Hydrobiologia 519: 19–27, 2004. © 2004 Kluwer Academic Publishers. Printed in the Netherlands.
19
A spatially explicit computer model for immature distributions of Glyptotendipes paripes (Diptera: Chironomidae) in central Florida lakes Richard J. Lobinske1,2 , Jerry L. Stimac2 & Arshad Ali1,2 1 University
of Florida, IFAS, Mid-Florida Research and Education Center, 2725 Binion Road, Apopka, FL 32703-8504, U.S.A. E-mail:
[email protected] 2 University of Florida, IFAS, Department of Entomology and Nematology, P.O. Box 110620, Gainesville, FL 32611-0620, U.S.A. Received 4 April 2003; in revised form 6 November 2003; accepted 11 November 2003
Key words: Chironomidae, population biology, distribution, model, nuisance midges, Glyptotendipes paripes
Abstract Density and spatial distribution data of immature Glyptotendipes paripes Edwards (Diptera: Chironomidae), in relation to selected water and sediment characteristics prevailing in three central Florida lakes, were used to develop and calibrate a qualitative model of G. paripes distributions with the purpose of developing more efficient sampling plans for research or population management. For the model, each lake habitat was defined by a twodimensional matrix, while a three-dimensional matrix was used to simulate the life stage-structured population. A Lefkovitch population matrix was used to project survival and development of the population in each spatial unit. The model incorporated lake bathymetry, sediment dry weight, water level, food availability (as a function of Secchi disk transparency) and strength of density dependence as influences on immature survival and adult fecundity. Temperature-dependent development data for G. paripes were incorporated to estimate development time. A simple redistribution function was used to simulate the dispersal of adults over the simulated habitat for oviposition. Immatures and relevant environmental data taken for eight dates, at each of two other central Florida lakes were used to validate the model. The mean correct prediction rate of the model for field density distributions within 0.5 log(n+1) immatures/m2 density in spatial habitat cells was 0.64 and 0.66 for the validation lakes. Presence/absence correct prediction rates were 0.61 and 0.76, while matching to a proposed sampling stratification based on a treatment threshold was 0.85 and 0.88. These values indicate that the model is efficient for preparing stratified immature G. paripes sampling plans for central Florida lakes.
Introduction The non-biting midge Glyptotendipes paripes Edwards (Diptera: Chironomidae) is one of the primary nuisance midge species in central Florida (Ali & Fowler, 1985). Large swarms of midges frequently emanating from lakes situated amid urban and suburban areas cause nuisance and economic problems that cost lakefront residents and businesses millions of dollars annually (Anonymous, 1977; Ali, 1995). Chironomid-related medical problems, primarily human allergies, arise from the larval hemoglobins remaining on the epidermis of adults, fragments of
which can be inhaled and cause conjunctivitis, rhinitis, hay fever, or asthma (Gad El Rab et al., 1980; Kagan et al., 1984; Giacomin & Tassi, 1988). Recently, the cholera pathogen has been isolated from chironomid egg masses (Broza & Halpern, 2001). A systematic research program on bionomics and management possibilities of nuisance midge populations in central Florida has continued for the past two decades (Ali, 1996). These studies indicated the need for a model of G. paripes immature populations in central Florida lakes as a tool for efficient sampling and management planning. Data on G. paripes distribution in three central Florida lakes (Dora, Yale and
20 Table 1. Matrix of relative habitat suitability by water depth and sediment dry weight used to generate base habitat maps for simulated immature Glyptotendipes paripes population model. Sediment % dry weight
Water depth (m) 7
0–10 10–20 20–30 30–40 40–50 50–60 60–70 70–80
0.043 0.003 0.010 0.150 0.876 0.864 0.600 0.443
0.310 0.077 0.091 0.061 0.078 0.227 0.493 0.922
0.321 0.018 0.002 0.002 0.046 0.002 0.007 0.090
0.400 0.002 0.001 0.001 0.002 0.001 0.002 0.050
0.367 0.001 0.004 0.005 0.008 0.002 0.001 0.001
0.001 0.001 0.001 0.001 0.001 0.001 0.001 0.001
0.063 0.034 0.354 0.428 0.480 0.813 0.807 1.000
0.192 0.111 0.181 0.333 0.296 0.650 0.533 0.877
Wauburg) (Lobinske et al., 2002a) were used to develop and calibrate a computer model of immature G. paripes spatial and temporal distributions, while G. paripes immature data from two other central Florida lakes, Jesup and Monroe (Ali et al., 1998) were used for qualitative validation of the model. Since insects develop through multiple immature life stages before reaching adulthood, life stagestructure was included in the model that considered survival and development of each life stage. To examine the heterogeneous lake ecosystem, the model included spatial information on habitat suitability for G. paripes (Lobinske, 2001; Lobinske et al., 2002a). Allen et al. (1996, 2001), Brewster & Allen (1997) and Brewster et al. (1997) developed techniques to incorporate specific habitat requirements and life stagestructure into computer models of insect populations. These models use matrix functions to provide discrete spatial information on habitat suitability and modeled populations. The current model expands on these techniques to include more detailed habitat information in the model that can be adjusted by initial conditions used for each simulation. The objective of this model was to qualitatively predict likely distributions of G. paripes immature populations in high densities (nuisance levels) in central Florida lakes. This would enhance the capacity of researchers and lake managers in planning efficient management of the pest. The model will provide a tool to develop stratified sampling plans based on existing lake conditions, requiring relatively little monitoring effort in areas with very low densities and greater monitoring effort in areas with high densities or the potential for high densities.
Materials and methods The current model used the spatially-specific matrix model techniques developed by Allen et al. (1996, 2001), Brewster & Allen (1997) and Brewster et al. (1997) with the computer software Matlab (The Mathworks Inc., Natik, MA). In general, a 50 × 50 matrix defined habitat suitability at each geographic location (each grid unit representing approximately 160 × 160 m area), and a 50 × 50 × 7 matrix simulated each life stage at each geographic location in the model (Fig. 1). The following model parameters must be entered for each simulation: Secchi disk transparency (cm), water temperature ( ◦ C) and deviation of water level from historic mean (m). For each iteration of the model (calibrated to represent three days), survival, development to next life stage, reproduction and dispersal were estimated for the simulated population. The habitat matrix was modified by lake water level and influenced the simulated G paripes survival, development and reproduction. Output included a final estimate of midge density throughout the simulated habitat. Larval and pupal G. paripes densities and distributions with associated selected water and sediment physico-chemical data collected over two years from Lakes Dora, Wauburg and Yale (central Florida), reported by Lobinske et al. (2002a), were used to calibrate the model. The relationship of G. paripes immatures (excluding eggs) density with sediment dry weight (DW), water depth and other benthic influences, such as presence of clay, detritus or rooted vegetation from these data were used to produce the habitat suitability maps used in the model. For simulations, lake bathymetry maps and sediment DW maps
21 Table 2. Correct classification rates of immature Glyptotendipes paripes distribution model for fit to immatures density [within 0.5 log(n + 1) G. paripes density], presence/absence [log(n+1) immature G. paripes density 1000 G. paripes immatures m−2 ) and those that could
potentially reach population management threshold (100–1000 G. paripes immatures m−2 ). Model output grid cells that predicted the low effort value and those that predicted high effort areas were compared to field data for matching within the same range.
Results Example result of simulation and field immature data for Lake Jesup during May 1996 are shown in Figure 2. The model identified the locations of relatively high-density areas to the southeast of the small island in the center of the lake as well as along sections of the northern shore and the northeastern tip. The simulated population had a wider geographic distribution than indicated by the field data. Comparison of geographic distribution of immatures between the simulated population and April 1996 field data from Lake Monroe is shown in Figure 3. The simulated and field populations both had moderate population densities (100–1000 immatures m−2 ) near the northwest shore and in the eastern bays, and lower densities along the northern and southern shoreline. The general spatial agreement of immature distributions in the simulated and field populations indicates that the model can predict likely distributions of immatures. Comparison of G. paripes density of individual grid cells (within 0.5 log unit) between field data and model projections for eight data sets each from Lakes Jesup and Monroe revealed mean ± SD values of 0.66 ± 0.12 matching (Lake Jesup) and 0.64 ± 0.09 matching (Lake Monroe) (Table 2). The mean ± SD values of correct classification rate for presence/absence by the model was 0.61 ± 0.11 for Lake Jesup simulations and 0.76 ± 0.03 for Lake Monroe simulations (Table 2). Mean ± SD correct classification rate between model and field data for sampling strata values for Lake Jesup was 0.85 ± 0.05, and for Lake Monroe 0.88 ± 0.08 (Table 2). For purposes of enhancing efficient use of sampling efforts to determine effective sample stratification plans in central Florida lakes, these data show an excellent fit and should prove useful.
Discussion A matrix model was successfully used to model the whitefly, Bemisia argentifoli Bellows & Perring, populations in a crop-system (Brewster et al., 1997). While
25
Figure 3. Comparison of model output (bottom) of Glyptotendipes paripes immatures distribution in Lake Monroe, central Florida, under matching conditions with April 1996 field collected data.
that model was applied to a terrestrial habitat complex, the core concepts of mapping habitat suitability are equally valid when applied to aquatic systems. The primary difference is that the Brewster et al. (1997) model used crop type as the primary factor of habitat suitability, whereas the current midge model uses a combination of sediment conditions and lake bathymetry as the primary factors. Otherwise, they are similar, with the immature insects developing in well-defined habitats that influence survival and global (evenly applied to all spatial units) parameters like
temperature that influence development rate. For both models, immatures remain in the same habitat unit through most or all of their development, and dispersal occurs primarily through adult flight and oviposition. The present model produced acceptable matching with field data for absolute density and presence/absence of immature G. paripes in the two central Florida validation lakes. It was efficient at predicting if immature G. paripes distributions in the validation lakes matched a proposed stratified sampling plan based on a population management threshold. However, there
26 remain some differences between the model and the validation data and the reasons for these are presently not fully understood. Nevertheless, it is reasonable to expect some degree of difference between a model and field data since any model will not be able to accurately include all influences on a simulated population. A preliminary investigation of fish predation on midge larvae in central Florida lakes (Lobinske et al., 2002c) indicated that bluegill (Lepomis macrochirus Rafinesque) was a generalist predator on midge larvae. Investigation is continuing to further elucidate predation pressure on G. paripes larvae to include in the model. Additional work for model improvement may include determining the actual carrying capacity of different sediments and examining linear and nonlinear density dependence effects, which would allow for a more precise definition of density dependence influence on survival. This model used simple assumptions on immature G. paripes mortality. Detailed development and use of a life table would be an enhancement in the further development of population models for this species. Also, operating the present model by using larger matrix maps to simulate the lake systems, such as 100 × 100 or 200 × 200 grid units, would possibly diminish or eliminate any scale-induced effects. Furthermore, calibrating grid units to different geographic areas will be investigated, as will more refining of the adult distribution function. This could be very important since adult midges are highly phototactic (Ali et al., 1984) and their dispersal behavior is often directly influenced by artificial lights. Quantifying the effects of artificial lights on adult dispersal and female oviposition may help to remove some of the current inconsistencies in the model predictions. Some of the other anthropogenic influences that should be investigated are shoreline alteration (planting or removal of vascular plants), water craft activity that may disturb sediments (e.g., shipping channels), and thermal effluents, such as from electric power plants that can locally effect development times of immature G. paripes. The current version of the model should work well in identifying the proposed sampling strata for immature G. paripes in central Florida lakes, given the availability of necessary bathymetric and environmental data. For researchers, this will assist in development of sampling plans that take into account the heterogeneous distribution of midge immatures, allowing for better population estimates of immatures in relatively less time. With enhanced population estimates, the ability to detect environmental influences on
the populations will increase. For lake managers, the model will indicate areas that will need closer monitoring for immature densities capable of producing adults at nuisance levels, and allow for the directed, more efficient use of resources for their management. In this context, targeted application of a larvicide to areas with high populations could reduce control costs considerably, and simultaneously diminish any possible adverse effects of the control agent on non-target biota. For the possible use of behavioral control methods, such as adult midge attraction to light (Ali et al., 1984), knowledge of high density areas in a lake will aid in the effective placement and execution of such measures. In some man-made or natural lakes where water levels can be manipulated, this knowledge may facilitate reducing areas of high immature populations by adjusting water levels.
Acknowledgements Gratitude is expressed to Dr Jon C. Allen for technical assistance and training with Matlab , Mr Robert J. Leckel Jr., for field and laboratory assistance, and Dr Jan Frouz for analytical and editorial assistance. This is Florida Agricultural Experiment Station Journal Series No. R-08709.
References Allen, J. C., C. C. Brewster, J. F. Paris, D. G. Riley & C. G. Summers, 1996. Spatiotemporal modeling of whitefly dynamics in a regional cropping system using satellite data. In Gerling, D. & R. T. Mayer (eds), Bemisia 1995: Taxonomy, Biology, Damage Control and Management. Intercept, Andover: 111–124. Allen, J. C., C. C. Brewster & D. H. Slone, 2001. Spatially explicit ecology models: A spatial convolution approach. Chaos Solitons and Fractals 12: 333–347. Ali, A., 1995. Nuisance, economic impact and possibilities for control. In Armitage, P. D., P. S. Cranston & L. C. V. Pinder (eds), The Chironomidae: The Biology and Ecology of Non-biting Midges. Chapman and Hall, London: 339–264. Ali, A., 1996. Pestiferous Chironomidae and their management. In Rosen, D., F. D. Bennett & J. L. Capinera (eds), Pest Management in the Subtropics: Integrated Pest Management – A Florida Perspective. Intercept, Andover: 487–513. Ali, A. & R. C. Fowler, 1985. A natural decline of pestiferous Chironomidae (Diptera) populations from 1979 to 1984 in an urban area of central Florida. Florida Entomologist 68: 304–311. Ali, A., W. D. Gu & R. J. Lobinske, 1998. Spatial distribution of chironomid larvae (Diptera: Chironomidae) in two central Florida lakes. Environmental Entomology 27: 941–948. Ali, A., S. R. Stafford, R. C. Fowler & B. H. Stanley, 1984. Attraction of adult Chironomidae (Diptera) to incandescent light under laboratory conditions. Environmental Entomology 13: 1004–1009.
27 Anonymous, 1977. Economic Impact Statement. Blind Mosquito (Midge) Task Force, Sanford Chamber of Commerce. Seminole County, FL. Brewster, C. C. & J. C. Allen, 1997. Spatiotemporal model for studying insect dynamics in large-scale cropping systems. Environmental Entomology 26: 473–482. Brewster, C. C., J. C. Allen, D. J. Schuster & P. A. Stansly, 1997. Simulating the dynamics of Bemisa argentifolii (Homoptera: Aleyrodidae) in an organic cropping system with a spatiotemporal model. Environmental Entomology 26: 603–616. Broza, M. & M. Halpern, 2001. Chironomid egg masses and Vibrio cholerae. Nature 412: 40. Fielding, A. H. & J. F. Bell, 1997. A review of methods for the assessment of prediction errors in conservation presence/absence models. Environmental Conservation 24: 38–39. Gad El Rab, M. O., D. R. Thatcher & A. B. Kay, 1980. Widespread IgE-mediated hypersensitivity in the Sudan to the ‘green nimitti’ midge Cladotanytarsus lewisi (Diptera: Chironomidae). II. Identification of a major allergen. Clinical and Experimental Immunology 41: 389–396. Giacomin, C. & G. C. Tassi, 1988. Hypersensitivity to chironomid Chironomus salinarius (non-biting midge living in the Lagoon of Venice) in a child with serious skin and respiratory symptoms. Bolletino First Seroterapico Milan 67: 72–75. Kagan, S. L., J. W. Yunginger & R. Johnson, 1984. Lake fly allergy: incidence of chironomid sensitivity in an atopic population. Journal of Allergy and Clinical Immunology 73: 187. Lefkovitch, L. P., 1965. The study of population growth in organisms by stages. Biometrics 21: 1–18.
Lobinske, R. J., 2001. Ecological studies of larval Glyptotendipes paripes (Chironomidae: Diptera) in selected central Florida lakes for creating an exploratory temporal and spatial model of nuisance populations. Ph.D. Dissertation, University of Florida, Gainesville, FL. Lobinske, R. J., A. Ali & J. Frouz, 2002a. Ecological studies of spatial and temporal distributions of larval Chironomidae (Diptera) with special emphasis on Glyptotendipes paripes in three central Florida lakes. Environmental Entomology 31: 637–647. Lobinske, R. J., A. Ali & J. Frouz, 2002b. Laboratory estimates of degree-day developmental requirements of Glyptotendipes paripes (Diptera: Chironomidae). Environmental Entomology 31: 608–611. Lobinske, R. J., C. E. Cichra & A. Ali, 2002c. Predation by bluegill (Lepomis macrochirus) on larval Chironomidae (Diptera) in relation to midge standing crop in two central Florida lakes. Florida Entomologist 85: 372–375. Ricker, W. E., 1954. Stock and recruitment. Journal of the Fisheries Research Board of Canada 11: 559–623. Stevens, M. M., 1998. Development and survival of Chironomus tepperi Skuse (Diptera: Chironomidae) at a range of constant temperatures. Aquatic Insects 20: 181–188. Xue, R. D., A. Ali & R. J. Lobinske, 1994. Oviposition, hatching and age composition of a pestiferous midge, Glyptotendipes paripes (Diptera: Chironomidae). Journal of the American Mosquito Control Association 10: 24–28.
