The timing and departure rate of larvae of the warble fly Hypoderma ( = Oedemagena ) tarandi ( L . ) and the nose bot fly Cephenemyia trompe ( Modeer ) ( Diptera : Oestridae ) from reindeer

The emergence of larvae of the reindeer warble fly Hypoderma (= Oedemagena) tarandi (L.) (n = 2205) from 4, 9, 3, 6 and 5 Norwegian semi-domestic reindeer yearlings (Rangifer tarandus tarandus (L.)) was registered in 1988, 1989, 1990, 1991 and 1992, respectively. Larvae of the reindeer nose bot fly Cephenemyia trompe (Moder) (n = 261) were recorded during the years 1990, 1991 and 1992 from the same reindeer. A collection cape technique (only H. tarandi) and a grating technique (both species) were used. In both species, dropping started around 20 Apr and ended 20 June. Peak emergence occurred from 10 May 10 June, and was usually bimodal. The temperature during the larvae departure period had a slight effect (significant only in 1991) on the dropping rate of H. tarandi larvae, and temperature during infection in the preceding summer is therefore supposed to explain the uneven dropping rate. This appeared to be due to the occurrence of successive periods of infection caused by separate periods of weather that were favourable for mass attacks by the flies. As a result, the temporal pattern of maturation of larvae was divided into distinct pulses. Departure time of the larvae in relation to spring migration of the reindeer influences infection levels. Applied possibilities for biological control by separating the reindeer from the dropping sites are discussed. K e y w o r d s : Parasite, reindeer warble fly, reindeer nose bot fly, throat bot fly, review, larval departure rate, control, Norway, Rangifer tarandus. Rangifer, 14(3): 113-122

Most populations of R. tarandus migrate in spring from wintering grounds inland to coastal areas where the animals spend the summer (Kelsall, 1968;Skjenneberg & Slagsvold, 1968;Skoog, 1968;Syroechkovskii, 1984;Boertje, 1985).The migration of semi-domestic reindeer in northern Norway starts in late March or April and the reindeer reach their summer ranges between early May and the end of June.The distances covered may be as much as 300 km.
The dropping of the larvae of H. tarandi and C. trompe coincides with the spring migration and, consequently, the parasites become separated by consi-derable distances from their host individuals (Winogradova, 1936).These parasites have an annual life cycle and, for a short period (from when the last larvae have dropped to when new infections start in July), the reindeer are almost free of these parasites.Thus, to continue their life cycle, the flies actively have to seek new host individuals in the summer.
The proportion of the larvae brought into the summer range, where they can infect their host, depends on the timing and speed of the migration of the reindeer in relation to the dropping of the larvae.The further the animals migrate, the further the larvae are left behind, and driving the reindeer greater distances between the time of larval departure and infection time has been recommended as a means of reducing infection by oestrid parasites (Hadwen, 1926;Palmer, 1934;Hearle, 1938;Saveljev, 1968;Washburn etal., 1980).Folstad etal. (1991) found that distance of the post calving migration was negatively correlated with mean intensity of infection of H. tarandi in semi-domestic reindeer in northern Norway.Kelsall (1975) showed that sedentary populations of caribou have higher infections levels of H. tarandi than migratory ones.
The migratory habits of the reindeer during the oestrids' free-living stages (dropped larvae, pupae and flies) influence the transmission rate of the parasites.Clearly, the fitness of the parasites would be increased by delaying their departure from their host until after the reindeer have reached their summer grazing areas.However, the potential for delaying eclosion is limited by the need for the flies to hatch early enough to reproduce before the short sub-Arctic summer is over.

Review of the known dropping time of the larvae of the reindeer oestrids
Figure 1 gives a summary of the published dropping periods of H. tarandi and C. trompe from Fennoscandia, Russia and North-America.Most reports agree that the dropping of larvae of H. tarandi starts in the latter part of April and ends in the latter part of June or the beginning of July.The peak of the dropping generally takes place between mid May to mid June.Gomoyunova (1976) mentioned that mass departure of larvae of H. tarandi from young reindeer starts 15 Apr -31 May, whereas in older reindeer this occurs between 10 and 20 May.Solopov (1989)    Sex: m = male; f = female.All reindeer were yearlings (= 1 year old), naturally infected the preceding year when 2-3 months old.Start: Start of sampling.Stop: Slaughter of host.Scars: Scars after H. tarandi warbles under skin + undropped (dead or alive) larvae at slaughter of host.Origin of reindeer: 1988: District 17 (Tromsdalen, n = 2) and 33 (Spalca, n=2);1989: District 33 (Spalca);1990, 1991, and 1992: District 37 (Skarfvaggi).those from young ones.The variability in dropping times also depends on climate, occurring latest in northern or more arctic climates.
The dropping periods of C. trompe have only been reported a few times (Figure 1).Generally, the departure time is similar to that of H. tarandi, lasting from end of April to the last part of June or beginning of July.Gomoyunova (1976) observed that the dropping of larvae of C. trompe (at Tsjuktsjer peninsula in Russia) was finished 10-15 July, somewhat earlier (15-20 days) in young reindeer.In her study, the dropping period lasted 63 and 71 days in 1968 and 1969, respectively.Solopov (1989) showed that the emergence of H. tarandi larvae was not normally distributed but occurred in a series of pulses.He suggested two reasons for this: 1) The rate of dropping might be temperature dependent and warm weather during dropping might stimulate higher dropping rates.
2) The temporal pattern of infection might be temperature dependent, so that the dates of emergence of larvae might be controlled by the weather conditions in the previous summer.
In the present study, we have investigated the temporal pattern of emergence of larvae and compared this with temperature records to test these hypotheses.

Material and methods
Naturally infected reindeer (age «11 months) were brought from their natural mountain pastures in late April in 1988April in , 1989April in , 1990April in , 1991April in and 1992 (Table 1).The animals were penned at Holt, Department of Arctic Biology, Tromsø, and given lichens and artificial fodder until slaughtering after the dropping period of the larvae.Two methods for collecting the larvae were used: 1.In 1988 and 1989, the reindeer yearlings were kept in a small outdoor corral.Pieces (1 x 1 m) of nylon mesh were fastened over the backs of the animals, so that all warbles of H. tarandi were covered.The mesh was fastened using Velcro strips, one of which was glued with Casco contact glue directly on to the fur while the other was sewn to the nylon mesh.When pressed together, these two strips adhere, and the mesh made a transparent bag into which the larvae collected when they emerged.The larvae were easily visible from the outside (Figure 2), and when present, could be collected by lifting the Velcro fastener.In this way, daily dropping from individual reindeer was registered.A similar technique has been used by Gregson (1958), Pfadt et al. (1975) and Barrett (1981) in collecting cattle grub larvae (Hypoderma lineatum and H. bovis).This technique will hereafter be denoted the «collection cape technique*.
2. The «collection cape technique* did not catch C. trompe larvae because these are expelled via the nose or the mouth.There were also problems with the glue sticking to the hairs (especially when the shedding started).Consequently, in 1990Consequently, in , 1991Consequently, in and 1992, larvae of both species were collecting using the «grating technique*.The reindeer were kept individually indoors in narrow (1 x 2 m) pens with slatted floors.They were not exposed to direct sunshine.Larvae fell through the slats and were collected on a wire mesh tray below, equal in size to each pen.Dropped larvae of both oestrid species were collected from each reindeer once a day.
The reindeer were slaughtered once dropping of larvae had stopped.Any remaining larvae of both species and scars from H. tarandi warbles under the skin were counted in each reindeer.In this way, a count of the total number of larvae (of H. tarandi) that had been present in the individual reindeer was obtained.There were no scars from previous infections, as might have been the case in adult animals, because the reindeer used here only had been infected one summer.Table 1 gives the sample size of the collected material and details about the host individuals.
In Figure 3 Larvae of H. tarandi remaining in the host as a function of date from 5 years of investigation.(Sample sizes, see Table 1).100 % is based on counting of larval scars at autopsy.

Results
The pattern of dropping of H. tarandi in different years is shown in Figure 3. Evidently sampling started too late to detect the start of dropping, except in 1989-The dropping is finished around 20 June all years.
Larval departure was highly variable in both 1991 and 1992 (Figures 4a and 4b).In both years there appeared to be two separate peaks of dropping, the first around 10-20 May the second around 1 -10 June.This bimodal pattern of dropping was synchronous among individual reindeer (inserted curves in Figures 4a and 4b).Pairwise comparison of daily dropping of larvae between individual reindeer gave the following Spearman's rank correlation coefficients (rs): 1991: mean 0.28 ±0.13 SD, range: -0.04 -0.48 (6 reindeer; 15 comparisons); 1992: mean 0.35 ± 0.16 SD, range: 0.09 -0.55 (5 reindeer; 10 comparisons).2. Temperature data from the infection area for July and August 1990 and 1991 (the infection period for the 1991 and 1992 sample, respectively) are depicted in Figure 6.Figures 5a and 5 b show the departure rate of C. trompe larvae for the years 1991 and 1992.A trend of bimodality in the departure rate may be traced as for H. tarandi.How many larvae were dropped before the sampling period is unknown because this species leaves no scars, but we assume that the sampling period covered the mass emergence.The first dropped larva were observed on 18th April, before the animals arrived at Tromsø.With one exception, no larvae remained at time of autopsy (Table 1).

Discussion
There are three reasons for the difference between the number of scars and number of larvae of H. tarandi (Table 1).First, larvae were dropped before the sampling period.Second, some larvae were lost.This was a problem particularly in 1988 and 1989 when the «collection cape technique* was used.Third, some of the counted scars originated from dead larvae and thus, the number of larvae that were able to drop was overestimated.In 1988In , 1990In , 1991In and 1992 we obtained the reindeer too late to observe the first larval departure.Inspection of the dropping curves for 1991 and 1992 shows that <20% of H. tarandi could have dropped before 1 and 5 May, respectively.In 1989, we obtained the reindeer 18 Apr, at a date when no emergence could be observed, see the horizontal start of the 1989-curve in Figure 3. First dropping occurred on 26 Apr.The highest dropping rate normally occurred from 10 to 20 May, but another high rate often took place later (around 10 June) (Figures 4a and 4b).In all years, the emergence is virtually completed around 20 June.
The observed departure rates of larvae of H. tarandi coincides well with results published by Solopov (1989) for reindeer calves (= yearlings) from the northern Taiga area of Russia, in which the dropping started around 20 Apr, with a mass departure from 10 May to 20 May.The end of departure in Russia (5 June) was, however, somewhat earlier than in our study (20 June).
The timing of the departure of C. trompe larvae coincides with the departure of H. tarandi, and dropping of C. trompe larvae is usually completed around 20 June (Figures 5a and 5b).Reindeer with high infection levels, may, however, continue to drop the larvae after this date.One reindeer harboured 42 third instar larvae as late as 28 June 1990 (Table 1).Some larvae may therefore drop during July.Larval development is density dependent (crowding effects) and influences the length of dropping period (Nilssen & Haugerud, accepted).Solopov (1989) found that the departure of H. tarandi larvae was dependent on the climate, with a later departure in a colder climate.In a region called «sub-arctic tundra», the larval dropping started around 15 May and lasted until 10-15 July.He also found that calves (= yearlings) dropped earlier than did older animals.This was not tested in the present study, as we only used yearlings.
Like Solopov (1989), we observed two separate, synchronous mass droppings of H. tarandi larvae (Figures 4a and 4b).In hypothesis 1) cited in the introduction, the departure rate is a function of both the date and the temperature during dropping.The number of remaining larvae determines how many larvae can be dropped and therefore declines with date (Figure 3).Consequently, Julian date and remaining larvae are highly correlated (R = 0.99), and date therefore equally well accounts for the decline due to decreasing number of larvae to drop.Date explained 36 % of the variance in the number of dropped larvae, whereas the mean temperature alone (when variance due to date has been accounted for) explains 13 % (p = 0.0015) and 3 % (p = 0.09) for the years 1991 and 1992, respectively (Table 2).We therefore conclude that the mean temperature has little effect on the departure rate.However, the reindeer in the experiment were not exposed directly to sunshine.Under natural conditions, there is still a possibility that the strong warming by sun radiation directly on the fur, and thereby the larvae beneath, accelerates the emergence of larvae.The statement of Skjenneberg & Slagsvold (1968) that the H. tarandi larvae preferably crawl out during warm, nice weather, may therefore be correct.In our study, the date (which was closely correlated with the number of remaining larvae) is the major determinant of departure rate.However, neither date nor the mean temperature during dropping explain the observed pattern of dropping peaks.
With the present data, it is not possible to test hypothesis 2) directly, where temperature during infection explains the separated peaks.Infection takes only place during warm days (above 12 -14° C, with mass attacks at higher temperatures).Attacks seldom occur before 10 July, and normally end around 20 Aug (unpublished results).As there will be a positive correlation between ambient temperature and attack rate, we expect that certain days (or periods of days) will give higher infection.In Figure 6 the daily mean temperatures in July and August (the main infection period) 1990 and 1991 are plotted for a weather station (Nordstraum in Kvaenangen) near the summer pastures of the reindeer used in 1991 and 1992.We observe that in 1990 there are three periods with high temperatures between 10 July and 20 Aug, namely around 20 July, 5 Aug and 17 Aug.These supposed periods of mass attacks may well be reflected in the uneven larval dropping observed in 1991 (Figure 4a).Similarly, in 1991 the supposed major attacks occurred around 16 July, and 1, 10 and 20 Aug (Figure 6), causing the separated peaks of dropping of larvae observed in 1992 (Figure 4b).
In our view hypothesis 2), with temperature during the infection period as the explanation for the uneven larval departure, is the most likely one.If there is a time difference of many weeks between the first and last infection, it is likely that the maturing time of the larvae will be different and consequently also the larval emergence.This is consistent with Saveljev (1968), who stated that time for removing the larvae in spring (for killing) should be postponed by approx. 2 weeks if the warble fly season was very late the previous year.Many other factors, such as density of larvae and condition of the host, may however be involved.There may also be a selection for early or late larval emergence dependent on factors like local climate and migratory habits of host.Only experimental infection and otherwise controlled conditions can give conclusive evidence.

Applied perspectives
The reindeer oestrids are responsible for economic losses in the reindeer management (Saveljev, 1968;Nordkvist et al., 1983), and different control measures have been applied.One «ecological» method is to separate the reindeer (in time and distance) from dropping sites of the larvae to infections sites (summer grazing areas).
We have established that the departure of both species of oestrids occurs between 20 Apr and 20 June.Thus, if the reindeer are moved to their summer pasture before 1 May, nearly all the larvae will be taken into the summer grazing land.The adult bot flies emerging in this area will consequently have short flying distances to find their host.Even if these flies are excellent fliers (Nilssen & Anderson, in press), distance per se is generally of importance for the infection levels (Folstad et al., 1991;Nilssen & Haugerud, accepted).
In northern Norway, some herds of reindeer spend the summer grazing on islands.The reindeer are brought there by ship.In current practice, this takes mostly place during the last part of April.Consequently, nearly all larvae are brought to the summer pastures, with possibilités for high infection levels of oestrids.If, on the other hand, this transport of reindeer could be postponed until 20 June, most of the larvae would be dropped far away from the summer ranges.Postponement of the spring migration could also be tried on herds with summer grazing ranges on the mainland.However, the migratory pattern is highly traditional and difficult to change.The cost and trouble of changing the spring migration time will probably be greater than the harm caused by the oestrids.A very late spring migration will also be in conflict with the current management strategy to reduce grazing in the limited spring and autumn pastures.In the old, traditional husbandry system, the reindeer spent more time to reach the coast in spring, leading to more larval dropping along the migration route instead of the summer area.Therefore, the traditional migration system undoubtedly was a better anti-oestrid strategy than current practice.Today, the only antioestrid «strategy» used in management is efficient medical treatment (e.g.ivermectin) (Nordkvist, 1984;Nordkvist et al., 1984), being carried out to an ever increasing extent (Heggstad, 1988;Haugerud et al., 1993).However, the need for treatment against oestrids can be reduced if the reindeer management make use of the «ecological» possibility to lower the infection levels of these two important parasites by altering the spring migration time.

Fig. 2 .
Fig. 2. The «collection cape technique* was used in 1988 and 1989 to collect H. tarandi larvae.The collection cape consisted of a nylon mesh fastened over the back of each animal.The mesh made a transparent bag into which the larvae collected when they emerged.The arrow points at one dropped larva.

Fig
Fig.4a and The multiple regression analyses of possible' effects of mean temperature (during the dropping Multiple daily dropping are given in Table

Fig. 6 .
Fig. 6.Daily mean temperature fluctuations for July and August 1990 and 1991 near the infection area for the reindeer used in 1991 and 1992.The highest peaks between 10 July and 20 Aug are supposed to be the periods with mass attacks by the bot flies.
also reported that the dropping time varies with age of host in that emergence from adult reindeer takes place 14 -19 days later than

Table 1 .
Host data and number of larvae of H. tarandi and C. trompe dropped from individual reindeer.

Table 2 .
Results from multiple regression of the influence of Julian date and mean temperature upon departure of H. tarandi larvae from host in 1991 and 1992.
Model: Number of larvae = Constant + Julian date + Daily mean temp.