Absolute Density Estimates Research Articles

Structure et dynamique d’une population d’Agama impalearis (Sauria : Agamidae) dans les Jbilets centrales, Maroc

A population study of Agama impaleari was conducted during two consecutive years (1993-1994) in a Jujube bush site located in an arid area at the Jbilet Mountains (Western Morocco). Lizards of both sexes required approximatively one year to reach sexual maturity. One to two clutches were produced per breeding season (May-August) and clutch size was significantly correlated with maternal body size. The annual reproductive output was in average of 20 eggs per female per year and did not exhibit a significant difference between the two years of study. Population size showed important seasonal changes throughout the study period and absolute spring density estimates were of 15 and 11 ind/ha, respectively in 1993 and 1994. In the post-reproductive period (September) the population density (excluding hatchlings) markedly declined and was only of 4-5 ind/ha. After the completion of hatching period (July to early November), the densities of hatchlings were 11 and 13 ind/ha respectively in 1993 and 1994 compensing for the population losses. The annual turnover rate of the population did not exhibit a significant yearly change and was of 77 %. Adult survivorship after the reproductive activity varied between years and juvenile survival during the first year was estimated to be 35 %. Maximum longevity assessed by means of squelettochro-nology was 5 years.

Estimating Populations of Aphidophagous Coccinellidae (Coleoptera) in Winter Wheat

Coccinellids are important natural enemies of aphid pests in wheat, Triticum aestivum em Thell, in the Great Plains of the United States. Coccinellid community development in wheat fields is unpredictable; therefore, precise, efficient sampling methods for coccinellids are needed for use in integrated pest management research and decision making. Our objectives were to compare removal sampling with quadrat sampling for estimating population density of adult and larval coccinellids in winter wheat, and to determine if timed count sampling and sweepnet sampling were useful for estimating adult and larval coccinellid densities. Removal sampling accurately estimated population density for adults of most species but consistently underestimated larval density. Timed count samples and sweepnet samples were significantly correlated with absolute density of both larval and adult coccinellids. Regression models were developed to convert estimates of relative to estimates of absolute population density. Depending on sampling method and life stage, models included the number of tillers per 0.3 m, wheat plant growth stage, plant height, and the number of aphids per tiller as variables; R 2 values ranged from 0.89 to 0.93. Sweepnet sampling was more precise per unit of effort than timed count, quadrat, or removal sampling for estimating adult coccinellid density. Removal sampling (removing all coccinellids seen during 2 successive 15-min searches from each of 9 plots (25 m 2 ) was the least efficient method.

Local movements and population density of Water RailsRallus aquaticusin a small inland reedbed

The elusive behaviour and preference for dense aquatic vegetation has resulted in a lack of information regarding most aspects of Water Rail Rallus aquaticus ecology. This paper reports on the results of a study of Water Rails within a small (0.5 ha) inland reedbed at Newbridge-on-Wye, mid-Wales, during autumn and winter 1993–94. Birds used a substantial proportion of the reedbed (captures made 30–¸60 m apart), occupying home ranges which appeared to overlap, although different individuals nevertheless favoured different areas of the reedbed. Evidence is presented that the site was deserted during a period of freezing weather, but that birds re-occupied their original home ranges upon their return. Frequent trapping of birds also permitted an estimation of absolute population density (14 birds per ha), a figure much higher than previously recorded at other sites where birds were counted only indirectly. These data indicate that trapping can yield valuable information on aspects of the ecology of this little-known species

Comparison of Sampling Techniques for Grasshoppers (Orthoptera: Acrididae) in Soybean

A 2-yr study was conducted for evaluating sampling techniques for the grasshoppers Melanoplus femurrubrum (De Geer) and M. differentialis (Thomas) in soybean. Objectives were to establish a technique for obtaining absolute estimates of density and to evaluate the potential for calibrating the sweep net (relative density estimates) for absolute density estimates. Assessments of absolute density were made at night by using a cage-bag method. Sweep samples were collected during both day and night. Night sweeping was more effective than day sweeping for capturing grasshoppers. Regression results of relative on absolute estimates for nymphs indicated that both day and night sweeps adequately reflected population change. In contrast, similar analyses for adults indicated sufficient fidelity for night sweeps only. Day or night, fidelity of sweeps was better for M. femurrubrum than for M. differentialis . Regression coefficients were significantly different between nymphs and adults, but not between species. Therefore, combined-species regression models were formulated for nymphs (day and night) and for adults (night). Satisfactory fidelity of species-specific and combined-species day (nymphs) and night (nymphs and adults) models warrants the use of these models for calibration. Despite both day and night sweeps being biased in estimating age-class structure for nymphs, nymphal models are useful for management.

Population Dynamics of Plasmodium falciparum Sporogony in Laboratory-Infected Anopheles gambiae

The population dynamics of cultured Plasmodium falciparum parasites was examined during their sporogonic development in Anopheles gambiae mosquitoes. Estimates of absolute densities were determined for each life stage, and life tables were constructed for each of 38 experimental infections. Macrogametocyte and ookinete mortalities contributed equally to the overall mortality. On average, there was a 40-fold decrease in parasite numbers in the transition from the macrogametocyte to the ookinete stage, a 69-fold decrease in the transition from ookinete to oocyst stages, and a total net decrease in parasite numbers from macrogametocyte to oocyst stage of 2,754-fold (i.e., multiplicative). There was no relationship between macrogametocyte and ookinete densities due to the inherent variability in fertility among different gametocyte cultures. There was a curvilinear relationship (r2 = 0.66) between ookinete and oocyst densities. Above a threshold of about 30 ookinetes/mosquito, the oocyst yield per ookinete became increasingly greater with increasing ookinete density. There was a linear relationship (r2 = 0.73) between oocyst and sporozoite densities, with an average of 663 salivary gland sporozoites produced per oocyst. Sporozoite production per oocyst was not affected by oocyst density and virtually all oocyst infections resulted in sporozoite infections of the salivery glands. This quantitative study indicates that the sporogony of cultured P. falciparum in laboratory-infected A. gambiae is an inefficient process and that the ookinete is the key transitional stage affecting the probability of vector infectivity.

Gypsy Moth (Lepidoptera: Lymantriidae) Egg Mass Distribution and Sampling in a Residential Setting

The spatial distribution of gypsy moth, Lymantria dispar (L.), egg masses was determined in 60 developed lots in a residential community in Greenbelt, Md. The community was divided into low and high gypsy moth density sections, with average densities of 393.3 and 2,656.3 egg masses per ha, respectively. In the high-density lots, the proportion of egg masses found on trees, man made objects, and houses was 73.9, 21.6, and 4.5%, respectively. The distribution in the low density lots was similar. Oaks, Quercus spp., received the highest proportion of egg masses out of all tree species subcategories at both low and high population densities. The cost effectiveness of a number of potential sampling units, including entire lots, back yards, fixed area plots, and individual trees, was evaluated. The entire lot sampling units required the fewest samples to achieve a predetermined level of precision. However, cost (= sampling time) for these sampling units was also the highest, resulting in relatively low cost effectiveness. Fixed-area plot samples (0.01 ha) were in general the most cost effective sampling units. A binomial (presence-absence) sampling approach, in which the percentage of trees in each lot with >5 egg masses is determined, was nearly as cost effective as fixed-area plots, but its usefulness is limited because it does not provide direct estimates of absolute gypsy moth population density.

Seasonal Distributions and Abundances of Loggerhead and Leatherback Sea Turtles in Waters of the Northeastern United States

Seasonal distributions and abundances of loggerhead (Caretta caretta) and leath- erback (Dermochelys coriacea) sea turtles in continental shelf waters off the coast of the northeastern United States were derived from more than three years of aerial and shipboard surveys. There were 3460 sea turtle sightings, including 2841 loggerheads, 128 leatherbacks, and 491 unidentified. Relative abundance patterns, corrected for uneven distribution of survey effort, demonstrated an extensive area of loggerhead distribution from near Long Island, New York, along the mid-shelf to near Cape Hatteras, North Carolina. Areas of high relative abundance of leatherbacks were more scattered, with clusters south of Long Island and in the central and eastern Gulf of Maine. Loggerheads occurred significantly farther south (x = 38?20' N) than leatherbacks (x = 40005' N), and in significantly warmer waters (mean surface temperatures of 22.2 C and 20.4 C, respectively). The two species did not differ significantly in water depth or bottom slope at sighting; the modal depth interval for both was 21-40 m. Patterns of distribution and concentration of sea turtles differed greatly from nearly all species of marine mammals, suggesting little overlap in resource utilization. Both relative and absolute density estimates were much higher for loggerheads. Overall mean relative densities were 21.6 loggerheads per 1000 km of survey track, and 6.85 leatherbacks per 1000 km. Absolute density estimates, derived from 454 special aerial surveys, ranged from 1.64 x 10-3 to 5.10 x 10-1 loggerheads km-2 and 2.09 x 10-3 to 2.16 x 10-2 leatherbacks km-2. The maximum densities for both species were higher than values reported for the Gulf of Mexico and off the eastern Florida shore. Total study area populations during the summer were estimated at 2200-11,000 loggerheads and 100-900 leatherbacks. These estimates are minimal because they are based on observations of turtles at the surface, and may represent much greater abundances. The patterns of relative abundance derived from this analysis could be used as a basis for the designation of critical sea turtle habitat off the northeastern U.S.

Sampling of Heliothis zea (Lepidoptera: Noctuidae) and Orius insidiosus (Hemiptera: Anthocoridae) in Grain Sorghum Panicles by a Modification of the Beat-Bucket Method

A modification of the beat-bucket sample method was evaluated for Helioth18 Zea (Boddie) and Orius insidiosus (Say) in grain sorghum panicles. Grain sorghum panicles were cut and twirled inside a 19-liter plastic bucket rather than beating the panicles against the bucket wall. Based on absolute density estimates, the modified beat-bucket method recovered > 50.0% of the H. zea larvae and O. insidiosus adults and 35.0% of the O. insidiosus nymphs in the grain sorghum panicles. The method provided reliable estimates of population trends of H. zea larvae and O. insidiosus nymphs and adults. The method was calibrated for H. zea larvae. Relative density estimates (y) may be converted to absolute density estimates (x) by the formula x = y/b (b , slope of regression line).

Estimates of high absolute densities and emergence rates of demersal zooplankton from the Agatti Atoll, Laccadives

Direct sampling of the sandy sub stratum of the Agatti Lagoon with a corer showed the presence of very high densities of epibenthic forms. On average, densities were about 25 times higher than previously estimated with emergence traps. About 80% of the organisms migrated into the water column within an hour after sunset, and maximal resettlement occurred soon after dawn.

ESTIMATING ADULT COCCINELLID POPULATIONS IN WHEAT FIELDS BY REMOVAL, SWEEPNET, AND VISUAL COUNT SAMPLING

AbstractRemoval sampling, sweepnet sampling, and visual count sampling conducted while walking at constant velocity through a field were used to sample populations of adults of five coccinellid species in plots established in spring wheat fields, Triticum aestivum L. Estimates of absolute population density obtained from two 20-min removal samples taken from each of six 5- by 5-m sub-plots per plot proved reliable and were used to convert estimates obtained from sweepnet sampling (180 sweeps per plot) and visual counts (36 min per plot walking at 10 m/min) into absolute estimates of population density. Population density estimates obtained by removal sampling were quite precise except for species with low capture efficiencies and low population densities.Crop plant growth stage influenced the numbers of beetles caught in sweepnet samples and was incorporated in regression models for converting sweepnet catch to absolute density. Values of R2 of regressions ranged from 0.51 to 0.90, depending on species. Visual counts were influenced by temperature and aphid density, and these variables were incorporated in regression models. Values of R2 for regressions relating visual counts to population densities ranged from 0.63 to 0.94, depending on species.

Density of Northern Spotted Owls in Northwest California

We estimated densities of northern spotted owls (Strix occidentalis caurina) over 4 years using intensive surveys of marked individuals on a 292-km2 study area in northwest California. For the 4-year sampling period, estimates from empirical and Jolly-Seber methods were in close agreement, yielding a crude density of 0.235 owls/km2 (95% CI = 0.214-0.256) and ecological densities of 0.544 (95% CI = 0.495-0.592) and 0.660 (95% CI = 0.601-0.719) owls/km2 of suitable habitat, using 2 definitions of suitable habitat. Increases in density over the 4 years were attributed to population processes, such as immigration, rather than internal increases within the sampled area. Densities in our area were relatively high when compared with other populations. Our study design should be useful for making geographic and habitat comparisons and for monitoring spotted owl populations. Estimates from our study and the U.S. Forest Service indicate a 60.0-82.5% reduction of current spotted owl populations on at least 1 National Forest under current management plans by the U.S. Forest Service. J. WILDL. MANAGE. 54(1):1-10 Over the past decade, the northern spotted owl has become the focus of controversial management policies in the Pacific Northwest because of the intimate association of the spotted owl with old-growth coniferous forests (Forsman et al. 1984, Gutierrez and Carey 1985, Dawson et al. 1987, Simberloff 1987, Thomas et al. 1988). Thus, an important element in their management will be estimating and monitoring changes in abundance of spotted owls (Dawson et al. 1987). Animal abundance can be measured as the number of individuals in the population (population size), the number of individuals per unit area (absolute density), or the density of 1 population relative to another (relative density) (Caughley 1977:12). Absolute density can be further subdivided into crude density-measured with respect to all of the area containing the population, and ecological density-measured with respect to a particular habitat (Johnson 1978:11, Tanner 1978:2). Absolute density is necessary for determining population and trophic dynamics (Verner 1985) and effective population sizes of demes (Barrowclough and Coats 1985). Relative density has been used to make comparisons between spotted owl populations and habitats (Forsman et al. 1977, Garcia 1979, Marcot and Gardetto 1980). However, these measures require standardization by some measure of absolute density (Caughley 1977:14, Verner 1985) and similar detection probabilities of individuals (Andersen et al. 1985) to ensure that relative measures provide an accurate index of abundance. Therefore, absolute density provides an initial step in estimating abundance. In this paper, we report empirical and markrecapture estimates of absolute density of northern spotted owls in northwestern California from 1985 through 1988. Our objectives were to estimate absolute density of northern spotted owls, to provide general guidelines for estimating density of northern spotted owls, and to evaluate our estimates in relation to current management plans. Our study would not have been possible without the field assistance of K. E. Young, C. A. Moen, J. A. Blakesley, T. J. Evans, and M. B. Kasper. C. A. Moen and J. A. Blakesley also assisted in data compilation. We would like to thank D. M. Solis, J. Kahl, D. Kudrna, J. L. Mattison, L. Roberts, A. C. Chrisney, M. Fox, and M. Smith of the Six Rivers National Forest for their help in providing information during this study. R. Rhino provided transportation. G. F. Barrowclough, J. A. Blakesley, J. A. Crawford, E. D. Forsman, W. S. LaHaye, J. D. Nichols, B. R. Noon, D. M. Solis, N. G. Tighlman, and J. Verner reviewed earlier drafts of this manuscript. Funding was provided by California Department of Fish and Game, Federal Aid

A Weighted Duration-in-Beam Estimator for the Volume Sampled by a Quantitative Echo Sounder

Estimation of absolute fish density or abundance using the echo-count technique requires estimation of the volume sampled. The unknown parameter in the volume sampled is the sine of the half-angle of the effective conical beam. We derive the minimum variance, unbiased, linear estimator of this parameter. This estimator is based upon weighted regression through the origin and reduces to a computationally simple form. We use this estimate and its variance to estimate absolute density, absolute abundance, and their variances. We use a bootstrap procedure to contrast the performance of our method against that of a contemporary method, which is based upon the averages taken over range strata. Although both of these methods are unbiased, we show that our estimator has a standard deviation that is 23.5% smaller than that of the average-over-strata method.

Density measurements of Hg(63P0,1,2) in a discharge using saturated laser absorption and hook methods

Two laser-based techniques are used to derive radial density distributions and absolute densities for the three 63PJ excited states of mercury in the positive column of a low-pressure Hg-Ar dc discharge. In the saturated laser absorption (SLA) method, a pulsed dye laser, traversing the discharge axially, is scanned through one of the 63PJ→73S1 transitions. At high laser powers the power absorbed from the beam on resonance approaches a constant saturated value directly proportional to the number density of Hg atoms in the 63PJ state selected. On-axis absolute density estimates and radial distribution results are presented for the three states at several discharge conditions. The hook method, a well-known interferometric technique, more accurately establishes the absolute densities on axis, and several off-axis measurements are compared with SLA results.

Visibility Bias in Aerial Surveys: A Review of Estimation Procedures

Aerial surveys, often used to estimate the density of wildlife populations, commonly underestimate population density because of animals being missed. This visibility can be serious. Methods of estimating visibility based on comparison with ground counts, use of a subpopulation of marked animals, mapping with multiple observers, line transect sampling, and multiple counts on the same area are reviewed. All of the methods are model based and the various model assumptions are discussed and compared. J. WILDL. MANAGE. 51(2):502-510 Aerial surveys are used to estimate the population size of many large birds (Caughley and Grice 1982), large terrestrial mammals (Rice and Harder 1977), and marine mammals (Burnham et al. 1980:76). They are used in conjunction with other methods of estimation or by themselves, either by choice or necessity. Visibility bias results from animals being missed and is exacerbated by factors such as dense vegetation, bad weather conditions, and observer fatigue. Beyond the logistics and dangers of flying surveys precisely, the principal difficulty in aerial censuses lies in estimating the visibility to establish a correction factor. Caughley (1977: 35) presents a table showing the visibility in a wide range of aerial surveys. Biologists tempted to minimize this problem should study this table where the visibility is often in the range of 30-50% animals missed! If absolute density estimates are to be reported there will always be the need for a correction factor. Here we review the various methods of estimating visibility from an applied statistician's perspective. Neither of us claim to be experts in the practicalities of running aerial

Predator Activity and Predation in Corn Agroecosystems

Predation of lepidopterous larvae in corn agroecosystems was examined by observing tethered live larvae over several 24-h periods. There were significantly more attacks on larvae and more larvae were consumed in no-tillage systems compared with conventional tillage treatments. Predator activity (number of attacks) and predation were greatest at night and least in early afternoon over the entire season and generally increased as the season progressed. The soil insecticide phorate significantly decreased predators and number of pest larvae consumed for ca. 2 months after its application. Large carabids (≥ 15 mm) were the most important predatory arthropods in these agronomic systems; however, noncarabid species also provided significant predation. Significant correlations were found among carabid activity, absolute density estimates of carabids, and number of lepidopteran larvae consumed in these corn agroecosystems. These high correlations indicate that the technique demonstrated the potential of endemic predatory arthropods to reduce soil larvae pests significantly.

Application of Dual-Beam Acoustic Survey Techniques to Limnetic Populations of juvenile Sockeye Salmon (Oncorhynchus nerka)

Two hydroacoustic surveys of Cultus Lake, British Columbia, were conducted in July 1983 and February 1984 to estimate the size and distribution of the juvenile sockeye salmon (Oncorhynchus nerka) population. The surveys demonstrated the application of a combined dual-beam/echo integration technique for obtaining accurate abundance estimates and other quantitative data. The dual-beam system measured in situ the target strengths (and backscattering cross section) of individual fish. For each survey the mean backscattering cross section of the sampled population was used to scale the integrator outputs to absolute fish density and abundance estimates. Based on previously derived empirical formulas, the mean target strengths compared well with the mean lengths of fish captured by trawl. For this monospecies population with a dominant single-size group of fish, the variance in integrator outputs was due almost entirely to the spatial distribution of fish and not to the variance in measured backscattering cross sections. The confidence intervals of biomass estimates depend on autocorrelation between consecutive acoustic samples, which is related to the spatial distribution of the surveyed population. Trawl catches indicated that about 95% of the fish in the lake were juvenile sockeye salmon. Population estimates made by hydroacoustics were consistent with other available biological data.

Sweep-net Sampling for Western Spotted Cucumber Beetle (Coleoptera: Chrysomelidae) in Snap Beans: Daily Fluctuations in Beetle Counts and Correlation Between Relative and Absolute Density Estimates

Sweep-net sampling was evaluated as a method of sampling populations of adult western spotted cucumber beetles (WSCB), Dtabrotica undecimpunctata undecimpunctata Mannerheim, in snap beans, Phaseolus vulgaris L. Each of 18 fields was sampled initially between 0800 and 1000 hours; 1–6 subsequent sets of samples were collected later on the same date. Field means of beetle counts in sweep-net samples collected after the initial sampling episode (subsequent means) were significantly lower than the field mean of initial samples collected in the same field (initial mean) in 16 of 42 comparisons. Subsequent mean/initial mean ratios were negatively correlated ( R 2 = 0.72) with the amount of increase in solar radiation intensity between initial and subsequent sampling episodes. When fluctuations in beetle counts in sweep samples were described in relation to an average daily radiation curve, estimated counts were lowest for sampling conducted at 1300 hours. Time based correction factors used as multipliers to stabilize relative density estimates according to a standard (0900 hours) sampling time ranged from 0.9 for 1800 hours to 2.5 for 1300 hours. The linear relationship between corrected counts of WSCB in sweep-net samples and measures of WSCB absolute density (beetles per meter of row) was described by the equation: y = 0.44 + 2.55 x ( R 2 = 0.69), where x = number of WSCB per meter of row and y = corrected number of WSCB per 10 sweeps.

Rates of Prey Consumption by Neotropical Robber Flies (Diptera: Asilidae)

Adult robber flies are sit-and-wait predators that attempt aerial capture of flying insects. Estimates of daily rates of prey consumption per individual were combined with estimates of absolute population densities to obtain daily, populational rates of prey consumption for nine robber fly species in a lowland neotropical forest. Estimates of biomass consumed/day/ha/population varied from 23 to 1425 mg (dry weight) among the nine species, and total daily consumption for all species was approximately 4600 mg (dry weight)/ha. This latter value was relatively low compared to estimates available for other insectivorous taxa. Although many studies have measured consumption rates of insect herbivores (Teal 1962, Van Hook 1971, Reichle et al. 1973) or detritivores (Kowal & Crossley 1971, McBrayer & Reichle 1971, Santos & Whitford 1981), few studies provide estimates of prey consumption (i.e., ingestion) rates by insect predators. I am aware of only two studies in nonagricultural ecosystems that estimate prey biomass consumed per unit area of habitat per unit time by populations of predatory insects, and both of these concern populations of nymphal stoneflies (Plecoptera) in alpine streams of western North America. Siegfried and Knight (1976) reported an annual prey consumption estimate of 3.7 g/m2 (dry mass) for Calineuria californica in a Californian stream. Allan (1982) studied three stonefly species in Colorado and estimated total annual prey consumption by all species to be 3.0 g/m2 (dry mass). Both studies suggested that predatory stoneflies consume a significant fraction of the prey biomass available. Previous reports (Shurovnekov 1962, Joern & Rudd 1982) also suggest that robber flies have a major impact upon prey populations. However, the data collected do not allow calculation of consumption rates in terms of prey biomass taken per unit area and consequently can not be directly compared with consumption rates for other predatory species. Here I present estimates of daily prey consumption rates for populations of nine robber fly species in a lowland neotropical forest. Adults of all nine species are visual, sit-and-wait predators that attempt aerial capture of flying insect prey. Following a successful capture, individuals return to a perch and consume the liquid contents of the prey. Since attack rate, proportion of sucI Received 16 October 1984, revision accepted 16 April 1985. 2 Present address: Pacific Biomedical Research Center, University of Hawaii at Manoa, Honolulu, Hawaii 96822, U.S.A. 166 BIOTROPICA 18(2): 166-17

Estimating Absolute Density from Relative Sampling of Lygus lineolaris (Heteroptera: Miridae) and Selected Predators in Early to Mid-season Cotton

Functional relationships of Lygus lineolaris (Palisot de Beauvois) and selected predator densities estimated with sweep net, drop cloth, whole-plant-bag samples (WPBS), and visual search were determined in early to mid-season cotton. Slopes of relative as a function of absolute densities were not different between years for most taxa. When compared against absolute density estimates (WPBS or visual search) drop-cloth estimates were generally more precise and captured a larger percentage of arthropods than 10-sweep samples of the sweep net. With few exceptions (notably Geocoris spp. nymphs) the proportion of the absolute density estimated by drop-cloth or sweep-net samples was low (ca. 10–30%). Reasonably precise relationships ( R 2> 0.50) of relative over absolute density were obtained for nymphs of Geocoris spp., L. lineolaris , and Orius spp.; Orius spp. adults; and spiders. Relationships using total predator counts generally had greater precision than those using individual species and stages. Adults of Geocoris spp., Coccinellidae, and possibly L. lineolaris , however, were not estimated with precision by drop-cloth or sweep-net samples.

Abundance and distribution of Halobates micans (Insecta : Gerridae) in the northwest Atlantic

A seven-year survey of the marine water-strider Halobates micans in the northwest Atlantic Ocean showed highest abundance of the insects in the Caribbean Sea and Gulf Stream. The Bahama Islands and northern Sargasso Sea regions yielded relatively low abundances, and few H. micans were collected north of 35°N latitude. Regional patterns of insect abundance appear to be directly related to surface water temperature, but biotic factors such as predation and abundance of gelatinous food organisms may be important. Water strider abundance does not appear to be related to regional differences in the abundance of floating substrata. Estimates of absolute density ranged from 0 to 53,000 H. micans km −2, with densities in tropical Caribbean waters averaging 7000 km −2. Using data from 9 to 48°N latitude as an extension of data from the central Atlantic ( Cheng and Schulz-Baldes, 1981, “Meteor” Forschungsgemeinschaft-Ergebnisse, 33, 17–21), density of H. micans approximates an inverse logarithmic function of latitude. In those geographic regions where repeated annual transects have been made, the year to year variations in standing stocks of the insect are low, except in the southwest Sargasso Sea.