‘Behavioural’ diapause in tropical ticks

As an adaptation to ensure that emergence coincides with optimal conditions for survival, many tick species have incorporated some form of diapause into their life cycles. In southern Africa, for example, recruitment of R. appendiculatus is not in fact continuous throughout the year (Rechav, 1981, 1982; Short & Norval, 1981; Pegram & Banda, 1990; Randolph, 1997). This species illustrates nicely the plastic, presumably adaptive nature of diapause. Diapause is confined to unfed adult ticks, which do not always quest and feed as soon as they have hardened after emergence from the engorged nymphal stage. In South Africa, those which emerge after July each year do not start questing until some time after November of the same year (Rechav, 1981; Randolph, 1997). This diapause appears to be induced by short day lengths (the southern winter solstice occurs in June) (Rechav, 1981; Short & Norval, 1981; Pegram & Banda, 1990; Madder et al. 1999). As a result, there are periods of near absence of each tick stage sequentially during the year (Fig. 4a), and the majority of the vulnerable egg-to-larval stage occurs during the warm wet season (December–May), which would favour rapid development and good survival, although excessively wet conditions are correlated with increased mortality (see below).

Fig. 4. Seasonal variation in the number of larval, nymphal and adult Rhipicephalus appendiculatus at (a) Gulu, South Africa (data from Rechav, 1982) and (b) Kirundo, Burundi (data from Kaiser et al. 1988). The grey arrows indicate which ticks of one life stage gave rise to which ticks of the next stage (according to Randolph, 1994a, 1997). In (a), ‘d’ marks the months of diapausing adults, and asterisks mark those adults and larvae which were counted after recruitment had apparently ceased (see text).

In contrast, in equatorial Africa, where long cold dry seasons do not occur, the same tick species does not show any diapause (Branagan, 1973). All tick stages feed throughout the year (Newson, 1978; Kaiser, Sutherst & Bourne, 1982, 1991; Kaiser et al. 1988), resulting in continuous overlapping generations each of c. 8–9 months duration (Fig. 4b). Marked declines in abundance of all stages occur simultaneously during the dry season. Based on laboratory experiments, Madder et al. (1999) concluded that there is a latitudinal gradient in the propensity to diapause, with ticks from Zimbabwe exhibiting diapause more consistently than those from Zambia. While long day length may trigger diapause termination in the south, further north in Zambia diapause may end gradually over periods of up to five months, perhaps in response to physiological ageing (Berkvens, Pegram & Brandt, 1995).

This variable pattern of diapause is significant not only for the survival of tick populations, but also for the epidemiology of associated infections. Norval et al. (1991) concluded that the occurrence of diapause in R. appendiculatus in southern Africa may account for the absence there of the virulent strain of Theileria parva that causes East Coast Fever in cattle in eastern and central Africa (see chapter by Bishop et al. in this Supplement). Only where there is sufficient overlap between the more or less continuous activity periods of the different tick stages can the virulent strain of T. parva be acquired from infected cattle by larvae and nymphs and transmitted onwards by nymphs and adults. If this is true, it suggests that inter-stadial transmission from larvae to nymphs, which do in fact overlap considerably more than nymphs and adults in South Africa (Fig. 4a), may not be sufficient to maintain cycles of virulent T. parva. The larval–nymphal route was indeed shown experimentally to be less efficient than the nymphal–adult route (Ochanda et al. 1996). The less virulent strain of T. parva that causes January or Corridor disease survives for longer periods in carrier cattle, and so can be acquired by nymphs and transmitted by adults where there is less inter-stadial overlap; hence its occurrence in southern Africa.

‘Morphogenetic’ and ‘behavioural’ diapause in temperate ticks

Temperate tick species have to deal with even more marked seasonality in environmental conditions. The inter-stadial periods between two or more stages of Ixodes species, including I. rubicundus from South Africa (Fourie, Belozerov & Needham, 2001), may be prolonged (i.e. adjusted) by a variable delay in the onset of development (Campbell, 1948; Kemp, 1968), commonly called ‘morphogenetic diapause’ (Belozerov, 1982). As this occurs in I. ricinus that are collected from the field after July, fed in the laboratory and then held in either constant laboratory conditions (Campbell, 1948; Kemp, 1968) or quasi-natural field conditions (Chmela, 1969; Cerny, Daniel & Rosicky, 1974; Daniel et al. 1976, 1977; Gray, 1982), diapause must be triggered by conditions experienced by the questing stages, or the difference between pre- and post-engorgement conditions. Belozerov identified photoperiod as the major cue for diapause, specifically the change from long (pre-feed) to short (post-feed) day length for both Palaearctic (I. ricinus and I. persulcatus) and Nearctic (I. scapularis) members of this species complex (Belozerov, 1998a; Belozerov & Naumov, 2001).

The exact light sensitivity of the photoreceptor cells on the dorso-lateral surface of all stages of I. ricinus (Perret et al. 2003) is unknown, but ticks have now been shown to respond to simulated natural changes in day length (Beattie & Randolph, unpublished observations), which are much more gradual (c. 20–30 minutes per week) than those used in Belozerov's experiments (typically an 8-hour difference within a few days, either side of feeding). Under natural conditions there is a latitudinal gradient in the date of diapause onset. It occurs in I. ricinus that feed from the end of July onwards in Scotland (c. 56° N), from mid-August in Ireland (c. 53° N) and from the end of August in the Czech Republic (c. 49° N). At this time of year, day length is longer but declines faster with increasing latitude, suggesting that the rate of day length change may be a critical cue. However, in light conditions that simulated rapid or slow day length decrease equivalent to July–October at St Petersburg (60° N) and Valtice in the Czech Republic (49° N), respectively, and under 18 °C constant temperature conditions, ticks entered diapause as photoperiods reached the same absolute length, c. 13 hours light, irrespective of the rate of decrease (Beattie & Randolph, unpublished observations). Therefore other factors, such as decreasing temperature, or even increasing age of the tick, both of which are known to enhance the diapause response (Belozerov, 1967, 1970; Fourie et al. 2001), may be responsible for its earlier onset further north, although on their own they would offer less reliable signals to ticks.

There is clearer evidence that temperature is instrumental in breaking morphogenetic diapause in I. ricinus. Ticks exposed to even a brief period of 0 °C show non-delayed development, similar to those fed in the spring (i.e. pre-July) (Campbell, 1948). This ensures that all ticks, whenever they feed after July/August, start development once temperatures increase after the winter (Fig. 3). Those which feed before July develop rapidly without delay. Under a wide range of conditions across the UK, the combination of diapause and seasonal temperature-dependent development rates acts to synchronise the emergence of virtually all unfed ticks of each stage to within a couple of months every autumn (Randolph et al. 2002). This conclusion was validated by examining the physiological state of questing ticks; emergence dates predicted from the day-degree summation development model (Fig. 5) coincided exactly with the first appearance in the field of questing nymphal and adult ticks with high fat content.

Fig. 5. Seasonal variation in the number of larval, nymphal and adult Ixodes ricinus in Dorset, UK. The arrows indicate which ticks of one life stage gave rise to which ticks of the next stage (according to Randolph et al. 2002). Asterisks mark show dates of first appearance of high-fat (newly recruited) nymphs and adults in the field.

Fat is a source of energy derived from each blood meal, that can be used as a marker of physiological ageing in the field (Uspensky, 1995; Walker, 2001). After emergence, fat is used to fuel a tick's locomotory and physiological activity involved in locating and ascending vegetation stems from which to quest, and descending to moist conditions at the base of the vegetation to restore its fluid content by passive and active uptake of atmospheric water (Knulle & Rudolph, 1982; Needham & Teel, 1991; Sonenshine, 1991; see chapter by Bowman & Sauer in this Supplement). Its natural rate of usage therefore varies with seasonal activity and climatic conditions (Steele & Randolph, 1985; Randolph & Storey, 1999), allowing a distinction between the calendar age and physiological age of ticks. At three field sites in the UK, at the end of the summer all nymphs and adults of I. ricinus had very low fat content until, in September or October, a large proportion of ticks had distinctly higher fat content, typically about three times as much when corrected for tick size (Randolph et al. 2002). At this point two overlapping cohorts of ticks were clearly present (Fig. 6). Thereafter modal fat content decreased throughout the following spring. Increasing numbers of ticks became active in the spring, but they had lower fat content than ticks seen in the previous autumn, indicating that they had endured a longer interval between emergence and questing than had the new recruits in the autumn. Through the summer, fat was gradually exhausted by ticks still questing for hosts, so that virtually no low-fat ticks persisted beyond October.

Fig. 6. Seasonal changes in absolute frequency distributions of questing nymphal Ixodes ricinus with different fat content indices counted per 100 m2 in Dorset, UK, starting with late summer at the top in each of two years. Data for two monthly samples are presented in each histogram, showing the earlier data of each pair as filled bars and the later date hatched. In September/October when old and new recruits occur within the same sample, new recruits are shown by open bars. The vertical dotted line is purely to guide the eye.

So it appears that under a wide range of climatic conditions in the UK, a single cohort of each stage of I. ricinus is recruited each year in the autumn and then survives for one year until the ticks have either fed or died of fat exhaustion. The same appears to be true for I. persulcatus, despite a prolonged total life cycle (up to six years) in more extreme climates (Filippova, 1985; Uspensky, 1995; Korenberg, 2000). This timing of recruitment does not, however, coincide with the timing of peak numbers of ticks seen questing in the field. Most newly emergent ticks delay their questing activity until the following spring in many parts of Europe, inter alia (Lees & Milne, 1951; Daniel et al. 1976; Gray, 1985; Randolph et al. 2002). This may be true ‘behavioural diapause’ (Belozerov, 1982), similar to that seen in adult R. appendiculatus and I. pacificus (Padgett & Lane, 2001), although the triggers will be species-specific. Belozerov (1982) concluded that unfed stages of I. ricinus entered diapause in response to short day length. In the field in mild climates, however, ticks may be active from January onwards (Fig. 7), suggesting that decreasing day length beyond a certain level, rather than absolute day length, may be the cue for diapause. Ticks also vary their questing activity in response to their immediate climatic conditions (see below). All-in-all, the pattern of seasonal questing by ticks that diapause in the unfed stage is the product of behaviour (and mortality) superimposed on the phenology. It is time to move to the second box of the tick population model (Fig. 1) and consider determinants of the probability of questing.

Fig. 7. Seasonal variation in maximum air temperature (Ta) and numbers of questing larval (○) and nymphal ([bull ]) Ixodes ricinus at three sites in the UK (note that larval density is presented at 33% of observed level). Dotted and dashed lines indicate the temperature thresholds for the onset of seasonal activity of larvae and nymphs, respectively.

Temperature and day length

Abiotic factors impose constraints on when, where and how ticks quest for hosts. Interacting with the putative inhibitory effects of decreasing day length, and the supposed converse permissive effects of increasing day length, low winter temperatures in temperate regions inhibit tick activity. Mild winters in southern UK allow us to disentangle day length and temperature as determinants of tick questing probability. In contrast to the north of the UK, Switzerland and Wales, where questing activity by nymphs and adults of I. ricinus does not begin in the spring until the weekly mean maximum temperature has reached 7 °C (Macleod, 1936; Perret et al. 2000; and Fig. 7c, respectively), in the south of the UK from 1998 to 2000 winter temperatures did not drop low enough to inhibit nymphal and adult questing activity (Fig. 7a, b). In the south, however, fewer questing nymphs (and adults – data not shown) were counted from October to December, when temperatures were well above threshold levels, than from January onwards. Thus we can propose that at the end of each year, decreasing day length reduces the probability of questing and low temperatures may inhibit activity altogether (see Wales, Fig. 7c), while at the start of each year, increasing day length is permissive, but only if temperatures are high enough. Gradual recruitment to the questing population will occur as temperatures reach this critical level in increasing parts of the ticks’ over-wintering micro-habitats (Eisen, Eisen & Lane, 2002). Because the use of field sampling data to impute questing activity may, however, be confounded by variable rates of disappearance from the questing population through variable feeding and/or mortality rates (see below), experiments under controlled conditions are essential to test this proposed explanation.

Tick size

Activity threshold temperatures vary inter-specifically, inter-stadially, and even intra-stadially, which appears to be size-dependent (Clark, 1995). In the UK, the onset of larval activity coincides with a threshold of 10 °C, rather than 7 °C, mean maximum temperature (Fig. 7). Furthermore, at our UK sampling sites, smaller nymphs (and adults – not shown) started questing later each year (Randolph et al. 2002). Each new cohort of ticks that appeared on the vegetation in the autumn showed a normal size (fat-free weight) distribution, virtually symmetrical about a median of 75–85 μg, which then changed seasonally (Fig. 8). A sub-set of larger nymphs started questing in February, with only 15–30% <80 μg and almost none <60 μg. As the year progressed, the questing population was composed increasingly of smaller nymphs, as small ticks appeared from March onwards.

Fig. 8. Seasonal changes in the size (i.e. fat free weight) composition of populations of questing nymphal Ixodes ricinus during the life-span of a single cohort on Exmoor, UK 1998–1999. Sample dates and sizes are shown in the key. With permission from Randolph et al. 2002.

Figure 8 also shows that once recruitment has ceased beyond April/May and tick numbers fall, the largest ticks (>100 μg) disappear first and only the smallest ones persist to September. Larger ticks arrive sooner and may find hosts more rapidly if their relatively higher energy levels (Van Es, Hillerton & Gettinby, 1998; Randolph et al. 2002) permit them to quest for longer periods. Smaller ticks (including all larvae – Fig. 7) that arrive later may find hosts more slowly if they are forced into temporary inactivity during the summer by the other major constraint of questing probability, moisture stress.

Moisture stress

Despite generic adaptations permitting them to inhabit dry environments, water loss remains the Achilles heel of all terrestrial arthropods, including ticks. Ticks that live in xeric habitats show special adaptations. They tend to be larger than average, thereby reducing their relative surface area, but still they must shelter in cracks and under stones. They cannot sit in the open and ambush passing hosts, but must hunt them actively, travelling fast from shelter to host when the opportunity arises. Sonenshine (1991) describes how ‘camel ticks, Hyalomma dromedarii, emerge from the barren sand around desert caravansaries and run across the hot, exposed ground to attack people and livestock when these hosts were resting near their shelters’.

Other tick species live in far less challenging habitats, but have lower tolerance to water stress. They can sit on vegetation to await passing hosts, but must still limit their host-questing activity to maintain their water balance. Questing behaviour therefore varies diurnally and seasonally with climate (Lees & Milne, 1951; Belozerov, 1982). Triggered by the onset of darkness, ticks appear to make use of moist night-time conditions to walk, thought to be related to selecting suitable questing locations (Perret et al. 2003). In experimental conditions, nymphs walked further after periods of quiescence (median 43 cm, max 9·7 m) than after questing (median 17 cm, max 2·9 m), while those denied atmospheric moisture walked between 5 and 31 m until they died.

Under favourable conditions, ticks may remain in questing positions on the vegetation for periods of several days (Lees & Milne, 1951; Loye & Lane, 1988), but they descend much more frequently in response to increased saturation deficit (SD – a measure of the drying power of the atmosphere that depends on both temperature and relative humidity – RH) so that, commonly, fewest ticks quest during the middle of the day. Macleod (1935) observed that I. ricinus demonstrates positive geotropism at saturation deficits above 4·4 mm Hg (equal to 80% RH at 24 °C, or 71% RH at 18 °C). In the field, reductions in the questing I. ricinus population coincided with maximum SD above 4·4 mm Hg (Perret et al. 2000), with adults less affected than nymphs. Likewise, experiments in quasi-natural arenas (Randolph & Storey, 1999) revealed that under such dry conditions questing activity was diminished more amongst larvae than nymphs, but for both stages it was reversible once moist conditions were restored, i.e. high SD, even up to 15–20 mm Hg, induced quiescence rather than direct mortality (Fig. 9). Nevertheless, high SD may shorten a tick's life span indirectly. In the dry arena, nymphs used up their fat twice as fast as those in the wet arena, presumably largely related to the increased metabolic costs of walking (Perret et al. 2003) and active water absorption (Fielden & Lighton, 1996; Gaede & Knülle, 1997), thereby reducing the estimated maximum questing period from 4 to 2 months (Steele & Randolph, 1985; Randolph & Storey, 1999). If ticks run out of fat before finding a host they will die. Conversely, under favourable warm, moist conditions, prolonged questing will increase the probability of finding a host, thereby exhausting the questing tick population more rapidly. Standard field sampling data can rarely distinguish between tick quiescence/mortality and tick feeding as a cause for the end of the active questing season (Eisen et al. 2002).

Fig. 9. Questing activity and attachment to rodents of Ixodes ricinus nymphs and larvae in relation to degrees of moisture stress in dry (left) and wet (right) experimental arenas. (a) maximum saturation deficit; (b) numbers of nymphs and (c) larvae counted per hour at 0900 and 2100 h (filled symbols) and 1200, 1500 and 1800 h (open symbols); (d) numbers of nymphs (circles) and larvae (diamonds) attached per rodent. Fresh ticks were added to the arenas on days 17 and 24 as indicated, and the dry arena was watered on day 24, both of which showed that quiescence rather than death was the cause of low questing activity. Data taken from Randolph & Storey, 1999.

Impact on host relationships

Adult ticks adopt species-specific preferred questing heights, loosely correlated with the size of their principal host animal (Lees & Milne, 1951; Gigon, 1985; Loye & Lane, 1988). Laboratory experiments with artificial ‘vegetation’, such as glass rods or wooden dowels, suggest an inherent preference (ibid), which may have been established under natural conditions through a response to scent-markings from glands on various parts of the host's body (see chapter by Sonenshine in this Supplement). Alternatively, the ‘choice’ of hosts may be determined purely mechanistically by the tick's questing height. A general feature for many tick species is vertical separation in questing positions between life stages, with sub-adults sitting nearest to the base of the vegetation, commonly with larvae lower than nymphs, and adults questing very much higher (Gigon, 1985). Inter-stadially, it seems as if ticks ascend as high as possible within the limitations of their size-related tolerance to moisture stress (Rechav, 1979), locomotory powers and costs, and energy reserves, but the benefits of height are not entirely obvious. Making contact with the upper body parts of the host will certainly facilitate attachment there, but large numbers of nymphal and larval I. ricinus, for example, attach and engorge successfully on the lower body and lower legs, respectively, of large ungulate hosts (Gilot et al. 1994; Talleklint & Jaenson, 1994; Ogden, Hails & Nuttall, 1998). Any increased ability of later life stages to exploit higher, unoccupied feeding niches is off-set by the lost opportunity to exploit smaller hosts, particularly rodents that are abundant and ubiquitous.

Stage-specific host relationships are affected by the impact of climate on questing heights. In the same arena experiments referred to above, dry conditions forced nymphs of I. ricinus to quest lower down the vegetation from where they contacted and attached to small rodents in greater numbers than in wet conditions (Fig. 9). Very few larvae in the same dry conditions fed on the rodents. Once the dry arena was watered, the same host relationships were established as in the wet arena, with the same low nymph[ratio ]larva ratio on rodents as is typically seen in the wild. These climatic effects will clearly have an impact on the potential for pathogen transmission between nymphs and larvae feeding on rodents, and perhaps also on other trans-stadial routes via other host species, and involving other tick species.

Natural field experiments

The final demographic process in any life history is death. Seasonally variable daily death rates are fixed by the specific local abiotic and biotic conditions, and then operate for the duration of each stage in the life cycle. While we commonly estimate daily rates empirically from observations on percent survival, in reality the latter is determined by the former together with the time-span, according to the simple equation:

where m is the daily mortality rate and t is the period in days over which that mortality acts. If, for example, hosts are scarce and host attachment rates are low, the period of questing will be prolonged; daily mortality will not increase (although possibly increases non-linearly with the age of the tick), but will operate for longer, and so the percentage survival will be reduced. To be of most versatile use, the daily probability of death should be quantified at as fine a resolution (temporal, spatial and phenomenological) as possible. Well-defined rates can always be combined subsequently for any particular use, but can rarely be dissected from data if the original measurements are too coarse. Ideally, seasonally variable daily mortality rates operating on each stage in each state (questing, feeding, developing) should be measured. This usually requires specific experimental, but natural, conditions, and has rarely been achieved.

By excluding sheep from one of the enclosures described above (Randolph & Steele, 1985), estimates could be made of the daily mortality rates for larval (0·015) and nymphal (0·031) I. ricinus questing on a Welsh hillside between June and August or September, respectively (Fig. 11). Normally, however, in the presence of hosts the nymphs would not suffer such high mortality, because this estimate includes some element of death by starvation for nymphs active from April onwards. For larvae not active until mid-June, this specific cause of mortality was probably minimal before mid-August, and the above estimate may be realistic – it would permit c. 40% of the unfed larval population to survive the 2 month questing season here. Although highly specific in place and time, these estimates highlight the potential importance of biotic agents of mortality, such as predators, parasites or microbial infections, in habitats where saturation deficit, and therefore abiotic stress, is typically low throughout the summer.

Field observations within containers

So that survivors may be recovered and counted, most estimates of tick mortality have involved the use of containers within natural habitats (Branagan, 1973; Daniel et al. 1976; Gray, 1981; Daniels et al. 1996; Bertrand & Wilson, 1997). These exclude most predators, but are designed so that ticks are fully exposed to ambient conditions through plastic or nylon mesh walls. Ticks, however, actively select their resting positions within their architecturally diverse microhabitats, which may well offer higher moisture and more buffered temperature conditions than are found within containers. An accident during observations on larval R. appendiculatus in Kenya (Branagan, 1973) illustrates this point convincingly. Larvae from one batch escaped from their container and survived about four times as long as those which did not. It appears that they positioned themselves relative to the stomata on the grass stems so as to benefit from the raised transpirational humidities both before and after temporary closure of stomata during brief periods of evaporative stress. These larvae eventually succumbed only during periods of more prolonged arrested transpiration. Branagan (1973) concluded ‘The association between the climatic recordings and survival within containers cannot therefore be related to the survival of emerged instars which have access to transpirational microclimates – i.e. those under natural conditions.’

Nevertheless, container-derived estimates are the most common tools currently available for comparing relative mortality rates for ticks at various stages of their development or activity, in different habitats and at different seasons. Examples from studies in the Czech Republic (Daniel et al. 1976, 1977; and see chapter by Daniel et al. in this Supplement) and Ireland (Gray, 1981) (Table 1) illustrate some general principles. For comparability across different seasons, durations and stages, the observations on percentage survival (or mortality) of each exposed group of ticks have been translated into daily mortality rates, using the equation described above. Where available, the estimates for each stage were very similar from the two sets of studies, apart from the very much higher over-winter mortality of developing larvae and developing eggs in the Czech Republic than in Ireland, presumably related to the harshness of the continental winter. Engorged nymphs appeared to suffer least from this, but this stage was consistently more robust (as are nymphs of I. scapularis when exposed to the cold – Vandyk et al. 1996). Unfed larvae exposed in July suffered very high daily mortality (0·0067) until October, but much lower mortality (0·0018) over-winter, as would be expected from both the greater tick activity and the moisture stress typical of late summer. Despite the very high percentage hatch of eggs laid by females in Ireland, only c. 40% of larvae became active and reached the top of the activity tubes a few weeks later (Gray, 1981, 1982). Also, as expected, the mortality rates of ticks exposed in open meadows, and even on forest edges, were consistently very much higher than those in forests (Daniel et al. 1976, 1977).

Overall, the mortality rates estimated from these studies are much lower than the theoretical requirements for population regulation. A working model can be based on the expectation of decreasing natural mortality rates with successive, increasingly robust, life stages. This was indeed observed for R. appendiculatus, whose daily mortality rates were 0·019–0·070 for questing larvae, 0·017–0·029 for questing nymphs and 0·005–0·011 for questing adults confined in quasi-natural conditions in Zimbabwe and Zambia (calculated from survival data given in Short et al. 1989; Pegram & Banda, 1990), and is not contradicted by the above estimates for I. ricinus. We can therefore postulate that, on average, for a tick species with a mean fecundity of 2000 eggs, 5% survival from eggs to larvae, 10% survival from larvae to nymphs and 20% survival from nymphs to fully reproductive adults would theoretically allow the equilibrium state, i.e. one egg-laying female tick replacing herself per generation (Randolph, 1998). Although these are merely conveniently rounded numbers, they do allow some idea of the average inter-stadial mortality necessary to prevent exponential population increase in such fecund organisms. 20% survival requires a daily mortality rate of 0·0176 over a 3 month nymph-adult inter-stadial period, decreasing to 0·0044 if the inter-stadial period lasts 12 months. Likewise, 10% survival requires daily mortality rates of 0·0252–0·0063 for 3–12 month inter-stadial periods. These are approximately an order of magnitude greater than any observed for I. ricinus in container experiments (Table 1). Either mortality from abiotic causes within containers in temperate climates is much lower than occurs naturally, in contradiction to Branagan's (1973) observations in the tropics, or the vast majority of ticks die during processes prevented by containers, such as feeding or falling prey to predators.

It is clear that, with no realistic empirical estimates of absolute mortality rates, we lack a fundamental component for use in population models. Fortunately, the continuous production, more or less throughout the year, of new recruits into the tick populations of tropical and sub-tropical regions (Fig. 4) allows the calculation, not of absolute mortality rates, but of monthly relative indices of mortality. Standard census data on either questing or cattle-feeding R. appendiculatus of all stages in natural open-field situations at 11 sites in East and southern Africa yielded monthly inter-stadial mortality indices in the form of k values (Randolph, 1994a, 1997). These are estimated by subtracting the monthly log₁₀(numbers+1) of one stage of ticks (e.g. nymphs) from the log₁₀(numbers+1) of preceding stage (e.g. larvae) with the appropriate time interval equivalent to the development period (Varley, Gradwell & Hassell, 1973). By seeking correlations with these mortality indices, it was possible to identify the seasonally variable drivers of mortality, not only climate but also those biotic factors responsible for tick population regulation.

Causes of aggregated distributions of ticks amongst their hosts

Higher infestation levels on larger hosts are common to many sorts of parasites, and this applies not only between, but also within, host species (Milne, 1949; Matuschka et al. 1992; Craine et al. 1995; Moore & Wilson, 2002). Size is just one of a nexus of host characteristics that cause aggregated rather than random distributions of parasites amongst their hosts, seen habitually in macro-parasites (reviewed by Shaw, Grenfell & Dobson, 1998). As mentioned above, these may be physical, behavioural or immunological, which frequently, but not invariably, operate synergistically. Thus within many wild populations, sexually mature male hosts tend to be larger, often range more widely in search of mates, and may be immunologically compromised by their hormones (Peters, 2000). All these features would promote higher tick infestation levels on this fraction of the host population, as has indeed been observed (Randolph, 1975; Matuschka et al. 1992).

Underpinning these particular host features are variable levels of testosterone, and this hormone has been shown experimentally to have a direct impact on resistance to tick infestations (Hughes & Randolph, 2001a). Testosterone levels were manipulated in voles, C. glareolus, and wood-mice, A. sylvaticus, within the normal physiological range seen in the wild. When exposed to repeated larval I. ricinus infestations, hosts with high hormone levels showed reduced innate and acquired resistance to tick feeding (Hughes & Randolph, 2001a). The innate resistance was manifested by the greater success with which larvae attached to high testosterone hosts independent of previous tick exposure. Successive infestations on voles were accompanied by a decrease in tick feeding success and survival, and this was significantly more marked in ticks fed on control voles than in those fed on voles implanted with testosterone (Fig. 14). Mice, which do not acquire resistance to I. ricinus, showed no such effects. Moreover, when reduced feeding success had been induced, by vaccination with tick salivary gland extract or four successive infestations, implantation with testosterone partially reversed that acquired resistance.

Fig. 14. Changes in mean (a) percent engorgement and (b) percent survival of Ixodes ricinus larvae with successive infestations on castrated voles (Clethrionomys glareolus) implanted with oil (○) or with testosterone ([bull ]). Data from Hughes & Randolph, 2001a.

These effects of testosterone will exacerbate the normal tendencies of acquired resistance to generate over-dispersed distributions of parasites (Anderson & Gordon, 1982; Pacala & Dobson, 1988), especially as those hosts with low resistance are also prone to high contact with ticks. High testosterone levels are also associated with increased locomotory activity in rodents (Randolph, 1977; Rowsewitt, 1986). Therefore, the same individual hosts are likely both to pick up most ticks and to feed them successfully. The ubiquitous presence of high testosterone levels in one fraction of wild populations, at least during part of each year, could offer a common basis for the typical over-dispersed distribution of diverse parasites amongst their hosts, and even amongst males themselves (Shaw et al. 1998).

In the case of pheasants as hosts, a field experiment showed that high infestation levels of ticks may themselves exacerbate further exposure to ticks via a negative impact on territoriality in male birds (Hoodless et al. 2002). Many sorts of parasites have an impact on male secondary sexual characteristics and female choice (reviewed in Clayton & Moore, 1997). When tick burdens on cock pheasants were reduced during March–August using a slow-release acaricide, the degree of wattle inflation was significantly improved (Fig. 15). This is an indicator of territorial status and a correlate of harem acquisition, and, correspondingly, harems were acquired by 44% of treated birds compared with only 22% of controls. Males that acquired females ranged over small areas on field edges, while those with no females ranged more widely in woods and the adjoining fields, thereby increasing their exposure to ticks. If non-territorial male pheasants do indeed feed a greater proportion of the tick population (which could not actually be observed in this experiment), ipso facto they support more tick-borne parasite transmission, which suggests therefore that ticks themselves could have a positive effect on the transmission dynamics of tick-borne parasites.

Fig. 15. Frequency distributions of the numbers of treated and control male pheasants that acquired different numbers of females, depending on their median wattle score (from 3 observations of each male in April). Score 1=no inflation of wattle, pinnae (ear tufts) not visible; 2=some inflation of wattles, pinnae discernible; 3=wattles fully inflated, pinnae erect. Number of females: open bars, 0; diagonal hatch, 1; stippled, 2; cross hatch, 3; black, 4–7. From Hoodless et al. 2002.

Consequences of aggregated distributions of ticks amongst their hosts

The population consequences of aggregated distributions of parasites amongst their hosts have been well revisited since Crofton's seminal work (Crofton, 1971a,b). With host mortality rates that increase with parasite burdens, and density-dependent constraints (immunological and other) on parasite survival, over-dispersion of parasites will stabilize the dynamical behaviour of host–parasite interactions (Anderson & May, 1978).

In addition, because ticks are also vectors, their over-dispersion increases the transmission potential of the dependent pathogens (Hasibeder & Dye, 1988; Randolph, 1995; Randolph et al. 1999) by focusing transmission efficiently within one highly infected fraction of the host population (Woolhouse et al. 1997, 1998). Quantitatively, this is an essential element in the persistence of those tick-borne pathogens that have very low R₀ values. Babesia microti, for example, is maintained in rodent populations despite low mean infestation levels of the vector, I. trianguliceps (and possibly I. ricinus – Gray, von Stedingk & Gurtelschmid, 2002), and the very narrow window of transmission (Randolph, 1995). In this system, testosterone has yet another effect, directly increasing the transmissibility of B. microti by causing prolonged and more intense infections (Hughes & Randolph, 2001b). This occurs in exactly the same fraction of the host population that carries most ticks through other, synergistic effects of testosterone.

Because tick-borne transmission of pathogens depends on at least two tick stages feeding on the same host, one to transmit and the other to acquire, we can predict that the transmission potential will be improved if both tick stages show coincident aggregated distributions. For pathogens such as TBE virus with short-lived, even non-systemic, infections, this coincidence must be more or less synchronous, whereas for longer-lived, systemic infections (e.g. B. burgdorferi sensu lato) it need not be so. Such a phenomenon would easily arise if the same causal factors of aggregated distributions apply to both tick stages, as would all intrinsic host factors. Analysis of tick infestations on over 2500 individual rodents trapped in western Slovakia over three years supported this prediction (Randolph et al. 1999). The distributions of larval and nymphal I. ricinus were highly aggregated and, rather than being independent, the distributions of each stage were coincident so that the same c. 20% of hosts fed about 75% of both larvae and nymphs (Fig. 16). As a result, twice the number of infectible larvae fed alongside each potentially infected nymph compared with the null hypothesis of independent distributions, and estimated R₀ values were >1, as required for virus persistence.

Fig. 16. The coincident aggregated frequency distributions of Ixodes ricinus larvae and nymphs on rodents (Clethrionomys glareolus and Apodemus flavicollis) from April to July in the Borská nízina lowland, Slovakia. At each intensity of larval infestation (X-axis), the numbers of hosts coincidentally feeding zero (open bars), 1–2 (light stippled), 3–4 (dark stippled) or 5–23 (black) nymphs are shown. Mean number of larvae co-feeding with a nymph=20. From Randolph et al. 1999.

The significance of this aspect of tick ecology in determining vector performance in nature is illustrated by the sympatric tick species in western Slovakia. Because of the different life cycle characteristics of D. reticulatus, the pattern of co-feeding larvae and nymphs does not apply to nearly the same extent as to I. ricinus. Development times of D. reticulatus are short, allowing the life cycle (apart from the egg stage) to be completed within a single year, with most larvae feeding one month before the nymphs. Thus, whereas only 3% of I. ricinus nymphs were recorded on hosts that were not carrying at least one larva, as many as 28% of D. reticulatus nymphs were not co-feeding with larvae of the same species (Randolph et al. 1999). Alternatively, as both species are competent vectors of TBE virus in the laboratory (Korenberg & Kovalevskii, 1994), the virus could potentially be exchanged between ticks of each species where they co-exist. These tick species, however, make differential use of voles and mice as hosts: more D. reticulatus were recorded on C. glareolus than on A. flavicollis, while I. ricinus showed the reverse host association (Randolph et al. 1999), so they do not co-feed to a significant extent. The principal host of I. ricinus, mice, are also the more efficient amplifiers of TBE virus (Labuda et al. 1993b, 1997). Factors such as these may explain why, amongst 18 species competent to transmit TBE virus in the laboratory (Korenberg & Kovalevskii, 1994), I. ricinus in Europe and I. persulcatus in Eurasia appear to be the only vectors and reservoirs responsible for long-term TBE virus survival in nature (Nuttall & Labuda, 1994). This is ecological specificity, leading to quantitative vectorial non-competence.

Of course, intra-specifically larvae are more likely to co-feed with other larvae than with nymphs, so could acquire the TBE virus from any trans-ovarially infected larva (Danielová et al. 2002). This, however, can only be an accessory route, insufficient on its own for two reasons. First, any resulting infected engorged larvae can only pass the infection on as they feed as infected nymphs on rodents, so we are back to nymphal-larval co-feeding transmission. Secondly, an essential quantitative element in transmission is amplification to offset the considerable natural mortality of ticks. If an infected nymph arising from a trans-ovarially infected larva, as above, feeds instead on a non-competent host but retains the virus and passes it trans-stadially to the adult stage, there would only be enormous diminution of infection prevalence in the tick population. Each larva, whether infected or not, on average gives rise to 0·01 adult female ticks (see above). Unless this were completely offset by trans-ovarial amplification, which is unlikely on the present evidence of 20% maternal transmission, 0·23–0·75% filial infection and the relatively low numbers of larvae that typically feed close together on any rodent (Danielová & Holubová, 1991; Danielová et al. 2002), TBE virus could not persist by this route. In fact, of 419 larval I. ricinus taken from 103 small mammals trapped in May and October within a TBE focus in the south of the Czech Republic, only 3 (1 unfed and 2 fed) were infected with TBE virus (Danielová et al. 2002).