Studies on the life budget of Aedes aegypti in Wat Samphaya, Bangkok, Thailand

Bull. Org. mond. Sante 1972, 46, 211-226 Bull. Wid Hith Org.J Studies on the life budget of Aedes aegypti in Wat Samphaya, Bangkok, Thailand T. R. E. SOUTHWOOD,1 G. MURDIE,2 M. YASUNO,3 R. J. TONN's & P. M. READER 5 For a complete understanding ofthe epidemiology ofa vector-borne disease, a knowledge of the bionomics of the vector is needed. The development ofAedes aegypti was studied in Wat Samphaya, Bangkok, Thailand, where work on the adult biology had been carried out the previous year (1966-67). Particular attention wasgiven to the variation in the numbersof immature stages of the mosquito in relation to the known seasonal incidence of dengue haemorrhagic fever. Of the three types of water container in the Wat, water jars were the main source of adults, flower pot plates were less important, and the contribution of ant traps was insignificant. The variation in the numbers ofemerging adults depended on changes in the mortality of the immature stages rather than on variations in the numbers of eggs laid. Both early and late larval instdr mortalities are important, theformer becoming more significant during the period March-August. The mortality between the eggs and second-instar larvae is density- dependent. There was no clear trend ofassociation between mortality and season exceptfor a fall in larval mortality in April-May preceding the increase in annual incidence of haemorrhagic fever, which usually occurs in June. After reviewing the information available on mos- quito ecology, aWHO Scientific Group on Mosquito Ecology (1967) reported that population data were lacking. It was suggested that life budgets (or tables) for various mosquitos should be constructed and analysed. As the WHO Aedes Research Unit (ARU) in Bangkok, Thailand, was already making measure- ments of the adult population by mark and recapture methods (Sheppard et al., 1969), it was proposed by the ARU working group for the research programme that life budget studies of the immature stages of Aedes aegypti should be started in Wat Samphaya, Bangkok. These studies were designed to provide absolute measures of populations as a basis for determining 1 Professor and Head, Department of Zoology and Applied Entomology, Imperial College, London, England. ' Lecturer, Department of Zoology and Applied Ento- mology, Imperial College. 'WHO Aedes Research Unit, Bangkok, Thailand. Pre- sent address: WHO Research Unit on Genetic Control of Mosquitos, Delhi, India. ' WHO Aedes Research Unit, Bangkok. Present address: East Africa Aedes Research Unit, Dar es Salaam, Tanzania. ' Assistant Experimental Officer, Department of Zoology and Applied Entomology, Imperial College. the effectiveness of various control measures. They were also intended to reveal the extent of natural mortality in the various stages under different condi- tions, as well as variations in the oviposition level. It was hoped that these data might provide an indica- tion of the way in which the environment could be modified to reduce the size of the mosquito popula- tion. Unfortunately, it was not possible to compare simultaneous data on the adult numbers from the mark and recapture study with those on pupal emergence. Ideally, this comparison would indicate whether the latter measure could be used to provide significant epidemiological information. THE STUDY AREA The work reported here was conducted at the Wat Samphaya which is situated near the Chao Phya river in Bangkok and is bounded by a slum area and by two-storey blocks of shophouses. This site was selected mainly in order to relate the life budget studies to the study of adult populations conducted there in 1966-67 (Sheppard et al., 1969). The area of the Wat is approximately 94 by 56 m. The 31 houses each have several rooms and are 2794 - 211 - T. R. E. SOUTHWOOD AND OTHERS occupied by about 100 priests and some schoolchil- dren. The major breeding containers for Ae. aegypti are about 100 water jars, 50 ant traps, and 50 flower pot plates, all of which can be regarded as good habitats for the larvae. There are very few other miscellaneous receptacles suitable as potential habi- tats. In the Wat, as in Bangkok as a whole, Ae. aegypti was the only mosquito breeding in the great majority of water containers. METHODS All water jars, ant traps, flower pot plates, and rooms in the study area were numbered. To estimate the numbers of larvae and pupae in the area, random samples of 10 water jars, 1 ant trap (later 5 ant traps), and 1 flower pot plate (later 5 flower pot plates) were sampled 3 or 4 times each week. All immature stages of mosquito in the water jars were sampled by means of a plankton net, and larvae in ant traps and flower pot plates were extracted with a pipette. After they had been counted, the larvae were replaced in the containers. The first- and second- instar larvae were not distinguished. Pupae were taken to the field laboratory in the study area and placed in plastic emergence cages measuring 10 by 12 by 17 cm. To estimate total production, 10 water jars, 20 ant traps, and 20 flower pot plates were placed beside randomly chosen receptacles of each kind. These experimental receptacles were left uncovered for 48 hours for oviposition by mosquitos. After this exposure, eggs laid in each water container were counted repeatedly by 2-4 persons and then each container was flooded after another 48 hours of incubation. The number of eggs hatching was deter- mined by extracting all newly hatched larvae each day, usually over a period of 10 days. Experimental containers for oviposition were moved to newly chosen sites in each experimental period. The mean number of eggs per container of each type multiplied by the number of containers of that type filled with water gave an indication of the total number of eggs laid in the Wat. To obtain data on the development pattern of each immature stage, four series of containers of each type were used. The first-instar larvae hatching in one day, usually the morning after each experi- mental oviposition container was flooded, were trans- ferred to a second series of containers. A total of 30 cement water jars was placed under the floor of a house to provide conditions similar to those offered by non-experimental water jars. Similarly, 60 ceramic ant traps and 60 earthenware flower pot plates were placed on the floor of a room in the Wat. Aged water taken from non-experimental containers of each type was used for rearing larvae in the experi- mental containers. After the initial oviposition period the access of further mosquitos to the experimental containers was prevented by the use of mesh covers. When the introduced larvae reached the third instar, they were transferred to the third series of containers and fourth-instar larvae were transferred to the fourth set. Pupae were transferred to emergence cages similar to those described above. The numbers of larvae and pupae transferred, and of emerged adults, were recorded daily until almost all larvae became adults. When few eggs hatched, first-instar larvae reared in the laboratory from mosquitos caught at the Wat were introduced into the contain- ers to provide the data sought. DEVELOPMENTAL VARIABLES A knowledge of the average time and time range occupied by each developmental stage is fundamental to the construction and analysis of a life budget. Although laboratory data have sometimes been used, they are unsatisfactory, as the present study clearly shows. Data from Christophers (1960) have been compared with the mean development times ob- served in the experimental containers (Table 1). These estimates were based on the time taken for half the population to moult to the next stage; this was determined arithmetically because graphs showed that the fall-off was not suitable for probit Table 1. Mean development times (days) of various immature stages of Ae. aegypti in different containers in the Wat Samphaya during November-August 1967- 68, compared with laboratory measurements made at 28°C (Christophers' data) Egg Larval instar TotalEgg ~ ~ ~ ~ Puadays fromContainer (after Pua egg toIflooding) I1+11 III IV adult water jar 1.45 5.20 3.21 6.51 2.20 18.57 ant trap 2.59 4.57 4.16 6.84 1.95 20.11 flower pot plate 2.88 3.98 3.32 5.49 1.28 16.95.~~~~~~~~~~~~~~ Christophers' (1960) data 3.10 2.90 0.84 1.00 2.00 9.84 212 LIFE BUDGET OF AEDES AEGYPTI 213 analysis, nor was it possible to make the calculation from successive peaks. One approach to the construction of a life budget proposed here (see below) requires information on the pattern of development, the percentage of a cohort entering each stage on each day of the total range of development time. In Ae. aegypti the eggs may hatch over a very long period, and the stimulus for hatching is apparently flooding (Christophers, 1960). In the present study, therefore, development time was calculated from the day of flooding. Occa- sionally some eggs would hatch before artificial flood- ing occurred, especially in the flower pot plates since these containers could be inundated by rain. The mean hatching patterns in each type of container are shown in Table 2. METHODS FOR THE CONSTRUCTION OF A LIFE BUDGET Most methods of population analysis involving the construction of life budgets are designed to be used for species that have discrete generations, yet the majority of insect species, particularly in the tropics, have overlapping generations or, like Ae. ae- gypti, breed continuously. Hughes (1963) devised a method for aphids that depended on the estimate of the rate of reproduction and its projection; this method is not entirely appropriate for Ae. aegypti studies, particularly as it is possible to measure natality (= oviposition in Aedes) in the field. Two simpler methods are therefore being used. Daily populations on a time-specific basis In this method daily values are calculated for the unique events, e.g., oviposition, hatching, and emer- gence, while the populations of the various develop- mental stages are expressed as numbers per median day. This is done by dividing an absolute estimate of the total population by the development time in days; the same approach is used in the graphical method to determine the total population of a stage of a discrete generation (Southwood, 1966). This concept assumes a steady mortality rate within each stage, and a constant level of recruitment over a period extending back in time to the date when the oldest pupa was recruited. In view of the fairly steady level over several weeks of the female population in the Wat, indicated by the mark and recapture experiments (Sheppard et al., 1969), the use of this method seems justified. At any time when the population is increasing or de- creasing rapidly, it will, respectively, over- and under- estimate mortality, and its value is thus limited. The life budgets calculated on this basis are given in Tables 3-5 and presented graphically in Fig. 4 and 5. As the maximum potential natality was not calculated, a modification of the method of Varley & Gradwell (1960), in which only mortality was consi- dered, was used for analysis. However, the role of natality can be investigated separately (p. 221). The actual population numbers are converted to loga- rithms and the various mortalities are expressed as the differences between the logarithms of the popula- tions under consideration. These are, of course, equi- valent to the ratio of one population to the other and are termed k values (Varley & Gradwell, 1960; Southwood, 1966). Because variations in actual numbers between the different occasions could be due to sampling errors, emphasis in the following discussion is placed on the comparison of these population ratios expressing mortality. The expected daily population on an age-specific basis The basis of this method is the determination of the number of individuals that would be expected in the field if there were no mortality during develop- ment, the starting-point being the number of eggs laid each day. If it is assumed that every individual has a constant and identical rate of development and that the third instar, for example, lasts 3 days and is achieved on the tenth day after oviposition, then the total number of third-instar larvae expected on a given day (n) would be the sum of the eggs laid on days n-9, n-10, and n-11. In fact, the development rate is not constant, but falls within a range (Table 2); therefore, the number of individuals of any stage on a given day will be the sum of contributions from oviposition on many days (see Fig. 1). Although the addition of the various components is theoretically straightforward, in practice they are so numerous that the summation must be done by computer. A computer programme was therefore devised.' The expected numbers can be calculated by means of this programme and the total mortality occurring between the egg and each developmental stage is found by subtracting the actual numbers from the expected numbers. The use of this method in this study assumes the following particular conditions: (1) the daily number of eggs laid in any type of container in the Wat is fully represented by the monthly sample; 1 An account of the computer programme has been de- posited in the WHO Library, and copies may be obtained on request from Chief Librarian, World Health Organization, 1211 Geneva 27, Switzerland. 214 T. R. E. SOUTHWOOD AND OTHERS Table 2. The development pattern of Ae. aegypti in different containers in the Wat Samphaya (percentage of the population moulting on given day) Day after flooding ontainer] oult 1 2 3 4 5 6 7 8 9 10 1 11 T 12 13 1 14 water egg-I 38.97 16.01 9.16 6.63 4.5 2.03 1.39 0.93 0.41 0.16 0.08 0.06 -ajars Il-lI 1.03 10.97 14.74 14.65 13.23 11.49 13.71 5.77 4.6 3.8 1.83 2.26 1.56 III-IV 1.87 3.22 11.25 14.82 12.02 6.6 12.98 7.45 5.95 4.87 3.71 IV-pupa 1.16 1.56 2.52 1.93 5.2 5.43 4.74 8.44 4.71 pupa-adult 0.76 2.09 2.25 1.63 4.31 6.05 6.01 ant traps egg-I 49.69 11.77 9.03 1.98 1.52 2.79 1.14 0.16 0.37 0.49 - 0.33 _b1l-lil 8.85 10.72 15.43 14.01 9.94 12.63 10.47 5.23 5.46 3.15 1.91 1.06 0.78 III-IV 2.57 5.9 9.56 9.61 9.98 6.67 5.72 12.68 5.72 8.92 2.12 IV-pupa 1.14 - - 0.31 1.45 5.09 3.34 23.61 13.55 pupa-adult 0.33 - 0.66 8.40 2.63 flower egg-I 33.96 23.77 16.09 3.04 1.12 0.47 0.11 0.43 - c pot Il-lI 8.64 14.07 21.85 15.86 13.71 8.85 7.41 3.13 2.27 1.78 0.66 0.29 0.84 plates III-IV 0.17 6.34 10.60 16.0 13.51 10.32 8.46 6.21 7.65 3.09 3.29 2.89 IV-pupa 1.36 3.07 7.61 12.6 7.55 6.85 6.86 4.42 6.04 pupa-adult 0.22 2.08 3.88 8.59 11.88 9.26 7.93 3.54 Day after flooding 15 16 17 18 19 20 21 22 23 24 25 26 27 28 water egg-Ijars Il-Ill - 0.18 0.18 III-IV 3.39 4.02 2.92 1.94 0.76 0.31 1.31 0.26 0.24 - - 0.04 - 0.04 IV-pupa 9.07 7.6 6.76 8.32 4.70 4.08 5.34 4.09 3.8 3.55 1.59 1.3 1.33 1.95 pupa-adult 9.65 4.94 10.07 5.24 5.83 7.38 3.87 4.7 6.08 3.99 4.35 3.21 2.38 1.33 ant traps egg-I Il-Ill - - 0.38 III-IV 3.56 1.79 2.0 1.63 1.6 2.53 5.31 0.67 0.36 - 0.39 0.21 - 0.51 IV-pupa 5.51 3.61 2.48 1.52 1.45 3.03 3.03 7.06 3.33 2.40 6.87 3.85 1.52 1.58 pupa-adult 23.99 15.93 6.22 2.27 5.16 1.47 2.29 1.58 4.85 5.9 5.48 1.91 5.15 0.33 flower egg-I pot l-l l - - 0.31 0.17 - - 0.09 plates III-IV 1.3 2.2 1.86 0.81 0.58 0.25 2.27 0.83 0.44 0.38 0.10 0.21 0.29 IV-pupa 4.01 6.61 5.45 3.87 2.61 3.78 1.59 1.5 2.04 1.28 2.24 2.5 0.99 1.5 pupa-adult 6.55 5.86 6.3 5.69 5.15 2.24 2.97 2.25 0.91 1.56 0.81 2.04 2.76 2.08 Day after flooding |___|___ _ 29 1 30 31 |32 33 34 |35 36 37 38 39 40 1 41 142...45 water egg-Ijars Il-Ill III-IV IV-pupa - - 0.31 0.33 - 0.21 pupa-adult 0.99 1.7 0.24 - 0.34 0.36 0.24 ant traps egg-I Il-Ill III-IV IV-pupa 0.45 - 2.5 - 0.45 - 0.45 0.45 pupa-adult - 5.46 flower egg-I pot Il-Ill plates IIl-IV IV-pupa 1.41 0.28 0.28 0.46 0.09 0.29 0.17 0.09 0.5 - - - - 0.09 pupa-adult 2.17 1.66 - 0.52 0.68 0.1 0.1 0.19 0.1 0.1 a 19-69 % hatched before flooding. 11 20.73 % hatched before flooding. c 21.9 % hatched before flooding. Table 3. Life budget and K values in water jars from October to December 1967 and from February to August 1968 Stage Number Log 1 1 Number Log k October 1967 November 1967 eggs 6898 3.83885 1.19906 6503 3.81311 1.23219 I + II 436.3 2.63979 0.11604 380.9 2.58092 0.02825 III 333.8 2.52375 0.13281 357.0 2.55267 0.25600 IV 245.9 2.39094 0.03301 197.9 2.29667 0.52362 pupae 228.0 2.35793 59.3 1.77305 K = 1.48092 K = 2.04006 December 1967 February 1968 eggs 4025 3.60477 1.12333 3499 3.54407 1.26487 + II 302.8 2.48144 0.03119 190.2 2.27920 1.94274 III 281.6 2.45025 0.18308 216.8 2.33646 0.39545 IV 184.6 2.26717 0.84229 87.3 1.94101 0.23259 pupae 26.6 1.42488 51.1 1.70842 K= 2.17989 K = 1.83565 March 1968 April 1968 eggs 4141 3.61710 1.03164 2554 3.40722 0.43548 I+ II 384.7 2.58546 0.46753 936.7 2.97174 0.44411 III 131.2 2.11793 0.25520 336.8 2.52763 0.43072 IV 72.9 1.86273 0.27727 125.0 2.09691 0.01234 pupae 38.5 1.58546 121.5 2.08457 K = 2.03164 K = 1.32265 June 1968 July 1968 eggs 4044 3.60681 1.19840 7280 3.86213 1.51383 + II 256.1 2.40841 0.18517 222.9 2.34830 0.06727 III 167.2 2.22324 0.71945 190.9 2.28103 0.47013 IV 31.9 1.50379 0.28107 64.7 1.81090 0.46455 pupae 16.7 1.22272 22.2 1.34635 K = 2.38409 | K= 2.51578 eggs I + 11 III IV pupae 3646 388.9 260.2 129.1 28.9 August 1968 3.56229 0.97234 2.58995 0.17465 2.41530 0.30437 2.11093 0.65003 1.46090 K= 2.10139 T. R. E. SOUTHWOOD AND OTHERS Table 4. Life budget and K values in ant traps from October to December 1967 and from February to August 1968 Stage Number Log k Number Log k October 1967 November 1967 eggs 92.2 1.96473 1.73148 90.0 1.95424 1.62996 + II 171.1 2.23325 0.88107 71.3 2.32428 0.85299 III 22.5 1.35218 0.35654 29.6 1.47129 0.34096 IV 9.9 0.99564 13.5 1.13033 pupae 0 0 0 0 December 1967 February 1968 eggs 114.75 2.06070 1.86703 149.9 2.17609 0.29471 I + II 156.2 2.19367 0.28145 76.1 1.88138 0.33731 III 81.7 1.91222 0.00805 35.0 1.54407 0.27690 IV 80.2 1.90417 1.26072 18.5 1.26717 0.86923 pupae 4.4 0.64345 2.5 0.39794 K = 1.41725 K = 1.77815 March 1968 April 1968 eggs 302.7 2.48144 0.50555 107.5 2.03141 0.17832 + II 94.6 1.97589 0.28836 71.3 1.85309 0.32289 III 48.7 1.68753 0.09424 33.9 1.53020 0.38719 IV 39.2 1.59329 1.76078 13.9 1.14301 0.76280 pupae 0.68 1.83251 2.4 0.38021 l______ K=2.64893 K=1.65120 June 1968 July 1968 eggs 85.7 1.93298 0.15338 60.7 1.78319 0.06802 + II 60.2 1.77960 0.25716 51.9 1.71517 0.08678 III 33.3 1.52244 0.63035 42.5 1.62839 0.44370 IV 7.8 0.89209 15.3 1.18469 0.12399 pupae 0 0 11.5 1.06070 K = 0.72249 eggs I + 11 III IV pupae 11.02 198.2 51.1 14.5 1.5 August 1968 1.04217 2.29711 1.70842 1.16137 0.17609 2.74506 0.58869 0.54705 0.98528 K = 0.86608 216 Table 5. Life budget and K values in flower pot plates from October 1967 to December 1967 and from February 1968 to August 1968 Stage Number Log k Number Log k October 1967 November 1967 eggs 38.7 1.58771 T.56240 90.0 1.95424 1.82391 I + II 105.6 2.02531 1.66928 134.6 2.13033 1.01974 III 226.5 2.35603 0.24882 12.9 1.11059 1.50421 V 127.9 2.10721 0.61306 40.4 1.60638 0.42169 pupae 31.2 1.49415 15.3 1.18469 K = 0.09356 K = 0.76955 December 1967 February 1968 eggs 31.5 1.49831 1.89734 90.45 1.95641 0.25140 + II 39.9 1.60097 0.68716 50.7 1.70501 1.32480 III 8.2 0.91381 1.69897 2.4 0.38021 1.47712 IV 16.4 1.21484 1.36727 8.0 0.90309 1.79929 pupae 70.4 1.84757 12.7 1.10380 K = 1.65074 K = 0.85261 March 1968 April 1968 eggs 638.2 2.80496 0.97694 95.6 1.98046 1.98570 I + II 67.3 1.82802 0.67573 98.8 1.99476 0.13144 III 14.2 1.15229 1.99092 73.0 1.86332 0.32803 IV 14.5 1.16137 1.11998 34.3 1.53529 0.04253 pupae 1.1 0.04139 31.1 1.49276 K = 2.76357 K = 0.48770 June 1968 July 1968 eggs 46.2 1.66464 1.84575 76.1 1.88138 0.68437 + II 65.9 1.81889 0.41577 15.7 1.19701 0.32311 III 25.3 1.40312 0.38609 7.5 0.87390 0.14881 IV 10.4 1.01703 1.61220 5.3 0.72509 1.92369 pupae 25.4 1.40483 6.3 0.80140 eggs + 1I III IV pupae K = 0.25981 II 226.1 65.9 19.4 17.9 22.8 August 1968 2.35430 1.81889 1.28780 1.25285 1.35870 0.53541 0.53109 0.03495 1.89415 K = 0.99560 K= 1.07998 T. R. E. SOUTHWOOD AND OTHERS Fig. 1. The contribution of a single egg cohort, produced at day 4, to a mixed population of overlapping genera- tions. The lines of different slope represent the fastest and slowest developers and the age extremes of the numbers of a single cohort. (2) the development times observed monthly in the experimental containers are true of the natural populations for the whole of that month; (3) the containers in the Wat dry out or are emptied, cleaned, and refilled with water at random. Unfortunately, these conditions are not com- pletely satisfied. The monthly egg counts cannot be accepted with full confidence as representing daily oviposition in the Wat. An alternative approach would be to interpolate between monthly counts to obtain daily estimates, but this was considered in- appropriate in the present study. However, by using the stable age proportions (Table 6) of each instar, based on the geometric mean development times implicit in the method, we can obtain a second series of time-specific estimates of mortality to compare with those of method 1. Comparison of mortality in water jars revealed by the two methods The expected numbers of larvae arising from me- thod 2 were calculated from the mean monthly counts of first- plus second-instar larvae and the expected proportion of pupae to first- plus second- instar larvae for each month (Table 6). The diffe- rences between log1o numbers of expected and ob- served pupae gives an estimate of cumulative mortal- ity (Table 7). The values, compared month by month, are of the same order of magnitude (Spearman's rank correlation r8 = 0.95; P < 0.01) but it is realized that these two methods are not completely independent since they are based on the same num- Table 6. Expected proportions of the populations in the various stages when a stable age distribution has been reached; computed from the developmental pattern for water jars Stage Experiment Date |+ IV |IPupa larvae larvae larvae Pupae 1 29 Oct. 1967 0.352 0.456 0.119 0.072 2 18 Nov. 1967 0.338 0.298 0.220 0.143 3 23 Dec. 1967 0.269 0.236 0.350 0.144 4 19 Feb. 1968 0.399 0.214 0.301 0.085 5 27 March 1968 0.257 0.162 0.466 0.115 6 29 April 1968 0.355 0.196 0.290 0.159 7 9 June 1968 0.240 0.197 0.450 0.113 8 13 July 1968 0.290 0.143 0.447 0.170 9 19 Aug. 1968 0.247 0.122 0.549 0.083 218 LIEFE BUDGET OF AEDES AEGYPTI Table 7. Comparison of mortality between larval stage ll and pupae expressed as difference between log populations, i.e., k2 + k3 + k4. Mortality Experiment Date Method 1 |Method 2 1 29 Oct. 1967 0.2818 -a 2 18 Nov. 1967 0.8079 0.8074 3 23 Dec. 1967 1.0566 1.1583 4 19 Feb. 1968 0.5708 0.2722 5 27 March 1968 1.0000 1.0244 6 29 April 1968 0.8872 0.9288 7 9 June 1968 1.1857 1.2332 8 13 July 1968 1.0020 1.0916 9 19 Aug. 1968 1.1291 1.0287 a Calculated mortality less than 0. ber of early stage larvae. No clear indication of which method is the more appropriate is available at this stage. However, method 2 is based on a complete X--X OCTOBER 1967 X-----zX NOVEMBER 1967 set of developmental information for each month, while method 1 utilizes 50% development times averaged over all the monthly experiments. This seems to favour method 2, but to use the method in this form, egg counts would have to be included. This would necessitate the use of 50% development times for the egg stage, and since the mortality estimates (Table 7) are so similar, it was considered that the use of a mixed model would not be justified. Therefore, estimates in the rest of this report are derived by method 1. SURVIVAL UNDER DIFFERENT CONDITIONS Gross comparisons of the numbers of the different stages in the various containers are made in Fig. 2 and 3. These numbers are not absolute populations since they have not been corrected to allow for the different development times of the stages. They do show, however, that with a few exceptions there is a considerable fall-off in gross numbers between the egg and pupal stages; survival in ant traps appears to be less than in the other types of container. When absolute population values from the life budget (Tables 3-5) are used (Fig. 4 and 5), it can be *----@ DECEMBER 1967 *-* FEBRUARY 1968 20 1O0h , l I411 m N PUPA '-ININSTAR-' 11 N5Z PUPA l~ NSTAR Is m 15 PUPA \INSTAR- "W' WHO 1079r8 Fig. 2. Monthly mean number of immature stages per container in the Wat Samphaya, 1967-68. A, water jars; B, ant traps; C, flower pot plates. w , I~- z 1( w 219 T. R. E. SOUTHWOOD AND OTHERS B I1Z m 3 PUPA x mrI a- lNSTAR -' '~- INSTAR-/ O-.O MARCH * * IJULY X-X APRIL S--S AUGUST X----X JUNE 20 - C l0 - .%4~. PUPA 14R m ]7 PUPA \- INSTAR -/ WHO 10799 Fig. 3. Monthly mean number of immature stages per container in the Wat Samphaya, 1968. A, water jars; B, ant traps; C, flower pot plates. seen that the shape ofthe survivorship curves for water jars appears to change in early March, which cor- responds to the beginning of the hot season. During the cool season from October to February there is relatively little mortality between the second and fourth instars (Fig. 4), whereas between March and August this plateau does not occur (Fig. 5). The data in Fig. 4 and 5 also show that there are more inconsistencies in the life budget estimates of the populations in ant traps and flower pot plates. The sources of variation are not obvious but are probably more the result of unstable conditions in the containers than of errors arising from the small number of samples. Clearly, these data must be interpreted with care, and they are therefore not used in the detailed analysis of the life budget in the fol- lowing sections. Gross comparisons between mean total mortalities in different containers were made using Student's t test. The mortalities in water jars were greater than in flower pot plates (0.01 > P > 0.002); other com- parisons were not statistically significant. CONTRIBUTION OF DIFFERENT TYPES OF CONTAINER TO THE POPULATIONS OF ADULTS IN THE WAT The importance of each type of container in the Wat depends on the number of containers, the num- ber of eggs laid in them, and the survival of the immature stages. The sum of the mean number of pupae for each type of container over the 9 experi- mental periods (Tables 3-5) multiplied by the mean percentage emergence gives the contribution made by each type (Table 8). This shows that water jars are the main source of the adult population and that the contribution of ant traps is quite insignificant. 220 LIFE BUDGET OF AEDES AEGYPTI X -X OCTOBER 1967 *-* NOVEMBER 1967 so0F B I I EGG 1+3I Xm m ' INSTAR PUPA X-----X DECEMBER 1967 ---- FEBRUARY 1968 30r C EGG I3 3 PUPA EGG "'- NSTAR- I+f m PUPA INSTAR-/ WMO 10800 Fig. 4. Monthly survivorship curves for the immature stages, based on total numbers corrected for development time, 1967-68. A, water jars; B, ant traps; C, flower pot plates. THE RELATIVE ROLES OF NATALITY AND MORTALITY A method for comparing the relative importance of natality (the number of eggs laid) and mortality is described by Southwood (1967). This can be applied to the data from the three types of container given in Tables 3-5, i.e., for the months ofNovember-August. This is done in Fig. 6, which clearly shows that the variations in the numbers of emerging adults are related to variations in mortality rather than natality. RECOGNITION OF THE KEY MORTALITY FACTORS From Fig. 7 it can be seen that in ant traps and from October to February in water jars the variation in total mortality (K) from egg to pupa is due mostly to variations in k,, i.e., the death of larvae between the fourth instar and pupation. From March onwards there is an interesting increase in the role of k1 for water jars, i.e., deaths between eggs and second- instar larvae, which account for the largest propor- tion of total mortality. The change in the key factor coincides with the beginning of the hot season and the change in survivorship curves noted earlier (p. 219). Unfortunately, only 12 of the 27 sets of life table data gave complete budgets (Tables 3-5). This is largely because none of the information for flower pot plates is acceptable, and no pupae developed in the ant traps in November, December, and July. A source of error in oviposition estimates could have been the use of clean experimental ant traps and flower pot plates to facilitate counting. Possibly they were less attractive as oviposition sites and led to eggs laid in non-experimental containers being under- estimated. The mortality factors k1 and k4 were analysed for density dependence by plotting each against the log density of the stage on which they acted (Fig. 8). Clearly k1 is density-dependent, the mortality rate increasing with density; this could result from com- petition between young larvae. The regression of k, on density with a slope greater than 1.0 indicates overcompensating mortality and hence an increase A4Or tn z 3C crLiJ z 221 260e T. R. E. SOUTHWOOD AND OTHERS Fig. 5. Monthly survivorship curves for the immature stages, based on total numbers corrected for development times, 1968. A, water jars; B, ant traps; C, flower pot plates. in the magnitude of the fluctuations in population size. Factor k4 acts independently of density but is an important component of the total, K (p. 221). Table 8. Comparison of contributions of the various types of container to the adult population emerging in the Wat Samphaya during the nine experimental periods Water Ant IFlowerjars traps | p___ot- total daily mean pupae 65.87 2.56 24.03 mean percentage emergence 83.25 84.07 83.55 total mean daily emergence 54.84 2.15 20.08 percentage of emerging adults 71.2 2.8 26.0 COMPARISON OF SEASONAL TRENDS IN MORTALITY AND ADULT NUMBERS Sheppard et al. (1969) have given an account of fluctuations in the numbers of adults in the Wat determined by mark, release, and recapture during the period in 1966-67 immediately preceding the present study. Their work revealed certain fluctua- tions and, in particular, a rise between April and July which they believed to be real. Their population estimates (Table 8 of Sheppard et al.) are plotted together with total mortality (K, on an inverse scale) against data by month in Fig. 9. It will be seen that although the data refer to successive years there is evidence of correspondence between peaks and troughs with the mortality curve slightly in advance. This appears to give independent support to the view that the population rise in April is a real phenomenon. It is particularly noteworthy that the present study suggests that the fall in mortality levels 222 LIFE BUDGET OF AEDES AEGYPTI 20 .-5 30 25 20 ? _ U ,,S o ,. ; < ZOIL - < \---1967--"-1968-/ XIXA xx x- 1.0 _ L IS5 I III 11 =D 2!- -Cu* -_ Uv u * a it <n<A O OILW< < \ 1967---'W-196802 WHO 10802 Fig. 6. Comparison of the importance of natality (PE) and mortality (K) in determining the number of adults (PR) in the Wat Samphaya, 1967-68. A, water jars; B, ant traps; C, flower pot plates. is associated with a change in the importance of early larval mortality, which supplemented late larval mortality at the onset of hotter weather. DISCUSSION AND CONCLUSIONS Fluctuations in the number of adult mosquitos emerging in the Wat are, on the evidence given here, determined by the mortality of immature stages, rather than by changes in the number of eggs laid. The most significant mortalities occur during the early (first and second) and last (fourth) larval in- stars. Many larvae seem to spend an excessively long period in the last instar and then fail to develop further. The mortality in the first instar is probably density-dependent. It seems, therefore, that in the absence of predators and disease both these mortal- ities could arise from competition effects, presumably resulting from food shortage. This point requires further study. From the beginning of the hot season in March, the survival of the early instars becomes more significant in the dynamics of the population and this leads to a rise in the numbers of adults emerging. It is pos- sible that this gives independent support to, and an explanation of, the rise in adult numbers between April and July suggested by the mark and recapture studies of Sheppard et al. (1969). Whether the fall in larval mortality in April-May is the trigger for a chain of events that leads to the rise in incidence of dengue in June (Wkly. Epidem. Rec., 1970) will depend not only on the establishment of the relation- ships suggested above, but also on the influence of the initially larger mosquito population on the biting rate, on the proportion of infected mosquitos in the population, and on the survival of adult mosquitos. 2-5 F PR 2-0 1-5_ 1-0 1-5 K 25 PE 4-0- ,3-5 II II l iI I II II ._ . >. X, &f u.u Z .'oXcX ""- 1967----1968 223 T. R. E. SOUTHWOOD AND OTHERS K 0/1X \/ '.5 XzXs X k1 10 _ k2 k4 0X O N D F M A J J A /\971\8 2B K xs o1 o x os5 - x x k3 _ X X 1-5 _ / wmo 10803 1.5 ki so Fig. 7. Comparison of changes in various component mortalities (kl-k4) with changes in total mortality (K). A, water jars; B, ant traps. k4 0.5 OL I 31 3.3 3.5 3.7 39 LOG NUMBER OF EGGS 0~~~ 0 N 1-2 1-4 1-6 1*8 2O 2*2 24 26 LOG NUMBER OF FOURTH-INSTAR LARVAE WHO 10804 Fig. 8. Relationship between mortality and the population density at the start of the age interval in which it operates. ki = Mortality from egg to second-instar larvae on numbers of eggs. k4 = Mortality from fourth-instar larvae to pupae on numbers of fourth-instar larvae. 224 LIFE BUDGET OF AEDES AEGYPTI Fig 9. Comparison of the population trends in adults during 1967-68 (from mark, release, recapture studies), and of seasonal variations in total development mortality in water jars (K being plotted inversely) during 1967-68. 5 0~~~~~~ 0 \_/\-J 30-0 x~~~~~ 0 N D F M A M J J A \-1967--/ 1968- WhO 101 RIaSUMt tTUDES SUR LE BUDGET VITAL D'AEDES AEGYPTI A WAT SAMPHAYA, BANGKOK (THATLANDE) Pour bien comprendre l'epidemiologie d'une maladie transmise par un vecteur, il est indispensable de disposer d'informations sur la dynamique de population de ce der- nier. Les 6tudes men&es sur Aedesaegyptiai Wat Samphaya (Bangkok) en 1967/68 ont port6 en particulier sur les variations numeriques des stades immatures ainsi que sur leurs relations avec les donnees ant6rieures concernant les populations d'adultes et les fluctuations saisonnieres de l'incidence du syndrome dengue/fievre hemorragique. On a estim6 le nombre d'aeufs, de larves et de nymphes d'Aedes presents dans les trois types de recipients les plus courants (jarres, pieges a fourmis, soucoupes placees sous les pots de fleurs) grace a des sondages pratiqu6s 3 a 4 fois par semaine. Le nombre d'aufs pondus a ete 6valu6 a l'aide d'ovipieges et les aspects du developpement de chaque stade immature ont ete dtudies en se basant sur l'evolution de larves d'age connu. On a constate que la duree d'evolution des differents stades etait plus longue dans le milieu naturel qu'au laboratoire, en raison proba- blement d'un apport insuffisant de nourriture. Deux methodes permettent d'etablir le budget vital d'une population d'insectes caracterisee par un chevau- chement des g6n6rations successives. La premiere consiste a calculer les effectifs quotidiens de chaque stade en divi- sant le chiffre estimn de la population globale par la dur6e d'evolution en jours. Dans la seconde, on evalue les effectifs quotidiens de chaque stade d'apres le nombre d'aeufs pondus chaque jour et le rythme de developpe- ment, la mortalite etant consider6e comme nulle. Les jarres representent le principal lieu de production d'Aedes adultes a Wat Samphaya. Dans les pieges 'afour- mis et les soucoupes, les taux de survie subissent des variations considerables, dues sans doute a l'instabilit6 des conditions de milieu (notamment la quantite d'eau disponible) dans ce genre de recipients. La mortalite glo- bale d'Aedes est significativement moins dlevee dans les soucoupes que dans les jarres. Dans les jarres, les variations de la mortalite contri- buent davantage aux fluctuations du nombre total des eclosions imaginales que les variations du nombre d'aeufs pondus. D'octobre 1967 a f6vrier 1968, le facteur deter- minant l'aspect de la courbe de mortalite globale a et la mortalite des larves entre le 4e stade et la nymphose (k), mais a partir de mars 1968 la mortalite des insectes entre le stade de l'ceuf et le 2e stade larvaire (kl) a &t6 en majeure partie responsable de la mortalite globale. L'importance de k, est fonction de la densite de l'effectif au stade sur lequel il agit. Il semble qu'en l'absence d'infection larvaire et d'intervention de preda- teurs la mortalite des larves durant les ler et 4e stades soit influencee par la concurrence entre individus, probable- ment due au manque de nourriture. S w05 225 226 T. R. E. SOUTHWOOD AND OTHERS A une mortalite accrue parmi les larves aux premiers stades, constatee en mars, succede, par un phenomene de corrpensation, une baisse de la mortalite larvaire globale en avril-mai. Cela pourrait expliquer I'augmentation du nombre d'Aedes adultes d'avril a juillet. II reste a etablir s'il existe un lien entre ces variations de la mortalite lar- vaire et la frequence plus elevee de la dengue/fievre hemor- ragique en juin et a etudier l'influence de l'augmentation du nombre et du taux de survie des vecteurs sur l'intensit6 de la transmission de la maladie. REFERENCES Christophers, S. R. (1960) Aedes aegypti L., the yellow fever mosquito, Cambridge University Press Hughes, R. D. (1963) J. anim. Ecol., 32, 393-424 Sheppard, P. M., Macdonald, W. W., Tonn, R. J. & Grab, B. (1969) J. anim. Ecol., 38, 361-702 Southwood, T. R. E. (1966) Ecological methods, London, Methuen Southwood, T. R. E. (1967) J. anim. Ecol., 36, 519-529 Varley, G. C. & Gradwell, G. R. (1960) J. anim. Ecol., 29, 399-401 WHO Scientific Group on Mosquito Ecology (1967) Wld Hlth Org. techn. Rep. Ser., No. 368 Wkly Epidem. Rec., 1970, 45, 201-208