JoURNALoF THE AMERIcANMosqurro ConrRor, AssocrerroN VoL.4,No.3

A DYNAMIC LIFE TABLE MODEL OF PSOROPHORACOLUMBIAE IN THE SOUTHERNLOUISIANA RICE AGROECOSYSTEMWITH SUPPORTINGHYDROLOGIC SUBMODEL. PART 1. ANALYSIS OF LITERATURE AND MODEL DEVELOPMENT1

D. A. FOCKS,' R. E. McLAUGHLIN2 A^ro B. M. SMITH3

InsectsAffecting Man and ResearchLaboratory, Agrirultural ResearchSeruice, U. S. Department of Agriculture, Gainesuillc,FL 32604

ABSTRACT. During the past decade,the rice agroecosystemand its associatedmosquitoes have been the subject of an extensive research effort directed toward the development and implementation of integratedpest management(IPM) strategies.The objectiveof this work was to synthesizethe literature and unpublished data on the rice agroecosysteminto a comprehensivesimulation model of the key elementsof the systemknown to influence the population dynamics of columbine.Subsequent companionpapers will presenta validation ofthese models,provide an in-depth analysisofthe population dynamics of Ps. colum.biae,and. evaluate current and proposed IPM strategies for this . This paper describesthe developmentof 2 models: WaterMod: Because spatial and temporal distributions of surface water and soil moisture play a decisiverole in the dynamics of Ps. colutnbioe,an essentially hydrological simulator was developed.Its purpose is to provide environmental inputs for a secondmodel (PcSim) which simulates the population dynamics of Ps. columbiac.WaterMod utilizes data on weather, agricultural practices, and soil charac- teristics for a particular region to generate a data set containing daily estimates of soil moisture and depth of water table for 12 representativeareas comprising the rice agroecosystem.This model could be usedto provide hydrologic inputs for additional simulation models of other riceland mosquito species. PcSim: This model simulates the population dynamics of Ps. columbioeby using the computer to maintain a daily accounting of the absolute number of mosquitoeswithin each daily age class for each life stage.The model createsestimates ofthe number ofeggs,Iarvae, pupae, and adults for a representative l-ha area of a rice agroecosystem.

INTRODUCTION Researchersof the Riceland Mosquito Man- agementProject (RMMP) and the closely asso- According to the Food and Agriculture Orga- ciated ARS/CSRS Regional Project on Riceland nization, rice is the staple food of half of the Mosquitoes(5-122) have been directing multi- human race with the projected demand for this institutional and multi-disciplinary researchef- cereal expectedto increaseby almost 90% from forts toward the development and implementa- 1975 levels by the year 2000 (Tsutsui 1984). tion of better integrated pest management Most rice-producing land in developing coun- (IPM) schemesagainst mosquitoesarising from tries is flooded for relatively long periods of time the rice agroecosystemof the United States since 1977.These efforts are basedon the prem- and much of it subsequentlyproduces disease "mosquito vectors, including primarily snails and mosqui- ise that populations associatedwith toes (Myers 1984). In the United States, rice such systems are ... among the most likely acreagehas increased ca. 55% during the past targets for the immediate application of IPM decade (N. G. Gratz, 1981, unpublished data, strategies since they have adapted themselves World Health Organization). This expansion to habitats which are already largely under hu- was accompaniedin many areasby a significant man control and have becomerather dependent increase in the abundanceof several speciesof upon man and his land use practices for their (Olson ricelandmosquitoes (Paine 1983). continuedexistence" 1983). A major objectiveof the RMMP and 5-122 programs is to develop a better understanding l This researchwas conductedby the U.S. Depart- of the rice agroecosystemas it pertains to mos- ment of Agriculture, Agricultural ResearchService in quito production by means of systemsanalysis; cooperationwith the USDA, CSRS Southern Regional this work representsa contribution toward that Project 5-122 involving State Agricultural Experiment goal. This objective stems from the belief that Station personnel located in Arkansas, California, improved control of riceland mosquitoes will Louisiana, Mississippi, and Texas. 2 come from a more rigorous understanding of the InsectsAffecting Man and Animals ResearchLab- population dynamics of the mosquitoescoupled ResearchService, U.S. Depart- oratory, Agricultural of new and/or improved ment of Agriculture,P.O. Box 14565,Gainesville, FL with the development 32604. control and survey techniques.Systems analysis 3Department of Neurosurgery,College of Medicine, through computer simulation is a useful and University of Florida, Gainesville,FL 32610. necessarytool in the understanding of the com- SEPTEMBER1988 PsoRopHoRA qoLUMBTAESrvur,errou Moonl,-PeRr I 267 plex interactions between environmental and (WaterMod) was developedto utilize data on biological factors of the rice agroecosystem.The weather, agricultural practices, and soil charac- developmentof models allows an integration of teristics for a particular region to generate a the current base of knowledge; it highlights data set containing daily estimates of soil mois- areas where additional information is required ture and depth of water table for representative and can be used to develop estimates of popu- areas comprising the rice agroecosystem.This lation parameters that are difficult to obtain model was also designedto provide hydrologic from field studies. Additionally, becauselarge- inputs for additional simulation models of other scale field trials are expensive,simulation is an riceland mosquito species. important adjunct to the development, evalua- In the present work, WaterMod provides en- tion, and optimizationof IPM schemes.Finally, vironmental inputs to a second model which the simulation model of Psorophora columbia.e simulatesthe population dynamics of.Ps. colum- (Dyar and Knab) could be used in an expert biae (PcSim). The major attributes of PcSim systemdesigned to aid mosquito abatementper- include: 1) developmentaltimes of immatures as sonnel with control decisionson a day-to-day a function of daily water temperatures derived basis. from air temperatures,2) survival of immatures Major rice growing regions of the United as a function of larval density and habitat. States include California and the southern elapsed time since a particular habitat was statesof Texas,Louisiana, Mississippi, and Ar- flooded, and the presenceof surface water, 3) kansas.In southern rice areas,the dark rice field survival of adults as a function of soil moisture mosquito, Ps. columbiae,is the major periodic and host density, 4) the proportion ofdiapausing (or floodwater) mosquito pest of humans and and non-diapausingeggs deposited in the fall as animals. It was a primary vector of Venezuelan determined by day length and air temperature, equine encephalitis (VEE) when this virus was 5) survival of eggsas a function of the diapause introduced into the United Statesin 1971(Sudia status ofthe egg and by the duration and sever- and Newhouse1971, Sudia et al. 1971,Olson ity of freezing temperatures during the winter, and Newton 1973). In addition, it is known to and 6) the habitat and elevation where oviposi- adverselyaffect livestock production (Steelman tion occurs as a function of land use patterns, and Schilling 1977, Steelman et al. 1972, L973). irrigation and drainage events, weather as it Becauseof its importance in southern rice areas, influences the spatial and temporal distribution this floodwater specieswas selectedas a primary of surface water and soil moisture. and the target of the RMMP/S-122 investigators. searchingability of femalesto locate appropriate In the rice agroecosystemof the southern ovipositionsites. USA, Ps. columbiaebreeds in temporary pools This paper describesthe developmentofthese such as swale areasand grassyroadside ditches, models from a synthesis of the literature and soybeanand rice fields,and pastures(Schwardt unpublished data. Subsequent companion pa- 1939,Horsfall1942, 1955; Meek and Olson1976, pers will provide validation of the models pre- 1977;Curtis 1985,Welch et al. 1986).The winter sentedhere, present an in-depth analysisofthe months and dry periods are spent primarily in population dynamics of Ps.columbioe, and eval- the egg stage.Oviposition occurs directly on the uate IPM strategiesfor this mosquito. earth and at locations and elevations where soil moisture ranges between 75 and 100% of field T[aterMod capacity(FC) (Olsonand Meek 1977,1980). The Severalcomprehensive models are able to pre. number of generations of Ps. columbiae each dict surface and subsurface hydrology on the seasonis limited primarily by the frequency of basis of soil characteristics, topography, inundation of eggswith water of summer tem- weather, and agricultural practices (e.g.,Knisel peratures (Horsfall 1955). While many factors 1980,Skaggs 1980). These models were deemed influence larval and pupal survival, a requisite unsuitable for our purposes becausethey were to immature survival and subsequent adult difficult to use by non-specialists and, more emergenceis the continued presenceof surface importantly, required prohibitive amounts of de- water until near the time of eclosion(AI-Azawi tailed, on-site data. However, a common feature and Chew 1959). of the soil profile of rice growing regions, the The major objectiveofthe effort reported here shallownessof the impermeablehardpan (Hors- was to develop a comprehensive simu-lation fall7942), allowed us to developa simplified but model of the key elementsof the rice agroeco- adequatemodel for this agroecosystem. system that influence the population dynamics Complete data requirements (inputs) for of mosquitoes.Because the spatial and temporal WaterMod are: 1) soil porosity, 2) depth of distributions of surface water and soil moisture hardpan, 3) average depth and area of small, play such a decisiverole in the dynamics of Ps. irregular surface depressions which can hold coLumbiae,an essentially hydrological simulator water (hereafter referred to as ponding area and JounNA,r,oF THEAupnrclN Moseurro Cowrnor,AssocrerroN VoL.4,No.3 depth), 4) dailv rainfall and pan evaporation the asynchrony in flooding, harvesting, etc. Un- rates, and 5) for each of the 10 representative Iess all of first crop acreageis reflooded for a rice fields, the date for the initial flood (used in secondcrop, the proportion of Iand in this hy- field preparation) and the cutting date for the drologic categorywill decline during the year as "Nom- first crop harvest (seethe sectionentitled first and secondcrop rice acreagebecomes fallow inal irrigation practices" below for details). The following harvest. The unique feature of the proportion of annual rainfall that is lost from RICE sub-area is the influence of water man- the area in the form of surface runoff was used agementpractices: levees,irrigation, and drain- to compare predicted and observedrates as an ing; 2) NON-RICE-refers to land usedfor soy- evaluation of the model during validation (Focks beans, sorghum, etc., and pastures, fallow or et al. 1988a). non-second-croppedrice fields.NON-RICE also The output of WaterMod is a data set con- contains an area considerednot suitable for use taining, for each day of the year, estimates of by Ps. columbiaeithis area would include areas the daily water table depths in 12 locations, 10 of paving, buildings, lakes, woodedregions, etc. within RICE, and one each representing the This non-suitable area is assumedto constitute NON-RICE and DITCH/SWALE areas. We two-thirds (67%\ of the land in the NON-RICE have retained the use of inches in this paper hydrologic area. The unique hydrologic charac- becausesoil hydrology continues to usethe Eng- teristic setting NON-RICE apart is the phenom- Iish system of measurement and the inch is a enon of runoff, a result of rainfall in excessof convenient degree of resolution for use in the that required to saturate the soil and fill small models. surface irregularities; 3) DITCH/SWALE-an Both WaterMod and PcSim simulate a rep- area which represents the swale areas of pas- resentative l-ha area of the rice agroecosystem. tures, soybeanfields, etc., and roadside ditches This area is divided into 3 sub-areason the basis and irrigation canals. Here, rainfall can collect of unique factors determining water balance in and, as a result of runoff from NON-RICE, each location. While the hydrological sub-area depths can becomegxeater than the amount of named RICE corresponds with the rice field rainfall. As will be seenbelow, while the 3 areas habitat, the other 2 divisions, NON-RICE and have specific hydrologic differences, they also DITCH/SWALE, do not corresponddirectly to share certain hydrologic similarities. any particular agricultural habitat. The divi- A description of the water balance calcula- sions,schematically represented in Fig. 1, are as tions used to estimate water table depths in the follows: 1) RlCE-represents land currently 12 representativelocations are as follows: used for rice production. This area is further Watergain. When the water table is below the subdivided into 10 rice fields to allow modeline surfacein any area,the initial fate ofadditional

/:/ 'artNoFI OITCH'SWALE

KEY EI suRFAcE WATER El s:i?'"'fi:., @ SATURAIEOSOiL Fig. 1. Schematic representationofthe hydrology ofthe rice agroecosystemwith sourcesofwater shown. SEPTEMBER1988 PsofuopHonA qoLUMBTAE SIIr,tuutIot MODEL-PART 1 269 water, whether from rain, flooding, or runoff ample, given a soil porosity of 0.01 and a hard- from another area, is assumedto infiltrate the pan depth of 50" (727 cm), even under the driest zonebetween the water table and the surfaceby antecedentmoisture conditions, in areas where an amount determined by the soil porosity and excesssurface water is lost as runoff (see de- the depth ofthe water table.In the presentcase, scription of NON-RICE areabelow) all rains > using a value of0.01 for porosity in southwestern ca. 0.5" (1.3 cm) bring the water table to the Louisiana (J, L. Fouss, personal communica- surface permitting any accumulated error in tion), a 0.1" (0.25cm) rain would be expectedto model estimation to be eliminated. raise the water table by 10" (25.4 cm). Water in Water loss.Water lossescommon to all 3 sub- excessof that required to bring the water table areas include surface evaporation and evapo- to the surface then accumulatesin small irreg- transpiration. As a simplification, reduction in ular surface depressions(ponds). For cultivated rates of loss from the water surface due to in- Iands in southern Louisiana, the averagepond creasing plant canopy was assumedto be asso- depth is ca.0.7" (1.8cm) and the combinedarea ciated with corresponding increases in evapo- of such depressionsis ca. 15% of the total area transpiration losses. Daily loss from surface (J. L. Fouss,personal communication); for rice water and soil moisture was programmed to fields that were rutted by equipment during first occur at 95 and 80%, respectively, of the pan crop harvest,values for pond depth and ponding evaporation rate (J. L. Fouss,personal commu- areawere estimatedto be 4" (10.2 cm) andtSVo, nication). Water loss in NON-RICE occursas respectively(based on Meek and Olson 1977). runoffon the day an excessis received.In RICE, The destiny of water in excessof that required an additional water loss occurs when a field is to satisfy soil porosity and surface depression drained or rain occurs when the leveesare cut. storagerequirements varies by sub-area.Excess Rice fields were programmed to drain a maxi- in NON-RICE accumulates as runoff in mum of 3' (7.6cm) per day (unpublisheddata). DITCH/SWALE; becauseNON-RICE servesas In addition to evaporation losses,the DITCH/ a watershed for the smaller DITCH/SWALE SWALE losseswere programmed to occur at a area, water depth in DITCH/SWALE is pro- rate proportional to water depth (25% per day) grammed to increase 3' (7.6 cm) for each inch to reflect greater flow in ditches with increased of runoff. In RICE, when the leveesare intact, water depth. In RICE and DITCH/SWALE, excess water either from flooding or rainfall when water depth equalspond depth, water bal- simply accumulates as surface water, i.e., a ance is calculatedas in NON-RICE. Because flooded rice field. When the Ieveesare cut, RICE the hardpan is near the surface,WaterMod as- behaveslike NON-RICE. sumesno lateral subsurfacemovement of water. Normally, a simplified model such as Nominal rice ircigation practices. WaterMod WaterMod would be subject to a significant generatesindividual water tables for 10 hypo- accumulation of error over the course of a few thetical first and second crop rice fields based months if it were used to predict hydrology for upon antecedent rain and the dates of initial areaswith a deepimpermeable layer and subject flood for field preparation in the spring and first to sigrrificant lateral movement of sub-surface crop cutting for each field. As some fields are water. However, the proximity of the hardpan planted soonerthan others, this feature permits limits the significance of sub-surface water simulating the asynchrony of water events as- movement and results in even modest rains sociated with rice production. With the excep- bringing the water table to a known point, i.e., tion of the wider latitude granted for water the surface.This secondramification of a shal- depths during the second crop, the scheduling low hard pan allows the model to accurately of events was programmed to follow the prac- reviseits estimate of the depth of the water table tices of a grower in Jefferson Davis Parish, La. every time there is a significant rain. For ex- (D. Hollier, personal communication).Briefly the nominal scheduleis as follows:

Date Day Agricultural Event WaterMod March 1 60 Flood for water leveling and Water depth increases2' (5.1 seed bed preparation in cm) per day to 4" (L0.2 cm). fields which have been Thereafter, additional water plowed earlier in the year. added when depth declines be- Water depth maintained at low 2" (5.1 cm) from evapora- 3-4" (7.6-t0.2 cm) for ca. 3 tion; depths >4" (10.2cm) from weeks. rain are programmed to be lost as over flow. 270 Jounxer,oF THEAunnrceN Mosqurro CoNrnor,Assocrnrron Vor,.4,

Date Dq Agricultural Event WaterMod March 25 84 Seedingby airplane.

March 26 6D The field is now drained but Drain programmed to reduce rnaintained in a moist con- water depth a maximum of 3" dition with rain or flushes if (7.6cm) per day until water ex- required to prevent the seed ists only in surface ponds. Ac- bed from cracking. tual water table (WT) depth then varies as a function of rainfall, irrigation, and pan evaporation rate. If, from rain, water depth ) pond depth, WT set to 0 (overflow). If WT <-30" (76 cm) from drying, WT set to 0 (flush).

May 1 127 First permanent flood put Water depth initially set to 3" on; depth maintained be- (7.6 cm). Additional water tween 2 and 3" (5.1 to 7.6 added as evaporation reduces cm) by rainfall and/or addi- depthbelow 2" (5.1cm).Depths tional irrigation water as re- >3' (7.6 cm) from rain are pro- quired. grammed to be lost as overflow the next day.

May 21 t41, Depth of permanent flood Water depth initially increased increasedto a nominal 6" to 6" (15.2 cm). Evaporation (15.2cm) and maintainedat Iosses trigger water depth in- this level via additional creaseto this depth at 4' (10.2 flooding or overflow when cm). Transient depths greater depths are 14" (10.2 cm) or than spillway height (6" or 15.2 >6' (15.2 cm), respectively, cm) due to rain are reduced to as a function of rain and spillway height the following evaporation. dav.

July 5 186 Levees cut to drain field As indicated for March 26 but prior to harvest. without addition of water.

July 18 200 Harvest. Pond depth increased to 4" (10.2cm) becauseof tire ruts.

July 20 202 Two days afberharvest, field As per the permanent, full flood is reflooded and water is al- of first crop except subsequent lowed to range betweenbot- flooding triggered only when tom of 4" -tire ruts (-10 cm) WT is slightly below bottoms of and 6" (15.2cm) abovepan tire ruts. Maximum depth is for 92 days for secondcrop. limited by spillway height of 6" Water depths not main- (15.2cm). tained as accurately as in first crop.

October 1 294 Levees cut to drain field See comments for WaterMod prior to secondcrop harvest. for July 5.

October21 315 Secondcrop harvest. See comments for WaterMod for July 5.

PcSim daily record of the estimated numbers of indi- viduals within each of the age classesof each We simulated the population dynamics of Ps. Iife stagefor each ofthe 12 representativeareas columbiaeby using the computer to maintain a of the rice agroecosystemmentioned previously SEPTEMBER1988 Psonop$ofuA )'LUMBIAE SIuuLenoN MoDEL-PART 1 271

EMERGENCE

ovtPostTtoN (o)

1)2***,.,

--ZorrcHlswALE AREA --:z NoN.RtcEaREA

KEY r,2,3,... OAILYAGE CUSES COMAINEOAGE CUSES x VARIABLEDEVELOPMENT TIME (D€PENOENT UPON TEMPEBAruRE) FECUNDIW(13) EGGS/FEMALE) SITEAND ELEVATIONOF OVIrcSITION(OEPENOENT ON SOILMOISruRE AND WATEF DEflH, DIAPAUSE@NDIION OEPENDSON PHOTOPERIODAND TEMPERATURE) H NON,DIAPAUSINGEGGS (HATCH WHEN INUNOATEO}

DAILYTRANSITION RATES (SUBSCRIPIS a, €, UpREF€R TO ADULTS,EGGS, ANO LARVAE/ruPAE, RESPECTIVELY) sa NOMINALADULT SURVIVAL (MODIFIED 8Y SOILMOISTURE ANO RATIO OF BLOOO.SEEKINGFEMALES TO HOSTS) se EGGSUFVTVAL (VAB|ES WtTH DTAPAUSE CONDTTTON OF EGG) silp SURVIVALOF ACTIVEIMMATURES (FUNCTION OF HABITAT,EUPSED TIME SINCE INUNDATION. LARVAUPUPAL OENSIW. AND AGEOF IMMATURE) Fig. 2. Schematic representation of the life cycle of Psorophnracolumbiae in the rice agroecosystem.The boxes portray the various daily or combined age classes accounted for by PcSim. Computationally, they represent array locations within the computer which contain the absolute number of individuals per hectare. For the sake of clarity, a secondegg array, identical to the one shown but containing diapausing eggs,is not depicted.

(10 rice fields of RICE, and the DITCH/ temperatures,the wide swings in survival due to SWALE and NON-RICE habitats; seeFig. 2). daily fluctuations in environmental parameters, The accounting was made dynamic by recur- and the need to predict daily populations for sively updating, in each area, each of the daily abatementpu{poses, a resolution ofone day was age class with the product of the number of usedin the model. individuals in the previous ageclass and transfer In addition to the daily hydrologic inputs from coefficients such as daily survival and fecundity, WaterMod, PcSim requiresdaily maximum and e.g., the number of 3-day-old females today is minimum air temperatures and, for each year of equal to the number of 2-day-olds yesterday simulation, the proportion of land devoted to times the proportion surviving from one day to first and secondcrop rice. These proportions are the next. The developmentof active immatures usedto determine the relative area of the NON- was handled somewhat differently and is de- RICE hydrologic area; as discussed below, scribedbelow. Much of the balanceof this paper DITCH/SWALE area is fixed and does not will discuss how the transfer coefficients were changewith changing RICE acreage. determined and how they are modified slightly on a daily basis as a function of environmental FACTORS INFLUENCING IMMATURE conditions. SURVIVAL AND DEVELOPMENT Termed appropriately a "dynamic life table" model by Haile and Mount (1987), this method Temperature. Using daily data from the has been used frequently in medical entomologl spring, summer, and fall of 1984from U.S. Geo- simulation studies (e.g., De Moor and Steffens logical Survey (USGS) and National Oceano- 1970, Haile and Weidhaas 1977, Focks et al. graphic and Atmospheric Administration 1978,Mount and Haile 1987).Because of the (NOAA) weather stations located within a rice rapid developmentof immatures at summertime field in Jennings,La. (D. Hollier, personalcom- 272 JouRNlr, oF TrrE AMERICANMosqurro CoNtRol, AssocrlrroN VoL. 4, No. 3

munication), the following expression was de- in daily simulation calculations.These daily val- veloped for use in PcSim to predict a daily ues are calculated for each of the 10 fields of average rice water temperature (AvgWater) RICE and for the NON-RICE and DITCH/ from the same day's maximum (MaxAir) and SWALE areasindividually to reflect local vari- minimum air (MinAir) temperature(.C): ations in the modifying factors listed above. The value usedfor MaxOSi in first and second A,vgWate4 : * 9.65 + 0.30 MaxAir. crop RICE is 0.60. This value was extrapolated from field data on + 0.48 * MinAirt R, : 0.80 the decline in dipper counts of Ps.columbioe between 2nd and 4th instars in For lack of data to enable us to do otherwise. second-croppedrice fields immediately after the this expressionwas also usedto predict NON- initial reflooding (Mclaughlin and Vidrine RICE and DITCH/SWALE water tempera- 1986);the calculation was made after correcting tures.The proportion oftotal developmentthat for age-relatedchanges in dipperefficiency (An- occurredeach day was calculatedusine the en- dis et al. 1983). These data were obtained at zyme kinetics model of Sharpeand DiMichele larval densities below that where densitv de- (1977) utilizing coefficientsdeveloped for ps. pendentmortality is thought to occur(Al-Azawi columbiaeby McHugh and Olson (1982).This and Chew 1959, Andis and Meek 1985) and, model predicts the observed development not most importantly, prior to colonization with only along the linear region at mid-range tem- predators (REM, unpublished data). This esti- peratures,but also along the nonlinear region at mate of 0.60 for MaxOSi in RICE is in close elevated temperatures with a high degree of agreementwith another field estimate of overall accuracy. The development time of larvae and immature survival under predator-free condi- pupae was made a function of temperature in tions: Using predator-exclusioncages in south- the model by accumulating this daily proportion western Louisiana rice fields, Andis and Meek of development until emergenceoccurred at a (1985) observed overall immature survivals to sum >0.95.With incrementsof ca. 70 to 20% of range between 0.50 and 0.75; the higher value total development typically occurring on each was associatedwith the period following reflood- day, programming emergenceto occur at a value ing of secondcrop rice. Values for MaxOSi for slightly less than one resulted in emergence the other areas,NON-RICE,0.12 and DITCH/ occurring on the average at 1.0 because the SWALE, 0.16,were a result of discussionswith accounting in PcSim is conducted on a daily M. D. Andis and J. K. Olson. basis. b. Predation. A number of workers have indi- Suruiual factors, Modeling the factors influ- cated that the predator complex is a major encing immature survival was difficult due to a source of irreplaceable mortality for immature paucity of information in the Iiterature dealing ricelandmosquitoes (Service 1977, Miura et al. directly with Ps. columbiaeimmature survival. 1978,Mogi et al. 1980a,1980b, 1984). Andis and The methods and factors employedin the model Meek (1985)estimated in the caseof Ps.colum- representa synthesisof the literature (Al-Azawi bioethat predation was also the most significant and Chew 1959, Andis and Meek 1985. Mc- mortality factor in southwesternLouisiana rice Laughlin and Vidrine 1986), discussions with fields, accounting for ca.80% of all losses.Un- Drs. J. K. Olson and M. D. Andis, literature Iike mosquitoes which are typically found in reports on permanent-water mosquitoesbreed- permanent water, temporary-pool-breeding ing in rice fields and temporary pools (Service mosquitoes are commonly in largest numbers 1977,Miura et al. 1978,Mogi et al. 1980a,1980b, immediately after inundation when the aquatic 1984), and in some instances,the results of habitat has yet to be (fully) colonizedby preda- simulation itself. tors. a. Nominal, habitat specific. As virtually all To reflect an accumulation of predators sub- literature presents immature survival estimates sequentto flooding in RICE and the attendant on a hatch-to-emergency basis, we began by reduction in survival, MaxOSi is linearly decre- assigninga maximum probability of overall im- mented over a 28-day period subsequentto in- mature survival (MaxOSi) from match to eclo- undation to a minimum value of 0.12or 20% of sion for eachhabitat. This value reflects habitat- the starting value of MaxOSi. We would expect specific biotic and abiotic factors and does not Iarvae to experiencethis lower survival in older include the effects of predation, density depend- and previously flooded fields where additional ent mortality, or death due to the disappearance irrigation water or rain had raised water levels of surface water prior to eclosion.As presented to the point of inundating eggspreviously laid in the following paragraphs,MaxOSi is reduced abovethe former water line. The value of 0.12, by these factors to produce a net overall imma- representing expected survival one or more ture survival which is in turn converted, based months after the permanent flood, is a some- upon the proportion of development occurring what higher value than an averageof 2 tentative on each particular day, to a value which is used field estimates for Ps. columbiae: 1) 0.10 for SEPTEMBER1988 PsonopHonA coLuMBrAEStvulluoN MoDEL-PART 1 overall immature survival in late, first crop rice (M.D.Andis, personalcommunication) and 2) 0.03for late, secondcrop rice (Andis and Meek 1985).The final, decrementedvalue of 0.12 for overall survival in RICE is, however, in almost perfect agreement with other reported values ranging between 0.01 and 0.07 (average0.035) for anopheline survival in permanently flooded rice fields (Service1977, Mogi et al. 1984,C. H. Schaefer, personal communication) when cor- rection is made for the unusually short devel- opment period of Ps. columbiae:an overall sur- vival of 0.12 translates into a daily survival of Fig. B. The ,"rurio".ill"'u'u'#enthe densityof 0.654 at a summer-time development rate of 5 activeimmatures and the densitydependent factor, a days,viz., O.!2Q/6).The correspondingdaily es- term usedin calculatingdaily immaturesurvival. timate for an overall survival of 0.035 is 0.658 if a field estimate of immature development of 8 nication) in ditches and swale areasas the hab- days is used (e.g., Anophcles qua.drimaculattn itat contracts from water loss. We therefore Say, J. A. Seawright,personal communication). createda density dependentfactor which ranges In the other 2 hydrologic areas, MaxOSi is between 1.0 at low larval densities (<1,000/m') similarly reduced daily until a minimum value to 0.01at higher densities(>5,000/m') (see Fig. of ca.20% of the initial value is reached.In all 3). In the model, these values rarely result in areas, each occasion that surface water disap- any density dependent reduction in Iarval sur- pears and inundation subsequently occurs, vival in rice fields. The factor does becomeop- MaxOSi is reset to its initial value and the daily erativein the DITCH/SWALE and NON-RICE decline begins anew. As would be expected,the areas where larval densities can occasionally frequency of the inundation/drying cycle and becomevery high. hence, the frequency of recolonization and, d. Drying of temporary pools. ln addition to therefore, reset in PcSim, is greater in NON- the factors mentioned previously, temperature, RICE than DITCH/SWALE and least in RICE. because it affects speed of development, and Becauseof differencesin the duration of surface hydrologic factors, becausethey determine the water, the most common value for overall sur- duration of surface water, interact to determine vival in RICE is 0.12; in contrast, the most the fate of a cohort of larvae. For example, Al- common values for the other 2 areas are much Azawiand Chew (1959)reportedthat Ps.confin- closerto their initial and uppervalues, MaxOSi. nis were found only during the months of June c. Density dependence.AI-Azawi and Chew through September in date groves in southern (1959) observedthat the size of 4th instars of California. The principal temporary pools were Psorophora confinnis (Lynch-Arribalzaga) irrigation lanes which were flooded on a bi- (membersof the Psorophoraconfinnis complex, monthly basis. It was only during a portion of as is Ps. columbiae) tended to vary inversely the year that temperatureswere high enough to with larval density in irrigation lanes of date allow developmentbefore the habitat dried out groves (an area corresponding to the DITCH/ about 5 days after flooding. Al-azawi and Chew SWALE areain the presentmodels). They con- (1959) observed that 3rd instars stranded on cluded that this probably indicated the effect of moist soil invariably died, probably from star- competition at high larval densities.In this hab- vation; when 4th instars and pupae were itat, Iarval densities ranged between 600 and stranded,44 and 91%,respectively, were able to 12,600/m2based on dipper counts converted to complete their development. absolute estimates after Andis et al. (1983). Suruiual calculations.Because initial overall While larval densities as high as 2,500-4,000/ immature survival varies by habitat and larval m2 have occasionallybeen observedin rice im- densities and elapsedtime since inundation are mediately after the reflooding ofa harvestedrice not necessarily the same for each site where field (Chambers et al. 1979, Andis and Meek larvae are present, daily survival values are cal- 1984, Mclaughlin and Vidrine 1986), a compi- culated individually for each ofthe 12 locations. Iation of all density estimates indicates that an Computationally, they are equal to the product averagemaximal value would be ca. 1,000/m2. of the density dependent factor and the decre- From laboratorystudies, Andis and Meek (1985) mented value of the appropriate MaxOSi raised concludedthat, at densities observedin rice, to the power of the proportion of development intraspecific competition had no effect on larval that occurred on that particular day as a func- survival. However, extremely high densities tion of temperature.If prior to emergence,sur- have been observedin Louisiana (unpublished face water had disappeared resulting in the data) and Texas (J. K. Olson,personal commu- strandingof immatureson moist soil, the num- JouRNer,oF THEAupnrcau Mosqurro CoNtRor,Assocrerron VoL. 4, bers emerging may be further reduced (depend- ing on the degreeof development that had oc- curred prior to stranding) by the amounts indi- cated in the preceding paragraph. Sex ratio of emergingmales and femalesis set to 1.11 and : ,^" 0.89,respectively (1.25) after Andis and Meek '6 (1985). E FACTORS INFLUENCING ADULT SURVIVAL AND REPRODUCTION Gonotrophiccycle and fecundity. Due to a lack 73 IOO of data, the length of the gonotrophic cycle of Soit moistur. (t fi.td copocity) Ps.columbi.ae in PcSim is modelledas constant Fig. 4. The relationshipbetween soil water condi- and independent of temperature (see Fig. 2). tions in the NON-RICEarea and the adult survival soil moisture One-half of the femalesare assumed factor,a term usedin calculatingdaily to oviposit adultsurvival. during the 4th day of adult life and one-half during the next; subsequentoviposition cycles are 3 or 4 days in length. Blood meals are as- that a cohort of Ps.columbioe adults disappears sumedto be taken on the day of emergenceand more rapidly when humidity conditions are low. on every other day thereafter; this results in 3 On this basis, we developeda factor related to meals being taken prior to the first oviposition soil moisture (Fig. a) which rangedbetween 0.97 and one meal during each subsequentcycle. It when the soil is dry @C < lS%) to 1.03 when is further assumedthat all females are insemi- surfacewater exists at a depth of >0. The water nated and lay a fixed number of eggsper gono- condition in the NON-RICE area was usedbe- trophic cycle (130). The valuesused for gono- causeit usually comprisesthe largestarea within trophic cycle, blood feeding, and fecundity were the rice agroecosystem.We considered using derived from discussionswith J. K. Nayar and saturation deficit as a modifier of nominal sur- from the literature (Horsfall 1955,Chapman and vival as this would allow the roles of temperature Woodard 1965, Nayar and Sauerman 19?5a, and humidity to be integrated.However, because L975b,1977); it is not possibleto completely adults typically rest in low-lying vegetation reconcilethe observationsof the previous work- (Schwardt 1939, Edman and Bidlingmayer ers with field observationsby J. K. Olson (per- 1969), preferring the underside of leaves or sonal communication) that females spend the stemsnear the ground (Horsfall 1955),and se- first 24-48 hr of adult life mating and migrating lect resting sites apparently in responseto soil without blood feeding. moisture during daytime when the deficit is Nominal adult suruiual. In a fashion similar usually greatest (Al-Azawi and Chew 1959), we to that used in the calculation of larval survival, reasonedthat adult survival is probably more the determination of daily adult female survival highly correlated with soil moisture than satu- begins with a nominal value (0.68) which is ration deficit. influenced by environmental parameters. The Influcnce of host density on adult surviual.The actual value for daily female survival used in purpose of including a second environmental calculations is the product of this nominal value variable was to make the model responsive to and 2 environmental factors that reflect the the influence of host density. As in the area of influence of soil moisture and host density on immature survival, there is very Iittle informa- survival. In simulation runs, during peak popu- tion useful for the quantification of this rela- lations when host density influences adult sur- tionship. We do know that cattle and to some vival (see below), daily survival values range degree,horses, serye as primary sourcesof blood between 0.59 and 0.63. These values are in fair mealsin Texas ricelands (Kuntz et al. 1982)and agreementwith unpublished estimates from 1) that the availability ofboth blood and carbohy- cagedinsects (J. K. Nayar, personalcommuni- drate strongly influence fecundity and adult sur- cation), 2) mean daily declines in sweep net vival of Ps. columbiae on a nutritional basis captures around cattle (unpublished data), and (Horsfall 1955, Nayar and Sauerman 1975a, 3) declinesin the daily capturesof8 New Jersey 1975b,1977).It is clear that there is a relation- light traps located in Acadia Parish, LA during ship betweenvariations in host density and mos- 1984and 1985(unpublished data). Male survival quito abundance: Al-Azawi and Chew (1959) is set to a constant 0.60 per day (J. K. Nayar, reported that adult densities averaged0.8/m2 in personalcommunication). irrigated groveswithout cattle as opposedto 9.8/ Influence of soil moisture on adult suruiual. m'(a 13-foldincrease) when (an undefinednum- The creation of a moisture factor was based on ber of) cattle were present. Meek and Olson the observation (J. Billodeaux, M. D. Andis, J. (1977)reported similar results: the averageden- K. Olson, and others, personal communication) sity of Ps. columbioeeggs was ca. 5 times greater SEPTEMBER1988 PsoRopHoRA qoLUMBIAE SIlrur,erroN MoDEL-PART 1 275

in fields with cattle than in fields where cattle fined cattle in Southern Louisiana (estimated were absent. A similar trend was reported by from Steelman et al. 1972,1973). Chambers et al. (1981) and Williams et al. Ouipositionsite selection.There are recordsof (1983).Mclaughlin and Vidrine (1987),using Ps.columbiae adults being found 8 or more miles regression,quantified the relationship between from their site of emergence (e.g. MacCreary dipper counts in rice fields and cattle density and Stearns 1937). However, a review of all within a l-mile radius of the larval habitat; they reports on the movement of this speciesindi- observeda ca. 2-fold increase in larval density catesthat typically, some90% of all adults never with an increaseof 10 cattle/mi' (3.9/km'). We more than a few miles from their larval also know that host defensivebehavior plays a habitat (Schwardt 1939,Horsfall L942,White- role in limiting accessto blood when Ps. colurn- head 1957).Examination of aerial photogaphs blae populations are high (HorsfaI| 1942, tg55; ofthe rice-growingregion of southernLouisiana Gahan et al. 1969). Finally, it would be under- suggestthat all habitat types are usually repre- standable if Iower host densities were to some sented within any area circumscribed by the degree associated with lowered rates of host flight range of Ps.columbioe (R. E. Mclaughlin, finding. unpublished data). On this basis and because The creation of a host density factor, the the consensusof many workers (Meek and Ol- relationship betweenadult survival and the ratio son 1976, Williams et al. 1984, Rankin and of blood-seekingfemales to hosts (biters/host), Olson 1985,Welch et al. 1986,Welch and Olson necessarilyrepresents a compromiseand a sim- 1987)is that, given adequatesoil moisture, there plification of what is known about the role of are locations within virtually all habitat types nutrition on fecundity and adult suwival for Ps. found in the rice agroecosystemthat are subject columbiae.We have assumed,as did the mos- to Ps. colurnbioeoviposition, we have assumed quito models cited previously, that sugar avail- the adult population to be mobile and capable, ability in the environment is not a limiting independent of their site of emergence,of ovi- factor. Furthermore, as we cannot partition the positing in any location as a function of water separate effects of host availability on fecundity depth, soil moisture, ovipositional habitat pref- and survival, we have chosen to attribute erence,and Iand use.We believea minor excep- changesin reproductive successas a function of tion to the adequacyof the searching ability of biters/host on the basis of its effect on adult Ps.columbiae females (discussed in paragraph c. survival alone, and not on its known role in below) occurs during dry periods in areaswhere influencing fecundity. We have assumedthat a rice acreage is very low and rice is the only rise in the density of biters per host is accom- habitat available for oviposition. panied by a decreasein female survival due to a. Di.stribution as a functinn of relntiue areas some combination of inadequate nutrition and and soil rnoisture.A soil moistwe >_757oFC, mechanical factors, viz., missed or incomplete correspondingto a water table that is within 80,, blood meals, death due to the increaseddefen- (76.2 cm) of the surface, is reported by Olson sive behavior of the host, and changes in the and Meek (1977) to be the range of adequate rate of host finding (Fig. 5). The values used in moisture. When all 3 hydrologic areasare suffi- Fig. 5 were developed by simulation so as to ciently moist to attract oviposition, PcSim par- producethe differencesin population abundance titions the total oviposition available on that similar to those seen in the field studies men- day as a function of each locations' relative area tioned above.These values are within the range in the environment.There are someminor ex- of Ps. columbiqe attack rates observedon con- ceptions to this as noted in the following para- graph. Should RICE be the only moist area, all ovipositionis placedin RICE. If NON-RICE is the only dry area,then oviposition is distributed among the RICE and DITCH/SWALE areasas a function of their 2 combined areas.The areas are calculatedas follows: 1) The amount of land planted to rice changesfrom .) 1.O first crop to second crop and from year to year dependingon a host of factors such as agricultural prices and govern- ment set asidepolicies; as a consequence,values for the proportion of Iand in first and second crop RICE for each year to be simulated are 5000 15000 entered as data; 2) the DITCH/SWALE area is Bit.B ,/ holt onimot set to a constantof l0% of the total area (J. L. Fig. 5. The relationship between blood-seekinsfe- Fouss,personal communication); 3) the area of males per host and the host density factor, a tlerm NON-RICE is the balanceof the land.However. used in calculating daily adult survival. for the purposes of distributing oviposition, a 276 JounNer, oF THE AMERTcANMoseurro Cor.rrnor,AssocrATroN VoL.4,No.3

portion of this area (called NOT-SUITABLE by vertical elevation in increments of one inch. and assumedtobe 67% of the NON-RICE area) The elevation ofegg depositionvaries by habitat is set asideas unsuitablefor utilization by ps. and in responseto soil moisture or the depth of columbiae regardless of soil moisture. This is surface water as the casemay be at the time of becauseNON-RICE includes,in addition to as- oviposition.In NON-RICE, eggsare expectedto ricultural fields,areas ofpaving, buildings, Iakei, be oviposited at the 0 elevation (i.e., within the lane without vertical relief, etc. If second crop small surface depressions) provided adequate rice acreage is less than first crop, the non- surface water and/or soil moisture exists. second-croppedacreage is movedfrom the RICE DITCH/SWALE oviposition occurs equally at to the NON-RICE category as a rice field with elevations0 to 4" (10.2 cm) above the water cut levees behaves hydrologically like NON- surface or (providing adequatesoil moisture) at RICE where, except for surface storage in tire the 0 elevation if there is no surface water (ex- ruts (correspondingto the irregular surfacestor- trapolating from Olson and Meek 1980). In ageponds of other NON-RICE areas),water is RICE, eggs are also uniformly laid between 0 not assumedto accumulate. and 4" (10.2 cm) abovethe water surface;with- b. Distribution rnodifiedby a preferencefor rice out surfacewater but with moist soil, 95% of the fields. An exception to a strict allocation as a eggsare laid in surface depressions(i.e., at the function of relative areaswhich are moist stems 0 elevation correspondingto oviposition on the from various accounts of a preference of Ps. surfaceof the pan) and 5% are placeduniformly colurnbioeto oviposit in harvestedfirst crop rice on the bottom 4' (10.2cm) of the levee(Olson fields. This preferenceis due to one or more of and Meek 1980).If the soil is dry, oviposition the following factors: 1) Increased relief is still occurs in RICE but only at the bottom of present as tire ruts from harvesting machinery the levee as this structure apparently retains "soil (Meek and Olson 1977),2) the becomes water and remains moist at the base during firmer sooner, and the surface layer dries to otherwise dry conditions (Olson and Meek dustiness much slower" and hence remains ad- 1980). equately moist for an extended period when d. Lossof ouipositiondue to dry soilconditinns. comparedto other areas(Horsfall 1955),also 3), As describedto this point, if the soil in all areas the fermentation of soil and rice stubble may becomestoo dry for oviposition (or everywhere influence site selection (Gerhardt 1959). For except RICE) and there are females ready to thesereasons, RICE was given a 30% preference oviposit, all eggsare programmed to be laid in for the month following harvest, a value deter- RICE. When rice acreageis not small, this prob- mined through simulation by comparing the rel- ably reflects reality: the searchingability of Ps. ative contribution ofthe various areas.Because columbiaefemales is sufficient to locate the only WaterMod predicts soil moisture, the relative available oviposition sites under dry conditions, attractivenessof rice vs soybeansas a function the rice fields. However under these circum- of soil moisture reported by Welch and Olson stances,when rice is a small componentof the (1987) required no special treatment. A second system,we believe it reasonableto assumethat exception arises from the need to give NON- an increasingproportion of females,after having RICE and DITCH/SWALE slightly fewer eggs Ieft their site of emergence to obtain blood, than would be expected on a strictly relative either die prior to locating (relocating in the area basis because of qualitative reports that case of females emerging from RICE) the rice roadside and drainage ditches were not as im- habitat or simply oviposit in inappropriate lo- portant an oviposition site throughout the year as rice (Horsfall 1942,Meek and Olson 1977). For example, if 50% of all moist land suitable for oviposition was RICE, 50% of all eggswould be oviposited in RICE without an ovipositional preference;a 10% preferencewould result in 50 * 10 or 60% of the oviposition occurring in RICE. For simulation, throughout the year, RICE is given a |Vo prcfercnceover the other 2 t areasand this is increasedtemporarily after first crop harvest for 36 days to 30Vo. c. Vertical distribution on leueesand ditches.

In addition to maintaining an accounting of the 0 0.15 1.OO distribution of eggsby area (the 10 representa- Propodon of lond plontcd to acc tive rice fields within RICE, NON-RICE, and Fig. 6. The relationship betweenthe lossofpotential DITCH/SWALE) and by diapause condition oviposition as a function of the proportion of land in (seebelow), PcSim keepstrack of egglocations RICE under dry soil conditions. Seetext for details. SEPTEMBER 1988 PsonopHogA gqLUMBIAESruur,luolt Moonr--Penr 1 277 cations. Figure 6 depicts the relationship used in PcSim to simulate this loss of potential ovi- position as a function of the proportion of land in RICE: Whenever RICE acreageexceeds 15% of the total, all oviposition occurs as described previously.However, under the conditions of dry soil as indicated above,an increasingpercentage of oviposition is lost when RICE acreagede- tr clines from LSVoto zerc. Diapausecondition, suruiual, and hatch of eggs. 0.o1 As an overwintering mechanism, Ps. colurnbine produces increasing proportions of diapausing con"""uti'l dcar* (c) doy: *l], ,-*tn eggsin the fall in responseto temperature and Fig.7. The relationshipbetween the product, degree photoperiod experiencedby the larvae, pupae, ('C) daysbelow freezing (DDBF), and a freezingfactor and adults.aPcSim inputs as data the proportion which is usedin calculatinga reducedoverwintering of eggs deposited in the diapause state; this eggsurvival. See text for details. proportion is determined manually using a model (Delorme et al. 1987) written for the of -5, -6, and -2"C, correspondingto 13 DDBF Texas and Louisiana area which integrates the would be sigrrificant as temperatures are low role of date and averageair temperature to pre- enough and for a period sufficient to fteeze a dict the proportion of eggswhich are laid in the proportion of the eggsin the environment. For diapausecondition each day. At the Iatitude of each day or series of consecutivedays with av- southern Louisiana, this proportion typically erage temperatures remaining below freezing, goesfrom 0 to L00%diapause during September. the number of degree days below freezing is Becauserice and other crops are usually ro- calculatedand the correspondingfreezing factor tated each year, we assumethat tillage destroys (as indicated in Fig. 7) is applied to the nominal or buries 85% of all overwintering eggsin RICE overwintering survival values presented previ- and ca. 67Voin NON-RICE. For the remainder ously. The values used in Fig. 7 do not result in ofthe eggs,survival rates were set to 0.9985per egg loss for most years in southwesternLouisi- day for diapause eggs (a value slightly lower ana. We recognize that this treatment is an than that estimated by D. K. Lee, Texas A and oversimplification of the real situation, however, M University, personal communication) and as elsewhere,data are not yet availableto permit 0.9970per day for non-diapauseeggs. Depending a more refined or sophisticatedtreatment. upon rainfall, it is possible for diapausing eggs Each day, all non-diapausing eggs4 or more that were depositedat the higher elevations not days old and therefore old enough to have com- to be flooded and hatched until late spring or pleted embryonation (Horsfall 1942), are as- early summer. To reflect the exhaustion of en- sumed to hatch when submerged (see Fig. 2). (after ergy reserves Olson and Meek 1979),etc., While we suspectthat temperature and increas- of older diapausing eggs, diapausing eggs are ing day length would influence the date ofbreak- assumedto survive at the 0.9985rate until mid- ing diapause,until we have data to allow a more June, at which time their survival rate changes sophisticatedtreatment, diapauseis assumedto (0.9970/day). to that of non-diapausingeggs break on the first of April when both types of Based on the premise that prolonged sub- eggsare permitted to hatch (J. K. Olson, per- freezing temperatures are lethal to eggs (J. K. sonal communication). A somewhat arbitrary Olson, personal communication), overwintering date of October 31 is set as the last date for non- survival was reducedby a factor reflecting both diapauseegg hatch (unpublished data). the duration and the severity of freezing tem- peratures,e.g., the number of degreedays below SUMMARY OF PROGRAM FLOW freezing (DDBF) (see Fig. 7). For example,a AND MODEL OUTPUT single day with an average air temperature of -3'C between warmer days would correspond WaterMod and PcSim were written and com- to 3 DDBF. Looking at Fig. 7 indicates that this piled with a version of BASICs that allowed the would not be expected to reduce egg survival. creation of highly modular and structured code However, 3 consecutivedays with temperatures on an IBM-AT6 personalcomputer. Our com-

a Delorme, D. R. 1984.Egg diapausein Psorophnra 5 BetterBASIC (version 2.1) by Summit Software columbio.e(Dyar and Knab): Developmentof a predic- Technology, 106 AccessRoad, Norwood, MA 02062. tive diapauseinduction model. Unpubl. Diss. Texas A 6International Business Machines Corporation, and M University. 95 p. P.O. Box 1328,Boca Raton,FL 33432. 278 Jounxer,oF THEAusnrcnN Mosgurro CoNrRor,AssocratroN VoL.4,No.3

puter was equipped with 3 M-bytes of random previousyear, a seriesof sequentialyears may accessmemory (RAM), a 60 M-byte hard disk, be simulated. An initial egg distribution was and an Enhanced Graphics Adaptor (board) developedby starting the model with a nominal (EGA) and EGA monitor; the programs may be egg distribution (100 eggs/m' in DITCH/ "suitable" run on any lBM-compatible machine with a SWALE and the areaof NON-RICE); minimum of 512K-bytes RAM and EGA sraph- the resulting eggs on the last day of this run ics capability. Lotus 1-2-3?was used for ihe were then used to initialize the next year, and creation of various input files and the display this year's the next, etc. After 5 years, the dis- and plotting of output files. Program listings of tribution of overwintering eggs remained con- both models may be obtained by contacting the stant from year to year and reflected the influ- senior author. enceof host density, weather, and land use.This For each year to be simulated, WaterMod is distribution is usedto start a seriesof simulation first run to create a file which contains dailv runs for the years of interest. estimates of water table depths for the 12 areas On a daily basis, PcSim reads as input a file used to represent the rice agroecosystem.This containing maximum and minimum air temper- output data set from WaterMod is later read as atures, and the proportion of eggsdeposited in input by PcSim where it is used to determine the diapause condition and the hydrologic file where oviposition occurs, where eggs hatch, producedby WaterMod. For eachday, the model where larvae are stranded, etc. At the start of then proceedsas follows: For immatures, 1) lar- each run of WaterMod, values for certain pa- vae and pupae are agedas a function of temper- rameterswhich will remain constant throughout ature using an enzyme kinetics model, 2) the the year are input, e.g., soil porosity, hardpan numbers of active immatures are adjusted in depth, ponding area and depth, and the dates each area as a function of habitat, age and for initial flooding and first crop cutting for each presenceof surface water, and larval density, 3) of the 10 representativerice fields. To model the hatch and emergenceare programmed to occur population dynamics around a single town or where appropriate, and 4) eggsare agedaccord- location, dates for 10 nearby rice fields are used; ing to their diapausecondition and the time of in this case, a particularly large field may be the year. Next, 1) oviposition occurs as a func- representedby 2 or more of the 10 hypothetical tion of current land use, soil moisture, and ovi- fields. Alternatively, to simulate the overall dy- position site preference and 2) adults are aged namics of a large area such as a parish, the dates using a common survival value which is a func- for each of the 10 representativefields will cor- tion of moisture and host density. respondto observeddates when 10, 20, .. . , 90, For the purpose of understanding the actual and 100% ofthe acreagein the parish have been population dynamics of Ps. columbioe,absolute flooded initially or cut. After initialization with densities of the various life stagesas predicted values for the parameters listed above, by PcSim are examined. However, for the pur- WaterMod then sequentially reads as input, a poseof comparing expectedadult densities from file containing observedvalues for rainfall and the model with light trap data from the field for pan evaporation for each day of the year. model validation, the influence of several addi- Similar to WaterMod, the output of PcSim is tional factors are considered: 1) The frrst is a also a data file containing daily estimates, in physiological factor. The model assumesthat this case, estimates for the entire year of the newly emerged adults (e.9., <24 hr of age) do absolutedensity of all life stagesof Ps.columbiae not come to light traps and therefore does not on a per m2 basis for each of the 12 locations. includel-day-old mosquitoeswhen making den- Each simulation run begins with an initializa- sity estimates which are to be compared with tion phase where data which remain constant light trap data. 2) Low night time temperatures throughout the simulation year are input. The are assumed to reduce trapping efficiency by first input file contains values for large either moving the swarming period of Ps. colum- host density and the relative proportions of Iand bioeadults forward in time and prior to nightfall used for first and second crop RICE, DITCH/ or by simply reducing flight activity. To respond SWALE, NON-RICE, and land "non-suitable" to this meteorological factor, model output is for oviposition by Ps. columbiae.A secondinput modified by a temperature factor (Fig. 8) which data set contains estimates of the number of rangesbetween 1.0at minimum night-time tem- overwintering eggsfrom the previous year. peratures>15.5"C to 0 at temperatures<5"C (J. By initializing with the distribution of over- K. Olson, personal communication). This factor wintering eggs found on the last day of the does not kill mosquitoes (i.e., the internal ac- counting of the actual number of adults on an area basis), it simply reducesthe proportion of 7Lotus DevelopmentCorporation, 161 First Street, adults expected to be seen in light traps as a Cambridge,M/.02142. function of lower temperatures. 3) The final SEPTEMBER1988 PsogopqonA IILIJMBIAE SItr4ur,ettolc MoDEL-PART I 279

New Orleans, LA for help with immature sur- vival values. The discussions and free access and interpretation to unpublished data of many RMMP/S-122 researchers is gratefully ac- knowledged and appreciated, especially that of J. K. Olson of Texas A and M University, Col- lege Station, Tx. Finally, the authors appreciate the manuscript review efforts of J. K. Olson and John Jackman, Texas Extension Entomologist, Texas A and M University.

REFERENCES CITED Fig.8. The r"r"ti""ili t.tween minimumair tem- peratureand a temperaturefactor used to modify Al-Azawi, A. and R. M. Chew. 1959. Notes on the expectedlight trap captureson coolernights. See text ecologyof the dark rice field mosquito, Psorophnra for details. confinnis, in CoachellaValley, California (Diptera: Culicidae).Ann. Entomol. Soc.Am. 52:345-351. Andis, M. D. and C. L. Meek. 1984.Bionomics of factor involves the scaling of light trap data. Louisiana riceland mosquito larvae. II. Spatial dis- These data are reported simply as the number persion patterns. Mosq. News 44:37I-376. of adults captured. However, PcSim outputs ex- Andis, M. D. and C. L. Meek. 1985.Mortality and pected adult densities on an area (per m2) basis. survival patterns for the immature stagesof Psoro- phnra For the purposes of comparing these disparate columbiae. J. Am. Mosq. Control Assoc. l:357-362. light trap captures series,we have assumedthat Andis, M. D., C. L. Meek and V. L. Wright. 1983. are linearly correlated with adult density in the "vicinity" Bionomics of Louisiana riceland mosquito larvae. I. of the trap, i.e., we assumethat light A comparison of sampling techniques.Mosq. News trap captures are a measureof density. To facil- 43:195-203. itate a visual comparison of trap data (ranging Carpenter,S. J. and W. J. LaCasse.1974. Mosquitoes betweena few and many thousandsofadults per of North America (North of Mexico). Univ. Calif. night) and predicted densities which rarely ex- Press.Berkeley 490 p. ceed100/m2, a scaling factor is used.This factor Chambers,D. M., C. D. Steelmanand P. E. Schilling. is simply the ratio of average captures during 1979. Mosquito speciesand densities in Louisiana ricelands.Mosq. News 39:658-668. the period of simulation to the average adult Chambers,D. M., C. D. Steelman and P. E. Schilling. density as predicted by the model. 1981. The effect of cultural practices on mosquito A validation of the modelspresented here and abundanceand distribution in the Louisiana rice- an in-depth analysisof the population dynamics land ecosystem.Mosq. News 41:233-240. of Ps. colurnbinein the rice agroecosystemand Chapman, H. C. and D. B. Woodard. 1965. Blood an evaluation of current and proposed and/or feedingand oviposition of somefloodwater mosqui- hypothetical IPM strategies for this mosquito toes in Louisiana: Laboratory studies. Mosq. News are presentedin subsequentcompanion reports 25:259-262. (Fockset al. 1988a.1988b). Curtis, G. A. 1985. Habitat selection strategies of mosquitoes inhabiting citrus irrigation furrows. J. Am. Mosq. Control Assoc.1:169-173. ACKNOWLEDGMENTS De Moor, P. P. and F. E. Steffens. 1970.A computer- simulated model of an -bornevirus trans- The authorswish to thank J. L. Fouss,Agri- mission cycle,with specialreference to chikungr,rnya cultural Engineer, Agricultural Research Serv- virus. Trans. R. Soc. Trop. Med. Hyg. 64:927-935. ice, U.S. Department of Agriculture, Baton Delorme,D. R., R. W. Meola and J. K. Olson.198?. A user's guide to predict diapause Rouge,LA for consultation on factors influenc- induction responses for Psorophoracolurnbiap:Texas and Louisiana. Tx. ing water balance. We would also like to thank Agric. Exp. Sta. Bull. In press. D. Hollier of Jennings, LA for discussionson Edman, J. D. and W. L. Bidlingmayer. 1969. Flight rice production practices and providing weather capacityof blood-engorgedmosquitoes. Mosq. News and water temperature data. The help of J. 29:386-392. Billodeaux, Director, Jefferson Davis Mosquito Focks, D. A., R. E. Mclaughlin and B. M. Smith. Abatement District, Jennings, LA and his staff 1988a. A dynamic life table model of Psorophora is greatly appreciated. Grateful acknowledge- columbiaein the southern Louisiana rice agroeco- system: ment is extended to J. K. Nayar of the Florida Part 2. Model validation and population dynamics.J. Am. Mosq. Medical Entomology Laboratory, Control Assoc.4;282-299. University of Focks,D. A. and R. E. Mclaughlin. 1988b.Computer Florida, Vero Beach,FL for discussionson fe- simulation of managementstrategies for P sorophora cundity and adult survival, M. D. Andis formerly columbiaein the rice agroecosystem.J. Am. Mosq. with the New Orleans Mosquito Control Board. Control Assoc.Accepted for publication. 280 Jounual oF THEAurnrcen Moseulro ConrRor,AssocrlrroN VoL. 4. No. g

Focks, D. A., J, A. Seawrightand D. W. Hall. 19?8. tera: Culicidae) Iarvae in fallow rice fields before Deterministic model for simulating the predation of summer cultivation. Trop. Med. 22:47-5g. Toxorlrytnrhites rutilus rutilus on Aedes aegypti. Mogi, M., A. Mori and T. Wada. 1980b.Survival rates (U.S. Dept. Agric., Agric. Res. Serv.) ARS-S-l?S. of immature stagesof Culextritaeniorhynchr.rs (Dip- 25 p. tera: Culicidae) larvae in rice fields under summir Gahan,J. 8., H. G. Wilson and D. E. Weidhaas.1969. cultivation. Trop. Med. 22:l7t-126. Behavior of Psorophnra speciesand Amphelzs quad- Mount, G. A. and D. G. Haile. 1982.Computer simu- rimaculntusinside and around untreated farm build- lation of area-wide management strategies for the ings in a rice growing area. Mosq. News 29:524-b82. Lone Star Tick, Amblyornrna arnericanurn (L.\ Gerhardt, R. W. 1959. The influence of soil fermen- (Acari: Ixodidae). J. Med. Entomol. 24:b2}-53t. tation on oviposition site selection by mosquitoes. Myers, L. E. 1984. Environmental management for Mosq. News 19:151-155. vector control in rice fields in the context of an Haile, D. G. and G. A. Mount. 1987.Computer simu- integrated approach. In; Environmental manage- lation ofpopulation dynamicsofthe Lone Star Tick, ment for vector control in rice fields. FAO Irrigation Amblyomma am,ericanum(L.) (Acari: Ixodidae). J. and Drainage Paper fi41. Food and Agric. Org. Med. Entomol. 24:356-369. Rome.152 p. Haile, D. G. and D. E. Weidhaas.1977. Computer Nayar, J. K. and D. M. Sauerman, Jr. 1975a. The simulation of mosquito populations (Arcphcbs al- effects of nutrition on survival and fecundity in bimarus) for comparing the effectivenessof control Florida mosquitoes.Part 2. Utilization of a blood technologies.J. Med. Entomol. 13:553-56?. meal for survival. J. Med. Entomol. 12:99-103. Horsfall, W. R. 1942.Biology and control of mosqui- Nayar, J. K. and D. M. Sauernan, Jr. 1975b. The toes in the rice area. Ark. Agric. Exp. Sta. Bull. No. effects of nutrition on survival and fecundity in 427,46 p. Florida mosquitoes.Part 3. Utilization of blood and Horsfall, W. R. 1955. Mosquitoes: Their bionomics sugar for fecundity. J. Med. Entomol. 122220-225. and relation to disease.Ronald Press. New York. Nayar, J. K. and D. M. Sauerman,Jr. 1977.The effects 723p. of nutrition on survival and fecundity in Florida Knisel, W. G., editor. 1980.CREAMS: A field-scale mosquitoes.Part 4. Effects ofblood sourceon ooc5rte model for chemicals,runoff, and erosion from agri- development.J. Med. Entomol. I4:L67-174. cultural management systems. (U.S. Dept. Agric., Olson, J. K. 1983.Final report of the Riceland Mos- Agric. Res. Sew.) ConsewationRes. Rpt. No, 26, quito Management Program for the developmentof 640p. strategies optimizing non-chemical approachesto Kuntz, K. J., J. K. Olson and B. J, Rade. 1982. Role managing mosquito populations in freshwater irri- of domestic animals as hosts for blood-seekingfe- gated cropping systemsusing the riceland agroeco- males of Psorophoracolumbiap and other mosquito system as a model. Texas A and M ResearchFoun- speciesin Texas ricelands.Mosq. News 42:202-210. dation, CollegeStation, Texas, 242 p. MacCreary, D. and L. A. Sterns. 1937.Mosquito mi- Olson, J. K. and C. L. Meek. 1977. Soil moisture gration acrossthe Delaware Bay. Proc. N, J. Mosq. conditions that are most attractive to ovipositing Extermin. Assoc.193?. 188-197. femalesof Psorophoracolumbioe in Texas ricefields. McHugh, C. P. and J. K. Olson. 1982.The effect of Mosq. News 37:.19-26. temperature on the development,growth and sur- Olson,J. K. and C. L. Meek. 1979.Potential longevity vival of Psorophoracolumbiae. Mosq. News 42:608- of overwintering egg populations of Psorophoraco- 613. lumbine(Dyar and Knab) in Texas ricelands.South- Mclaughlin, R. E. and M. F. Vidrine. 1986. A more west.Entomol. 4:163-166. accurate method for assessmentof Psorophora co- Olson,J. K. and C. L. Meek. 1980.Variations in the lumbiaelawicidal tests.J. Am. Mosq. Control Assoc. egg horizons of Psorophoracolwnbiap on leveesin 2:362-363. Texas ricelands. Mosq. News 40:55-66. Mclaughlin, R. E. and M. F. Vidrine. 1987.Psoro- Olson,J. K. and W. H. Newton. 1973.Mosquito activ- phora columbiaelarval counts in southwesternLou- ity in Texas during the 1971outbreak of Venezuelan isiana rice fields as a function of cattle density. J. equine encephalitis (VEE). I. Vector potential for Am. Mosq. Control Assoc.3:633-635. VEE in the Texas rice belt region. Mosq. News Meek, C. L. and J. K. Olson. 1976.Oviposition sites 33:553-559. used by Psorophoracolumbiae (Diptera: Culicidae) Paine, E. O. 1983. Major riceland mosquitoes: An in Texasricelands. Mosq. News 36:311-315. annotated bibliography. Texas A and M Research Meek, C. L. and J. K. Olson. 1977.The importance of Foundation, CollegeStation, Texas, 252 p. cattle hoof prints and tire tracks as oviposition sites Rankin, S. E. andJ. K. Olson.1985. Effects of sprinkle for PsorophnracoLumbiae in Texas ricelands. Envi- irrigation on the occurrenceand abundanceof Pso- ron. Entomol.6:161-166. rophara columbineeggs in ricelands. J. Am. Mosq. Miura, T., R. M. Takahashi and F. S. Mulligan III. Control Assoc.1:454-.457. 1978. Field evaluation of the effectivenessof pre- Schwardt,H. H. 1939.Biologies of Arkansas rice field dacious as a mosquito control agent. Proc. mosquitoes.Ark. Agric. Exp. Sta. Bull. No. 377. 22 Calif. Mosq. Control Assoc.46:80-81. p. Mogi, M., I. Miyagi and B. D. Cabrera.1984. Devel- Service, M. W. 1977. Mortalities of the immature opment and survival of immature mosquitoes(Dip- stagesof speciesB of lhe Anophelesgambiae com- tera: Culicidae) in Philippine rice fields. J. Med. plex in Kenya: Comparisonbetween rice frelds and Entomol. 2l:283-291. temporary pools, identification of predators, and Mogi, M., A. Mori and T. Wada. 1980a.Survival rates effects of insecticidal spraying. J. Med. Entomol. of immature stagesof Culer tritaeniorhynchus (Dip- 13:535-545. SEPTEMBER 1988 Psonopuone joLUMBTAESruur,arroN MoDEL-PART I 28r

Sharpe,P. J. H. and D. W. DeMichele.1977. Reaction rus. IL Experimental vector studies. Am. J. Epid' kinetics of poikilotherm development. J. Theor. 93:206-211. Biol. 64:649-670. Tsutsui, H. 1984.Rice production and water manage- Skaggs,R. W. 1980.A water managementmodel for ment. In: Environmental management for vector artificially drained soils. N.C. Agric. Res. Service control in rice fields. FAO Irrigation and Drainage Tech.Bull. No. 267,N.C. StateUniv., Raleigh,NC, PaperNo. 41. Food and Agric. Org' Rome.152 p. 54 p. Welch, J. 8., J. K. Olson and M. M. Yates. 1986. Steelman,C. D. and P. E. Schilling. 197?.Economics Occunence of Psorophnracolumbiae eggs in differ- of protecting cattle from mosquito attack relative to ent field types comprising a Texas ricefield agro- injury thresholds.J. Econ. Entomol. 70:1,5-17. ecosystem.J. Am. Mosq. Control Assoc.2:52-56. Steeknan, C. D., T. W. White and P. E. Schilling. Welch, J. B. and J. K. Olson. 1987.Oviposition habi- 1972. Effects of mosquitoes on the average daily tats of Psorophnracolumbine in soybeanfields of a weight gain of feedlot steers in southern Louisiana. Texas agtoecosystem.J. Am. Mosq. Control Assoc. J. Econ. Entomol. 65:462-466. 3:251-258. Steelman,C. D., T. W. White and P. E. Schilling. Whitehead, F. E. 195?.Flight range of rice field mos- 1973. Effects of mosquitoes on the average daily quitoes. Ark. Agric. Exp. Sta. Bull. 590. 7 p. weight gain of Hereford and Brahman breed steers Williams, D. C., C. L. Meek and V. L. Wright. 1983. in southern Louisiana. J. Econ. Entomol. 66:1081- Abundanceof mosquitoeggs in a permanent pasture 1083. and effects of cattle movement and hooforint den- Sudia,W. D. and V. F. Newhouse.1971. Venezuelan sity on eggdistribution. Southwest.Entomol. 8:273- equine encephalitis in Texas. Mosq. News 31:350- 278. 351. Williams, D. C., C. L. Meek and V. L. Wright. 1984. Sudia, W. D., V. F. Newhouse and B. E. Henderson. Comparisonof mosquito oviposition in a fallow rice 1971. Experimental infection of horses with three field and a permanent pasture in southern Louisi- strains of Venezuelanequine encephalomyelitisvi- ana. Southwest.Entomol. 9:319-325.