Layer 3) are present, is therefore

image(4)

Let psuccess = probability of the infected nymph developing into an infected adult and subsequently finding a livestock host to feed on

image(5)

Thus, pinc, the overall probability of incursion of CCHFV in one or more livestock animals in the grid cell is given by:

image(6)

where NPosTicks is defined by eqn (1) using the bird density data in Layer 1. This assumes that given a CCHFV-infected adult tick has bitten a livestock animal, that animal will be infected. This is a worst-case assumption, although there is evidence that 11 adult H. marginatum rufipes ticks, a proportion of which were positive for CCHFV, were sufficient to infect a calf (Lee and Kemp 1970). Predicted absolute risks (pinc) are calculated for each of the 25 × 25 km grid cells comprising Europe for a given month. Relative risk maps are produced by subtracting pinc for the 2005–2014 decade from that for the 2075–2084 decade.

Calculation of annual number of grid cells across Europe in which at least one incursion occurs  Due to the fact that the number of cells is large (14 282) and the individual cell probabilities (pinc) are small, the average number of grid cells across Europe in which at least one incursion of CCHFV occurs in livestock per year during any month can be approximated by the sum of the individual probabilities.

Results

Prediction of absolute risk of CCHFV incursion in livestock in Europe

The predicted absolute risks of CCHFV incursion in livestock in 25 × 25 km grid cells are plotted in Fig. 2 for March, April and May under current conditions (decade 2005–2014) and taking the predicted consequences of the climate-change scenario into consideration (decade 2075–2084). The predicted absolute risks are influenced by month increasing both in magnitude and in spatial distribution across Europe from March to May. The highest predicted risk of at least one incursion in livestock over all three spring months combined in a single 25 × 25 km grid cell in the current decade is 0·0017 per year. This is a very low risk and represents one incursion in one or more livestock animals on average every 588 years in that grid cell, which was located on the north-west coast of France. The predicted numbers of 25 × 25 km (625 km2) grid cells across Europe in which at least one incursion occurs in livestock per year through the four bird species are presented in Table 5. Across Europe as a whole, the risk of incursion in one or more livestock animals is on average 7·3 × 10−5 per 25 × 25 km grid cell per year in the current decade (Table 5). Most of the incursion-positive cells are predicted to occur in the month of May despite most birds’ arriving in April (Table 3). The predicted risks are lowest in March both currently and with the climate-change scenario over the 3 months studied.

Figure 2.

 Predicted absolute risks (per year) of incursion of Crimean-Congo haemorrhagic fever virus in livestock in the months of March (a,b), April (c,d) and May (e,f) in 2005–2014 (a,c,e) and 2075–2084 using HadRM3 medium emissions scenario (SRESA1B) (b,d,f). (inline image) 0.00000; (inline image) 0.000001–0.000011; (inline image) 0.000012–0.000032; (inline image) 0.000033–0.000065; (inline image) 0.000066–0.000114; (inline image) 0.000115–0.000188; (inline image) 0.000189–0.000330; (inline image) 0.000331–0.000574; (inline image) 0.000575–0.000909; (inline image) 0.000910–0.001452 and (inline image) 0.001453–0.002067.

Table 5.   Predicted number of 25 × 25 km (625 km2) grid cells across Europe in which at least one incursion occurs in livestock per year through the four species of migratory bird
MonthCurrent decade 2005–2014Decade 2075–2084 with climate-change scenario
  1. *Calculated as total annual number of incursion-positive grid cells divided by the total number of grid cells comprising Europe (14 282).

March0·0100·024
April0·1910·372
May0·8390·631
Total1·041·03
Average risk per grid cell per year*7·3 × 10−57·2 × 10−5

The predicted impact of climate change on bird abundance and the nymph moulting criterion

The contraction in the ranges of three of the four bird species predicted by Huntley et al. (2007) is projected into a decline of 34% in the total bird numbers within Europe in the climate-change scenario in the model here (Table 2). This projected decline in abundance is particularly marked for the Willow warbler. The number of 25 × 25 km grid cells comprising Europe which meet the criterion for moulting of the nymph (>8°C for 15 continuous days) increases between the current decade and the 2075–2084 decade for all 3 months reflecting global warming. This is shown for the month of April in Fig. 3. In March, the regions mainly affected in 2075–2084 are coastal, with significant increases in inland areas in April predicted for that decade (Fig. 3). By May, large areas across Europe meet the criterion both in the current decade and in the 2075–2084 decade.

Figure 3.

 Predicted number of years per decade for which the criterion of 15 consecutive days at >8°C is met (at least once) in the month of April in 2005–14 (a) and 2075–2084 using HadRM3 medium emissions scenario (SRESA1B) (b). Number of years: (inline image) 0; (inline image) 1; (inline image) 2; (inline image) 3; (inline image) 4; (inline image) 5; (inline image) 6; (inline image) 7; (inline image) 8; (inline image) 9 and (inline image) 10.

The predicted impact of climate change on the risk of CCHFV incursion in livestock in Europe

Overall, the number of 25 × 25 km grid cells across Europe in which at least one incursion of CCHFV occurs in livestock per year through immature ticks carried by migratory birds is predicted to remain virtually unchanged with the climate-change scenario used (Table 5). The highest predicted risk of at least one incursion in livestock in the three spring months combined in a single 25 × 25 km grid cell was 0·0030 per year over the 2075–2084 decade. This represents a very low risk, that is, one incursion in one or more livestock animals on average every 329 years in that grid cell which was located in south-west Ireland. Difference maps for the effect of the climate-change scenario on the risk of incursion of CCHFV in livestock in Europe are presented for the months of March, April and May in Fig. 4. Spatially, there is an apparent redistribution of the risks within Europe with the climate-change scenario. Thus, large parts of central and southern Europe are predicted to have reduced risks in the 2075–2084 decade, while the risks increase in northern Europe and in particular in parts of the United Kingdom (UK), Denmark and the Baltic coast. The difference is greatest for the month of May although coastal areas of north-western Europe, mainly the UK and France, are affected in March. There is, however, considerable local variation in the trends. This is particularly noticeable for the UK, France and Spain where some areas are predicted to experience increased risks while other adjacent areas have decreased risks.

Figure 4.

 Impact of the climate-change scenario on the predicted risks of incursion of Crimean-Congo haemorrhagic fever virus in livestock in the month of March (a), April (b) and May (c). Difference calculated as predicted risk in 2075–2084 minus that in 2005–2014. (inline image) −0.001079 to −0.000638; (inline image) −0.000637 to −0.000280; (inline image) −0.000279 to −0.000108; (inline image) −0.000107 to −0.000024; (inline image) −0.000023 to −0.000001; (inline image) 0.000000; (inline image) 0.000001; (inline image) 0.000002–0.000023; (inline image) 0.000024–0.000107; (inline image) 0.000108–0.000279; (inline image) 0.000280–0.000279–0.000637 and (inline image) 0.000638–0.001452.

Discussion

Risk equations have been developed to quantify deterministically the absolute risks of incursion of CCHFV in livestock. These equations do not predict the actual numbers of infected livestock. Instead, the probability of one or more livestock becoming infected in each 25 × 25 km grid cell is estimated.

The model accommodates almost 120 million individual birds entering Europe each year over the months of March, April and May under current conditions (Table 2). Despite this large number of birds, the overall risk of incursion of CCHFV in livestock across Europe through the four species of migratory bird considered here under current conditions is very low (Table 5). Introduction of CCHFV into Turkey by migratory birds has been argued against by Randolph and Ergonul (2008). Furthermore, despite the predicted higher temperatures, the predicted overall risk in the 2075–2084 decade with the climate-change scenario is unchanged compared with that for the current decade (Table 5). Thus, although increasing temperatures in the climate-change scenario enhance the moulting of the nymphs on their arrival, there is a 34·4% decline in the predicted numbers of nymphs coming in on these four species of birds. This reflects the predicted decrease in the ranges of some of the bird species considered here, notably the Willow warbler, together with the accompanying projected decline in their abundance (Table 2). While breeding populations of Common quail have remained stable, those for Northern Wheatear, Tree pipit and Willow warbler have undergone a moderate decline from 1990 to 2000 (BirdLife International 2004).

The increase in predicted absolute risk from March to May (Fig. 2 and Table 5) for both the current and future scenarios reflects both the increasing temperature over this period of the year and the proportion of migratory birds arriving in each of the 3 months (Table 3). Despite the fact that most of the birds enter Europe in the month of April (Table 3), most of the predicted CCHFV incursions in livestock occur in May (Table 5). This reflects the higher temperatures predicted in May and hence the greater probabilities of meeting the criterion for moulting of the ticks.

It is interesting to note that while the predicted number of incursion-positive grid cells per year almost doubled in the months of March and April in the 2075–2084 decade (with the climate change scenario) compared with the current decade (Table 5), they declined by 25% in May. The predicted fall in May between the current decade and the future decade reflects the impact of the projected decrease in bird abundance in Europe, while the increases in March and April are driven by temperature.

The difference maps in Fig. 4 suggest for the climate-change scenario, that the risk will increase across northern Europe relative to the current risk, but decrease in central Europe and parts of southern Europe. This may reflect the fact that only four species of bird are considered here, and that their breeding ranges are predicted to move northwards with climate change (Huntley et al. 2007), leaving vacant many of those areas in central and southern Europe which are currently occupied. No allowance is made here for the possibility that other bird species could fill those vacant niches.

It is assumed here that all those H. marginatum ticks on birds which enter Europe are nymphs (and not larvae). This simplifies the model in two major ways. First, data on the survival of larvae and the trans-stadial transmission efficiency from the larval to nymph stages together with data on the impact of temperature on moulting from larva to nymph are not required. In this respect, the predictions from the current model may be too high because no allowance is made for failure of some of those larvae to survive and moult. Second, layers on the habitat suitability for H. marginatum (layer 3 in Gale et al. 2010) and on the density of small mammal hosts, for example, hares, (layer 4 in Gale et al. 2010) are not required for this incursion model.

The value of pfind will vary depending on the terrain and will also be affected by the species of birds which introduced that tick and in particular its favoured habitat and hence proximity to livestock. Unlike the Cattle egret (Bubulcus ibus), which is not a trans-Saharan migrant (EBCC 1997), none of the four species of birds studied here specifically associate with livestock. There are opposing effects of temperature on pfind because while survival rates for the adult H. marginatum tick decrease at higher temperatures, questing activity increases (Estrada-Peña et al. 2011). The magnitude of pprev may change reflecting future changes in the distribution and prevalence of CCHFV in Africa.

The data of Molin et al. (2011) for the numbers of Hyalomma ticks on birds in spring are used here instead of those of Hoogstraal et al. (1964) because more birds were trapped (7453 individuals compared with 867) and the study was performed more recently (2009 compared with 1962) and in European countries (namely Italy and Greece compared to Egypt). In the spring study of Hoogstraal et al. (1964), the average number of H. marginatum rufipes ticks per Northern wheatear was 0·44 which is nine-fold higher than the value used here. However, in a study in Cyprus in spring 1968 (Kaiser et al. 1974), the average number of H. marginatum rufipes ticks on Northern wheatears was just 0·012 per bird. Taking the Egypt (Hoogstraal et al. 1964) and Cyprus (Kaiser et al. (1974) spring studies together, the average is 0·062 nymphs per Northern wheatear, which is comparable to the 0·049 per bird used here (Table 1). Over the 35 species of migrant bird trapped in Cyprus in spring 1968, Kaiser et al. (1974) recorded an average of 0·037 H. marginatum rufipes ticks per bird which is comparable to the 0·049 Hyalomma ticks per bird used here. Of interest is the finding that the predominant tick on spring migrants trapped in Norway was I. ricinus (Hasle et al. 2009). This suggests different patterns for tick species on spring migrants in northern and southern Europe perhaps reflecting the effect of longer distances to northern Europe resulting in higher drop-off of ticks during the journey.

Based on the four species of bird studied here, the absolute risk of incursion of CCHFV in livestock through immature ticks carried on migratory birds is very low. Despite expected increases in temperature owing to climate change from now to the 2080s, the risk of CCHFV incursion in livestock in Europe is predicted to be little affected because of the predicted decrease in the range and abundance of the four bird species studied. The predicted risk of incursion of CCHFV, however, increases in northern Europe with climate change but decreases in southern and central Europe. The opposing effects of climate change on the number of nymphs introduced by birds and on their moulting efficiency highlight the importance of developing this spatial approach further. It is important to consider more species of migratory bird and other routes of introduction (e.g., livestock trading and importation of wildlife including hares for hunting) of CCHFV into Europe for a full assessment of the risk of incursion in livestock.

Acknowledgements

This work was supported by the EU Network of Excellence, EPIZONE (Contract no. FOOD-CT-2006-016236). We thank our colleagues in EPIZONE Work Package 7.4 including Paul Phipps, Tony Fooks and Emma Snary (Animal Health and Veterinary Laboratories Agency, UK), Anthony Wilson (Institute of Animal Health, UK), Aline de Koeijer (Central Veterinary Institute of Wageningen, The Netherlands), Gioia Capelli (Istituto Zooprofilattico Sperimentale delle Venezie Viale dell’Università, Italy), Helen Wahlström (National Veterinary Institute, Sweden), Michele Dottori and Giulia Maioli (Istituto Zooprofilattico Sperimentale della Lombardia e dell’Emilia Romagna, Italy), Anja Globig (Friedrich-Loeffler-Institut, Germany) and Hong Yin (Lanzhou Veterinary Research Institute, China). We thank Professor Brian Huntley and Dr Yvonne Collingham of the School of Biological and Biomedical Sciences, Durham University (UK) for providing data in GIS format from their ‘Climatic Atlas of European Breeding Birds’. We thank Dr Ruud Foppen and Henk Sierdsema of the European Bird Census Council and Dr John Gloster and Ag Stevens of the UK Met Office for their help.

Ancillary

Advertisement