On the vectors of cutaneous leishmaniasis in the Central Amazon of Brasil . 3 . Phlebotomine sand fly stratification in a terra firme forest 0 )

Stratification studies showed that the vast majority of sand fly species in the Manaus region are found in tree canopies (15m), and that there are two dominant species which are the vectors of Leishmania braziliensis guyanensis. It took 43 weeks of trapping to capture all 50 species of sand flies collected.


INTRODUCTION
The systematics, ecology and behaviour of sand flies in areas where Leishmania braziliensis guyanensis is endemic, has been the subject of much research in the past decade.
Recently the taxonomic status of the principal vector species of "pian boib" was clarified (Ware & Fraiha, 1977) and Lutzomyia umbratilis was indicated as being the principal vector both in the Monte Dourado and Manaus areas of northern Brazil (Lainson ef al., 1976;Arias & Freitas, 1977b).In the Manaus area, Arias & Freitas (1977a) indicated L. anduzei as a secondary vector of Le( 3 ) • b. guyanensis while Lainson el al. (1981b) indicated L. whitman/ as a secondary vector in the Jari area.
Even though the range of L-anduzei does extend to the Jari region, and promastigotes have been found in its digestive tract, this sand fly species has not been shown to be a secondary vector in Jari (Lainson, Shaw & Ready, personal communication).L whitmani, on the other hand, has not been taken in the Central Amazon around Manaus.
Other studies of Arias ef al. (1981), Arias & Naiff (1981) and Lainson ef al. (1981aLainson ef al. ( , 1981b) ) have shown that the natural reservoir host of this Leishmania are arboreal mammals, as suggested by Arias & Freitas (1978).Sand fly stratification in other regions of the New World has been done by Disney (1963), Williams (1970), Chaniotis ef al (1971aChaniotis ef al ( , 1971b)), Shaw ef al. (1972), and near Manaus by Arias & Freitas (1977b).All authors showed that there were stratification differences between sand fly species, some species being caught predominantly in the canopy and other species being caught predominantly on the forest floor.highway.This is a terra firme forest that is frequently entered by man and is maintained as a biological research station.A more detailed description of this area can be found in Penny & Arias (1982).1978), from wild animals (Arias et al., 1981;Arias & Naiff, 1982), and man (Arias, unpublished data) from this area studied.

Leishmania
All sand flies were sorted at INPA in Manaus, slide mounted in Berlese solution, and identified according to the systematics proposed by Lewis et al. (1977) (").

RESULTS
Figure 1 shows the period of time that it took us to capture 100% of the flies taken in light traps.The first week we captured 46% of all species we found at the Ducke Reserve at all heights during the 62 weeks.By the second week we had taken over 50% of the species.We surpassed the 75% mark after che fifth week of captures, the 80% mark after 12 weeks, the 90% mark after 22 weeks, 95% mark after 33 weeks, and only after 43 weeks did we take all species captured during this study.No previously uncollected species were taken during the last 19 weeks of the study.L anduzei. -1 •T •ox"r   At the one meter height, there were 5 dominant species which comprised almost 65% of the total sand fly catches.These were: L. anduzei, which comprised 14.82%, L rorotaen sis, 12.49%, L ruii, 8.50%, L. umbratilis 14.50% P. s. SQuamiventris 14.48%.
On the other hand, there were only two species which could be considered dominant at the 15 meter capture sites.These were.L. anduzei, which comprised 41.06% and L umbratilis, which comprised 36.01 %.
Of the total sand fly population captured at both heights, we can see that there are two dominants, L. anduzei and L umbratilis, having captured 37.56% and 33.08% respectively Figure 3 shows the incidence of all sand flies per trap per week as caught at the one and 15 meter heights.We see here that there appears to be a high initial population which drops readily after 5 weeks of trapping and h.38 peaks during the end of the month of November and month of December, 1977 and a larger peak during the month of May, 1978 to reduce the size of the graph and maintain the smaller pe,';ks.It can be seen here that the initial population appears to drop rapidly after an initial peak, slowly rising around No vember of 1977, dropping at the end of De cember, and having another large peak in May of 1978.The one meter population appears to be at its highest during the lirst part of the collection period, dropping down and main taining a low level of incidence throughout the collection time.
Figure 5 shows the activity of P. ayrozai This species had two peaks of activity around December, 1977 andMarch, 1978 at the 15 meter height.At the one meter height this species was not present at ail times and had a slight peak of activity during March, 1978. Figure 6 shows the population activity of P. davisi.
This species has a main peak of activity at 15 meters during the month of February and lesser peaks of activity through out the year.At one meter the activity of this species appears to be slight throughout the capturing period.solid bars across the graphs at three different intervals represent periods of time when twice as many traps (4) were set at both heights to assess "trapping out", which did not appear to occur.The peak of sand fly activity at both heights during the end of November and throughout the month of December may be due to the second complement of traps set.This may account for a large existing adult popu lation (particularly L. anduzei and L. umbratilis) which is taken initially, and subsequent captures represent the forest productivity of sand flies.One second possibility for an increase in population at this time is the change in season, beginning of rainy season, which has been shown to trigger activity in some insect families (Penny & Arias. 1982) .
Figure 4 shows the seasonal incidence of L. anduzei.It must be noted here that the graph is shown with two scales on the left hand side Figure 7 shews the activity of P. guyanen sis- At both the 15 and the one meter heights this species was nonapparant until November, 1977.This activity is noted until around August of 1978 when it stops.At one meter this species appears in low number and follows approximately the same activity but at greatly reduced levels; it was not captured again after July, 1978.The low periods of activity of P. guyanensis correspond directly with the dry season in our area (Penny S Arias, 1982).
Figure 8 shows the population activity of P. paraensis.
Again at the 15 meter level the peak activity is during the month of February.This species has two subsequent peaks of activity during the latter part of the year.The initial population, like that of L. anduzei and L. urr.bratilis, started high only to drop rapidly.At one meter height the population activity was low, sporadically appearing in the captures.
Figure 9 shows the population activity of L. roroiaensis.
This species was one of the few species which was considerably more active at the one meter level than at the 15 meter level.At the 15 meter level this species was present at a low incidence, having a slight peak of activity during the end of 1977.At the one meter height traps it was collected in great numbers and showed a small peak of activity during October, 1977.It had its greatest peak of activity during November-De cember, 1977.
After this, the population dropped only to have a slight peak of activity during late June of 1978.
Figure 10 shows the activity of a new species of sand fly in the subgenus Trichopho romyia (Arias & Young, 1982), which for now is called P. p3raensis and /... umbratilis, this species shows great ac tivity at the beginning of the trapping program and drops after the first 5 weeks.The two following peaks of activity at the 15 meter level occur during the months of November. 1977 andMarch. 1978.This is one of the few species which showed considerable activity at both the one and 15 meter levels, even though there was greater activity at the 15 meter level Most of the one meter level activity peaks correspond with those at the 15 meter level.
Figure 11 shows the activity of P. s. squomiventris- The 15 meter activity was low and fairly constant throughout the capture program.However, at the one meter level the activity was more pronounced.Here we find activity peaks during late November -early De cember, 1977 andlate March andearly April. 1978.This species, once considered to be P. maripaensis by Arias & Freitas (1977a, 1977b. 1978), has recently been shown to be P. s. squamiventris by Ready et zl. (1982) .
Figure 12 shows the activity of L. umbra tilis.As mentioned before, this species showed The seasonal populational activity for the remaining species was so low that only sporadic captures were found and no con clusions could be drawn.

DISCUSSION
At least 22 weeks of trapping were required to get 90% of the species of sand flies utilizing CO" 2 baited CDC light traps.Even though we ran our traps for 62 weeks, and we had captured all the species by the 43rd week, we did not capture all the species found in the area.Other studies in the same Ducke Re serve area have produced at least 5 other species which we did not capture here, and we expect that in the future we will find more species as new capturing techniques are applied; collections of some sand fly species being very specific as to capture method.L-dendrophila is a species that is very commonly taken in tree base captures, yet we only took 6 individuals in light traps and these were taken at 15 meters.L. spinosa, another species found resting on tree bases, was also only taken in traps at 15 meters.
The relatively small number of commonly captured sand flies, and the large number of infrequently taken species reflects what appears to be the trend of species compo sitions in the Amazonian forest (New, 1979: Meinander & Penny, 1982) .Very few species are taken very commonly and many species are hard to find.The capture method utilized also strongly influences the species taken.
However, summarizing all methods of capture we have utilized in the past (Arias & Freitas. 1977a, 1977b, 1978), there still exists a vast number of species that are not at all frequent.
Since the two dominant species of sand flies captured during this program were the vectors of "pian bois" in the Manaus region of the Amazon (Arias & Freitas, 1977a, 1978),  1982) .Human bait captures in terra firme forest near Manaus also show that even though L. umbratilis is a causual man biter at the forest floor, it is the principal man biter at the 15 meter level (Arias & Freitas. 1977b;Ready & Arias, unpublished data) .Vector and anthropophilic species have been shown to have different biting and population peaks at different heights in the forest from Belize, British Honduras (Disney, 1968;Williams. 1970), andPanama (Johnson ei al., 1963;That cher, 1968;Chaniotis ef al., 1971aChaniotis ef al., , 1971b) ) to Brazil (Shaw ef al., 1968;Arias & Freitas 1978).Usually it has been noted that the biting habits of the vector species is very closely related to the habitats of the natural reservoir host.If the natural reservoir host is a ground dweller, the vector species usually is a ground level biter; and if the natural reservoir host is a tree climber, the vector species is a canopy biter (Shaw et al., 1968;Arias & Frei tas, 1978).
Even though there is a stratificational preference of sand flies, it appears from these results that the sand flies in the Northern Central Amazon prefer the canopy to the forest floor.Of all the species, only five were more frequently caught on the one meter traps.This appears to be in contrast with the results of light trapping in Panama (Chaniotis et al.. 1971b) where only two species were more frequently captured in the canopy light traps.In our work, almost 90% of all sand flies taken were in the canopy light traps, while in the Panama work less than 25% were taken in canopy light traps.Maybe the influence of C0 2 , which was present in our traps and not in theirs, accounts for these differences.In another work (Chaniotis et el. 1971a), they showed that with man biting captures almost 50% were taken at the canopy level.
Several times our study showed that there was an initial population peak that sharply fell during the first five weeks.Subsequently there was a smaller peak during the month of October.It is believed that the original popu lation is high and the light traps rapidly reduce this, in the immediate area and the subsequent collections are indicators of lateral migration through the forest and emergence of new sand flies.We feel this is the case because the smaller peak in October corresponds to the introduction of the 4 additional traps that were set cut to see if we were "trapping out" the population.The first few catches in the extra sets of traps were usually higher than the traps that remained throughout the entire study.
The large catches which correspond to the very large peak in May, 1978 may have resulted from a "bonanza" population (Penny & Arias. 1982) .There was only one trap at 15 meters which had this exceptionally high number of sand flies, particularly L. anduzei and L. umbra tilis; this was during two consecutive weeks the first week with over 4000 sand flies in the trap.P-s.squamiventris is the most avid man biter at ground level, yet does not appear  to go up into the canopy except in relatively small numbers.The peak activity in the traps corresponds with those of Arias & Freitas (1977b) when doing horse and human bait studies.
The most frequent peak of specific activity is that in late November and throughout December.This may be due to one of two factors.
First, that the second complement of traps encountered an older population that had not been previously taken.If this were true, it is logical that the peak would not be as accentuated as the initial peak because a) there is a dilution factor of 2 other traps (those which had been running for a longer period of time) and b) because of the use of a 4 week running mean, which tends to reduce extreme "peaks and valleys".The second possible explanation for this peak is that there is a slight emergence of insects in general at this time of year (Penny & Arias, 1982) which closely corresponds with the onset of the rainy season.

ACKNOWLEDGMENTS
We would like to thank João F. Vidal for his field and laboratory assistance, and Artemio Coelho da Silva for the drawings.
CDC miniature light traps were utilized throughout the study.The slight modifications included replacement of the collector sack with one of a finer mesh screen and by adding a small styrofoam container containing approximately 800g. of dry ice.The styrofoam box was completely taped closed to allow slower release of the C0 2 gas, and then taped to the side of the traps.The power source for each trap was two rechargeable 6-volt motorcycle batteries.Two traps were set at 15 meters and two at 1 meter above the forest floor.Each of the collection sites remained the same throughout the study, and were located at INPA's Ducke Forest Reserve, 26 kilometers NE of the city of Manaus on the AM-010

Figure 2
Figure 2 shows the population density of the different sand fly species taken.Even though a great variety of species were taken, the majority of individuals collected were two species.These two species are L. umbratilis

Fig. 3 -
Fig. 3 -Seasonal distribution of all sand flies captured in light traps over a 62 week period at the Ducke Forest Reserve, Manaus, Brazil (from Penny & Arias, 1982).The curves represent 4 week moving averages as in Chaniots et al. (1971b).

Fig 4 -
Fig 4 -Seasonal distribution of L. anduzei captured in light traps over a 52 week period at the Ducke Forest Reserve, Manaus, Brazil.The curves represent 4 week moving averages as in Chaniotis et al. (1971b).

Fig. 5 -Fig. 6 -
Fig. 5 -Seasonal distribution of P. ayrczai captured in light traps over a 62 week period at the Ducke Forest Reserve, Manaus, Brazil.The curves represent 4 week moving averages as in Chaniotis et al. (1971b).

Fig. 7 -Fig. 8 -
Fig. 7 -Seasonal distribution of P. g-jyanensis cap tured in light traps over a 62 week period at the Ducke Forest Reserve, Manaus, Brazil.The curves represent 4 week moving averages as in Chaniotis et al. (1971b).

Fig. 9 -
Fig. 9 -Seasonal distribution of L. rorotaensis cap tured in light traps over a 62 week period at the Ducke Forest Reserve, Manaus, Brazil.The curves represent 4 week moving averages as in Chaniotis et al. (1971b).

Fig
Fig. 10 -Seasonal distribution of L. 1.20.1.20cap tured in light traps over a 62 week period at the Ducke Forest Reserve, Manaus, Brazil.The curves represent 4 week moving averages as in Chaniotis et al. (1971b).

Fig. 11 -
Fig. 11 -Seasonal distribution of P. s. squamiventris captured in light traps over a 62 week period at the Ducke Forest Reserve.Manaus, Brazil.The curves re present 4 week moving averages as in Chaniotis et al. (1971b).
Fig. 12 -Seasonal distribution of L. umbratilis captured in light traps over a 62 week period at the Ducke Forest Reserve, Manaus, Brazil.The curves represent 4 week moving averages as in Chaniotis et al. (1971b).