Southeastern Naturalist J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 138 2016 SOUTHEASTERN NATURALIST 15(1):138–152 Seasonal Specialization and Selectivity of the Eastern Mosquitofish, Gambusia holbrooki, toward Planktonic Prey Joseph M. Dirnberger1,* and Julia Love2 Abstract - Gambusia holbrooki (Eastern Mosquitofish) are often characterized as opportunistic surface feeders. We examined seasonal shifts in prey use, feeding specialization, and prey electivity to determine whether a feeding bias existed toward small planktonic prey. We sampled invertebrates (from the water column and from near plant surfaces) and Eastern Mosquitofish over 4 seasons in a wetland (Cobb County, GA). Gut analysis indicated seasonal shifts from large to small prey, particularly toward cladocerans, even when larger prey remained abundant. Small planktonic prey were consumed by all sizes of Eastern Mosquitofish. Electivities for cladocerans tended to be positive, whereas electivities for other prey, and especially copepods, were negative. Prey that would be expected to aggregate at the air–water interface were consumed in much greater proportions than those sampled elsewhere in the environment, suggesting that Eastern Mosquitofish are able to consume cladocerans caught in the surface tension. Introduction Mosquitofish of the genus Gambusia are often the numerically dominant fish within shallow, macrophyte-dominated areas (Blanco et al. 2003, Garćia-Berthou 1999, Pyke 2008). Their success as one of the most extensively distributed freshwater fish genera is likely due to a combination of attributes including rapid maturation after live birth, multiple broods within a season from stored sperm, tolerance to a wide variety of physical conditions, and generalist diets (Peck and Walton 2008, reviewed in Pyke 2005). Dietary habits of Gambusia have been difficult to characterize. Some studies indicate predation focused on mosquito larvae (based on reduction in larval density in the presence of mosquitofish; e.g., Hoy et al. 1972, Peck and Walton 2008) and on other macroinvertebrates (e.g., Oliver 1991, Specziár 2004). Though mosquitofish are reported to feed on insects at the water surface (Pyke 2005) using a dorsally oriented mouth situated on a flattened head, several studies note predation on microcrustaceans and rotifers (e.g., Blanco et al. 2003, Gkenas et al. 2012, Mieiro et al. 2001) that typically are an order of magnitude smaller in length and are often planktonic (e.g., Lair et al. 1996, Pennak 1966). Some evidence suggests that despite a morphology adapted for surface feeding, mosquitofish can be effective planktivores. Experimental pond and enclosure studies suggest that mosquitofish can alter zooplankton assemblages, tending to reduce abundances of cladocerans relative to other taxa (Blanco et al. 2003, Hurlbert and Mulla 1981, Ning et al. 2010, Peck and Walton 2008). However, it 1Department of Ecology, Evolution, and Organismal Biology, Kennesaw State University, Kennesaw, GA 30144. 2Department of Biological Sciences, Boise State University, Boise, ID 83725. *Corresponding author – jdirnber@kennesaw.edu. Manuscript Editor: Nathan Dorn Southeastern Naturalist 139 J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 is unclear the degree to which such changes are the result of direct consumption of planktonic taxa or indirect effects such as shifts in abundance of invertebrate predators and competitors, and alteration of food resources (Hurlbert and Mulla 1981, Ning et al. 2010, Peck and Walton 2008). Reduction in abundance (Gliwicz and Rykowska 1992, Jakobsen and Johnsen 1987, Lair et al. 1996, Taleb et al. 1994), body size (Gliwicz and Rykowska 1992, Hall et al.1979, Werner and Hall 1988), and clutch size (Gliwicz and Rykowska 1992) of zooplankton in littoral areas compared to open water have been attributed to several other fish that are common among macrophytes, like Rutilus rutilus (L.) (Common Roach) or Lepomis macrochirus Rafinesque (Bluegill Sunfish). These fish possess morphologies including forward-oriented mouths that are characteristic of fishes that feed away from the water surface (Keast and Webb 1966). For example, the success of Bluegills as planktivores is attributed to strong suction created by the structure of a protrusible premaxilla (Stabb et al. 2012). In mosquitofish, the protrusible premaxilla is dexterous so that mouth position can be altered to allow feeding on prey in front and below as well as at the surface, though this morphology comes at a cost of decreased strike velocity (Ferry-Graham et al. 2008) that would likely diminish the success of planktivory. In this study, we documented seasonal changes in abundance of invertebrates and in gut content of Gambusia holbrooki Girard (Eastern Mosquitofish) in situ in a small wetland. We examined shifts in prey use, feeding specialization, and prey electivity. We evaluated whether a feeding bias toward smaller-bodied prey that inhabit the water column exists for a predator that is well-suited for feeding at the surface and is capable of taking much lar ger prey. Field Site Description The study was conducted in a wetland (1.5 ha) formed by Castor canadensis Kuhl (American Beaver) impounding the upstream section of the Ragsdale Creek embayment on Lake Acworth, Cobb County, GA (34°02'57"N, 84°41'11"W). The rooted floating aquatic fern Marsilea mutica Mett. (Australian Water Clover) and emergent macrophyte Typha latifolia L. (Broadleaf Cattail) dominated shallow (less than 1 m) areas near shore, and surrounded a 25 m x 30 m patch of open water with an unvegetated silt bottom. Lake Acworth lies within the Etowah River watershed, part of the Alabama- Coosa-Tallapoosa drainage basin, placing the study site near known introgression zones for G. affinis (Baird and Girard) (Western Mosquitofish) and Eastern Mosquitofish (Angus and Howell 1996, Walters and Freeman 2000). At our site, no Western Mosquitofish were found, nor any evidence of hybridization based on fin-ray count criteria used by Angus and Howell (1996) (n = 20 individuals). Methods We sampled potential prey from the water column and plant surfaces using 2 independent methods. Because most devices for sampling invertebrates within Southeastern Naturalist J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 140 vegetated areas are likely to capture both truly planktonic organisms as well as those closely associated with macrophyte surfaces (e.g., tow nets and tube samplers that come in contact with macrophytes; McGavigan 2012; Paggi et al. 2001; Pennak 1962, 1966), we used funnel traps to sample invertebrates that occur in the water column for all or part of the diel cycle. Funnel traps were placed >5 cm away from macrophyte surfaces and from the bottom. Taxa caught in traps and that are small enough to be susceptible to transport by currents are by definition planktonic. Funnel traps consisted of clear plastic funnels (large opening = 79 cm2, small opening 2.0 = cm2) attached to the opening of wide-mouth transparent jars (modified from Whiteside and Williams 1975). At each location, we oriented 1 trap upward and 1 downward to capture plankton moving vertically. We suspended funnel traps by tripods at 3 locations near the shore within Australian Water Clover-dominated beds at mid-depth (0.2–0.4 m), and also from floats at 3 open-water locations at middepth (0.5–1.0 m), ~2 m from macrophyte beds. At the end of the 24-h sampling period, we filtered trap contents through 80-μm mesh and preserved retained organisms in 70% ethanol for identification and enumeration in the lab. We calculated abundance for each taxon in the water column as the mean number of individuals from all 12 traps (3 replicates for 2 orientations at 2 locatio ns). We quantified invertebrates on or near macrophyte surfaces using clear 2-liter plastic bags (11 cm diameter). While bags may also have collected individuals from the water column in addition to those near plant surfaces, individuals that stay on or near macrophytes would not be collected by traps placed away from macrophytes. Each bag was submerged sideways within 2 cm adjacent to clusters of Australian Water Clover, and rapidly pulled over plants down to the base of the plant. We severed macrophyte stems near the base and closed the bag (Campbell et al. 1982, Quade 1969). We collected 3 bag samples during each sample period, except in September 2013 when Australian Water Clover was absent in the study area. We shook bags to dislodge organisms from plant surfaces, then poured the water and suspended organisms onto an 80-μm–mesh filter and preserved the retained organisms in 70% ethanol for identification and enumeration in the la b. During each sample period, we used dip nets to sample Eastern Mosquitofish at multiple locations among macrophytes in the same general areas as where we collected bag and trap samples. For gut analyses, we selected 11 Eastern Mosquitofish that represented the broadest range of body lengths observed on each date (range ≥ 11.5 mm on each date, mean length for each date varied from 23.7 mm in January to 33.1 mm in May). Due to difficulty in determining the sex of juveniles in the field, we did not use sex as a criterion when selecting fish, but adult males made up 25% of all individuals sampled. Open-water areas were not sampled because a seining and electroshocking assessment as well as visual observations indicated that Eastern Mosquitofish were extremely rare in open water. In the laboratory, we measured total length of fish using a Vernier caliper. We dissected the entire intestinal tract from the collected Eastern Mosquitofish, preserved them in 70% ethanol, and examined the contents under a compound microscope at 100x. The post-abdomen of cladocerans tended not to be digested, thereby allowing identification to genus and Southeastern Naturalist 141 J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 species for most cladoceran prey. Using an ocular micrometer, we measured total body length for 10 intact individuals per prey taxon (or however many available if less than 10). We calculated the mean length of all prey in each mosquitofish gut weighted by the abundance of each prey taxon. Every 6–8 weeks during 2013, we collected invertebrates by funnel traps set out over 24 h and retrieved in late afternoon. We also collected invertebrates near macrophyte surfaces and sampled Eastern Mosquitofish in late afternoon. In September 2012, we captured Eastern Mosquitofish 5 times over 24 h (0900 h, 1700 h, 2100 h, 0600 h, and 0900 h) to determine the sampling time when guts would be most full. Mosquitofish guts had the highest number of prey at 1700 h, and were empty after the 9-h sampling interval that preceded day break (data not shown), supporting findings from other studies that Gambusia feed primarily during daylight hours (Oliver 1991, Pyke 2005). Gut contents of Eastern Mosquitofish sampled in late afternoon and early evening in September 2012 were used for comparisons with other sampling dates. We used a graphical analysis of feeding strategy to assess degree of feeding specialization by plotting prey-specific abundance against frequency of occurrence (Amundsen et al. 1996). Prey-specific abundance was calculated as all individuals of a given prey taxon found in the guts on a given date divided by all individuals of all prey taxa found in the guts of those Eastern Mosquitofish that consumed that prey taxon, and expressed as a percentage (plotted along the vertica l axis). Frequency of occurrence was figured as the proportion of individual Eastern Mosquitofish that have consumed that prey taxon (plotted along the horizontal axis). According to Amundsen et al. (1996), greater “specialization” is indicated for prey taxa located toward the top of this 2-dimensional representation. Prey-specific abundance multiplied by frequency of occurrence yields the percentage that a given prey type makes up of all prey taken by all Eastern Mosquitofish sampled on a given date. We also wanted to determine whether particular types of prey were selectively targeted or avoided on different dates. Ivlev’s electivity index (E) was used to compare consumption of a prey category (r) relative to abundance of that prey category in the environment (p) (Ivlev 1961): Ei = (ri - pi) / (ri + pi) The value of pi is the mean number of individuals of prey category i sampled from a given habitat on a given date divided by the sum of the mean number of individuals from all prey categories sampled in that habitat on that date. The value of ri is the mean number of individuals of category i found in all guts on a given date divided by the sum of the mean number of individuals from all categories found in all guts on that date. The value of E can range from -1 to 1, with a value of 0 indicating food is taken in proportion to its abundance and positive and negative values indicating consumption of a prey category at a greater and lesser proportion, respectively, than occurs in the environment. We used only those taxa or groupings of taxa that made up >10% of all individuals from all taxa present in a given habitat on a given date (i.e., pi ≥ 0.1) for analysis to avoid error in estimating electivity associated with low Southeastern Naturalist J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 142 abundance (Lechowicz 1982). So that data for most taxa could be included using this criterion, we grouped rare taxa into larger prey categories based on size and on taxonomic similarity where multiple taxa within the same order were identifiable (number of individuals for all taxa within a category were summed prior to calculating pi and ri). The Ivlev’s index was chosen in this study because of its conceptual simplicity while providing reliable rank-order comparisons within multispecies assemblages (Lechowicz 1982). An inherent flaw in electivity analyses is that the researcher can rarely be sure that habitats sampled are representative of the habitats from which the predator is taking prey. In our study, electivity could be calculated based either on prey abundance in the water column (trap samples) or near macrophyte surfaces (bag samples). However, feeding location of individual Eastern Mosquitofish could not easily be observed directly because the mosquitofish were at times spooked by observers, waters were turbid, and prey were small. Because a central question in our study was whether Eastern Mosquitofish have a positive bias toward small prey that inhabit the water column, we reported electivities for each prey category on each date only for the habitat in which the category occurred in the greatest abundance relative to the habitat’s total prey assemblage (pi). This approach provides the most conservative estimate of positive feeding bias. The electivity calculated from the habitat where pi is lower relative to the other habitat will always yield a higher electivity value because electivities based on abundances from either habitat are calculated from the same value of ri. Such calculations based on the habitat where relative prey abundance is lower will be an overestimation of electivity if the predator actually feeds mostly from the habitat where that prey is relatively more abundant. Because electivity derived from the habitat where prey are relatively less abundant cannot yield a lesser value, in cases where all habitats sampled yield positive electivities, bias toward that prey is the most likely explanation based on the habitats sampled. While interpretation of electivity is limited because of unknowns associated with feeding location, in instances where electivities are extremely high in all habitats sampled, the possibility that predators may have taken prey from habitats not considered in the sampling design must be suspecte d. Results We identified 41 taxa within the guts of Eastern Mosquitofish, representing 68% percent of all taxa found away from surfaces in funnel traps, and 76% of taxa found near plant surfaces collected by bag (Table 1). During the first autumn (September 2012), Eastern Mosquitofish consumed more dipteran larvae and pupae (of the orders Chironomidae and Ceratopognidae) than other taxa (Fig. 1A). Relatively few prey items were consumed in January 2013 (only 2% of prey taken over all sampling dates), with Ceriodaphnia (81% of large cladocerans consumed) and the small cladoceran Chydorus sphaericus (hereafter Chydorus) being most frequently taken. In March and May, Chydorus constituted up to 78% of prey items taken by all Eastern Mosquitofish on a single date (variation among gut contents was high, ranging from 17 to 155 Chydorus per gut). In June, total number of prey consumed Southeastern Naturalist 143 J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 was low (only 2% of prey taken over all 8 sampling dates), with terrestrial insects, rotifers, cyclopoid copepods, and aquatic mites being the most frequent diet items. Prey composition shifted to dominance by another small cladoceran, Bosmina longirostris (hereafter Bosmina), making up to 93% of prey items taken in late summer and September of the second year (ranging from 0 to 402 Bosmina per gut). Seasonal trends in the 2 small cladocerans, Chydorus and Bosmina, consumed by Eastern Mosquitofish paralleled changes in their abundance within the water column over the study period. Chydorus was first abundant near plant surfaces in winter (Fig. 1C), and became more abundant in the water column and in mosquitofish guts during the spring (Fig. 1A, B). On dates when Bosmina were abundant in the water column, Bosmina became a dominant part of the diet of Eastern Mosquitofish (e.g., September 2013) (Spearman rank-order correlation of mean Bosmina abundance in the water column versus gut: rs = 0.97, P = 0.00007, n = 8 sampling dates). In contrast, copepod nauplii made up the majority (61%) of other smaller taxa in the water column during the first autumn, but were rarely consumed by Eastern Mosquitofish (mean Table 1. Total number of all prey taxa found in Gambusia holbrooki (Eastern Mosquitofish) guts (total for all 8 sampling dates, 11 fish per date). Electivity indices were calculated for taxa and categories with “*” (rarer taxa were grouped into larger categories based on size and taxonomic similarity so that data for most taxa could be included using the sample-size criterion discussed in the text; values within parentheses indicate mean length and standard deviation for categories with multiple taxa). Items listed under “miscellaneous taxa” differed from taxa not included in this category because either they do not actively move and hence would not be effectively sampled by plankton traps (ephippial eggs, algae and detritus), or they could not be confidently identified as free-living prey (nematodes and acanthocephalans). Taxa # found Taxa # found Bosmina longirostris (O.F.M.) (0.29 mm)* 799 Dipteran larvae and pupae (4.53 mm + 1.47)*: Chydorus sphaericus (O.F.M.) (0.29 mm)* 907 Chironomid larva 55 Large cladocerans (0.78 mm ± 0.32)*: Ceratopognid larva 23 Daphnia sp. 2 Ceratopognid pupa 3 Diaphanosoma sp. 1 Dipteran pupa 19 Ceriodaphnia sp. 208 Other large taxa (2.49 mm + 0.97)*: Eurycercus lamellatus (O.F.M.) 41 Odonata larva 1 Camptocercus sp. 109 Ephemeropteran larva 14 Scapholeberis mucronata (O.F.M.) 222 Trichopteran larva 1 Leydigia quadrangularis (Leydig) 4 Terrestrial insect 70 Simocephalus serrulatus (Koch) 17 Terrestrial spider 1 Alonella sp. 9 Amphipod 12 Alona sp. 5 Coiled snail 2 Sida crystallina (O.F.M.) 54 Oligochaete annelid 11 Unidentified cladoceran 24 Unidentified annelid 1 Cyclopoid copepod (1.03 mm)* 144 Miscellaneous taxa (0.38 mm ± 0.30): Ostracod (0.50 mm)* 8 Ephippial egg 48 Other small taxa (0.41 mm + 0.36)*: Alga (Closterium sp.) 28 Calanoid copepod 5 Alga (Micrasterias sp.) 1 Harpacticoid copepod 7 Unidentified alga 2 Copepod nauplius 2 Nematode 5 Mite (Arrenurus sp.) 29 Acanthocephalan 11 Rotifer 17 Plant detritus 2 Southeastern Naturalist J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 144 abundance in the water column versus gut: rs = -0.25, P = 0.547, n = 8 sampling dates). Dipterans and other large-bodied taxa (80% of which were annelids) were relatively more abundant near plant surfaces (48% of individuals sampled) than in the water column (less 1%), showed less seasonality (Fig. 1C), and were not strongly linked to Eastern Mosquitofish diet. Number of dipterans and other large-bodied individuals per bag was not correlated with mean number consumed by all Eastern Mosquitofish (P = 0.645 and 0.638 for dipterans and other large-bodied taxa, respectively, n = 7 sampling dates), nor with mean number consumed by Eastern Mosquitofish larger than the median size of 29.5 mm (i.e.. those that were more likely to prey on these larger taxa; P = 0.702 and 0.900 for dipterans and other large-bodied taxa, respectively, n = 7 sampling dates). Bias toward small prey was suggested by the tendency of all sizes of Eastern Mosquitofish to consume small prey. Smaller-bodied specimens tended to take primarily small prey (less than 1.0 mm), while larger bodied individuals were capable of taking large prey (>1.0 mm) as well (Fig. 2). Only on 1 date was mean prey length significantly related to mosquitofish length (linear regression for September 2012, P = 0.013, r2 = 0.512, n = 11). On several dates larger Eastern Mosquitofish consumed mostly small prey taxa, even though densities of large-bodied prey in the environment (e.g., dipteran larvae; Fig. 1) were similar to those at other times of year. Feeding-strategy analyses indicated that when guts contained the largest number of items (March and September 2013), Bosmina and Chydorus constituted large Figure 1. (A) Mean number of prey consumed by Gambusia holbrooki (Eastern Mosquitofish) by sampling date (error bars are 1 standard deviation to illustrate variation in prey taken among fish). (B) Mean invertebrate abundances in the water column by sampling date. (C) Mean invertebrate abundances near plant surfaces by sampling date. Marsilea mutica (Australian Water Clover) died off during late summer of 2013 so invertebrates near macrophyte surfaces could not be sampled by bag in September of that year. Abundance is plotted on a log-10 scale so that rare prey categories are more easily compared. Southeastern Naturalist 145 J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 portions of prey taken (78% and 84%, respectively) and were taken by most individuals (Fig. 3). During those times, prey taxa other than Bosmina and Chydorus tended to constitute smaller portions of prey items in the diet compared to dates when overall food intake was lowest (January and June 2013; Fig. 3). When food intake was lower, use of those same taxa (e.g., copepods, mites, and large-bodied taxa such as chirnomids, oligochaetes, and terrestrial insects) increased such that diets were more generalized; Eastern Mosquitofish did not show strong specialization toward any particular prey species (with Ceriodaphnia being the taxon that they consumed in the greatest portion at 20% of prey taken in J anuary). Electivities tended to be positive for Bosmina and larger cladocerans, but mixed for Chydorus depending on date (Fig. 4). Electivities were typically strongly negative for copepods, ostracods, other small taxa (composed mostly of nauplii), dipterans larvae and pupae, and other lar ge taxa. Discussion In this study, mosquitofish consumed a wide variety of prey types that varied dramatically from season to season both in terms of planktonic behavior and body size by over an order of magnitude in length (which would be approximately 3 orders of magnitude in biovolume). Ontogenetic shifts from consumption of cladocerans and rotifers by younger individuals toward much larger insect prey by larger, Figure 2. Mean length of prey consumed by individual Gambusia holbrooki (Eastern Mosquitofish) as a function of total body length. The line of best fit is included for September 2012, the only date when mean prey length was significantly related to mosquitofish length (see text for details). Southeastern Naturalist J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 146 older individuals have been observed for mosquitofish (e.g., Blanco et al. 2003, Garćia-Berthou 1999). Mieiro et al. (2001) observed a decrease in insect prey and an increase in microcrustacean consumption by mosquitofish in non-summer months related to shifts in the abundance of these prey types, though Garćia-Berthou (1999) and Gkenas et al. (2012) noted little seasonal variation. In our study, shifts in prey types reflected shifts in abundance of cladocerans. Ontogenetic shifts did not appear to be strong in our study system, as both small and large-bodied Eastern Mosquitofish commonly consumed small prey (mostly microcrustaceans) over most dates, even though larger prey were also present at these times. Some laboratory experiments note a similar phenomenon. Mattingly and Butler (1994) did not consistently find a selection bias for larger prey by larger Crenicichla alta Eigenmann (Pike Cichlid), and Bence and Murdoch (1986) observed that patterns in profitability as a function of prey size did not differ among Western Mosquitofish that differed in body length. In our study, energetic cost associated with capturing a high number of small prey (Brooks and Dodson 1965, Werner and Hall 1974) is presumably offset by favorable attributes (such as accessibility) associated with cladocerans, or by lower success Figure 3. Feeding strategy of Gambusia holbrooki (Eastern Mosquitofish) over 4 dates. Feeding strategy is evaluated as a prey taxon’s contribution to diet as a function of the proportion of individuals that consume that taxon (see text for details). Prey taxa with gut abundances greater than 1% of all individuals consumed on a date (calculated as prey-specific abundance multiplied by frequency of occurrence) are labeled. Dates were selected to compare the dates with the emptiest guts (44 individuals consumed in January and 63 in June) with dates with the highest numbers of prey per gut (804 in March and 657 in September). Southeastern Naturalist 147 J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 rates and longer attack times associated with larger prey (Bence and Murdoch 1986). As indicated in the graphical analysis of feeding strategy, there were times when Eastern Mosquitofish strongly specialized on cladocerans while other prey taxa including larger taxa tended to be taken less often and were taken by fewer individuals. The actual biovolume of prey consumed was not evaluated in our study, so specialization in the context of our study is a reflection of the number of prey captures (which relates to energetic costs) rather than contribution of calories. Cladocerans consumed in our study generally shared attributes of size (ranging 0.2–1.0 mm) and shape (compressed laterally and roughly round in profile). Larger cladocerans, for which electivities were mostly positive, were frequently found in the water column. Though Chydorus sphearicus is typically described as littoral (e.g., Pennak 1966, Ward and Whipple 1918), it is reported to consistently leave its substrate (Whiteside et al. 1978). In our study, Chydorus were rarely taken in January 2013 when abundant near macrophyte surfaces, but were taken frequently in subsequent months when they became abundant in the water column (Fig. 1). Bosmina was rarely found near macrophyte surfaces (less than 1% of all occurrences) and is widely reported in the plankton of open waters (e.g., Geraldes and Boavida 2004, Taleb et al. 1994). The specialization and high importance of Bosmina in some seasons indicates the Eastern Mosquitofish took smaller prey at a distance from plant surfaces. Regardless of where Eastern Mosquitofish were taking these prey, a feeding bias existed for prey that constitute plankton assembla ges. Figure 4. Electivity for prey categories consumed by Gambusia holbrooki (Eastern Mosquitofish) by sampling date. Electivity compares prey consumed to occurrence of prey in the environment. Black bars are electivity based on abundance of prey in the water column, and gray bars are electivity based on abundance of prey near macrophyte surfaces. No electivities were zero, so lack of bar indicates that the prey was not included in the electivity analysis for that period (electivity was calculated for only the habitat with the highest relative abundance and only for those prey categories that constituted >10% of total abundance in the habitat; see text for details). Southeastern Naturalist J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 148 Being planktonic and small did not in itself result in a positive feeding bias. Despite an abundance of copepod nauplii in the water column in September 2012, no nauplii were consumed, suggesting that nauplii (typically less than 0.2 mm) were energetically unfavorable prey. Cyclopoid copepodites and adults were frequently found in the water column, but consistently occurred less frequently in Eastern Mosquitofish guts. Ostracods were also rarely consumed despite extremely high abundances most of the year both near macrophytes and in the water column. Biases toward one prey type versus another as indicated by electivity can be the result of several factors associated with predator and prey attributes including ability of prey to escape, ability of predators to detect prey (i.e., handling and search time), and active choice based on palatability or energetics and density of prey (e.g., Lewis 1977, Werner and Hall 1974). Alternatively, electivity can be an artifact of sampling design if locations where prey were sampled do not accurately reflect abundances of prey where predators actually fed. The air–water interface was not sampled in our study, so relative availability for prey items that are common on the surface was underestimated, potentially inflating our estimates of electivity for these items. Of all invertebrate taxa consumed by Eastern Mosquitofish in our study, the 3 prey that were most rare in the water column and near macrophyte surfaces relative to their occurrence in the gut were terrestrial insects, ephippia, and the cladoceran Scaphaloberis mucronata (O.F.M) (mean electivities over all dates and both habitats were 0.98, 0.97, and 0.86, respectively). There is good reason to believe that Eastern Mosquitofish take these prey at the air–water interface. Terrestrial insects are likely be caught on the water surface to which they fall, ephippia are known to float (Ślusarczyk and Pietrzak 2008), and S. mucronata inhabit the underside of the surface tension created at the air–water interface (Gladyshev 2002, Ward and Whipple 1918). A congener, S. ramneri Dumont & Pensaert, was the most common prey item in the guts of Eastern Mosquitofish studied by Garćia-Berthou (1999). Mosquitofish have long been reported to feed at the surface (Hildebrand 1919) as would be expected given their dorsally oriented mouth situated anteriorly on a flattened head. Cladocerans have been observed caught in the surface tension. Rose et al. (2012) noted Daphnia on the surface in mesocosms, and Gladyshev (2002) reported Bosmina, Chydorus, and Alona attached to the surface film, but not copepods or other non-cladoceran taxa. Cladocerans (particularly Bosmina) tend to clump together when caught by surface tension (J.M. Dirnberger, pers. observ.), as do other negatively buoyant objects (Vella and Mahadevan 2005), potentially making small prey easier to spot, and foraging energetically more efficient. Consumption of such rafts would explain the poor relationship between prey size and predator size, and the higher electivities for cladocerans, as well as the high variability in number of Bosmina and Chydorus consumed among individual Eastern Mosquitofish because dense, isolated patches would be encountered only occasionally. While positive feeding bias by Eastern Mosquitofish toward cladocerans found in our study suggest that it has the potential to alter zooplankton assemblages, the impact of this species on zooplankton assemblages would depend in part on the degree to which Southeastern Naturalist 149 J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 it takes planktonic prey from the water surface as opposed to from the water column. Because mortality is presumably imminent for most zooplankton caught in the surface tension (Gladyshev 2002), surface-feeding predation would not directly impact those prey populations. Mosquitofish have been widely introduced for mosquito control, and where introduced are often implicated as the agent responsible for declines in native amphibian and fish populations, earning them an alternative name, plague minnow (Pyke 2008). Predation has been implicated as a factor in these declines (e.g., Meffe 1985), though the ability of mosquitofish to displace native vertebrates has been called into question in some systems (e.g., Ling 2004). Mosquitofish efficacy for mosquito control has also been seriously questioned in many systems because earlier evidence was largely anecdotal and has been often contradicted by more recent studies using larger sample sizes and randomized treatments (Pyke 2008). Experimental manipulations of prey densities may elucidate whether conditions that favor the entrainment and subsequent concentration of planktonic prey at the surface alter the impact of mosquitofish on mosquito and other larval insect populations, or whether plankton are an alternative prey choice only after mosquitofish have reduced populations of larger prey. Quantifying available surface prey in heterogeneous shallow ponds and lakes will be a future challenge for understanding the effects of mosquitofish on prey assemblages. Acknowledgments We thank Marina Kasearum, Eric Iversen, Marielle Kromis, Lauren Lee, Nicole Lynch, Ryan McWilliams, Marjan Mirkheshti, Elena Petra, Daniel Rhiner, Nazanin Gerami Sarabi, and Amber Wilson for assistance in the field and lab, and William Ensign, Amy Whitney, and 2 anonymous reviewers for comments on the manuscript. Literature Cited Amundsen, P.A., M. Gabler, and F.J. Staldvik. 1996. A new approach to graphical analysis of feeding strategy from stomach contents data: Modification of the Costello (1990) method. Journal of Fish Biology 48:607–614. Angus, R.A., and W.M. Howell. 1996. Geographic distributions of Eastern and Western Mosquitofishes (Poeciliidae: Gambusia): Delineation of ranges using fin-ray counts. Southeastern Fishes Council Proceedings 33:1–6. Bence, J.R., and W.W. Murdoch. 1986. Prey-size selection by the mosquitofish: Relation to optimal diet theory. Ecology 67:324–336. Blanco, S., S.Romo, M.J. Villena, and S. Martínez. 2003. Fish communities and food-web interactions in some shallow Mediterranean lakes. Hydrobiologia 506:473–480. Brooks, J.L., and S.I. Dodson. 1965. Predation, body size, and composition of plankton. Science 150:28–35. Campbell, J.M., W.J. Clark, and R. Kosinski. 1982. A technique for examining microspatial distribution of Cladocera associated with shallow-water macrophytes. Hydrobiologia 97:225–232. Ferry-Graham, L.A., A.C. Gibb, and L.P. Hernandez. 2008. Premaxillary movements in cyprinodontiform fishes: An unusual protrusion mechanism facilitates “picking” prey capture. Zoology 111:455–466. Southeastern Naturalist J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 150 Garćia-Berthou, E. 1999. Food of introduced mosquitofish: Ontogenetic diet shift and prey selection. Journal of Fish Biology 55:135–147. Geraldes, A.M., and M.J. Boavida. 2004. Do littoral macrophytes influence the crustacean zooplankton distribution? Limnetica 23:57–64. Gkenas, C., A. Oikonomou, A. Economoue, F. Kiosse, and I. Leonardos. 2012. Life-history pattern and feeding habits of the invasive mosquitofish Gambusia holbrooki in Lake Pamvotis (NW Greece). Journal of Biological Research-Thessalon 17:121–136. Gladyshev, M. 2002. Biophysics of the Surface Microlayer of Aquatic Ecosystems. International Water Association. London, UK. 160 pp. Gliwicz, Z.M., and A. Rykowska. 1992. “Shore avoidance” in zooplankton: A predatorinduced behavior or predator-induced mortality? Journal of Plankton Research 14:1331–1342. Hall, D.J., E.E. Werner, J.F. Gilliam, G.G. Mittelbach, D. Howard, C.G. Doner, J.A. Dickerman, and A.J. Stewart. 1979. Diel foraging behavior and prey selection in the Golden Shiner (Notemigonus crysoleucas). Journal of Fisheries Research Board Canada 36:1029–1039. Hildebrand, S.F. 1919. Notes on the life history of the minnows Gambusia affinis and Cyprinodon variegatus. Annual Report for the United States Commission on Fish 6:3–15. Hoy, J.B., E.E. Kauffman, and A.G. O’Berg. 1972. A large-scale test of Gambusia affinis and Chlorpyrifos for mosquito control. Mosquito News 32:161–169 . Hurlbert, S.H., and M.S. Mulla. 1981. Impacts of mosquitofish (Gambusia affinis) predation on plankton communities. Hydrobiologia 83:125–151. Ivlev, V.S. 1961. Experimental Ecology of the Feeding of Fishes. Yale University Press, New Haven, CT. 302 pp. Jakobsen, P.J., and G.H. Johnsen. 1987. The influence of predation on horizontal distribution of zooplankton species. Freshwater Biology 17:501–507. Keast, A., and D. Webb. 1966. Mouth and body form relative to feeding ecology in the fish fauna of a small lake, Lake Opinicon, Ontario. Journal of the Fisheries Research Board of Canada 23:1845–1874. Lair, N., T.H. Taleb, and P. Reyes-Marchant. 1996. Horizontal distribution of the rotifer plankton of Lake Aydat (France). Aquatic Sciences 58:253–268. Lechowicz, M.J. 1982. The sampling characteristics of electivity indices. Oecologia 52:22–30. Lewis, W.M. 1977. Feeding selectivity of a tropical Chaoborus population. Freshwater Biology 7:311–325. Ling, N. 2004. Gambusia in New Zealand: Really bad or just misunderstood? New Zealand Journal of Marine and Freshwater Research 38:473–480. Mattingly, H.T., and M.J. Butler. 1994. Laboratory predation on the Trinidadaian Guppy: Implications for the size-selective predation hypothesis and guppy life-history evolution. Oikos 69:54–64. McGavigan, C. 2012. A quantitative method for sampling littoral zooplankton in lakes: The active tube. Limnology and Oceanography: Methods 10:289–295. Meffe, G.K. 1985. Predation and species replacement in American southwestern fishes: A case study. The Southwestern Naturalist 30:173–187. Mieiro, L., J.A. Cabral, and J.C. Marques. 2001. Predation pressure of introduced mosquitofish (Gambusia holbrooki Girard), on the native zooplankton community. A case study from representative habitats in the lower Mondego river Valley (Portugal). Limnetica 20:279–292. Southeastern Naturalist 151 J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 Ning, N.S.P., D.L. Nielsen, T.J. Hillman, and P.J. Suter. 2010. The influence of planktivorous fish on zooplankton communities in riverine slackwaters. Freshwater Biology 55:360–374. Oliver, J.D. 1991. Consumption rates, evacuation rates and diets of Pygmy Killifish, Leptolucania ommata, and mosquitofish, Gambusia affinis, in the Okefenokee Swamp. Brimleyana 17:89–103. Paggi, J.C., R.O. Mendoza, C.J. Debonis, and S.B. José de Paggi. 2001. A simple and inexpensive trap-tube sampler for zooplankton collection in shallow waters. Hydrobiologia 464:45–49. Peck, G.W., and W.E Walton. 2008. Effect of mosquitofish (Gambusia affinis) and sestonic food abundance on the invertebrate community within a constructed treatment wetland. Freshwater Biology 53:2220–2233. Pennak, R.W. 1962. Quantitative zooplankton sampling in littoral vegetation. Limnology and Oceanography. 7:487–489. Pennak, R.W. 1966. Structure of zooplankton populations in the littoral macrophyte zone of some Colorado lakes. Transactions of the American Microscopical Society 85:329–349. Pyke, G.H. 2005. A review of the biology of Gambusia affinis and G. holbrooki. Reviews in Fish Biology and Fisheries 15:339–365. Pyke, G.H. 2008. Plague minnow or mosquito fish? A review of the biology and impacts of introduced Gambusia species. Annual Review of Ecology, Evolution, and Systematics 39:171–191. Quade, H.W. 1969. Cladoceran faunas associated with aquatic macrophytes in some lakes in northwestern Minnesota. Ecology 50:170–179. Rose, K.C., C.E. Williamson, J.M. Fischer, S.J. Connelly, M. Olson, A.J. Tucker, and D.A. Noe. 2012. The role of ultraviolet radiation and fish in regulating the vertical distribution of Daphnia. Limnology and Oceanography 57:1867–1876. Ślusarczyk, M., and B. Pietrzak. 2008. To sink or float: The fate of dormant offspring is determined by maternal behaviour in Daphnia. Freshwater Biology 53:569–576. Specziár, A. 2004. Life-history pattern and feeding ecology of the introduced Eastern Mosquitofish, Gambusia holbrooki, in a thermal spa under temperate climate, of Lake Hévíz, Hungary. Hydrobiologia 522:249–260. Staab, K.L., R. Holzman, L.P. Hernandez, and P.C. Wainwright. 2012. Independently evolved upper-jaw protrusion mechanisms show convergent hydrodynamic function in teleost fishes. Journal of Experimental Biology 215:1456–1463. Taleb, H., P. Reyes-Marchant, and N. Lair. 1994. Effect of vertebrate predation on the spatio-temporal distribution of cladocerans in a temperature eutrophic lake. Hydrobiologia 294:117–128. Vella, D., and L. Mahadevan. 2005. The “Cheerios effect”. American Journal of Physics. 73:817–825. Walters, D.M., and B.J. Freeman. 2000. Distribution of Gambusia (Poeciliidae) in a southeastern river system and the use of fin-ray counts for species determination. Copeia 2000:555–559. Ward, H.B., and G.C. Whipple. 1918. Freshwater Biology. John Wilely and Sons, Inc. New York, NY. 699 pp. Werner, E.E., and D.J. Hall. 1974. Optimal foraging and the size selection of prey by the Bluegill Sunfish (Lepomis macrochirus). Ecology 55:1042–1052. Werner, E.E., and D.J. Hall. 1988. Ontogenetic habitat shifts in Bluegill: The foraging rate– predation risk trade-off. Ecology 69:1352–1366. Southeastern Naturalist J.M. Dirnberger and J. Love 2016 Vol. 15, No. 1 152 Whiteside, M.C., and J.B. Williams. 1975. A new sampling technique for aquatic ecologists. Verhandlungen des Internationalen Verein Limnologie 19:1534–1539. Whiteside, M.C., J.B. Williams, and C.P. White. 1978. Seasonal abundance and pattern of chydorid, Caldocera in mud and vegetative habitats. Ecology 59: 1177–1188.