Macroconditions of Channel Segments Utilized by Culex Pipiens Pallens Immatures in Saga Citx Southwest Japan

Journal of the American Mosquito Contol Association, ll(4):448453, 1995 Copyright O 1995 by the American Mosquito Control Association, Inc.

MACROCONDITIONS OF CHANNEL SEGMENTS UTILIZED BY CULEX PIPIENS PALLENS IMMATURES IN SAGA CITX SOUTHWEST JAPAN

M. MOGI,I T SOTA'2 AND M. MOTOMURA3

ABSTRACT. The relation between macroconditions of water channels and the breeding of Culex pipiens pallens in Saga City were examined, based on weekly census data over 3 years for ca. 2,OOO channel segments. Mosquito breeding activity was greatest in channel segments between building lots and field lots in the suburban area. The segments involved ranged from narrow roadside ditches (width < I m) to wide irrigation canals (width > 10 m), all of which were utilized by mosquitoes. There was a weak tendency for more mosquito breeding in narrower segments. Water flow indices ranged from OVo (always stagnant) to lNVo (always flowing). Mosquito breeding tended to be suppressed with an increase in flow indices. Mosquito breeding was confirmed in ca. 35Vo of segments yearly, and such mosquito- productive segments tended to persist during 3 study years.

INTRODUCTION drains are the common habitat for Culex quin- quefasciatus Say at various localities in South- One feature characterizing Saga City in west- east Asia. Following the widespread develop- presence ern Japan is the of channel networks. ment of insecticide resistance in this mosquito The origin of the channels dates back to those (World Health Organization I 986), environmen- constructed on reclaimed land in the 7th century. tal management measures, such as improvement channels Since then, have been developed as of water flow and quality by cleaning and design sources of water for irrigation and domestic use modiflcation, have been recommended for per- (Saga as well as drains for rainwater and waste manent control of the mosquitoes (World Health The City 1977). channels supply water for ag- Organization 1988). However, the feasibility of ricultural fields in rural and suburban areas of environmental management for Cx. quinquefas- the city and, in the rainy season, protect the ur- ciatus control in large areas has not been re- ban area lying at sea level from flooding. Excess ported. sewage from an increasing human population Channel networks in Saga City collectively "creeks" since the middle 2oth century has polluted the have been called by the citizens. In water. Consequently, numerous segments along reality, they are extremely variable in their lo- the channel networks have turned into favorable cation and size, ranging from rural irrigation ca- habitats for larvae of the mosquito Culex pipiens nals >10 m wide to urban roadside ditches

448 DncBwen 1995 Cx. p. pnttgt'ts Lenv,ct- Hngtrn.rs N Unsex CHer.tIEI-s 449

60 ters of the human population lives in suburban and rural areas. The annual mean temperature is l6.l'C (for 50 1961-90) with the monthly minimum of 5.0"C in January and the maximum of 27.6"C in Au- 40 gust. The mean annual precipitation is 1,836 mm with a monthly minimum of 62 mm in January and a maximum of 341 mm in July. 30 In Saga City, Cx. p. pallens overwinters as adult females in diapause. Biting activity starts 20 in March and continues until November with the predominant biting season being from May 10 though September (Mogi 1987a). Larvae some- times are found in artificial containers around human houses, but, quantitatively, the role of 0 ',2 such habitats in the reproduction of Cx. p. pal- 3 lens is negligible compared to the omnipresent Environment water channels. Census: Water channels form intricate net- 60 works. A census unit consisted of one segment along the network. One segment typically was a 50 part of a network between 2 junctions, where the channel width was constant. When the width changed within a part between 2 junctions (often at turns), each subpart with a constant width was goo regarded as a separate segment. The census was 430 conducted from 1986 through 1988. Before the tr 1986 mosquito breeding season, 2,135 segments were listed and numbered as potential larval 920 habitats. Excluded from this list were cement- E o lined segments with constantly flowing clear wa- 'i 10 ter. Segments located >10O m from the nearest houses in the rural area also were excluded due to labor limitations. In succeeding years, some 0 of initially registered segments were deleted from the list because census results in the pre- ceding year(s) indicated little possibility ofthose Creekwidth (m) segments being utilized by mosquitoes. Thus, 1,737 and 1,698 segments were inspected in 1987 and 1988, respectively. The mean length of registered segments was ca. 150 m. Registered segments were surveyed weekly from May through September. Presence or absence of water flow, irrespective of flow speed, was recorded by visual inspection, if necessary

a Mogi, M. 1987. Report of basic studies for mosquito control, Saga City.

Fig. l. Macroconditions or channel segments classified by environments (A; 1 = between field lots, 2 : between a field lot and a building lot, 3 = between building lots), widths (B), and flow indices (C; no. censuses with flow/total no. censuses per year), Class values indicated by ranges include the lower limit but exclude the upper limits. Vertical lines on the bar in- Flow rate (%) dicate I SD for 3 vears. JounNaI- oF THE AMERICIN Mosquno CoNrnol AssocrarloN Vol

Thble 1. ANCOVA surnmarv table. Source of Sum of Mean Year variation df squares square F 1986 Environment 2 0.95 o.47 g.g5*l width I r.42 1.42 29.81* Flow index 1 11.58 I 1.58 243.44* Error 2,126 lol.o9 0.05 t987 Environment 2 1.35 0.68 15.63* width I o.64 o.64 14.95* Flow index I 8.20 8.20 189.46* Error r,527 66.12 0.04 1988 Environment 2 1.27 0.63 15.96* width I 1.53 1.53 38.52* Flow index I 5.54 5.54 139.56* Error r,567 62.24 0.04 r*indicatesP<0.0001.

with the aid of paper pieces placed on the water vegetable cultivation. Building lots consisted of surface. Presence or absence of culicine mos- houses, shops, schools, temples, factories, car quito immatures, including egg rafts, larvae, and parks, and office buildings. The flow index at pupae, was inspected by dipping. Stagnant or each segment was calculated as no. of flow-pos- slowly running areas along each segment where itive censuses/total no. censuses per year. mosquitoes might occur were sampled by a dip- Analyses: An analysis of covariance (AN- per (capacity 500 ml). When emergent plants, COVA) was performed for each year with the such as Phragmites communis (Gramineae) and general linear models procedure of SAS (SAS O enanthe j av anic a (U mbelliferae), were present, Institute 1985), with mosquito-positive index areas among plants also were sampled by dip- (no. mosquito-positive censuses/total no. census- ping. Sampling was discontinued when the pres- es per year) as a dependent variable. The inde- ence of mosquitoes was confirmed, irrespective pendent variables were the environments (1, 2, of their density. Due to time constraints, a max- 3), with segment widths and flow indices as cov- imum of 20 dips per site was established. Spec- ariates. imens were not preserved. Previous surveys indicated that Cx. p. pallens was the dominant cul- RESULTS icine species breeding in urban and suburban segments (>95Vo), although Culex tritaenior- Channel macrocondition: More than 5OVo of hynchus Giles may also be abundant in some channel segments occurred between building rural segments near rice fields in midsummer lots (Fig. 1A). Combining these segments with (Mogi 19874). segments between a building lot and a field lot, Among macroconditions that may influence more than 9OVaof registered segments were lo- mosquito breeding, general environments cated between or by the side of building lots. around segments, segment widths, and flow in- Segment widths ranged from (l to >10 m, but dices were selected for analyses. These condi- 5O7o of segments fell into the class of l-3 m tions varied widely among segments but could wide (Fig. lB). Flow indices showed a bimodal easily be quantified. All segments were more or pattern with frequency peaks in both the classes less exposed to direct sunlight and none were of constantly stagnant segments and of constant- completely covered by trees. Most segments ly flowing ones (Fig. lC). At ca. 4OVo of the were lined with cement with only a few being segments, water was always or usually stagnant unlined. Though vegetation varied among seg- (flow indices < 25Vo), whereas at another 4OVo, ments, quantification of the vegetation cover water was always or usually flowing (flow in- was difficult due to seasonal changes and move- dices ) 75Vo).'t\us, in Saga City, stagnant or ment of floating vegetation by winds and water usually stagnant segments of narrow or medium flow. The environment of each segment was widths exist very close to human residential or classified into one of 3 categories: I = between working places. field lots, 2 : between a field lot and a building Mosquito breeding: All 3 macroconditions af- lot, and 3 = between building lots. Field lots fected mosquito breeding significantly through 3 were pieces of land used for rice, barley, and study years (Table 1). The mean mosquito-pos- DBcnNasen 1995 Cx. p. pettevs Lnnvnl Hnsnnrs IN UneeN CSINNBI-s 451

20 itive index for segments between building lots and field lots was l1%o, which was higher than that for segments between field lots and also that for segments between building lots (Fig. 2A). 15 Thus, a doughnut pattern was evident, that is, mosquito breeding activity was greatest in the circular zone surounding the central urban area. 10 Mosquito-positive indices tended to decrease with an increase in segment widths but segments of all width classes were utilized by mosquitoes (Fig. 2B). Mosquito-positive indices decreased 5 rapidly with an increase in flow indices from the mean )2OVo for stagnant segments to ca. ZVo for segments that were always flowing (Fig. 2C). 0 Thus, mosquito breeding activity was greatest in 12 narrow, stagnant channel segments in the sub- 3 urban area. Within each year, mosquito breeding Environment was conflrmed for ca. 35Vo of segments (Fig. 3). Segments with higher mosquito-positive indices bq 20 were fewer, btt ca. TOVoof segments exceeded () a mosquito-positive index of 507o. Mosquito TE breeding was not confirmed for 42Vo of seg- c, 15 ments throughout the study years, whereas 2OVo .: of segments were utilized by mosquitoes every .= year (Fig. 4). Frequencies (P,) of mosquito-pos- U' o itive years (r) expected when mosquito-positive CL 10 and -negative segments alternate randomly by oI year can be calculated from the binomial distri- = as: t bution ct p,: o 5 ,C;O.35,.0.653_,, o where 0.35 = mosquito-positive index per year. E The frequency was higher than expected tr actual (E 0 values for segments where mosquito breeding o was either not confirmed or was confirmed in all = the 3 study years (G-test,G*i: 51.576,dt :2, P < 0.001). Thus, mosquito-negative or -posi- Greekwidth (m) tive segments tended to persist through the study years.

DISCUSSION Mosquito-productive channel segments typically were stagnant and narrow, located between building lots and field lots. Approximately 35Vo of segments were responsible for mosquito production each year, and such mosquito-productive tend to persist at least during 3 study

:t-"n,r

Fig. 2. Relationship between mean mosquito-positive indices (no. censuses when mosquito immatures were confirmed/total no. censuses per year) and environments (A), widths (B), or flow indices (C). For definition of environment classes and flow indices, see explanations ofFig. 1. Class values indicated by ranges include the lower limit but exclude the upper limit. Vertical lines on the bar indicate 1 SD for 3 years (not Flow rate (o/o) shown if smaller than the circle radius). 452 JounN,qlop rse AuenrceN Mosqurro CoNrnoL AssocnrroN VoL.ll, No.4

be I o o c tr o o J = at E o o L lr lt

0123 No.rnosquito-positive years Fig. 4. Channel segmentsclassified by the number Mosquito-positiverate (7o) of years when mosquito breeding was confirmed. The line graph indicates frequenciesexpected when mos- Fig. 3. Channel segmentsclassified by mosquito- quito-positiveand -negativesegmenls alternate ran- positive indices.For definition of mosquito-positivein- domly by year. dices,see explanation for Fig. 2. Classvalues include the lower limit but exclude the upper limit. Vertical lines on the bar indicate I SD for 3 years. quito breeding in the urban area have been suppressed by a higher prevalence of modernized sewerage and increased channel cleaning by in- years. Some of these features may be character- dividual homeowners compared to the suburban istic to Saga City but some (e.g., stagnancy) are area. People living in new houses in the subur- certainly common to other localities where open ban area are mostly young-generation newcom- drains are the most important breeding habitat ers. Consequently, community-wide activities, for the Culex pipienr complex. In view of the such as channel cleaning by community partic- importance of this species complex as urban fil- ipation, are more difficult to organize and im- ariasis vectors worldwide (White 1989), and plement. Without such suppression, mosquito also as nuisance pests, the scarcity of compara- breeding in the urban area would have been as tive data may warrant further study. Character- high as or even higher than that in the suburban ization of local conditions through comparison area. This means that mosquito production in the with other localities would facilitate the estab- suburban area also can be suppressed by an im- lishment of control strategies appropriate to each provement of sewerage and increased commu- locality. nity-wide activities. Extensive mosquito breeding in the transition Wider channel segments are usually deeper zone between urban and rural areas does not and less polluted. Furthermore, wider segments necessarily indicate the presence of intrinsic are inhabited by various kinds of larvivorous conditions facilitating mosquito breeding in this fish (Mogi et al., unpublished data). If there is zone. Waste discharge, which tends to slow wa- rich emergent vegetation, areas among plants ter flow and promote eutrophic conditions, is may be suitable habitats for Cx. tritaeniorhyn- more extensive in the densely populated urban c/zzs immatures, but usually such habitats are area. Dogs and humans are the most important less utilized by Cx. p. pallens. Wider segments blood meal hosts in Saga City, and their densi- supply fish to narrow connecting segments. In ties are higher toward the central urban area. fact, native predators, including fish and preda- Adult resting habitats also are abundant in the cious insects, cause a substantial mortality in C-x. urban :uea. Among others, covered roadside p. pallens larvae in polluted urban channels in ditches offer the best resting place for Cx. p. Saga (Mogi and Okazawa 1990). Although it is pallens. Though 9OVoof the houses are equipped impractical to widen channel segments for mos- with window screens (Motomura et al. 1995), quito control, it is a reasonable strategy to main- under-floor spaces and storerooms can also eas- tain channel widths and to avoid destroying wid- ily be utilized as resting places by this endo- er segments so as to preserve fish diversity and philic species. Potential risks of intensive mos- abundance. Dpcerr,rsnn1995 Cx p. pAtr,ENsLanvnr- Hnsrrnrs N URBAN CHANNELS

Conspicuous effects of water flow on mos- Kruse, C. W. and M. L. Reynaldo. 1955. Automatic quito breeding indicate the possibility of sup- siphon for the control of Anopheles minimus vat. }{yg. 6l:349- pression of mosquito breeding in channel seg- flavirostris in the Philippines. Am. J. 36r. ments by increasing water flow. Improvement in Mogi, M. and T Okazawa. 1990. Factors influencing stagnant segments (flow in- always or usually development and survival of Culex pipiens pallens ( dices 25Vo) would especially be effective. larvae (Diptera: Culicidae) in polluted urban creeks. Howeve! there are still problems to be exam- Res. Popul. Ecol. (Kyoto) 32:135-149. ined. First, more infonnation about aquatic eco- Motomura, M., K. Ohgushi, T, Sota and M. Mogi. systems in channels is required. As mentioned 1995. A questionnaire study on mosquito bites in above, these channels are habitats of various Saga City, Japan. Jpn. J. Sanit. Zool. 46 (in press). aquatic animals, some of which may be serious- Oomen. J. M. D.. J. de Wolf and W. R. Jobin. 1990. ly affected by an increase in water flow. Second, Health and irrigation: incorporation of disease-con- measures in irrigation, a multi-faceted task in it is necessary to establish flow speeds required trol design, construction, operation, Volume 1. Interna- for mosquito control in urban channels. Water tional Institute for Land Reclamation and Improve- discharge or flow speed required for riverine an- ment Publ. 45, The Netherlands. opheline control in hilly areas has been studied Saga City. 7977. History of Saga City, Volume 1. (Kruse and Reynaldo 1955, Oomen et al. 1990), Saga City, Saga. [In Japanese.] but no such information is available for culicine SAS Institute. 1985. SAS user's guide: statistics, ver- mosquitoes breeding in water channels in flat al- sion 5 ed. SAS Institute, Cary, NC. luvial plains. Because water is a valuable re- Subra, R. 1981. Biology and control of Culex pipiens source, the most cost-effective manner of water quinquefasciatas Say, 1823 (Diptera, Culicidae) special reference to Africa. Insect Sci. Appl. l: use for mosquito control should be investigated. with 3 19-338. Suenaga, O. and T. Itoh. 1973. Studies on the filarial ACKNOWLEDGMENTS prevalence among dogs and the mosquito vectors in Nagasaki City, westem Japan. Trop. Med. 15:131- We express our sincere appreciation to the au- 140. thorities of Saga City for permitting the analyses Suenaga, O. and T, Itoh. 1976. Relative feeding pref- of the mosquito control records of Saga City. erences of mosquitoes for man and dogs. Trop. Med. This study was supported by Nippon Life Insur- 18:59-63. ance Foundation (C89ll0) and Saga City. White, G. B. 1989. Lymphatic filariasis, pp. 23-34. 1n.' Geographical distribution of arthropod-borne diseases and their principal vectors. World Health REFERENCES CITED Organization, Geneva. Karoji, Y., O. Sasaki,A. Kuroda and T. Karaki. 1980. World Health Organization. 1986. Resistance of vec- Host-feeding patternsof Japanesemosquitoes II. A tors and reservoirs of disease to pesticides. W.H.O. host-blood identification study of daytime-resting Tech. Rep. Ser. 737. Culex pipiens pallens in Kyoto, Japan.Jpn. J. Sanit. World Health Organization. 1988. Urban vector and Zool. 3l:289-295. pest control. W.H.O. Tech. Rep. Ser. 767.