.. I Reprinted from TRXNSACTIOSS OF THE ROYALSOCIETYOF TROPIUL ~UEDICIX~AXD HYGIEXE Vol. 49. No. 6. pp. 505-527, 1953 ANOPHELES GILES GArMBIAE COASTAL AKD AREA ANOPHELES OF LIBERIA, MELAS WEST AFRICA THEOBALD IN A t BY HENRY ;_Tatiotlal Institutes of Health, hl. GELFAND, Laboratory of Tropical 3I.D.’ Diseases, Bethesda, Maryland. Anopheles gambiae Giles has long been known to be one of the most common mosquito species in West Africa, and the one mainly responsible for the transmission of malaria in that part of the world. Only one year after its original description in 1902, a variety melas was described for an extremely dark variation bred from brackish water in Gambia (THEOBALD, 1903). B_UBER and OLINGER (1931) extended knowledge of the variety with observations and experiments in Ngeria. Since that time, this designation has been applied to females of A. ganzbiae which possessedmelanic coloration and an additional dark band in the terminal white area of the palps, and which were associated with brackish water in coastal regions (EVANS, 1931, 1938). Th e variety has been found in numerous localities along the coasts of West and East Africa (MUIRHEAD-THOMSON, 1947, 1951a ; and HOLSTEIN, 1951). It is worthy of note that var. ;meZas was never found in Brazil during the invasion of that country by A. gambiae (SOPER and WILSON, 1950). During World War II, the greatly increased interest in the epidemiology and control of malaria in West Africa resulted in a great extension of our knowledge of this variety. As a result of work in Sierra Leone (RIBBANDS, 1944a ; MUIRHE.~D-THOISOX, 1945), it was elevated to specific rank, and important differences in the bionomics of the two species were discovered. In Liberia, YOUNG and JOHNSON(1949) recorded the finding of Anopheles melas larvae on Bushrod Island, near Monrovia. I am unaware of other references to this species in that country. The present report is based on work done in Marshall Territory, Liberia from January, 1952 through July, 1953:. Over 11,000 adult specimens of the A. meZasA. gambiae complex were collected, and over 3,400 egg batches identified. t This work was done as part of the United States Technical Co-operation Administration Pro-. gram in Liberia, and was a collaborative project of the National &IicrobioJogical Institute of the U.S. Laboratory Public Health Service and the National Pubiic Health Service of the Republic of Liberia. facilities were made available by the Liberian Institute of the American Foundation for Tropical: Medicine. and American, without whose co-operation. * I am greatly indebted to many persons, Liberian and assistance this work would not have been possible. I wish specifically to mention The Hon. R, Francis Okai, Superintendent of LIarshall Territory, Justice Williams, proprietor of Cow Farm, and Mr. Joseph K. Davis, faithful laboratory assistant. 1 Routine collections for June and July, 1953 were continued after my departure, under the direction of Dr. Thomas A. Burch for whose assistance I am greatly indebted. -‘ -._-. 509 Desuiption Anopheles gambiae AP?D A. melas IN LIBERIA of the area. Marshall Territory is a small district on the Liberian Coast, lying between the counties of hlontserrado and Grand Bassa, about 30 miles east of Monrovia, comprising the area on both banks of the This is a tidal, brackish-water stream lined for many miles by mangrove sxvamp. lower Du River. Marshall City, the principal town, is situated at the end of a long, narrow peninsula, betsveen the river and the Atlantic Ocean. Numerous small villages are scattered along the river edge and throughout There are few roads : travel is mainly by foot trails the “ bush ” extending away from the river. and canoe. The one motorable road leads from the river opposite Marshall, two and a half miles to the Liberian Institute, thence to Roberts Field, Firestone Plantation, and hlonrovia. The predominant vegetation bordering the river is the red mangrove, Rlzizoplzo~a Taceznosa, &ich is restricted to the margin subject to daily tides. This band, of variable width, is a dense tangle of mangrove aerial roots growing in bare, black muck. Scattered specimens of the white mangrove, FIGVRE1. Lower Du River, Marshall Territory, Liberia. Avicennirz nitida, are occasionally seen on higher ground where no mosquito breeding occurs. In almost all places investigated, the slope of the land rises sharply, and the fringing mangrove swamp In relatively few isolated areas, the slope of the river bank is more gives way to dry, sandy grassland. This second zone is above the gradual, and a zone of salt-marsh grass, Paspalmz aaginatztm appears. level of the daily tides, but is flooded twice monthly by the “ spring ” tides, which, at the lower Du These zones and their influence on mosquito River, average about one foot higher than “ neap ” tides. breeding have been fully described by Mr;uEtE_4D-THOhlSoK (1945) for Freetown, Sierra Leone, and The Marshail area differs only in detail from both by BRUCE-CHWATT (1949) for Lagos, Nigeria. those places. Figure 1 shows the general area of the lower Du River and the towns and villages where the studies Marshail City, including Krutown, Fantitown, and Freetown, to be reported below were conducted. where both native thatched huts and two- or is a semi-urban community of some 500 inhabitants, The six villages-on the north bank of the three-storied wood and corrugated iron houses are found. river, opposite Marshall, are small, primitive communities of the predominant Bassa tribe with a popuLloydesville and Wehn are somen-hat larger villages of about 100 lation of only 25-35 people each. people each. ?eather. However, such data are Ko meteorological data are collected anywhere in alarshall Territory. These may differ somewhat available for Roberts Field, located about seven miles from the Du River. Table I summarizes from conditions in Marshall in day-to-day detail, but give a correct general picture. H. M. GELFAXD 510 the available information by months during the period of this study.” in 1952 was about average and the monthly distribution was typical. were not marked. The total of 138 inches of rain Daily temperature variations The mosquitopopulation. Table II lists all the adult mosquitoes collected during 17 months in Marshall Territory by nighttime human bait-collectors, in man-baited traps, and in native houses in the early morning. It gives, therefore, some indication of the relative numerical importance of the anthropophilic night-feeding species during the period of this study. Day-feeding mosquitoes were very infrequently encountered in this area and were never troublesome. TABLE I. Rainfall, temperature Rainfall Month (inches) and relative humidity 1 Average air temperature i i Maximum I at Roberts (“F) Field, Liberia, 1952-53. 1 Average relative humidity Minimum Maximum Minimum 88.6 ,! 73.0 89.2 j ; 73.7 >, 88.8 73.3 >, 60.0 89.3 73.9 >) 63.1 87.6 74.0 ,, 68.4 72.9 ,, 68.4 81.0 72.7 ,, 75.2 32.48 79.4 71.8 ,, 78.7 Sept. 34.54 81.7 72.6 ,, 81.2 Oct. 18.13 84.3 72.4 I, 84.6 Nov. 7.44 85.7 73.0 9) 79.6 Dec. 2.46 86.8 72.2 ,> 76.0 Jan. 0.17 88.9 73.0 f, 63.2 Feb. 1.66 89.0 73.2 >, 60.2 March 3.76 90.3 73.3 t) 58.4 April 9.48 89.8 73.1 ,, 63.4 May 16.53 86.9 73.5 >, 69.6 June 29.80 July 60.23 Jan. 1.15 Feb. 3.67 March 7.13 April 2.05 May 12.30 June 31.93 85.7 JOY 33.41 Aug. * Provided by the Manager, is gratefully acknowledged. Ii 1 Pan American j 1 i I I 1 World Airways, Saturated Roberts Field, Liberia, (%) ( I 1 65.1 59.9 whose courtesy Anopheles gambiae AKD A. 511 melas IN LIBERIA Table II indicates the importance of the A. melas-A. gambiae complex. These species were the main pest mosquitoes, in addition to their importance as disease vectors. Also notable in this table is the absenceof Anopheles funestus. This species, an important vector of malaria in most parts of equatorial Africa, and present in large numbers only a few miles away from the Du River, was collected only twice in Marshall Territory during this time. TABLE II. Anthropophilic, night-feeding mosquitoes collected by all methods in Afarshall Territory, 3nc9 c7 17JL'JJ. Anopheles melas-Ano-pheles gambiae CztLextJmllasizrs Theo. compler: 11,331 1,113 Anopheles hancocki Ed\v. 521 Mamonia 103 (Man.) ztn<fomis (Theo.) Aedes spp. Other CzJex spp. Other Mansonia 180 78 spp. 52 Other Anopheles spp. 26 The identification of A. melas and A. gambiae : In a detailed study, RIBBANDS (1944a) showed conclusively that the melanic “ 4-banded palp,” so long used for the identification of ” var. melas,” is a genetically determined familial characteristic, and that the percentage of proved A. melas demonstrating it varies from locality to locality. This was investigated in a small series in Liberia. Of 206 A. melus identified by eggs in Marshall Territory, 20, or 10 per cent. had a recognizable extra band in the terminal white area of the palps. In addition, 14 others showed a few black scales in this site (grade 1 of RIBBANDS). Of 149 A. gambiae none had a recognizable extra band, and only one had a few black scales. RIBBANDS (1944a) also reported the finding of constant differences in the larval pectens of fresh-water “ gambiae ” and brackish-water “ ntelas,” correlated with differences in larval reaction to salinity which was made the basis of a physiological test to differentiate between larvae of the two groups. In Liberia, both criteria were used in examinations of hundreds of field collections of larvae, and of larvae reared from the eggs of wild-caught females. Both characters were found to be absolutely reliable, and to correlate with each other and with the morphology of the eggs from which the larvae hatched, if known. In addition, in our laboratory, A. ganzbiae was maintained in a continuous colony for 9 months, and A. nzelusfor 4 months, by use of fresh water only for rearing both ; and there was no change in the morphological appearance of the pectens. Eggs from the A. melas colony could be hatched and the larvae successfully reared to adulthood in sea-water ; eggs from the A. gambiae colony hatched poorly, and larval and pupal mortality was high, in dilute sea-water. &IUIRHEAD-THOMSON (1945) reported the finding of constant differences in the eggs of the two species. CHWATT (1945) reported that 8.6 per cent. of egg batches at Lagos, Nigeria, In the series from Marshall Territory, of were intermediate and could not be identified. 3,400 egg batches, only eight or 0.2 per cent. presented difficulty, and most of these contained markedly abnormal eggs. In the great majority of cases, there was no question whatsoever, H. 512 &I. GELFAND and a single egg was sufficient for positive identification. This difference in the eggs provides the only reliable character now known for separating wild adult females of the two species, and has been used in the work reported below. Eggs of the two species from Liberia have been illustrated and described by me (GELFAXD, 1954) as follows : A gambiae - platform narrow, owing to the closer approximation the float is separated from the frill by a distinct space. A. melas - platform wide, and the float touches, or is very narrowly distinct space. of the frills separated on either from, the frill side ; by a The morphology of the eggs was found to be a stable character in laboratory colonies maintained under identical conditions for the above-mentioned time. Infectivity of A. melas and A. gambiae for malaria and bamroftian Jilariasis : Prior to World War II, dissection series of “ Anopheles gambiae ” in coastal areas of WTest Africa undoubtedly included numbers of A. melas as well. The comparative role of the two species therefore required evaluation. This was done for malaria infections in Sierra Leone (TREDRE, 1946) and Nigeria (MUIRHEAD-THOMSON, 1945). In the latter report, based on 3,494 dissections, the over-all sporozoite rate for A. melas was 3.5 per cent. and for A. gambiae 10.0 per cent. Apparently no work had previously been done on the comparative capacity of the two species to serve as vectors of filariasis. In Liberia, there have been no reports comparing A. melas and A. ganzbiae as vectors of malaria. The vectors of filariasis in Marshall Territory have recently been the subject of a separate report (GELFAKD, 1955), and are fully discussed there. However, the results of dissections of wild-caught females of these two anopheline species are included here in Table III for comparative purposes. Since it is not possible to make a specific identification based on filarial larvae from the mosquito host, their designation as Wirchmeria banmofti is only an assumption. However, since A. n&s and A. gambiae were both readily infected in the laboratory by W. banmofti from human volunteers, and since these mosquitoes are markedly anthropophilic (see below), worms agreeing with the morphological description of this parasite were so designated. Table III summarizes the dissections of both species, as well as of unidentified members of the complex, performed throughout the year in Marshall Territory. These mosquitoes were collected in the same localities throughout the study but, because of the seasonal variations in incidence (see below), not necessarily at the same time, Because of the average delay of almost 3 days necessitated by identification by eggs, the percentage positive for malaria may be slightly higher than would be found in mosquitoes dissected immediately. However, in the case of filariasis, the age of the worms was estimated and, if it appeared that a mosquito was not infected prior to the date of capture, it was not This was necessary because some of the mosquito trappers were counted as “ positive.” themselves infected with IV. bancrofti. All identifications and dissections were performed by myself. For malaria, these results are in general agreement with those from Lagos. The sporozoite rate for A. gambiae (5.7 per cent.) is four times as high as that for A. melas (1.4 per cent.). Nevertheless, A. melas is a vector of considerable importance, and its significance is increased by the prevalence of this species in the dry season when A. gambiae may be almost absent. For filariasis, the infectivity rate in A. melas appears to be significantly higher than that in A. gambiae considering both total infections (27.1 per cent. vs. 19.5 per cent.) and mature proboscis infections (3.6 per cent. vs. 1.9 per cent.). Anopheles gambiae AND A. 513 TABLE III. Dissections of A. melas and A. gambiue, Marshall Territory, malaria or filariasis, or both. Filariasis NO. NO. dissected positive 3G6 I 1 369 83 27.1 262 Salivary gland infections only. Dissected on the day of capture of the two species. I 1 A. melas/gambiaeT _L NO. I positive 21 1 5.7 1142 27 2.4 51 19.5 592 132 22.3 4 1.5 11 1.9 3.6 and therefore for either I I NO. positive 1.4 No. with mature larvae * t /O 1952-53, No. Y 0. dissected’ I 01 Liberia, A. gambiue 6 427 total T A. melas ~~ Malaria* melas IN LIBERIA not identified $ ‘6 specifically. dissected, j This group % is a mixture In order to understand these discrepant results, which show A. gambiae to be a more effective vector of malaria and a less effective vector of filariasis than A. meh, the infectability of both species for the two diseasesmust be studied under experimental conditions. ROBERTSON (1945), in a very small series, found that A. melas seemed to be as susceptible to experimental infections with malaria as A. gambiae. This work should be repeated on a larger scale. In a limited experiment (GELFAND, 1955), I found no significant difference in the laboratory susceptibility of both species to infections with IV. banmuftiwhen fed simultaneously on the same human volunteer. If both these findings are verified, then the difference in natural infectivity rates must be explained on a bionomical rather than on a physiological basis. Several possibilities suggest themselves, but differences based on the degree of association with man appear to be ruled out because the rates were based on dissections of the same individuals in most cases. Hypothetical factors in the environment which differ seasonally, or which are different because of responses in behaviour of the two species during the resting periods between blood meals, may be considered. An attractive possibility, worthy of investigation, is a difference in the feeding preference of the two mosquito species for persons of different ages. MUIRHEAD-THOMSON (195 lb) found a marked preference for feeding on older people in A. aZbimanus in Jamaica. If the same were true for A. meh, and if by contrast, A. gamhiae exhibited a preference for infants, it would provide a simple and biologically very interesting explanation. However, THOWYS (195 1) in Sierra Leone has indicated that A. gambiae has a host-age feeding preference similar to that of A. albimanus, and I\%IRHEAD-THOMSON (1954) in a later report assembled data on filariasis and its vectors which suggest epidemiologically that anophelines in general may have a host-age preference of this sort in contrast with a less discriminating tendency on the part of culicines. Obviously a great deal more work is required on this entire subject. Seasonal incidence of A. melas and -4. gambiae : Soon after the recognition of A. meh species distinct from A. gambiae, it was realized that there were marked seasonal fluctuations in the populations of both species. For the 18 months, February 1952 to July, 1953, routine adult mosquito collections were made in three Du River villages on the north bank of the river, and for a shorter time in Marshall City. From these data and the results of as a H. M. GELFAND 514 concurrent egg identifications, it has been possible to describe a rhythm of seasonal population rise and fall in one complete annual cycle and in part of a second. Of the various collection techniques that were used, the one that seemed most reliable was the use of supervised two-man teams of bait-collectors. These consisted of two Africans, equipped with kerosene lantern, flashlight, test tubes and small cages. The men were stationed outdoors under the sheitering roof of an inhabited native hut, and worked for the 1Zhour period 7 p.m. to 7 a.m. They had a regular schedule of stations, and always used the same hut in each locality. Every few minutes throughout the night they searched each other for the presence of feeding mosquitoes, which were captured by test tube and transferred alive to a small cage. The results mere then expressed as “ females per night per two-man bait The mosquitoes were carried to the laboratory in the morning, and placed in tubes team.” for oviposition. This technique for estimating two mosquito populations is valid only if c I 0 Marshall 25 20 I 0 2 City = A. melos = A. gombioe I5 \ z 0 i= u IO 5 0 Du River Villages FIGURE 2. Seasonal incidence of A. melns in &la&all Territory, and A. gambiae Liberia and monthly rainfall at Roberts field, Liberia, February, 1952 to July, 1953. 0 50 v) W I v z - Rainfall c - Roberts Field 40 30 20 IO F fs! A M. J. J. 1952 A S. 0 N D J. F Fri. A. M. J. J. 1953 both attack. man to the same extent under these conditions. That both A. nzelas and A. gambiae do readily feed on man outdoors will be shown in a later section. One bait-collector team operated on the north bank of the river. It had a varying schedule which always included, however, one night each week at each of the three Du River Table IV gives the combined average villages, Taryeweh, Dyochewehn, and Cow Farm. collections per night of A. ~neZu.s and A. gamtiae in these villages, expressed on a monthly basis. A second team operated on the south bank of the river, spending one night each week in Krutown, Fantitown, Marshall City proper, Freetown, and two tiny villages between Table V gives the combined average collections per night for these the city and Freetown. six localities on a monthly basis. This material, plus monthly rainfall, is graphically presented in Figure 2. It is evident that there is a marked seasonal alternation in species dominance in Marshall Anopheles gambfae 515 A. AND melas IN LIBARIE Territory, and that this is related to rainfall. The direct and indirect effects of variation in rainfall on larval breeding of the t\vo species are discussed in a later section. The relative importance of these factors may vary from locality to locality. In Marshall Territory, A. meLasappears to be almost completely absent during the rainy season, to reappear during the dry season, to be most abundant during the period of intermittent showers at the beginning of the rainy season, and to decline in numbers with the onset of heavy continuous rainfall. TABLE IV. Average / / Month 1952 Feb. nightly collections Average nightly collection of A. melas and A. gambiae by bait-catcher Identified I 1 A. melas 27.4 54 by eggs. per night I j ya A. melas i A. n&as c A. gambiae ’ Average teams, Du River 1 98.2 A. gambiae //I 26.9 0.5 1 39.0 0.0 / 16.4 March 1 39.0 70 0 100.0 April / 18.1 49 5 90.7 12.1 19 12 61.3 34.9 2s 42 17.2 13 61 17.6 3.0 14.2 62 6.1 0.7 11.5 0.0 May June i July Aug. / 12-2 Sept. j 8.0 0 102 7.0 0 92 0 72 0 27 Oct. Kov. I I I ] Dec. 8.0 1 1 I 2.3 4 ) 1 I I I 1 40.0 / 7.4 I ; 14.0 1.7 4.7 ) 20.9 0.0 1 8.0 0.0 I 7.0 0.0 1 8.0 0.0 0.0 j 2.3 1.5 0.0 0.0 I 1953 Jan. 1 ! 2.3 / 20 10 66.6 Feb. I 1 30. I 34 2 94.4 1 4.6 i 120 4 I 96.8 4.5 0.1 3 I i ] 97.4 10.5 0.3 89.5 18.8 2.2 March April May I [ j 10.8 21.0 June 16.3 July 10.8 I 112 I 153 1 44 1 4 I8 104 1 128 I I / I / 2.8 0.8 , 0.2 29.7 j 4.8 11.3 0.3 j 3.2 10.5 This is quite different from data presented by RIBB.GDS (1944b) from Aberdeen, Sierra Leone, which show the A. melas population at its maximum during the rainy season and from those of MUIRHEAD-THOMSOX (1948) f rom Lagos, Nigeria which showed considerable variations bet\veen different collecting stations, the peak months being different and not necessarily related to rainfall. Furthermore, in Liberia A. melas was abundantly found in H. AI. GELFAND 516 iMonrovia in October and November, 1951, by me, and in November, 1952, by Mr. C. E. KOHLER (personal communication) at a time when it was essentially absent from the Du River. It is apparent, therefore, that the varying combinations of environmental conditions which will occur even in the limited area of the Liberian coast necessitate a study of the local conditions in each place where work is contemplated. rA.EtLE \'. i\lonth Average I -- 1952 Sept. Oct. nightly Average nightly collection coIlections of A. melas and A. gambiae by bait-catcher Identified A. melas -I, I I NOV. I Dec. 1953 Jan. Feb. March April May June July 0 0 0 2 1 , 1 I I ! I I I ! j A. gambiae i 3; A. melas -_ 1.5 0.5 1;‘: 2615 29.5 22.2 14 8 10 58 69 33 5 / / I per night. -j A. gambiae _/_ 0.0 0.0 0.0 32 20 11 2 0.0 0.0 0.0 50.0 0.2 5.2 4.0 1.7 0.2 2 1 12 34 38 276 324 87.5 88.9 45.5 63.0 64.5 10.7 1.5 1.3 0.4 1.2 11 .o 17.1 3.2 0.3 0.2 0.1 1.4 6.4 9.4 26.3 21.9 i - A. melas I I 5.2 4.0 1.7 0.3 Average _. _/_ I I by eggs. teams, hIarshal1 City. j , I - - It is also worth while to note the variations that may occur in the same locality from year to year. In 1952, March was the month of peak abundance of A. melas but in 1953 the peak occurred 2 months later and was much lower. This indicates that even a prolonged study of this sort cannot serve as a guide to forecasting, and that great caution must be observed in interpreting reductions in mosquito incidence following the institution of control measures. In contrast to A. melas, A. gambiae is most abundant during the rainy season and declines to low population levels during the dry season. This is the usual observation in many parts of Africa (DE MEILLON, 194’7). Distribution of A. melas : In order to determine the limits of the area infested by A. melas and its dissemination away from its brackish-water breeding grounds, collections and ovipositions were obtained from villages at increasing distances from the Du River during months xvhen A. melas T;siaspresent in large numbers. Table VI presents these data and indicates that A, llzelns becomes progressively less abundant in comparison with A. ganzbiae TABLE VI. Collections Village Du River villages Gai-bli Samuel Zopablo Liberian Institute of A. txeZas and A. gambiae at increasing distances from the Du River, July, 1952 and January-May, 1953. Distance from Du River (miles) 0.0 0.8 1.0 2.0 2.4+ A. gambtie A. melus 493 16 40 2 0 i I 146 24 224 79 40 April- o/o A. melas I I ’ 77.1 40.0 15.2 0”:; 517 AnopheZes gambiae AXD A. melas IK LIBERIA one moves inland, and will probably not be found beyond two miles from brackish water. This is in agreement with observations in other countries. as To determine the range of A. rrze2a.s up a brackish river, collections and identifications by eggs were made at various localities on the Du River during April and May, 1953, when this species was abundant (Table VII). All these villages are within the limits of mangrove swamp and brackish water, analyses for XaCl at Wehn showing at least 25 per cent. sea-water at high tide during these months. This Table indicates that the importance of A. melas rapidly declines away from the mouth of the Du River. The extreme range had not been reached, however. Feeding habits of A. melas and A. gambiae : a. Tims of biting To obtain information about relative feeding activity at various hours during the night, the teams of bait-collectors described above were used. Mosquitoes were collected without interruption throughout the night, but were placed in separate small cages marked in 2-hourly intervals : 7 - 9 p.m., 9 - 11 p.m., 11 p.m. - 1 a.m., 1 - 3 a.m., 3 - 5 a.m., 5 - 7 a.m. In this latitude, sunset and sunrise are nearly uniform throughout the year. The men collected mosquitoes which , actually landed on their bodies and attempted to feed. These two-man groups were stationed outside inhabited native dwellings, on the sheltered “ piazza.” In this site, the factor of time of entry into houses was eliminated. Because the number of specimens suitable for positive identification by egg examination was too small for an hourly analysis, for each species those months of the year were chosen when it was preponderant in Du River villages, the number of specimens of the minority species being so small that the error thus introduced was insignificant. TABLE VII. Collections Village Wehn Lloydesville Poortown Freetown Cow Farm Fantitown of A. meZa.sand A. gambiae at increasing May, 1953. Distance COW Farm from (miIes)i A. melas 6.5 5.5 4 1 1.5 3.0 -1.0 3; 261 43 - distances up the Du A. gambiae 4-4 45 67 24 17 10 River, o/o A. April- melds 9.1 2.2 9.5 57.9 93.9 81.1 Table VIII shows the total number of mosquitoes of each species collected during each time period. When only a few anophelines were collected per night, they were randomly distributed through the night. As the total number increased, there appeared to be a tendency in each species for an increasing percentage to be collected later during the night, but this tendency was never marked, and the activity of these anophelines throughout the night is (1943) found the notable. Working with these two species in Nigeria, MUIRHEAD-THOMSON period of greatest feeding activity inside huts to be between 4 a.m. and dawn. HADDOW et al. (1947) working with A. gambiae only, in uninhabited forest, also found a peak in the few hours before dawn. b. Outdoor L ‘S ’ _indoor feeding - The preference of A. melas and A. gambiae for feeding outdoors and indoors was investigated by stationing the same bait-collecting team, on alternate nights, outside on the “ piazza ” of an inhabited house, and inside one of the sleeping H. hf. GELFAND 518 rooms (vacated by other persons for this occasion) of the same house. Table IX shows the results from Cow Farm village and for comparison, collections of A. hancocki in Samuel Village are also included. These data indicate no marked preference : the three species fed readily outdoors if a suitable host was available, and all moved indoors without hesitation in search of a blood meal. Observations in other villages, and during periods when both species were present, lead to the same conclusion. TABLE VIII. Percentage of A. nteZasand A. gambiae collected by human bait-collectors at 2-hour intervals throughout the night, Du River villages, 1952-53. L I I Average i ) x0. of Species 1collections I Time per. j collection 7-9 j 9-11 j M / 1-3 j 3-5 ii-7 -‘ A. melas A.gambiue 1 24 16 4 Total 9.2 19.6 64.5 38 7.8 36.6 17.2 21.6 14.7 18.2 20.6 9 .3 16.3 To& TABLE IX. 1 17.2 17.8 20.1 18.4 i 15.4 19.1 34.8 23.2 I, I 19.2 12.0 16.5 15.4 16.9 12.0 14.9 10.1 18.6 13.3 Outdoor and indoor night collections by bait-collector team, COW Farm village, 1952-53. i I Outdoors Species A. melas A. gambiae Indoors No. of collections Average per collection No. per collections Average per collection 7 35.3 7 39.3 13 8.4 I2 8.8 10.1 I4 10.1 I A. hancocki 14 1 A brief investigation was made of the vertical distribution of these anophelines in dwellings of more than one floor. HADDOW et al. (1947) sh owed that A. gambiae was found at all levels up to 82 feet in the uninhabited Semliki forest in Uganda. In Marshall City in May, 1953, six collections were made in the same two-floored house. First-floor and second-floor collections were made on alternate nights by the same team of two men. A total of 41 A. melas and A. gambiae was collected on the first floor and 28 on the second. Of five identifications of egg batches from the second floor catches, two were A. ganzbiae and three were A. m&s. Although too limited for statistical significance, the results do suggest that these species will readily ascend to upper levels in a house. c. Host prefuence - A limited experiment was designed to compare the relative attractiveness to A. melas of man and the other animal species commonly found in villages in this area when all were available simultaneously in separate traps. For this purpose, small stable traps were constructed of wood. They were 5 feet long x 24 ft. wide, without flooring, with a removable, slanting, tarpaper-covered roof, and with a single door at one end protected Anopheles gambiae AND A. melas IX LIBERIA 519 with a heavy cloth curtain on the inside. Mosquito entry lx-as possible through a plasticscreening louvre running the entire length of both sides, which followed the “ Egyptiantype ” design of BATES (1949). The experiment was run for 18 nights in March, 1953, in Cow Farm village. Six different species of animal bait were used -man, monkey, goat, hog, dog, and chicken. No cattle nor horses are normally present in the area. The traps were placed in a radiating pattern, like the spokes of a wheel, at one edge of the village. The anthropophilic tendency Because of the difficulty in procuring and keeping of the anophelines was striking (Table X). animals here, the volume of bait in each species was not comparable. However, man attracted 207 anophelines compared to only 12 for all other bait species. These results apply only to A. TFzelas since of 65 identifications by eggs, 64 were A. metas and only one was A. gambiae. was not possible to study entrance of anophelines Activity iu houses : a. &try -It into habitations by direct trapping, but a few points of interest were established by other methods. Both A. melas and A. gambiae freely enter African huts - rather flimsy affairs in Marshall Territory with woven “ mat ” or thin mud walls and a thatch roof which is not Openings in the eaves are often a foot or more in width, and, despite tightly tied to the v-a&. doors tightly closed and windows shuttered at night, almost unrestricted entrance is provided for mosquitoes. Entry can be readily inhibited, however, by reducing the apertures available to hungry mosquitoes. For other studies (see below), huts were constructed into which mosquito entry was permitted only through numerous small holes, about one half inch in diameter, punched through the mud at the eaves. Although numerous elsewhere in the village, never more than one or two A. ntelas were found in these baited huts in the morning. The openings were then enlarged to chinks about two inches in diameter, and mosquitoes became numerous inside the huts. To determine to what extent A. melas vi-as still being excluded, on four consecutive nights in March, 1952, the same t\\-o men bait-collected inside the experimental hut at Cow Farm village. On two nights the door was kept closed and a window exit trap was left in place all night (the eave openings thus providing the only means of entrance). On two alternate nights, the door and window were left open until just before dawn, and then the door was closed and the trap put on the window. Table XI shows the results of these four collections, and, incidentally, provides some information about the percentage of the mosquitoes attracted to them which are actually captured by the baitcollectors. This table shows that, even with openings of this size, mosquito entry was reduced by two-thirds. TABLE X. Mosquitoes attracted to different baits offered simultaneously March, 1953. ! Species Man Goat Ho& Chicken Monkey Dog No. of individuals in stable traps, A. melas/A. ganzbiae No. of nights TotaI 16 18 10 13 11 5 207 9 2 1 : I Average per 1 I night 12.8 0.5 0.2 0.1 0.0 0.0 Du River, I Culicines total 6 16 0 0 2 0 H. M. 520 GELFAX’D Time of entry was indirectly studied by all-night collections by the bait teams stationed inside native sleeping rooms. The results showed no significant difference from the hourly distribution of outdoor collections. No detailed records were kept of moonlight, but no obvious effect was noted (see RIBBAXDS, 1946). b. EL&t - Because large numbers of freshly-fed A, gambiae can often be found resting in the morning in dark native huts, it has been generally assumed that this is a predominantly house-haunting species, using outdoor resting places to only a slight extent. More recent work, however, has demonstrated that this is not true, and that A. gambiae females, and A. melas females to an even greater extent, leave the house at some time after feeding to spend a part of the gonotrophic cycle outdoors (see review by MUIRHEAD-THOMSON, 1951~). TABLE XI. Collections made in an experimental hut with restricted mosquitoes, Cow Farm, Du River. I Door Open and window (1) (2) Average Closed (1) (2) Average / , ( 1 I Number collected Number trapped 58 and unrestricted entry for Total 89 67 78 60 59 147 127 137 35 7 21 13 25 19 48 32 40 To study mosquito exit under controlled conditions, five similar experimental huts were constructed, one each in TaryeTveh, Cow Farm, Fantitown, Marshall City, and Freetown. Modified from the general design of MUIRHEAD-THOMSON (1948), these were built to simulate African huts as closely as possible, but with mosquito exit limited to one window, to the outside of which a trap was tightly fitted. These huts were just sufficient in size to accommodate txvo beds, with a low, tightly-thatched roof tapering to a central point. There was a door in the western end, made light-tight by a double layer of heavy cloth on the inside, and a one foot x one foot lvindow in the eastern end. The walls were made of woven reed “ mat ” ; two were lined with mud, and two with old newspaper and cassava starch in the local manner. In the Taryeweh and Cow Farm mosquito houses, the space between wall and roof was filled in with mud, into which chinks about two inches in diameter were made at intervals. In the Marshall huts, this space was left open and instead, a mat louvre was made on the inside, leaving a one inch entry space all around the walls close to the roof. A half-ceiling of mat was built into the eastern end of the hut just above the window, producing a dark, undisturbed recess, seemingly ideal as a mosquito resting place. In the dim light of the moon, at dusk and at dawn, the only light visible was that coming in through the window opening. Even during the day, the recess above the window was still dark. The removable exit traps were one foot cubes, screened except for the surface opening into the window. At this face, a wire screen cone was attached, tapering into the trap and slightly upward, to a point about two inches from the far surface, with an opening about one cm. in diameter. Mosquitoes attracted to light readily entered this trap, and were unable to leave until removed through a cloth sleeve in one wall. A series of observations, by using empty mosquito houses, showed that only rarely did mosquitoes enter these huts for shelter alone. By employing two volunteer “ sleepers,” 521 Anopheles gambiae AND A. melas IN LIBERIA it was possible to determine daily entry, exit at any time-interval desired, and resting preferences. To estimate the time elapsed following the last blood meal in female anophelines, the scheme was used. Laboratory observations had sho1v-n that the gonotrophic following cycle lasts 2 days in the great majority of A. meZasand A. gambiae, never less, It was therefore possible, by examining the female abdomen, to estimate the stage of the mosquito in the gonotrophic cycle. JT7henexamined in the morning, those which were both unengorged and non-gravid were “ stage 1.” Freshly engorged females had fed during the night before, Gravid or part-gravid females had fed no more recently than and were called “ stage 2.” 2 nights before, and were “ stage 3.” To determine how long A. melas and A. gambiae remain inside an undisturbed hut, the Cow Farm hut was occupied by “ sleepers ” every night for periods of time, with a window exit trap attached continuously except when changed once daily. The contents of the trap were examined each morning, and the anophelines were “ staged ” according to the scheme outlined above. It was assumed that all freshly-fed females (stage 2) had departed during the night of feeding or at dawn, and that all gravid and part-gravid (stage 3) had been in the hut for at least one full day. Stage 1 was ignored in the calculations. At times, mosquitoes were collected by hand from the walls in the morning, after the trap was changed. When this was done, those females were included in the stage 3 class for the day following. Almost 90 per cent. of the female A. melas which fed indoors left an undisturbed hut on the night of feeding to seek outdoor shelter (Table XII). There were very small numbers of A. gambiae, but these indicated that only 40 per cent. left the hut on the night of feeding. To determine the time of departure of the freshly-fed and gravid A. melas, during 12 nights in April, 1953, the window trap was changed at intervals during the night. A trap was left in place from 7 a.m. to 7 p.m. Since there appeared to be no tendency for mosquitoes to leave during the daytime after 7 a.m. this trap collected those females leaving at dusk. The trap was then removed and replaced at 9 p.m., 11 p.m., 1 a.m., 3 a.m., 5 a.m. and 7 a.m. Table XIII summarizes the results by Z-hour periods. Here it can be seen that the majority of all A. melas females left at dawn. Of those which had fed on the night in question, 72.7 per cent. left at dawn ; of those that remained in the hut during the day, 91.8 per cent. left at dusk that evening. Of the engorged females that did not leave the hut during the night of feeding or at dawn, i.e. 10 per cent. of A. nteh and 60 per cent. of A. gambiae, the percentage that left at dusk on the following night can be estimated in another way, by analyzing the morning collections in native huts. The mosquitoes were “ staged ” in the same manner. Table XIV shows TABLEXII. Window exit trap collections, Cow Farm experimental hut, 1952-53. I Stage (1) (2) (3) Total Unengorged, non-gravid Freshly engorged Gravid and part-gravid (2 + 3) 76 Remaining I A. melas A. ganhiae 76 1 16 27 745 43 10.2 1 H. TALE XIII. Time of departure M. GELFAND of A. mela-s from experimental Stage 1 Hours hut, Cow Farm, Total 2 3 April, 1953. Per cent. (2 + 3) 2 (2 + 3) I_ 7 7 9 11 1 3 5 a.m. p.m. p.m. p.m. a.m. a.m. a.m. - 7 p.m. 9 p.m. 11 p.m. 1 a.m. 3 a.m. 5 a.m. 7 a.m. 0 0 0 2 2 0 4 1: 7 : Total 22 1 0 0 0 1 4 lo9 20 /i 0 1 151 i I 24 22 I 10 7 1;: I 12.6 0.6 2.9 5.7 4.0 62.8 11.4 72.7 175 100.0 the results of collections in Du River villages. Assuming that, on the average, equal numbers of mosquitoes enter every night, the ratio gravid : fed would be 1 : 1 if all remained for the entire Z-day gonotrophic cycle. However, for A. m&s, the ratio found was 1 : 18. Therefore, 17 out of every 18 which remained in the hut in the morning subsequently left at dusk later that day. , Thus only 0.57 per cent. of A. melas females spent the entire gonotrophic cycle indoors in this study. Similarly, 8.6 per cent. of A. gambiae remained in the hut for the entire cycle. Therefore, both species appear to be exophilic in resting habits in this area. TABLE XIV. Morning collections in huts, Du River Stage (1) (2) (3) A. melas Unengorged, non-gravid Freshly engorged &avid or part-gravid Ratio gravid : fed villages, A. gambiue 4 304 17 1 : 17.9 1952-53. 28 204 28 1 : 7.3 c. Resting places in houses- Two walls of the experimental huts were surfaced with mud, and two were covered with old newspaper applied with cassava starch. The thatch roof was left bare. A half ceiling of “ soft mat ” covered half the length of the beds. The beds were of rough wood, the eaves (in Marshall) were of “ hard mat,” and the door was protected by a cloth screen. Therefore a great variety of surfaces was availables for mosquitoes that remained in the huts in the morning. Of 181 females of mixed A. melus and A. gambiae collected in all five huts, the surface distribution was as follows : paper-covered walls - 70, soft, mat ceiling - 59, mud walls- 30, mat eaves - 11, thatch - 6, under beds - 5, cloth door cover - 1. Statistically these results prove little, for the surfaces were not equal in area and the ease of collecting (or missing) on the different surfaces was not taken into account. Nevertheless, the contrast between thatch roof and “ mat ” ceiling, and between mud and paper-covered wall, was marked. In African huts, the preference of these anophelines for resting on “ soft mat ” and paper-covered walls when available was often noticed. Outdoor resting places of A. melas : It is apparent that both A. melas and A. ganzbiae use outdoor resting places to a considerable extent. Some effort to locate those resting places in Marshall Territory was made, but was almost completely unsuccessful. Only 523 Anophles gambiae A-ND A. melas IN LIBERIA a few unfed females were found under rotting logs near Cow Farm. However, in this area, there are few specific sites where a concentration of resting adults might be expected. In the enormous area of surrounding mangrove swamp, many thousands of mosquitoes could be dispersed with little likelihood of discovery. That such dispersion does occur is suggested by night bait collections on the uninhabited island just offshore from Cow Farm. This island consists almost entirely of red mangrove swamp, and no larval breeding has been found there. Nevertheless, large numbers of A. ~zelnswere attracted to human bait stationed on the island, probably to remain there after feeding. In Monrovia, outdoor resting A. melas including both freshly engorged and gravid females, were readily found in large numbers at the buttressed bases of kapok trees and in old termite hills within or near native villages. A few were even collected from the walls of crab holes near the mangrove swamp. I@Lence of rainfall and tides on the adult population of A. melas : RIBBANDS (1944b) made a detailed statistical analysis of this relationship at Freetown, Sierra Leone. He . concluded that the incidence of A. n&s is controlled seasonally by rainfall, and this has been shown to be true also in Marshall Territory. On a short-term basis, however, tidal fluctuations were the controlling factor. For Marshall Territory, this is graphically demonstrated in Figure 3, and the time relationships between peaks of mosquito abundance (A. nzekasonly) and the occurrence of the spring tides are shown in Table XV. The mosquito population was measured at Cow Farm by the use of an experimental hut baited with two human sleepers. To minimize the irregularities resulting from adventitious daily fluctuations, each day’s collection was expressed as a 3-day average of the catch the day preceding, the day in question, and the day following. The tide was recorded by daily observation of an arbitrary scale at 5 8 9 6 % II 7 4 2 0 5 IO 15 20 25 JANW FIGURE 3. Rainfall I * 3', i 4 I 3 'L&_lf-LI 14 i9 24 FEEFiUARf tides, r. !_ /_ I _'L I ni, I 6 I! I6 21 26 MARCH and the adult 1 Al 31 5 P'r) I j n f! IL IO I5 20 25 30: 5 fiPR!L female population Liberia, 1953. ~,n,& 0 15 20 25 Y) MAY 4 9 14 19 24 JUNE of Anopheles melas, Cow 29 4 Farm 9 bt 24 JULY Village, The daily high tides were plotted to give the alteraa dock on the Du River near Taryeweh. From this figure and table, it is very apparent that in the tions of spring and neap tides. absence of rain, the A. ?neZa.s population was totally dependent upon the bi-monthly spring tides, reaching a peak about 11 days after each high tide. With the advent of occasional showers, the effect of the tides was reinforced by the prolonged submersion of the larval H. M. 524 GELFAIW breeding areas, and the adult population peaks were higher and longer. After the last distinct spring tide on May 10, a period of continuous heavy rainfall set in, and no further tidal fluctuations were discernable, the level of the river being dependent wholly upon rainfall. The larval breeding areas T-t’ere then continuously submerged, breeding almost ceased, and the population of A. m&s rapidly declined, never to rise again during the period of this study. TABLE XV. Relation between matimum tides, rainfall, and peaks in the population of A. melas, Cow Farm, 1953. 1 Date Date of matimum tide 1 l/14 l/27 2j13 2125 3/14 3131 4/10 4528 5110 I I average of of population 3-Day I 1 population at peak. peak following. 1128 214 2120 316 3/26 4/15 (?) 4/24 517 5126 5.7 10.0 4.0 16.7 16.7 24.7 46.3 27.7 55.5 Interval between maxi. mum tide and population peak (days) 14 8 7 9 12 15 14 1: Lamal breeding of A. melas and A. gambiae : The importance of the topography of the land, rainfall, and tidal fluctuations on the larval breeding of A. melas have been suggested above, but can be summarized for this area as follows : No breeding occurred in the swamps of red mangrove where tidal influence was felt twice daily throughout the month. Repeated collecting in water-holding depressions around mangrove roots and in crab holes was always unsuccessful. The second tidal zone, that of salt-marsh grass, was above water level during neap tide, but was submerged by the spring tides which occurred twice a month at the new and full moons. A. melas larvae promptly appeared after flooding, and development was rapid. During the dry season, brackish water remained in pools and low areas for about a week after the tide receded, and only one mosquito generation was possibIe. Larvae in collections at this time were all of approximately the same age. Early in the rainy season, the occasional showers caused the flooded areas to remain submerged for a longer period of time. Several generations of larvae were possible, and the total adult population was much greater. As the rainy season advanced, rain occurred almost every day and the volume of fresh water coming down the river became of greater significance than the pressure of the ocean tides. Fluctuations at a greater level still occurred, but depended upon the amount of rain rather than upon the phases of the moon. The river overflowed the grassland areas completely, and larval breeding was prevented. The adult population then rapidly declined, This situation is unlike that encountered in Freetown ( RIBBANDS, 1944b) and Lagos ( MUIRHEADIn both those places, complete flooding of the breeding area did not take THOMSON, 1943). place, and usually breeding was maximal in the rainy season. Elsewhere in Liberia, limited collections in the Monrovia area indicated two distinct A. meZas cycles. On Bushrod Island, where breeding occurred in extensive Paspalum grass marshes reminiscent of Lagos lagoon, it was maximal in the rainy season. In the Sinkor district, where breeding occurred in large, permanent, brackish pools which have lost their connection with the sea, it was maximal in the dry season. Anophdes 525 gambiae AND A. meIas IN LIBERIA Extensive collecting in Marshall Territory failed to locate the breeding of A. nzel~.~ in other than brackish water. In Monrovia, however, larvae of this species were found occasionally in salt-free rainwater puddles on the shores of the ocean and the Mesurado River. The larval breeding of A. gambiae was almost restricted to the rainy season in Marshall Territory. Larvae of this species were found only in the “ typical ” breeding places, i.e. small, open, sunlit, freshwater pools. These included : rainwater puddles, hillside seepages, semi-permanent pools in a sand quarry, hog wallows, and discarded calabashes. A few larvae were found in water tanks inside a brick pumphouse. These types of water collections are almost absent during the dry season, and A. ganzbiae adults were then rare. No larvae of this species were ever found in brackish water in bIarshal1 Territory. FIGURE 4. Comparison - - - - - .. ... . . l of adult mosquito collecting techniques, Du River Villages. Night bait - collector team (Females per night per two-man team). Human-baited hut traps (Females per night per two-man team). Morning hut collections (Females per 10 sleeping rooms). The selective breeding of A. melas in brackish water is interesting, since larvae develop equally well in fresh and saline water in the laboratory. It is apparently not due to conscious selection by the gravid female of saline waters. This was experimentally tested in our laboratory colony. Two similar vessels for oviposition were placed in the cage on several nights, one containing distilled water and the other 37 per cent. sea water. Of 4,031 eggs, 48.7 per cent. were laid in fresh water. In a similar test in the A. gambiae cage, 75.0 per cent. of 5,538 eggs were laid in fresh water. The selection of breeding places must be made by the ovipositing female, but on some basis other than that of salt content. Contpartion of adult collecting techniques: During the course of this study, three different techniques were used for the routine collection of adult anophelines, i.e. two-man teams of bait-collectors, human-baited hut traps, and hand collecting in native huts by means of glass tubing aspirators. When these three techniques were used in the same locality at the same time, an excellent opportunity was provided for comparing them as population sampling devices. Figure 4 shoxs the collections of the A. melas - A. gambiae complex obtained in Du River villages. The mosquitoes caught by bait-catchers and baited huts were expressed as the average number of females (A. meh - A. gambiae complex only) per night per two-man H. M. GELFAND 526 team, and those collected in huts in the mornings as the number of females per 10 sleeping rooms. The striking similarity in the general character of the three curves is most interesting. Different individuals were involved in each collecting method, who were unaware of each other’s results. Since the three methods seem to support each other, any one n-ould appear to be a valid sampling device for this group of mosquito species. Each will have its own special field of usefulness. %JMMARY (1) The bionomics of Anopheles llzelas and Anopheks gambiae were studied in Marshall Territory, a rural, coastal district of Liberia, West Africa, from January, 1952, through July, 1953, (2) The A. mehs - A. gam-biae complex constitutes the most abundant group of mosquitoes in this area, as well as being the main vectors of malaria and bancroftian filariasis. (3) A. melas and A. gambiae are distinct though closely related species, separable by well marked and constant characters in the egg and larval stages. Adults cannot be identified directly. (4) The over-all infection rate for malaria (salivary gland dissections) was 5.7 per cent, in A. gambiae and 1.4 per cent. in A. hzelas ; for filariasis it was 19.5 per cent. in A. gambiae and 27.1 per cent. in A. melas. (5) There was a marked seasonal periodicity in both species in Marshall Territory, A. melas being most abundant in the late dry season and early rains (January - June), and A. gambiae being most abundant during the rains (June - November). The seasonal incidence is a local characteristic and will probably vary from place to place in Liberia. (6) A. melas was restricted to coastal districts. It probably does not occur inland beyond two miles from brackish water, and, in Marshall Territory, was not common beyond three-four miles from the mouth of the Du River, although it occurred in small numbers at least 10 miles upriver. (7) Both species fed readily outdoors but entered human habitations without hesitation. Feeding occurred at all hours of the night, from dusk to dawn. A. melas was shown to be markedly anthropophilic. (8) Window exit trap studies showed that 90 per cent. of the A. melas females feeding in a hut during the night left for outdoor resting places by dawn, the majority at dawn. Of those remaining, 90-95 per cent. left at dusk later the same day. For A. gambiae, the figures were 40 per cent. and 85 per cent. Few females of either species spent the entire gonotrophic cycle indoors. (9) Outdoor resting places could not be located in Marshall Territory. In Monrovia, freshly-fed and gravid A. melas females were found abundantly on the buttressed trunks of kapok trees and in old termite hills. I I I I I II (10) Routine adult collections of A. melas showed a marked correlation of peaks in the adult population with tidal fluctuations, population peaks occurring about 11 days after each “ spring ” tide in the dry season. With the onset of the rains, the effect of the tides was reinforced. During the height of the rains, the river rose, tidal influence vanished, and the population of _&I.szrplasdeclined.