Three Anopheles species have been reported from Finland 9. Anopheles beklemishevi has a northern distribution in Finland, while the other common species, Anopheles messeae, is dominant in the southern part of the country 17. Anopheles claviger is recorded from the Aland Islands and was also found by the authors in the south-western archipelago. In December 2004 and January 2005 flying specimens of an unidentified Anopheles (cf. messeae) were observed in summer cottages in two different localities in southern Finland.

Because of many taxonomical problems in the Anopheles genus, there is conflicting information on the details of the life cycles of the separate species. A general feature of the northern species is that the adult female hibernates. There are possibilities of a summer generation, at least in warm summers as in 1901 in Finland (possibly A. messeae)912. A. claviger hibernates as larvae 33 and, at least in England, seems to be bivoltine 34. In Finland, the majority of the hibernating females reach the adult stage in the middle of August. The male is short lived and dies soon after mating. The female may suck blood or honey-dew before seeking shelter in sheds and houses for hibernation. Because of the cold climate the female cannot leave the hibernation site before spring. In the traditional agricultural society of Finland many of the mosquito females spent the main part of their adult life together with man and domestic animals. In warm conditions the female may take several blood meals during the hibernation, but it will not lay eggs before the spring 13.

A. messeae has been considered an important vector of historical endemic malaria in several parts of Europe and was also an important vector in the Soviet Union 415. In the northern parts of Russia and along the Baltic coast, A. beklemishevi was an important vector 35. A. claviger has also locally been reported as a vector 15. Some recent research, however, has questioned A. messeae as a vector of malaria 36. Jaenson & Ameneshewa 37 reported that blood-feeding was not a prerequisite for hibernation of A. messeae. This does not exclude the possibility of repeated indoor blood-feeding in warm indoor conditions. There is still unresolved and conflicting information on the feeding, resting and hibernation habits of the female (exophagy or endophagy, zoophily or anthropophily, exophily or endophily, complete diapause or semiactive winter habit). The anophelines in the northern region, however, must be flexible in their habits to survive the strong annual and seasonal fluctuations in temperature (-51°C to +33°C in Finland and rain/snow). In any case, one or more species, distributed over the whole of Finland, have been vectors of malaria in Finland. Yet it is likely that one or more additional species, still not reported from Finland, will be found as four additional species have been reported from Estonia and Sweden 1011. For the time being it is not possible to define which mosquito species was important for the malaria transmission in Finland.

The Anopheles species, which have been vectors of malaria in Finland, presumably have existed here since prehistoric times. It is, however, probable that malaria reached Finland only during the 17^th century. There are several cases of malaria in the Mälaren region to the west of Stockholm in Sweden during that time. Both 1692 and 1693 are reported as years of severe fever 38, which spread rapidly also to the east. A total of 1,803 persons died of malaria in the western parts of Finland and in the south-western archipelago during the years 1751–1773 23. Haartman 21 reports severe epidemics in the region of Turku in the years 1774–1777 and the physician F.W. Radloff mentioned that malaria was very common in the Aland Islands in 1795 39.

From 1800 to 1870, there were at least 5,431 death from malaria in Finland. During this period the Finnish population grew from 832,700 to 2,032,700 people 40. In a few epidemics, the mortality can be estimated to 0.85–3.0% 6. The number of deaths from malaria increased when the number of malaria cases increased and thus a change in virulence of the Plasmodium is not to be assumed. The total number of malaria cases in the middle of the 19^th century may tentatively be estimated to be 100,000–300,000 or about 7–20% of the whole population.

Although the source value of both the death statistics and the medical reports can be discussed separately, they are strengthened when compared with each other. As sources they are completely independent, and they partly overlap the period 1850–1870, showing similar trends. The district physicians reported 174 local epidemics during 1854–1862 and during the same period the deaths from malaria rose to 1,687 cases. In the years 1873–74 and 1877, the number of epidemics also rose and the same trend can be seen in the parish death records.

The worst malaria year in Finland was 1862. The eastern parts of Finland were particularly affected, including Karelia. Several physician's reports reveal how severe the situation was. In the district of Mikkeli over 4,000 persons became ill. Moreover, in the district of Joensuu 4,000 persons suffered, in Rautalampi several thousands and in the worst parishes in the districts of Viipuri and Muola every third person became ill. In eastern Finland the summer was cold and damp. Malaria epidemics even broke out in the north and in Sotkamo in the Kajana district more than a hundred people became ill.

The correlation between temperature and the number of malaria deaths was tested on annual, seasonal and monthly levels. The seasons were interpreted as follows: winter (DJF), spring (MAM), summer (JJA) and autumn (SON). The malaria cases were tested against the temperatures of both the current malaria year and the preceding year. In Table 4 the calculations of monthly correlations were repeated with a square root transformation of the number of malaria cases to avoid biases caused by strong peaks during epidemics. Malaria deaths peak late in the spring but for completeness the calculations are extended further to the autumn of the malaria year. The results are presented in Tables 1,2,3,4 and Figure 2.

Annual mean temperatures were obviously not significant in the correlations. On the seasonal level the summer of the preceding year is highly significant (0.1% risk level). In all other seasons the risk level was more than 10%. Renkonen 26 tried to show a correlation between malaria epidemics and the temperature of April or spring. The correlation coefficient for both April and the entire spring were close to zero and thus Renkonen's hypothesis can be rejected.

June and July of the preceding summer were very significant with a faint "tail" in August. The weak correlation of October temperatures in the St. Petersburg series (5% and 10% risk level respectively in Tables 3,4) can safely be considered stochastic noise with no support from the other series.

Death caused by what were called 'intense fever', hetsig feber (48853 cases), cold, influenza, flussfeber (7670 cases), fever, feber (64604 cases) and cold, förkylning (1381 cases) were tested for correlations with both temperatures and with malaria in 1800–1850, but no correlations were found. The words expected to indicate malaria are well-defined and separated from other kind of fevers and they can safely be analysed as a homogeneous disease.

The correlation is strongest with the season when the vector is in the larval stage. Thus the number of malaria cases correlates with the warmest time of the summer when there is no Plasmodium present in connection with the mosquito. The conclusion is that fluctuations in summer temperatures regulate the number of mosquitoes reaching the adult stage and not the development of Plasmodium. Because malaria epidemics covariate with the summer temperatures, the sporogony inside the mosquito may be treated as a constant parameter. Thus the sporogony of Plasmodium in Finland must have taken place in reasonably stable temperature conditions.

The sporogony is temperature dependant and the duration of the sporogony of P. vivax is 7–8 days in 30°C, 8–10 days in 28°C, 15–16 days in 20–21°C and 55 days near 16°C with development stopping completely below 16°C 4. When the mosquito reached the adult stage in August, the temperature was already falling. If the mosquito took a blood meal from a carrier of malaria late in the summer or at the beginning of autumn, the malaria parasite would have had no time to complete its sporogony in the vector in outside conditions. August was the warmest month the adult mosquito female encountered during its life. The mean temperatures of August and May during the years 1829–1870 were in Helsinki 15.53°C (range 12.3–20.6°C) and 7.34°C respectively and the corresponding temperatures in Tornedalen 13.15°C (range 9.9–16.7°C) and 4.53°C. These values give a realistic picture of the temperature conditions within the area where endemic malaria occurred in Finland. The summer season in Finland was within a wide range clearly unsuitable for the development of Plasmodium (Figure 3).

The principal explanation for continuation of endemic malaria must be the choice of hibernation shelter by the female mosquito, within a warm environment in close proximity to humans. Such places in previous centuries were animal shelters and human dwellings. An indoor transmission cycle was first suggested by Robert Koch in the 1890's 25 and Swellengrebel and de Buck 41 gave a detailed description of indoor infection by hibernating mosquitoes in the Netherlands. The situation in Finland was different. Sivén 25 noted during the epidemic in Helsinki in 1902, that many infants who got malaria in the spring, were born in the winter. In the years 1750–1850 there were about 40–50 corresponding death cases. As a consequence hibernating mosquito females were active and capable of infecting humans during the whole winter.

The relative proportion of annual deaths in different age groups of the total population was compared in Figures 4,5,6, for malaria and seven other diseases in the years 1750–1850. The age group structure of malaria deaths is very similar to some other fevers and approaches that of the total population. Deviations in the total population are mainly explained by increased (>50%) omissions of the cause of death in children under two years old in the parish records. Malaria seems to spread with equal probability among all age groups. If the majority of infections had occurred during a limited time in late summer or the beginning of autumn, the resulting age group distribution of malaria deaths would have been different. In the traditional agricultural society, certain age groups used to live apart during the summer, usually in unheated buildings. In October-November, decreasing temperatures forced them to move back into heated buildings 42. Thus, the most conservative explanation is an extended cold season infection time when all age groups congregate in heated buildings.

Hernberg 4344 studied the malaria epidemic in Finland during the years 1941–1945, resulting from infections acquired during the war. He concluded that P. vivax has a variable incubation time of 5–12 months or more. Another conclusion was, that the spring peak of malaria was a result of an infection in the previous year, because he assumed that no mosquitoes were present during the winter. He obviously did not note Sivén's observation of winter infections. Winter infections were probably the reason why Renkonen tried to correlate the infections with the mean temperature of March 26.

According to Swellengrebel and de Buck 41, the main infection time of malaria in the Netherlands was from the middle of August to the middle of November and there were practically no infections after November because of degenerating sporozoites. The degeneration of sporozoites late in the autumn was due to low temperature in unheated bedrooms. In Finland, however, hibernating mosquitoes have been able to infect humans during the whole winter because of heated bedrooms. The importance of the dormancy of the Plasmodium is pronounced in the summer, as the parasite needs a prolonged dormant stage in humans to survive the summer. The summer temperature conditions are unstable and highly variable and there are no or too few anopheline females during a considerable part of the warm season. The winter temperature conditions are stable inside human dwellings and the mosquito female can repeatedly produce sporozoites and infect humans. Some of the sporozoites immediately initiate schizogony while most of them remain as dormant hypnozoites in the hepatocytes. Another source for schizonts may be gradual activation of dormant hypnozoites from the preceding winter. As a consequence, during the erythrocytic cycle some gametocytes are produced throughout the winter in high latitudes. It has been suggested that hormonal changes in the host, controlled by the changing light-cycle, induce mass activation of the hypnozoites in the spring 45. The spring peak of hypnozoites could be of no importance for the Plasmodium in Finland. Some of the hypnozoites, however, were activated only in the following autumn or later. Thus, paradoxically, summer dormancy was crucial for maintaining the Plasmodium until the next generation of anopheline females were present. Such a pattern had already been suggested, in parts, by Reiter 1.