The Norwegian School of Veterinary Science, Department of Morphology, Genetics and Aquatic Biology, PO Box 8146 Department, 0033 Oslo, Norway & University of Oslo, Department of General Physiology, PO Box 1051, Blindern, 0316 Oslo, Norway
1During the brief growing season in Arctic and high mountain ecosystems, undisturbed grazing is crucial in order to maximize growth and fattening. During summer 1997 we investigated the influence of weather and insect harassment on the behaviour and group dynamics of reindeer ( Rangifer tarandus tarandus L.).
2Climatic data, activity of parasitic flies, and female reindeer behaviour were recorded from two wild reindeer populations in Southern Norway. Temperature and solar irradiation were good predictors of oestrid fly activity. Throughout the warm summer, reindeer were exposed to vigorous oestrid fly harassment, which caused dramatic decrease in feeding and lying, and increase in walking, running and standing. This behavioural change may compromise the physical condition of individuals entering winter.
3Mosquitoes had little influence on reindeer activity patterns. In the absence of oestrid flies, weather parameters had no influence on reindeer activity pattern. Even during the warmest days, no signs of heat stress were recorded. Accordingly, snow patches, marshes and windy mountaintops were used primarily to avoid oestrid fly harassment. Thus, most disruptions of feeding that are often reported on warm days are responses to oestrids, not thermal stress.
Although, ‘insect harassment’ of Rangifer is widely reported in the literature (e.g. Kelsall 1968; White et al. 1975; Thomson 1977; Reimers 1980; Helle & Tarvainen 1984), few authors clarify which insects were active in their study. Mosquitoes (especially Aedes sp., Culicidae), blackflies (Simuliidae), horseflies (Tabanidae), warble flies (Hypoderma tarandi L., Oestridae) and nose bot flies (Cephenemyia trompe L., Oestridae) (hereafter referred to as oestrid flies), commonly attack reindeer in Norway. However, Bergman (1917), Downes, Theberge & Smith (1986), Mörschel & Klein (1997) and Anderson, Nilssen & Hemmingsen (2001) have concluded that oestrids are the primary tormentors of reindeer and caribou and that mosquitoes and tabanids only cause a minor annoyance. Moreover, it has been proposed that oestrid flies cause considerable losses in reindeer husbandry (e.g. Erne & Nordkvist 1970). However, there exists a widespread but poorly documented belief that mosquitoes are responsible for all, or much of the insect harassment of reindeer and caribou. Many reports of insect harassment are based only on mosquitoes attacking humans. Although repeatedly cited, few manage to document a clear impact on reindeer behaviour by mosquitoes alone. Mosquitoes and a few species of blackflies are well known to travellers in Arctic and Subarctic areas. Hence, when sampling insects around the observer it is easy to conclude that mosquitoes and blackflies are responsible for the observed harassment on reindeer. Most observations of reindeer and caribou aggregating into large herds, exhibiting various panic behaviours, and dispersing to mountain tops and snow patches, are reported to occur on warm sunny days. However, under these conditions, mosquitoes have been shown to be less abundant (Anderson & Nilssen 1998).
Even if there is little direct evidence for heat stress in Rangifer, high temperature has often been proposed to influence population dynamics of reindeer and caribou, and cause them to seek heat-relief habitats such as snow patches (Downes, Smith & Theberge 1986; Ion & Kershaw 1989; Anderson & Nilssen 1998). High temperatures in combination with intense solar radiation could potentially lead to heat stress, especially in the Arctic part of Russia and North America, where summer temperatures occasionally can raise above 30 °C. However, except from extreme temperatures in laboratory experiments (Rosenmann & Morrison 1967; Yoesef & Luick 1975), no study has yet managed to reveal heat stress from temperatures alone.
The objectives of this study were to determine and evaluate: (1) the relationship between the weather parameters and the parasitic flies, and establish if weather parameters are good oestrid fly predictors during the insect season; and (2) the impact of oestrid flies and mosquitoes on reindeer activity pattern, and to what extent summer weather parameters influence reindeer behaviour in the absence of parasitic flies.
Our study was conducted in 1997 in southern Norway with a population of semidomestic reindeer in Norefjell-Reinsjøfjell (60°25′N; 9°15′E; 308 km2) and a population of wild reindeer in Rondane North (62°N; 9°45′E; 1441 km2). The two areas were selected because the present study assisted in a study of vigilance on two reindeer populations with a different history of human influence and animal body condition (Reimers & Svela 2002). Both areas are alpine and located within the continental part of Norway; precipitation ranges between 400 and 600 mm. Hunting has maintained the winter population in Norefjell-Reinsjøfjell between 500 and 600 animals and that in Rondane North around 1200 animals (Reimers, unpublished data).
PARASITIC FLIES IN THE AREAS
Adult mosquitoes emerge in spring, mate, and females seek a blood meal from any available vertebrate (Schmidt & Roberts 1981) to maximize egg production. Although extreme numbers of 6000–10 000 have been estimated on individual reindeer (Syroeckovskii 1995), few reliable recordings of mosquito abundance in Rangifer areas are found in the literature.
Both warble- and nose bot flies are host-specific endoparasites in reindeer and caribou, and have fairly similar 1-year life cycles. Between early May and June, warble fly larvae leave their host through their breathing hole in the skin (Bergman 1917; Kelsall 1975; Solopov 1989), while nose bot fly larvae are coughed and sneezed out (Espmark 1968). On the ground the larvae pupate, and after a few weeks they emerge as flies, normally from the beginning of July, occasionally during warm summers already in the end of June (Nilssen & Haugerud 1994, 1995; Nilssen 1997a,b). After mating, females seek reindeer, and each warble fly may attach 500–700 eggs to the hair of the reindeer fur (Espmark 1968; Savaljev 1968) whereas nose bot flies spray live larvae onto the reindeer muzzle (Anderson & Nilssen 1990) from where they crawl into the nasal cavities and later into the larynx and pharynx regions. The oestrid flies are strong and fast fliers, and have a great flight capacity (Nilssen & Anderson 1995). This enables the flies to locate, track and follow reindeer easily. Both the direct parasitic load (e.g. Zumpt 1965) and indirect effect of disturbance during summer grazing (e.g. Helle & Tarvainen 1984) might increase mortality and reduce weight and condition. Nose bot flies can restrict breathing of the host (Skjenneberg & Slagsvold 1968) and during fast running, larvae might be inhaled into the lungs, causing pneumonia and even death (Skjenneberg & Slagsvold 1968).
Materials and methods
Fieldwork was conducted during two periods in 1997: Norefjell (28 June–7 July and 28 July–5 August) and Rondane (12–21 July and 10–17 August). The observation periods were scheduled relative to the peak calving date (the date when 50% of the pregnant females have calved), which was 7 May in Norefjell and 22 May in Rondane (Reimers 1997). Reindeer were observed with binoculars, telescopes and a video camera. Except for a few close-range observations, we assume the reindeer were not aware of our presence during sampling.
Eight climatic parameters were recorded mainly every 15 min during reindeer behaviour sampling; ambient temperature, temperature in shade and sun at ground level, relative humidity, sun irradiation, wind speed, wind direction and degree of cloud cover.
The three recordings of temperature were measured with a hand-held digital humidity and temperature instrument (Vaisala HM 34) with a recording accuracy of 0·1 °C. Ambient temperature in the shade was measured at 2 m above ground level [Norwegian Meteorological Institute (DNMI) standard]. A board was held over the thermometer to provide shade. Temperature at ground level was measured by laying the thermometer in the vegetation or on the bare ground, but never on stone due to its high thermal conductivity. Temperatures on the ground were taken in both direct sun and in the shade. This was carried out to document rapid temperature changes caused by appearing and disappearing direct sunlight. Wind speed (m/s) was measured at 2 m above ground level with a hand-held anemometer (Lambrech). Speed values were recorded with an accuracy of 1 m/s except for the value 0·1 m/s given for minor air movement. Sun irradiation (global radiance) was measured with a solar meter instrument (HÆNNI Integrator Solar 118), set to measure W/m2. Additional reindeer behavioural data in relation to weather conditions, such as being in direct sun or shade, were noted.
RECORDING OF PARASITIC FLY ACTIVITY
Only activity of warble flies, nose bot flies and mosquitoes (Culicidae) were recorded. Other known parasitic flies such as blackflies and tabanids were not recorded, as methods for recording the presence of these flies in a mobile herd are not available. Their direct impact on reindeer is unclear.
Oestrid fly activity
The presence and relative activity levels of oestrid flies were inferred strictly from reindeer annoyance behaviour (Espmark 1968), and was recorded for every scan recording. Stereotypical behaviour, such as leg-stomping (Thomson 1977) was used as an indicator for the presence of warble flies, while sudden nose dropping, rubbing of the muzzle in the ground and violent sneezing (Espmark 1968) were used to record the presence of nose bot flies. If animals displayed one of these behaviours clearly during sampling, oestrid flies were considered as present in the herd, otherwise as absent. Although violent body shaking, head shaking and swift movements of the head are considered as specific defence reactions to oestrid flies, none of these factors alone were used as indicators, since these reactions also can be caused by other parasitic flies (e.g. Anderson & Nilssen 1998).
Both oestrid flies were frequently active at the same time and their activity level was categorized into four categories; 0 = no harassment, 1 = light harassment, 2 = moderate harassment and 3 = severe harassment. Light harassment was recorded if only one or a few individuals in the herd displayed annoyance behaviour. The response would be relatively mild, and the herd still maintained ‘normal’ behaviour. At moderate behaviour, reactions to oestrids were more violent and many of the animals would be stomping their legs or rubbing their muzzle in the ground. At this stage the herd was clearly disturbed, feeding activity was disrupted, and moving and standing increased. At severe harassment, the whole herd was severely affected, resulting in practically no feeding. All individuals would be alert, and herds were either standing tightly or running in panic. Hence, the harassment level was not set by amount of animals harassed alone, but also by the intensity and violent reactions to the flies.
Because there are no suitable methods for measuring mosquitoes in a mobile herd, abundance around the observer was recorded instead. Recordings were carried out on a strictly subjective basis on a scale of 0–4 (0 = no mosquitoes, 1 = low abundance, 2 = moderate abundance, 3 = high abundance and 4 = extreme abundance). As for oestrid flies, mosquito abundance was recorded for every reindeer observation. No sweep-net samples or similar collecting techniques were employed. Our scale was set on basis of our own earlier experience with mosquitoes in Northern and Southern Norway. Although abundance around the observer could be considered to represent the general abundance of mosquitoes in the area, it is conceivable that large reindeer herds could attract considerably more mosquitoes than two humans.
REINDEER BEHAVIOUR OBSERVATIONS
Reindeer behaviour was categorized for scanning and divided into five categories; feeding, walking, running, lying, and standing/other. Feeding was defined as the act of ingesting forage with muzzle down. Walking occurred when the head assumed a normal upright position and the reindeer walked. Running was defined as moving fast with the head in an upright position. Trotting and other forms of fast pacing were included in running. Standing/other was incorporating activities such as alert- and insect-defence posture, normal standing, nursing, social activities, fighting, cleaning, scratching, drinking and playing.
When possible, herds were scanned every 15 min and the behaviour of each animal in the herd (excluding calves) was tape-recorded. When scan sampling became difficult because of aggregation of larger herds and more movement and stress caused by the insects, proportion of animals in the different behavioural categories was estimated to the nearest 5%. The estimates were first performed independently, and then compared between the observers, before the values were noted. The simplified scans proved to be reliable, although less detailed than the ordinary scans.
We used SAS 6.12 (SAS Institute Inc.) for statistical analyses, and Microsoft Excel for graphical presentation of the data. The distribution and variance of the data sets was tested before choosing parametric statistical tests such as t-test, anova and linear regression (Fowler & Cohen 1993; Fry 1993). If the requirements for parametric tests were violated after various transformations, we used non-parametric statistical tests such as the Wilcoxon rank sum test (Mann–Witney U-test), the Kruskal–Wallis test and Spearman's rank test (Siegel & Castellan 1988). For logistic regression, we used logit as the link function. These tests will be referred to hereafter as t-test, anova, Linear reg., Reg., Wilcoxon, Kruskal–Wallis, Spearman and Log. reg., respectively. A significance level of P < 0·05 was set for the selection criterion on all tests.
Although weather parameters remained uncategorized for analyses of insect activity, most graphical presentations are presented for mean categorized values (SE, not scatter-plots that would be more correct). Most analyses of insect activity contained more than 600 recordings and scatter-plots became too cluttered. Analyses of the effect of weather parameters on reindeer behaviour were divided into presence and absence of oestrid flies in order to separate the effects of oestrid fly harassment and stress caused from weather alone. For all regressions, we tested whether there was a linear response to the predictor variables. Where there appeared to be a polynomial response, we fitted a model (x + x2 or x + x2 + x3) for the regressions. Adjusted R2 (adj. R2) is provided where models are used for analyses. Analyses were performed for both areas separately to test if differences in their response were present. If not stated, no correlation or differences of importance were present between the two areas. A total of 764 scans were used for analyses of oestrid fly activity levels, mosquito abundance and activity budgets.
RECORDING OF PARASITIC FLY ACTIVITY
Several close-range observations down to 2 m allowed us to observe which insects actually attacked the reindeer. The (bumblebee-like) oestrids were observed frequently attacking reindeer, landing on nearby rocks and caught by the observers. Horse flies that normally attack humans when present were observed and caught only once during fieldwork.
Oestrid fly activity
Oestrid fly presence was recorded from 3 July–17 August. Fieldwork began too late and ended too early to determine the beginning and end of the season of the two oestrid fly species. Activity levels of oestrid flies had a strong correlation with time of day (time + time2) (Reg., n= 758, P < 0·001, adj. R2 = 0·39), with their highest activity between 11.00 h and 15.00 h (Norwegian summertime). The earliest and latest observations of oestrid fly harassment were recorded at 07.10 h (4 August) and 23.20 h (17 July), respectively. Severe harassment (level 2 + 3) was observed only between 09.02 h and 20.13 h. Oestrid fly activity levels in relation to time of day corresponded with ambient temperature and ground temperature in the shade and in the sun. Solar irradiation appeared to have the strongest influence on the oestrid fly activity during time of day (Fig. 1).
Oestrid fly activity level was strongly positively correlated with ambient temperature at 2 m in shade, temperature at ground level in shade, temperature at ground level in sun, and solar irradiation (Fig. 2). The lowest ambient temperature measured for a recording of oestrid fly harassment was 6·9 °C (recorded once). Harassment during 7·0–7·9 °C was recorded on two different dates, 8·0–8·9 °C on three different dates, and 9·0–9·9 °C on nine different dates. Recordings of oestrid fly activity with temperatures between 6·9 and 9·0 °C were made with solar irradiation of 92–463 W/m2. Similarly, observations of harassment were made when sunlight was low or absent, with ambient temperatures above 11·0 °C. Above 14·0 °C, oestrid fly activity was always recorded. On several occasions we observed reindeer herds seeking shade to avoid direct sunlight. Oestrid fly harassment increased in intensity when reindeer were exposed to direct sunlight, compared to when they remained in the shade.
During periods with oestrid fly harassment, reindeer also moved to wind-exposed sites, and remained where wind exposure provided sufficient relief. Nevertheless, analyses showed no negative effect of wind on oestrid fly activity. On the contrary, wind speed showed a weak positive correlation with oestrid fly activity (Linear reg., n= 661, P= 0·015, R2 = 0·01). Logistic regression of absence and presence of oestrid flies gave significant relationship from all analyses of weather parameters. Analyses of humidity and cloud cover are presented in Hagemoen (1999).
Mosquito abundance was low throughout summer, with level 1 (low abundance) recorded as the highest abundance level during reindeer observations. Mosquito abundance was correlated with the time of day (time + time2+ time3) (Log. reg., χ2 = 45·37, P < 0·001), with a decrease in abundance from early morning until midday, followed by an increase towards evening. There was a positive correlation between mosquito abundance and ambient temperature at 2 m (Log. reg., χ2 = 48·87, P < 0·001), with a decline at temperatures higher than 16 °C, and a negative correlation with solar irradiation and wind speed (Fig. 3). Mosquitoes were not observed at wind speeds above 7·5 m/s. Even weak wind (0·5–1 m/s) appeared to lessen mosquito activity.
REINDEER BEHAVIOUR IN RELATION TO PARASITIC FLY ACTIVITY AND WEATHER PARAMETERS
The many close-range observations and video analyses allowed us to record how individual reindeer or herds responded to parasitic fly attacks. Reindeer response to oestrid fly harassment was visibly distinct, typical and frequently violent during the summer. Tiny blackflies (unidentified, possibly Muscidae) were on observed several occasions in great numbers around the eyes and on the nose and antlers. This was obviously unpleasant, and reindeer were seen repeatedly shaking their heads due to these flies. However, their response to small flies seemed mild compared to the response from single oestrid flies approaching the animals. Reindeer responses to oestrids did vary, given different oestrid fly activity level and intensity of attacks. Under severe harassment reaction was spontaneous and violent, often ending in jumping and running, and panicked, disrupted herd behaviour. During cloudy days, when oestrid fly harassment was low, reindeer could shake off oestrids and continue to feed, rather unaffected by the oestrids. However, the response would be more violent than their response to the numerous tiny flies. Oestrid fly defence movements of the calves were similar to adult females, but reactions were more nervous and violent.
During oestrid fly harassment, reindeer were observed seeking wind exposed hilltops and ridges. On less windy days, reindeer herds could be running for hours seeking refuges, and running up and down hilltops. Individuals would occasionally pause to eat quickly and then rejoin the herd. Practically no feeding was conducted during such conditions and the animals often covered vast distances during the day. For example, at 10.16 h on 6 July, herds aggregated on the highest peak in Norefjell and ran continuously up and down every hilltop and through every pond in the area until 18.30 h. Only a few minutes of feeding and some short visits to snow patches interrupted this activity. On many such days, reindeer in Norefjell and Rondane were observed running from the first sign of severe harassment in the morning until late in the evening.
During intense harassment reindeer could stay on snow patches or wet marshes for hours. On snow patches, tightly aggregated reindeer were often seen running back and fourth from one end of the patch to the other. Only rarely, reindeer would seek the edges for feeding, at which point they would be forced back onto the snow patch or the marsh. If weather changed or shadows emerged, harassment could cease, and reindeer would immediately leave the snow patch or marsh for grazing. Apart from young calves and 1·5-year-old-males, reindeer were seldom observed lying on snow patches during oestrid fly harassment. Herds often formed spiral-shaped aggregations, with the outermost animals seeking their way into the middle. Under intense harassment, animals could be running at high speeds around and around in circles in these spiral-shaped aggregations.
In the absence of oestrids, we recorded no specific behavioural response to either mosquitoes or weather parameters. Reindeer would feed undisturbed, even during warm and sunny conditions. We did not observe reindeer on snow patches in the absence of oestrids.
EFFECTS OF OESTRID HARASSMENT ON ACTIVITY BUDGETS
Oestrid fly harassment caused a substantial decrease in feeding (Kruskal–Wallis, n= 760, P < 0·001) and lying (P < 0·001), and an increase in walking (P < 0·001) (decrease from level 2–3), running (P < 0·001) and standing/other (P < 0·001) (Fig. 4). From harassment levels 2–3 running often replaced walking. Analyses of absence and presence of oestrid flies (Wilcoxon) revealed the same trends with P < 0·001 for all categories.
Effects of mosquito harassment on activity budgets
Only a decrease in lying (Wilcoxon, P= 0·010) and an increase in walking (P = 0·002) were correlated consistently with mosquito presence when analysed alone. In absence of oestrid flies, lying decreased (P = 0·001) and standing/other increased (P = 0·001) with mosquito presence. In the presence of oestrid flies, only walking increased with mosquito presence (P = 0·001). Although significant, the magnitude of the changes was low compared to the response to oestrid fly harassment.
Effects of weather parameters on activity budgets
Reindeer responded differently to increasing temperatures and solar irradiation, depending on presence or absence of oestrid flies. In the absence of oestrids, we recorded no significant response to solar irradiation (Fig. 5). In the presence of oestrids, feeding decreased (Spearman, P < 0·001) and standing increased substantially (P < 0·001). With increasing ambient temperature at 2 m, feeding decreased (P < 0·001), and standing increased with oestrid fly presence (P < 0·001). In the absence of oestrid flies, only a minor increase in standing occurred with increasing ambient temperature (P = 0·023).
RECORDING OF PARASITIC FLY ACTIVITY
Although body-shaking, tail-flicking and other movements are caused frequently by oestrids, recordings of oestrid presence are probably more safe by using only leg stomping and sneezing and sudden nose dropping as indicators. Karter & Folstad (1989) demonstrated in an experiment that violent sneezing and closing of the nostrils were elicited only when exposed to the nose bot flies. During our fieldwork, leg stomping was observed only when oestrids, or when observed from a distance large flies, landed on reindeer. Beside oestrids, horseflies are the only large flies attacking reindeer at high elevations in the Norwegian Mountains. Arne C. Nilssen, who has published several papers on parasitic flies, has seldom observed that horse flies cause oestrid fly defence behaviour of reindeer when oestrid flies are not present (personal communication). He believes that reindeer are normally able to distinguish between oestrids and horse flies, although in stressful situations during oestrid fly harassment, attacking horse flies might cause reindeer to display oestrid fly defence behaviour. Because only one horsefly was caught during fieldwork, and horsefly abundance at the high-elevated study areas normally are low, most occurrences of leg stomping in our study must have been caused by oestrids.
Oestrid fly activity
Oestrid fly activity levels were recorded on the basis of annoyance behaviour observed in the herd, and not only from annoyance behaviour of the single focal animals, as in Mörschel & Klein (1997). The presence of oestrid flies affects the whole herd rather than only the attacked individual, as alertness and notice of the presence of the flies is spread within the herd. Individual reindeer might be stressed and affected by harassment in the rest of the herd, even if this individual is not attacked directly.
Most observations of oestrid fly presence were recorded between 08.00 h and 21.00 h. During favourable weather conditions, oestrids could be present from earlier and later than these hours. Several observations of oestrids were made at low ambient temperature, while sun intensity was relatively high. Similarly, at temperatures above 11·0 °C oestrids could be active when sunlight was low or absent. These recordings and visual observations reveal a coexisting correlation between ambient temperature and solar irradiation that sets the condition for the flying capability of oestrid flies. According to Breev (1951), direct solar radiation can raise temperature of flies by 7–9 °C, and therefore decrease the lower critical ambient temperature for flight. Like Mörschel (1996), we observed that reindeer defensive reactions ceased almost immediately after thick clouds shaded them, and resumed once the clouds cleared. Oestrids are not expected to be active during the night, as temperature declines and the sun is absent. We observed oestrid fly activity on only 4 days after 21.00 h, and none after 23.22 h.
Oestrid fly activity was correlated strongly with all temperature measures. Temperature measurements on ground and solar irradiation explained (two times more of variation in activity levels than did ambient temperature. Hence the temperature at ground level and solar irradiation is probably more critical for activity of oestrid flies than ambient temperature. The lowest ambient temperature for oestrid fly presence was 6·9 °C, which supports observations by Downes et al. (1985). Others have reported threshold temperatures of 10 °C (Anderson, Nilssen & Folstad 1994; Mörschel 1996), 13–17 °C (Helle & Tarvainen 1984) and 13–15 °C (Kelsall 1975). Breev (1956, 1961) reported that the warble fly would fly at 7–8 °C and nose bot fly at 9–10 °C if exposed to direct sunlight. Different recording methods and number of observations probably explain most of the variation. We recorded ambient temperature at 2 m, while measurements in the literature vary between 30 and 40 cm (Nixon 1991) to 1·5 m above ground (White et al. 1975). Temperatures measured at 2 m could be between 9 °C and 25 °C lower than temperatures at ground level in the shade and the sun, respectively. The closer the ground, the more the recordings are influenced by thermal ground effects. Most likely 6–7 °C is the lower threshold temperature, although this depends on the presence of sunlight. Because oestrid flies were always active above 14 °C, regardless of other weather conditions, this temperature might be the threshold for continuous flight in absence of direct sunlight.
During temperatures lower than the threshold for flight, oestrids exposed to sunlight on the ground may heat up, raising body temperature to allow take-off. The ambient temperature might eventually cool the fly as cool air passes through the tracheal system and the fly will be forced to land. This would explain why oestrids were observed flying out from snow patches, warming up on rocks, then flying back to reindeer on the snow patch.
Although wind speed is reported to have a limiting effect on their flying capability, oestrid activity was not influenced statistically by wind speed. Strong wind will make manoeuvring difficult and cool the fly. Oestrid fly activity was recorded in wind speeds up to 12 m/s, compared to limits of 8–9 m/s (Kelsall 1975), 6–8 m/s (Anderson et al. 1994) and 11 m/s (Mörschel 1996). At high wind speeds, in our study oestrids flew close to the ground where wind was less severe. During severe harassment, reindeer occupied mainly the windiest sites. Reindeer were observed on hilltops only during oestrid fly harassment.
A lower threshold temperature of 7 °C concurs with numerous other studies with threshold temperature at 6 or 7 °C (White et al. 1975; Helle & Tarvainen 1984; Nixon 1991; Mörschel 1996; Anderson et al. 2001). Threshold temperature measurement probably depends on the recording distance from the ground. A recorded increase in mosquito abundance, followed by a decrease above 16 °C, corresponds with Nixon (1991), who observed an increase to about 18 °C, and then a decrease. Taylor (1963) hypothesized the existence of a lower and upper temperature threshold for insect activity. The decrease of mosquito abundance with solar irradiation may indicate that mosquitoes become overheated and inactive during midday and under sunny conditions. Thus, it is likely that there exists an upper critical temperature for the mosquitoes.
Wind speed was the single most important parameter for mosquito activity, and analyses revealed that the small and light mosquitoes were vulnerable to even low wind speeds. We made only six observations of mosquito activity at wind speed 5–7·5 m/s. Mosquitoes at this wind speed (at 2 m) were observed close to the ground, and thus exposed to lower wind speed than the one recorded. Others have noted threshold values for 6 m/s (White et al. 1975; Nixon 1991; Mörschel & Klein 1997) and 8–10 m/s (Anderson et al. 2001). Analyses of oestrid flies and mosquitoes revealed that their preference for weather conditions was different. Oestrids prefer warm, sunny conditions that are present mainly at mid-day, while mosquitoes prefer mild conditions (above a certain threshold temperature) present mainly in the mornings and afternoons.
EFFECTS OF OESTRID FLY HARASSMENT ON ACTIVITY BUDGETS
The effect of insect harassment on activity budgets has been reported in several studies, indicating similar trends; a decrease in feeding and lying and an increase in moving and standing (e.g. Thomson 1973; Curatolo 1975; White et al. 1975; Roby 1978; Downes et al. 1986; Ion & Kershaw 1989; Nixon 1991; Mörschel & Klein 1997). The dramatic decrease in feeding in this study supports the results of Thing (1984) who reported a five- to sixfold decline in feeding intensity between periods of no harassment and severe harassment. Helle & Tarvainen (1984) found that reindeer calf weights in the autumn were correlated negatively to the insect harassment the preceding summer. This supports results from Setesdal Ryfylke, Norway, where significantly lower carcass weights were recorded after the warm summer of 1997 compared to the cold summer of 1998 (Colman 2000). The same study documented that animal did not compensate through increased grazing time or intensity during night-time when insect harassment was low. Indeed, the Svalbard reindeer (R. t. platyrhynchus) make reference to an undisturbed life in the absence of insects and predators. During summer in these islands the reindeer spend 62% of the day grazing and 34% lying (Reimers 1980).
Most of the harassment studies fail to record which parasitic flies that has caused the recorded harassment and altered behaviour. However, Anderson et al. (2001) have, with the use of host-mimicking trap catches, determined which parasitic flies that actually attack reindeer. They concluded that oestrids were the primary cause of panicked individual and herd behaviours of reindeer. These results concurs with our study were oestrid fly harassment had the greatest effect on activity budget of reindeer. Curatolo (1975), Downes et al. (1986) and Mörschel & Klein (1997) reported that the activity budget of caribou changed significantly with the appearance of oestrid flies. Our results are identical to those reported in Thomson (1973) and White et al. (1975), with a dramatic decrease in feeding and lying, and increase in walking, running and standing. Description of reindeer annoyance behaviour in these studies indicates that oestrid flies caused severe harassment, while mild harassment could have been caused by any bloodsucking fly.
EFFECTS OF MOSQUITO HARASSMENT ON ACTIVITY BUDGETS
Mosquito abundance had no important effect on the activity budget in our study, where only a small decrease in lying and an increase in walking were evident. Although mosquitoes increased avoidance and annoyance behaviour, the direct effect on feeding from mosquitoes alone seems unclear. Neither Curatolo (1975), Roby (1978) nor Mörschel & Klein (1997) could record a significant decrease in time spent feeding with the presence of only mosquitoes. Roby (1978) and Mörschel & Klein (1997) concluded that oestrid flies had a larger influence on activity budget of caribou than the presence of mosquitoes. Roby (1978) observed a higher percentage of standing and running and a lower percentage of lying for oestrid fly harassment, compared to similar levels of mosquito harassment.
EFFECTS OF WEATHER PARAMETERS ON ACTIVITY BUDGETS
The important behavioural responses that correlated with high temperatures and intense solar irradiation during summer were mostly indirect responses to oestrid fly harassment. In the absence of oestrid flies, no consistent reaction to these weather parameters was evident. Snow patches, marches, shadow areas and windy spots were visited frequently during warm days, but only during oestrid fly attacks. Yousef & Luick 1975) found that that reindeer appeared to be as heat-tolerant as some domestic cattle and wild African ungulates. With access to water, respiratory rate, heart rate, rectal temperature and other physiological measurements did not increase before ambient environmental temperature of 30–35 °C (Yousef & Luick 1975). Rosenmann & Morrison (1967) concluded that reindeer have a good capacity for heat resistance when water is available, but very poor resistance to water deprivation with or without heat stress. Several ungulates in the northern hemisphere are reported to be less active during warm summer days, or to select sites to avoid heat stress as, e.g. moose (Alces alces) (Shwab & Pitt 1991). In Norway, wild mountain reindeer seldom stay at lower altitudes than 1000 m during the summer. Above this altitude ambient temperature rarely exceeds 20 °C, while temperatures of 25 °C almost never occur. While ambient temperatures should normally not pose any thermal threat to Norwegian reindeer, the effect of direct solar radiation on reindeer is not known, and should be investigated further. If there is a behavioural response to increasing temperature and solar radiation, reindeer should lie, minimizing energy expenditure. However, energy demanding behaviour such as walking and running often increases with increasing temperature, due to attacking oestrids. During the summer of 1997, the warmest summer in Southern Norway in 51 years, no signs of heat stress or direct unpleasantness from high temperature or direct sunlight were observed during our study.
We would like to thank Sigurd Svela for his stamina and invaluable help during preparation and fieldwork. Thanks to Eric Post, Pernille Døving, Frank Mörchel and Stein Kristiansen (DNMI) for help and inspiration during preparation, Dag Hjermann, Hege Gundersen, Gry Gundersen and Atle Mysterud for statistical help, Jonathan Colman and Tom Warren for proof-reading and Arne C. Nilssen for guidance regarding parasitic flies.