Il]]]

452 JounNe,r,oF THEArrannrclN Moseurro CoNrnol AssoctarroN VoL.6,No.3

DISTRIBUTION AND ABUNDANCEOF LARVAL COQUILLETTIDIA PERTURBANSIN A FLORIDA FRESHWATERMARSH1

C. D. MORRIS',J. L. CALLAHAN3nNo R. H. LEWIS

Polk County Enuironmental Seruices,P.O. Box 39,Bartow, FL 33830

ABSTRACT. All instars of larval Coquilletti.diaperturbans were found in the samehabitats, but early instar larvae were more aggregatedthan later instars. Larvae were most numerous in areas dominated by arrow-arum (Peltandra uirginba) and maidencane(Panicum hemitomon),less so in areasdominated by sedges(Corer spp.) and miscellaneousmixed vegetation,and least abundant in pickerelweed(Ponte- deria cordata) areas. Larvae were uncommon in open water or in areas dominated by small floating plants such as water fetn (Saluinin rotund.ifolin),duckweed (Lemrw mirwr) and fern (Azolla caroliniana). Larval concentrationswere $eatest in water 35-70 cm deep.There was also a tendency for them to concentratein areasbeyond 25 m from shore. Larvae were log-normally distributed in favorable sites and becameprogressively more aggregatedas sites becameless favorable.

INTRODUCTION This treatment is economicalbecause larvae in these mats occur only within 1 m of the cattail perturbans The widely ranging Coquillettidia edgesor openings(Batzer and Sjogren 1986). (Walker) pest perma- is an important near its Larval distributions in other than floating cat- nent freshwater marsh breeding sites, and it is tail mats in Minnesota have not been examined. associatedwith intimately epidemic and epior- If larvae in Florida aggregate as they do in nithic foci of Eastern equine encephalomyelitis then perhaps briquettes or other (Morris Minnesota, virus in the easternUnited States 1988). control methods could be economically applied perturbans The larvae of Cq. attaeh to the sub- in areasof high larval concentration. merged roots of emergent aquatic vegetation. Thus, studies were begun to: 1) describe the Larvae occur primarily in the benthic zone of spatial distributions of all larval instars of Cg. well established marshes and swamps which perturbans, and 2) determine the factors which characteristicallyhave low dissolvedoxygen, low influence these distributions. pH and a substantial organic detrital bottom cover(McNeel 1932,Brower 1953,Callahan and Morris 1987a).They are typically associated MATERIALS AND METHODS with cattails (Typha spp.) and sedges (Core.t spp.), and in temperate climates are often abun- The 4.5-ha study site was part of an extensive dant on floating mats (Allan et al. 1981,Batzer freshwater marsh system covering approxi- and Sjogren1986, OIds et al. 1989). mately 10,000ha northeast of the city of Lake In Florida, floating cattail mats produce more Alfred, FL. The site was dominated by maiden- Cq. perturbans than rooted cattails (Slaff and cane grass (Panicurn hemitornonSchult.) inter- Haefner 1985) but the speciesis also a prolific mingled with bladderwofts (Utricularia gibba breeder in many other southern plant commu- Linn. and U. inflnta Walt.). False maidencane nities (Bidlingmayer 1968,Lounibos and Escher (Sacciolepisstriata (Linn.) Nash), arrow-arum 1983,Slaffand Haefner 1985). (Peltandrauirginica (Linn.) Kunth) and sedges It is difficult and expensiveto larvicide effec- (Carex spp.) were also found in large stands. tively in Cq. perturbans habitats. Nonetheless, The onshore vegetation included swamp-bay the Metropolitan Mosquito Control District of (Persea palustris (Raf.) Sarg.), Florida maple St. Paul, Minnesota, controls larval Cq.pertur- (Acer barbaturnMichx.), pine (Pinus spp.),saw bons with methoprene briquettes placed at the palmetto (Serenoa repens (Bartr.) Small), edges of cattail mats and in holes within the water-primrose (Ludwigia spp.) and dog fennel mats (R. D. Sjogren,personal communication). (Eupatorium capillifolium (Lam.) Small). Plant terminolory follows that of Dressleret al. (1987) and Tarver et al. (1979). The study site was divided into 70, 25 x 25 m I Institute of Food and Agricultural Sciences,Uni- sections along north to south and east to west versity of Florida Experiment Stations; Journal Series transects. Each section was subdivided into 4, R-00293. (NE, 2 12.5x 12.5m quadrats NW, SE and SW). Current address: Florida Medical Entomology This resulted in 280 quadrats. Samples were IFAS-University of Florida, 200 9th Laboratory, every 3-5 weeks from approximately SE, Vero Beach,FL32962. collected Street, quadrats 3 Current address:Florida Department of Natural 5O%of the water-covered during each Resources,Bureau of Mine Reclamation, 1677 High- sampling period from February 2t to December way 17 South, Bartow, FL 33830. 1?, 1984. On January 15, 1985, samples were l SnprpN{snn 1990 Cg. ennrunnnNs LARVALDrstnreufloN 453

collected from all water-coveredquadrats. Col- oxygen from 0.6 to 1.4 ppm. The marsh bottom lections were made from the NE and SW quad- temperature ranged from 10 to 25.5"C. rats of each section during the first (February Seasonalabundance and uoltinismr On the 13 21) and subsequentodd-numbered sampling pe- sampling dates 15,458larvae of 5 genera were riods, and from the NW and SE quadrats during collected(n: 1,179).Of these, L5,093(97.6%) the second(April 2) and subsequenteven-num- were Cq. perturbans (Table 1), with an average bered sampling periods. If one of the preselected of 0.72 first instars, 2.18 secondinstars, 4.46 quadrats had insufficient water to sample, the third instars and3.42 fourth instars per sample. nearest available quadrat within the same sec- We also collected 11 pupae, 185 Uranotaenia tion was selected. spp., 79 Atnphcles spp., 66 Mansonia spp. and Water/detritus samples potentially contain- 35 Cul.exspp. These were excludedfrom further ing Iarvae were collected from an airboat using analysis. the probe and pump system of Morris et al. The number of winter diapausing fourth in- (1985).The probewas insertedthrough the veg- star Iarvae remaining after early April was low, etation so that the tip was resting on the bottom as indicated by the drop in averageinstar num- of the marsh; the pump was turned on for 90 ber on April 24 (Lounibos and Escher 1983) sec. Preliminary work with the system showed (Table 1). Apparently there was no significant that approximately 25 liters of material were oviposition by adults until after early April. At pumped in 90 sec(Morris et al. 1985).The water this time, the number of larvae increasedwith a line was clearedbefore the pump was shut off to concurrent drop in the averageinstar number. prevent carryover of larvae from one sample to From late April until the end of the study, the another. average instar number gradually and steadily Samples were hand sorted for larvae in the increasedwhile the number of larvae decreased. Iaboratory on the day of collection. Larvae were Statistical distributions: For each of the 13 identified to and Cq. perturbans lawae sampling dates the variance exceededthe mean were tabulated by instar. Water depth and dom- by a factor ranging from 5.6 to 74.7, indicating inant aquatic plant speciesin the vicinity of the nonrandomness. The maximum likelihood probe were recorded. Water depth and water method (Elliott 1971) of estimating the aggre- temperature were also recorded at a baseline gation coefficient, k, produced values ranging station. from 0.073 to 0.323. This range of k indicates Water samples were collected for chemical stability and a potential for using the common analysis from 12 randomly selected sites at k (0.205)to developsequential sampling meth- monthly intervals from June 4 through Novem- ods for population density determination. This ber 5, 1984. The pH was recorded in the field study did not address this potential. The chi- "goodness-of-fit" with a FisherAccumet model 320 expanded scale square test for of k to the research pH meter. Dissolved oxygen was also negative binomial (or log-normal) distribution recorded in the field with a Yellow Springs In- was acceptedfor all 13 sampling dates. strument model5T oxygenmeter. The remaining water quality parameters were measuredin the laboratory. Total alkalinity was determined Table1. Summaryof Coquillettidiaperturbans lawae using the American Public Health Association foundin 1779samples collected from a marshnear mrc nft"a,n".U (APHA) (1981) procedures. Orthophosphate was determined on a Bausch and Lomb Spec- Cq.perturbans Average tronic 20 using methods described by APHA No. of instar (1981)that weremodified for usewith prepack- Date samples Total Mean number aged chemical reagents(Hach Company, P.O. Feb.21 I44 119 0.8 3.50 Box 389,Loveland, CO 80539). Apr.2 I47 38 0.3 3.92 The procedures of Norusis (1g86, 1988a, Apr.24 145 444 3.1 2.I3 1988b), Taylor (1984) and Elliott (1921) were May 14 136 2,L33 L5.7 2.2I used to analyze the data on an IBM PC XT June 11 129 1,899 r4.7 2.57 computer. July 9 141 1,805 r2.8 2.57 Aug.6 143 r,597 1t.2 2.74 Sep.4 1,37 7,782 13.0 2.92 RESULTS Oct.2 126 708 5.6 2.94 Oct.29 116 1,294 7t.2 3.06 Water chemistry and temperature: Between Nov.26 t1.4 844 7.4 3.59 June 4 and November5, 1984,the averagepH Dec.17 107 916 8.6 3.67 in the study area ranged from 3.6 to 4.9; ortho- Jan.15 194 r,5L4 7.8 3.60 phosphateranged from <0.001 to 0.14ppm; total alkalinity from 0 to 5.96 ppm; and dissolved Total I,779 15,093 8.5 2.86 JounNlr- oF THE ANrsRrceNMosqulro Cot*tRor, AssocrerroN VoL.6.No.3

Data were then examined using Taylor's Relationship to tall uegetation:Early observa- powerlaw (Taylor 1961,Elliott 1971).The value tions suggestedthat larvae may be more abun- of b in power transformation analysis (slope of dant in areasadjacent to trees and shrubs which the regressionof log-variance on log-mean for protrude abovethe marsh. To test this hypoth- each sampling date) is an index of aggregation esis, mosquito production in quadrats within similar to k. For total Cq. perturbans larvae, 12.5m of vegetationtaller than 2 m (n: 162) Taylor'sb was2.19 (Table 2, all instars),slightly was compared to production elsewhere(n : abovethe b : 2 expectedfor a log-normaldis- 118). Larval density near tall vegetation (x : tribution. The Iinear regression between log- 9.5, n :953) was not significantly higher than variance and log-mean of total Cq. perturbans near shorter vegetation (i : 7.5, n : 826) in was highly significant (r : 0.97,P < 0.001,n : ANOVA (P : 0.41, n : 7t79). The frequency 13). distribution of samples that contained larvae, However, when the appropriate power trans- based on proximity to tall vegetation, was not formations were calculatedfor individual instars significantly different from the distribution for rather than the total, there was a downward all samples(chi-square : 0.298,df = 1, P : trend in b as larvae aged. This indicates less 0.585). aggregationin later instars (Table 2). Again, the Vegetation:The distribution of sampleswhich regressionsbetween variance and mean were contained Cq. perturbans, by vegetation, was highly significant for all instars (Table 2). significantly different from the distribution for Distancefrom rnarshedge: Ofthe 280 quadrats all samples (chi-square : 89.5, df = 15, P ( sampled, 700 (35.7%) were within 25 m of the 0.001). There were more samples with larvae marshedge, ll5 (41.I%) werein the 26 to 75 m than expectedin maidencane(506 vs. 446), ar- range, and 65 (23.2%) were 76 to 125 m from row-arum (106 vs. 79) and sedge(87 vs. 71). the edge. Twice as many larvae were found in There were fewer positive samples than ex- quadrats located beyond 25 m, but this differ- pected for all but one of the remaining plants ence was not significant (Table 3). The fre- (water-primrose,10 vs. 9). The greatestdiscrep- quency distribution of samples containing lar- ancies occurred with fragrant water-lily (6 vs. vae, based on distance from the edge, was sig- 37), water fern (4 vs. 23), pickerelweed(53 vs. nificantly different from all samples(chi-square 69) and duckweed(3 vs. 16) quadrats. : 14.7,df : 2, P : 0.001).There were more There were also significant differencesin lar- than expectedin the 26 to 75 m range (243 vs. val density by vegetation (P < 0.0001,ANOVA) 231) and 76 to 125 m range (456 vs. 415) but (Table 4). The number of differences increased fewer than expectedin the 0 to 25 m mnge (224 as the Iarvae aged;4 differencesfor first instars, vs.277). 13 for seconds,19 for thirds, and22 for fourths.

Table 2. Power transformation statistics.

Index of aggregation Power Mean-variance (Taylor's b) transformation" regression (r)b Cq.perturbans 1st instars 3.39 x-o t 0.97 2nd instars 2.75 x -' 0.99 3rd instars 2.63 x-- 0.98 4th instars 2.44 x-o., 0.92 All instars 2.r9 x "- 0.97 Vegetation Miscellaneous I.Jb xo3 0.67 Maidencane 2.t4 x-o l 0.98 Sedges 2.L6 x-0 1 0.97 Arrow-arum 2.22 x-0 1 0.96 Pickerelweed 2.99 x -- 0.97 Water depth (cm) <35 2.48 x "- o.92 35-55 2.23 x-o t 0.96 60-70 2.09 xo 0.98 >70 2.5L x "" 0.97 'An exponent of: 0.5 is a square-roottransformation (Poisson);0.0 is a logarithm transformation (negative binomial); -0.5 is a reciprocal-of-the-square-roottransformation; -1.0 is a reciprocal transformation. oP < 0.02in all cases. Ssprprr,rgnn1990 Cq. eenrunneNs LARvALDlsrnrsutlol,l

For all instars, auow-arum, maidencane,false variability of the larval numbers of the former maidencaneand sedgeshad significantly more group comparedto the latter group. Iarvaethan 10, 9, 4 and 4 other plants, respec- Foi further analysis, each quadrat was as- tively (Table 5). Bladderworts had significantly signed to 1 of 5 vegetation communities based more larvae than fragrant water-lily. on the dominant plant speciesin that quadrat. Smartweed,spikerush, and red-root had Iarval The 5 communities were maidencane, arrow- densities comparable to arrow-arum, maiden- arum, pickerelweed, sedgesand heterogeneous cane,sedges and false maidencane(Table 4) but areas consisting of various other plant species. no significant differences were found between Maidencanedominated 120(42'S%) of the quad- these 3 speciesand other plants. This was due rats, arrow-arum dominated'40 (l4.3Vo),pick- to the very small number of samples and high erelweed35 (l2.\Vol, sedges28 (10%) and mis- cellaneousvegetation 57 (20.4%). Based on the averagenumber per collection, Table 3. Density of Coquillettidia perturbans latvae approximately two-thirds of the larvae were by distance from the marsh edge. found in the arrow-arum and maidencaneareas Cq.perturbans and one-third in the miscellaneous and Carex zones (Table 6). There were no significant dif- Distance Arithmetic Log from marsh No. of ferences among the densities in these 4 com- edge(m) samples Mean (%) Mean'(%) munities. Less than 3% of the larvae were found 0-25 534 4.6(18.6) 1.2A (20.3) in pickerelweed and the density was signifi- 26-75 444 10.1(40.7) 2.2A (37.3\ cantly lower than in the other four groups(Table 76-125 801 10.1(40.7) 2.5A (42.4) 6). Total r.779 25.6 1.43 Grouping quadrats by vegetation community " Values followed by the same letter are not signif- permitted us to also quantify the spatial distri- icantly different, P: 0.19 in ANOVA, Student-New- butions by vegetation.Larval distributions were man-Keuls. log-normal in arrow-arum, maidencane and

Table 4. Density of Coquilbttidia perturbans laivae in 1,779samples dominated by 22 vegetationtypes and in open water. Mean number of Cg Perturbans No. of Vegetation samples Arithmetic Log Best Arrow-arum (Pe ltand r a uirginba) 151 13.8 5.5 Maidencane (Panic um hemito mo n) 856 12.2 3.8 Smartweed(Polygonum hyd.ropiperoi.des) 11 t4.5 .1. r Good Red-root (Lachnanthescarolininna) 9 5.6 2.8 False maidencane(Sacciolcpis strinta) t64 5.5 2.7 Sedges(Carer spp.) 1an 3.6 2.7 Spikerush (E leocharisbaldw inii) 19 b.l- 2.7 Fair Bladderworts (U t ricular ia spp.) 66 6.5 2.3 Water-primroses (Ludwigia spp.) 18 aa 2.3 Spatterdock (Nuphar luteum) 10 2.7 1.9 Pickerelweed(Pontederin cordata) r32 2.0 t.7 Water pennywort (Hydrocotyleumbellnta) oa 1.3 Poor Swamp-bay(Persea palustris) 2 0.5 t.4 Saw grass(Cla.dium jarnaicense) 2 0.5 L.4 Buttonbush (Cephalnnthusoccidentalis) I7 0.5 t-J Water fern (Salu inia rotundifolin) 45 0.7 t.2 Common duckweed(Lemna minor) 30 0.2 1.1 Fragrant water-lily (Ny mphaeaodorata) 72 0.2 1.1 Mosquito fern (Azolla caroliniana) z 0.0 0.0 Bog-mats (Wolffielln spp.) 0.0 0.0 Brazilian-pepper (Schinus terebinthifolius) 2 0.0 0.0 Ferns (Pteridophytes other than Saluinincea) 0.0 0.0 Open water I 0.0 0.0 Unknown I 0.0 0.0 456 JouRNlr,oF THEArraenrcex Mosqurro Cournol Assocrnrrox Vor,.6, No. 3

Table 5. Pairs of plant specieswith significantly (P < 0.05,ANOVA-SNK) different populations of larval Coquillettidia perturbans. Arrow- False arum Maidencane maidencane Sedges Bladderworts Fragrant water lily 2,3,4,4 r,2,3,4,4 3,4,4 3,4,4 Duckweed 2,3,4,4 2,3,4,4 A A Water fern 2,3,4,4 2,3,4,4 4,4 4,4 Pickerelweed 3,4,A r,2,3,4,4 4,4 4,4 Sedges 2,3,4,4 t,2,3,4,4 False maidencane 2,3,4,4 r,2,3,4,4 Bladderworts 2,3,4,4 3,4,A Water pennyworts 2,3,4,4 3A Buttonbush 3,4,A A Maidencane A " 7,2,3 and 4 : larval instar number,A : all instars.

Table 6. Density of Coquillettidiaperturbans larvae, Table 7. Density of Coquill,ettidiaperturbans larvae by -a.jor r"getatt"n by water depth when the samplewas taken. Cq.perturbans Mean numberof Water Cq.perturbans Arithmetic Log depth No. of No. of (cm, samples Arithmetic Log Vegetation samples M""" (A Mean" (%) 108 0.0 0.0 Arrow-arum 256 11.8 (33) 3.3 (34) A 15 t2 0.0 0.0 Maidencane 816 11.9 (34) 2.eA (30) 20 29 0.3 0.2 Miscellaneous 283 7.8 (22) 1.8A (1e) zb bb 0.2 0.1 Sedges 183 3.0 (8) (14) 1.3A 30 24 0.9 0.2 Pickerelweed 24I 1.0 (3) (3) 0.3B 35 33 7.0 1.4 Total 1,779 35.5 2.2 40 45 6.2 r.4 " Values followed by the same letter are not signif- 45 53 4.7 1.3 icantly different, P : 0.0001, Oneway ANOVA, Stu- 50 61 6.4 2.0 dent-Newman-Keuls. 55 90 8.9 1.6 60 r2r 10.9 3.1 65 273 t4.7 3.8 70 363 13.0 3.2 sedgedominated communities,even more aggra- 75 360 5.6 1.5 gated in pickerelweed,but were nearlv random 80 180 4.0 1.3 85 in miscellaneousvegetation areas (taUte Z). 50 1.9 1.1 Water depth: 900 Mean water depth ranged from 95 31 77 1.5 1-0 to 52 cm and changedprimarily in iesponse 100 0 to rainfall patterns. The preadult stagesof Cg. 105 5 0, 0.0 perturbans. develop slowly and, apparently, based on this study, can take up to-a y.", io Total 1,779 develop. Assuming larvae migraled httie from sites of oviposition, the immatures experienced widely fluctuating water levels duringtheir de- velopment. Therefore, it was necessarvto ana- Table 8. Density of Coquill.etti.diaperturbans larvae lyze the relationship between Cq. pirturbans abundance and water depth in 2 ways. The first Density of Cq.perturbans was to examine abundance based on the water Water depth when the Arithmetic Log samplewas taken. In the second depth No. of analysis each quadrat was assignedto one of 4 (cm) samples Mean (%\ M."tr: (?r) water depth groupsbased on its seasonalaverage <35 tr\ (11) depth. This permitted the comparison 156 0.6 1.2A across 35-55 2r7 (33) (27) these groups and the quantificalion 9.2 2.98 of spatial 60-70 744 13.8 (50) 4.5B (41) distributions in waters of various depths. >70 662 4.1 (15) 2.38 (21) Larvae were abundant in 3b to ?0 cm deep Total 7,779 34.2 3.0 water, with 60-70 cm being optimum (Table Z). " Values with the same letter There are not significantly were significantly fewer Iarvae in areas different,P: 0.0001,oneway ANOVA, Student-New- with less than 35 cm of water (Table 8). Based man-Keuls. 457 SBprnrrrspn1990 Ce. psnrunatNs LARvALDrstRIsutroN

on the frequenciesof all samplescollected, there The depression in both larval numbers and were also substantially fewer sampleswith lar- the rate of increaseof the averageinstar number vae than expectedin the 0 to 35 cm range (19 in early October may indicate a minor fall emer- vs. 81) but substantially more than expectedin genceihat occasionallyoccurs in central Florida the 60 to 70 cm range (477 vs.386) (chi-square (Lounibos and Escher 1983). This seemsun- : 71..L,df: 3, P < 0.0001). likely since there was a decreasein numbers of all instars. Statistical distributions: The most appropriate DISCUSSION transformation method fot Cq.perturbons larval data would be the power law on an instar by Water chemistry: The low pH and dissolved instar basis. That is, replace each instar count - oxygen conditions observed in this study are by Xp, where p = 1 (bl2) and b is Taylor's b typical for Coquillettidiabreeding sites in central for that instar. The b values observed,however, Florida (Lounibos and Escher 1983, Callahan were only slightly greater than the b value of a and Morris 1987a),elsewhere in North America negative binomial (i.e., 2). For practical pur- (Allan et al. 1981,Batzet and Sjogren 1986,and poses,Cq. perturbans lawae are distributed ap- others) and in southern France (Guille 1976). proximately as a negative binomial; therefore, Seasonalabundance and uoltinisrn:In Florida, sample numbers can be normalized using log the annual number of generationsof Cg.pertur- transformations. This practice should already bonstypically rangesfrom one in the temperate be routine when analyzing trapping data north to 3 in the tropical south (Provost 1976)' (Wadley1967, Steel and Torrie 1960). While there can be both spring and fall emer- gencesin central Florida (Provost 1976,Louni- bos and Escher 1983),in 1984only the spring brood was apparent at the Lake Alfred site. Theoretically, assuming equal sampling effi- from the egg raft from which they hatched. The ciency for all instars, the lowest averageinstar relative magnitude of the changesin b between number should occur simultaneously with the instars suggeststhat essentially all of the dis- highest number of larvae. This was not the case persion occurredin the secondinstar. One could in our study. The lowest averageinstar number argue that, since the index never approached (2.13) occurred in Iate April, one sampling that of a random distribution (i.e., 1), Cq.per' period before peak numbers of larvae in May. turbans larvae must have dispersedonly short This seems higher than expected considering distances. that 3 weeksbefore (April 2),92% ofthe larvae Distance from marsh edge:There were more collected were fourth instars and yet, on May Iarvae, and more sampleswith larvae than ex- 14, 3 weeksafter the averageinstar number low pected,in the 76 to 125 m zone,but fewer than point, the averageinstar number had risen only expectedin the 0 to 25 m zone. Using the ra- 3.8% (Table 1). These observationssuggest that tioiate of the edge effect theory, one would either larvae remained first instars for only a expect greater numbers of larvae in the 0 to 25 short period of time or first instar larvae were m'zone: Guille (1976)explained his lower den- underrepresentedin the collections. sities at the edge of his marsh as the result of The 350 pm mesh screen used in the probe unsuitably high- soil temperatures produced-by and pump systemwas larger than the 260 to 300 evaporation and drought dtyitg orrt the edge' prmhead capsulewidths of first instar Cq. per- Lounibos and Escher (1983) found equal num- turbans (Nemjo and Slaff 1984). This would bers of Cq. perturbans near and away from the allow some first instars to pass through the edgeof a vegetationcovered phosphate pit which sieve. Head capsulewidths of secondto fourth had steepsloped edges not susceptibleto drying instar larvae in the Nemjo and Slaff study all out. exceeded4I0 pm; theoretically, all of these lar- Vegetation: There was no evidence in this vae would have been collected. Field trials of study that tall vegetation either in the center or screenswith mesh sizeless than 350 pm resulted at the edgeof the marsh influenced oviposition in unacceptableplugging problems. The lower site selection. In a later study Callahan and than expected numbers of first instar larvae Morris (1987a)found that mosquito larvae of 5 could also be attributed to the greater difficulty genera, including Coquillettidia, were more in finding these small, transparent in the abundant in lake waters adjacent to tall shore- often murky collection water. This collection Iine vegetationthat apparently servedas diurnal bias means the averageinstar number in Table resting refuges for gravid adults. In an earlier 1 is an overestimateof age,at least through mid- study on the distribution of Mansonia dyari July while first instars were still abundant. (Belkin, Heinemann and Page) and Ma. titillans 458 JounNeL oF THE ANanRrcer.rMoseurro CoNtRor, AssocrerroN VoL.6,No.3

(Walker) larvae on water lettuce (pistia stra_ best larval habitats. These habitats provide ob_ tiotesLinn.), we also observed higher larval den_ vious barriers to predators, especiallyfish. sities in waters adjacent to tall shoreline vege_ While plant speciescan be a major parameter tation (Morris et al. 1986). to use in characterizing the breeding potential Arrow-arum, maidencane and smartweed of a marsh or other vegetatedwetland, caution pqke up a group of plant species "best" that were is advised.For example,in this study, smartweed indicators for the habitats for "best', Cg.pertur_ was classified in the category. yet, in a bons larval-developmentand survival (faUte a). preliminary- study of several lake littoral zones, Redroot, false maidencane, sedgesand spikerush and in a subsequentstudy of 2 marshes and 2 were indicators of "good', habitits. Communities additional lake littoral zones, we never found dominated by these Z speciestypically had very Iarvae associatedwith this plant (Callahan and densevegetation, often with a floating mat com_ Morris 1987b). The absenceof a detrital laver ponent. The subsurfacestructure also often in_ in the latter casesprecluded larval survival. cludedsubmergent plant species. Pickerelweedwas classifiedas onlv a,,fair,, Bladderworts, water-primroses, spatterdock, indicator of abundance in the .urr"ttt rt td.rr. pickerelweed and water pennyrrort "fair" supported However,Bidlingmayer (1g50)found pickerei- only Iarval populations. "best" Mosquitoes were weedto be one of the hosts in a marsh in seldom found in monotypic stands of bladder_ Leesburgh,FL, just 50 miles north of our Lake worts, a submergedplant, but were often abun_ Alfred site. dant where bladderworts were mixed with one A third case in point pertains to Cq. pertur- or more emergentplants. . In these casesblad- boru larvae which are highly associ"t"a *itt derworts may have provided larvae additional cattails (Typha spp.), particularly in temperate protection from predation, eventhoueh bladder_ climates. Throughout its range, Cq. pertirbans worts can themselves be larvivorous{Baumgart_ Iarval densities are higher on floaling mats of ner 1987).Dense mats ofbladderwort reducJnor cattails than on rooted plants (Batzeiand only predation, Sjo- but also gravid females' access gren 1986,Slaffand Haefner 198b). to the water surface for oviposition, and fewer One is well advised to heed the words of larval attachment sites. RobertArmstrong (1941)who wrote:,,In search- In contrast, spatterdock and pickerelweedoc- inq f9r breeding places [of Monsonia perturbansl, cuned i1 m9r9 sparsely populated monotypic it is better to be guided by the appeara.rceand stands that did not exclude predators. Larvae texture of the marsh than by a list of host found associated with water-primroses were plants." probably from eggsIaid near other plants since depth: Our findings of larval density in water-primroses ,Water do not possess appropriate to water depth agrees precisely with roots lelation for larval attachmenr. Guille (1976),who also found the greatest den- pl.ant The 10 speciesin the last group indi- sities of Cq. richiardii in the 60 to Z0 cm range. cated "poor" larval survival areas-arrd *e." He also found less than half as many at depths either 1) woody plants without suitable larval below 30 cm and above T0 cm, althougli he attachment sites (swampbay, buttonbush and explained the lower densities at the greater Brazilian-pepper);2) small floating plants which depths as an artifact of his collection rnethod, tend to exclude oviposition (watei fern. duck- Since the collection efficiency of our sampling weed, mosquito fern and bog-mat) (Hobbs and method was independent of water depth and we Molina 1983);3) monoculturesin deep waters also found progressivelyfewer larvae at greater subject to wave action and/or which allow easv depths, we do not believe Guille's results are accessby predators (saw grass,fragrant water-- erroneous. pickerelweed Iily, and spatterdock); or 4) were Basedon Taylor's index ofaggregation,larvae temporarily inundated terrestrial non-salvini- were log-normally distributed in the optimum aceae ferns located at the edge of the marsh water depth zones (35 to 70 cm) and slightly which tended to dry up. more aggregatedin less favorable depths (Table In general, Cq. perturbans Iarvae "fair" were most 2); a trend similar to that seen earlier with aggregated-in plant communities, and log- vegetation. "best" ..good', normally distributed in and com- munities. The near randomnessobserved in the CONCLUSION miscellaneous plant zones probably resulted from plant the mosaicof tlpes which were them- While larval aggregationwas related to vege- selves more or less randomly distributed. tation, water depth and proximity to the marsh We (1926) concur with Guille's observations edge, but not proximity to tall vegetation, the (Ficalbi) on _Cq.richiardii that higher density aggregationwas insufficient to pinpoint control greater and diversity of vegetation provide thl efforts such as is done in Minnesota. However. I

SEPTEMBER1990 C8. psnrunatNs LARVALDrsrRrsutloll

Leesburg some control could be achieved by managing biology of Mansonia mosquitoes in the Assoc.21:1-4. shore and aquatic vegetation, and water depth. area.J. Fla. Anti-Mosq. W. L. 1968. Larval developrrent of approach may not be possible in Bidlingmayer, While this mosquitoes in central Florida. Mosq. possible in manmade,re- Mansonia natural marshes,it is News28:51-57. stored and enhancedwetlands projects, includ- Brower, L. P. 1953. The distribution of Mansonia ing lake shores. Such activities are on the in- perturbans (Walker) in Morris County. Proc. N. J. crease in Florida and elsewhere,and environ- Mosq.Exterm. Assoc.40:147-149. mental engineers are quickly learning how to Callahan, J. L. and C. D. Morris. 1987a.Habitat design and build productive marshes.Unfortu- characteristics of Coquillcttidia perturbarrc in cen- nately, designs are often unintentionally ideal tral Florida. J. Am. Mosq. Control Assoc.3:176- for the massproduction of mosquitoes. 180. J. L. and C. D. Morris' 1987b.Survey of 13 Vegetation and water depth can be easily ma- Callahan, County, Florida, lakes for mosquito (Diptera: waste water retention Polk nipulated in storm and Culicidae) and midge (Diptera: Chironomidae) pro- and detention ponds. These facilities are ofben duction.Fla. Entomol. 70:47l-478- desigled and constructedin urban areaswithout Dressler,R. L., D. W. Hall, K. D. Perkins and N. H. consideringthe mosquito producing potential. Williams. 1987. Identification manual for wetland Wildlife managementareas often include wet- plant speciesof Florida.Univ. Fla. Inst. FoodAgric. lands ideal for Cq.perturbans production. Man- Sci.,Gainesville. agement plans for these wetlands, when man- Elliott, J. M. 19?1. Some methods for the statisticaL agedfor waterfowl, usually include lists of desir- analysis of samples of benthic invertebrates. 25. species,slope ratios and water Freshw.Biol. Assoc.Sci. Publ. able vegetation eco-ethologiquessur Co- "desirableplant" lists should Guille, G. 19?6.Recherches depths.Making up quillctti.dia(Coquillettidia) richiardii (Ficalbi)' 1889 also take mosquito breeding potential into con- (Diptera-Culicidae)du littoral Mediterraneen Fran' sideration. The incidence of mosquito transmit- cais. II. Milieu et comportement.Ann. Sci. Nat ted wildlife diseasessuch as avian malaria and Zool., Paris 18l,5-112. EEE may also be reducedif mosquitoesin these Hobbs,J. H. and P. A. Molina. 1983'The influenceof areasare reduced. the aquatic fern Saluinia auriculata on the breeding of Anophelesalbimanus in coastalGuatemala. Mosq. News 43:456-459. L. P. and R. L. Escher.1983' Seasonality ACKNOWLEDGMENTS Lounibos, and sampling of Coquillettid'iaperturbans (Diptera: in south Florida. Env. Entomol. 12:1087- not have been possiblewith- Culicidae) This work would 1093. out the technical assistanceof Marc Slaff, John McNeel, T. E. 1932.Observations on the biology of Haefner and Lottie Jordan. Thanks are also due Mansonia perturbans (Walk.) Diptera, Culicidae. Philip Lounibos and Jorge Rey for reviewing an Proc.N. J. Mosq. Exterm. Assoc.19:91-96. earlier version of this manuscript. Morris, C. D. 1988.Eastern equineencephalomyelitis, pp. 1-20. In: T.P. Monath (ed.),The arboviruses: epidemiologyand ecology,vol. 3. CRC Press Inc., REFERENCES CITED Boca Raton, FL. Morris. C. D., J. L. Callahanand R. H. Lewis' 1985. Allan,S. A., G. A' Surgeoner,B. V. Helsonand. D' H' Devicesfor sampling and sorting immatwe Coquil- Pengelly.1981. Seasonal activity of Mansoniaper- lettidia perturbons. J. Am. Mosq. Control Assoc. turb-onsadults (Diptera: Culicidae) in southwestern 1:247-250. Ontario.Can. Entomol' 113:133-139' Morris. C. D., M. Slaff and R. Parsons.1986. Investi- AmericanPublic Health Association.1981. Standard gations of methodologiesfor management of mos- methodsfor the examinationof water and waste- quito populations in phosphatemining areas.Publ. water,15th ed.APHA-AWWA-WPCF, Washing- 03-015-043,Fla. Inst. PhosphateRes., Bartow, FL' ton,DC. Nemjo, J. and M. Slaff. 1984.Head capsulewidth as Armstrong,R. L. 1941.Mansonia perturbans (Walk.) a tool for instar and speciesidentification of Man' on Cape Cod. Proc. N. J. Mosq. Exterm. Assoc. sonia d.yari, M. titillians and Coquillettidin pertur- 28:184-188. bons(Diptera: Culicidae). Ann. Entomol. Soc.Am. Batzer, D. P. and R. D. Sjogren. 1986.Larval habitat ??:633-635. characteristicsof Coquillettidiaperturbans (Diptera: Norusis,M. J. 1986.SPSS/PC+ for the IBM/PCIST/ Culicidae)in Minnesota.Can. Entomol. 118:1193- AT. SPSSInc., Chicago. 1198. Norusis,M. J. 1988a.SPSS/PC+ V3.0 Update manual Baumgartner, D. L. 1987. Laboratory evaluation of for the IBM PCIXT/AT and PS/2. SPSS/PCInc., the bladderwort plant, Iltricularia uulgaris (Lenti- Chicago. bulariaceae),as a predator of late instar Culexpip- Norusis, M. J. 1988b.SPSS/PC+ Advanced statistics iens and assessmentof its biocontrol potential' J' V2.0 for the IBM PCIXT/AT and PS/2. SPSS/PC Am. Mosq. Control Assoc.3:504-507' Inc., Chicago. Bidlingmayer, W. L. 1950. Some observationsof the Olds. E. J.. R. W. Merritt and E. D. Walker. 1989. 460 JouRNu, oF THE AlrnRrclN Mosqurro CoNrnol AssocllrroN VoL.6,No.3

Sampling, seasonalabundance and mermithid par- proceduresof statistics,with specialreference to the asitism of lawal Coquill.ettid,iaperturbans in south- biologicalsciences. McGraw-Hill, New York. central Michigan. J. Am. Mosq. Control Assoc. Tarver, D. P., J. A. Rodgers,M. J. Mahler and R. L. 5:586-592. Lazor. 1979.Aquatic and wetland plants of Florida. Provost, M. W. 1976.Mansonia mosquitoes:genera- FIa. Dep. Nat. Res.,Tallahassee, FL. tions per year in Florida.J. Fla. Anti-Mosq. Assoc. Taylor, L. R. 1961. Aggregation,variance and the 47:25-27. mean.Nature, Lond. 189:732-735. Slaff, M. and J. D. Haefner. 1985.Seasonal and spatial Taylor, L. R. 1984. Assessing and interpreting the distribution of Mansonia dyari, Mansonia titiltans spatial distributions of insect populations. Annu. and Coquillettidiaperturbans (Diptera: Culicidae) in Rev. Entomol. 29:321-357. the central Florida, USA, phosphateregion. J. Med. Wadley, F. M. 1967.Experimental statistics in ento- Entomol. 22:624-629. mology. Graduate School Press, USDA, Washing- Steel,R. G. D. and J. H. Torrie. 1960.Princioles and ton. DC.