We studied parental behavior in six syntopically breeding species of centrarchid fishes to determine whether energetic costs could contribute to our understanding of the diversity of parental care. We used a combination of underwater videography, radio telemetry and direct observation to examine how the cost of parental care varied with both its duration and intensity. Duration of parental care, activity patterns, and energetic costs varied widely among species. Overall, the duration of care increased with parental size between species. When energetic costs were adjusted for species-specific differences in the duration of parental care, the cost of parental care also increased with mean size of the species. Species with extended parental care exhibited stage-specific patterns of activity and energy expenditure consistent with parental investment theory, whereas fish with short duration parental care tended to maintain high levels of activity throughout the entire period of parental care. The only apparent exception (a species with brief parental care but stage-specific behavior) was a species with multiple breeding bouts, and thus effectively having protracted parental care. These data suggest that some species with short duration parental care can afford not to adjust parental investment over stages of offspring development. Using our empirical data on parental care duration and costs, we reevaluated the relationship between egg size and quality of parental care. Variation in egg size explained almost all of the observed variation in total energetic cost of parental care, and to a lesser degree, duration—the larger the eggs, the more costly the parental care. This research highlights the value of incorporating energetic information into the study of parental care behavior and testing of ecological theory.
Subjects and study site We considered parental care to include the period from fertilization until the parent deserts the offspring. We recognize that parental investment can include nest construction, courting, and the actual spawning event, all of which can be costly (Cooke et al. 2001). However, it was not possible to quantify these variables in this study. This study was conducted in Lake Opinicon (4433¢30¢¢N, 7620¢00¢¢W), Ontario, from 1 May to 9 July 2001. In addition, we supplemented the radio telemetry study of black crappie by monitoring six additional fish from 16 May to 7 June 2002 during a period when water temperatures were the same as in 2001. Lake Opinicon has served as the focus for a great deal of previous research on the reproductive biology, including parental care and early life-history, of centrarchid species (e.g., rock bass, Gross and Nowell 1980; pumpkinseed, Colgan and Gross 1977; black crappie, Colgan and Brown 1988; bluegill, Gross 1980; smallmouth bass, Philipp et al. 1997; Cooke et al. 2002; and largemouth bass, Brown 1984; Colgan and Brown 1988; Cooke et al. 2002). The life-history traits of these species are well documented and have been the subject of an entire book (Carlander 1977). However, one common pattern is the fact that there is extreme variation in lifehistory traits among populations of the same species (and across latitudes). Consequently, it is not possible to present a brief summary of the life-history characteristics for all six species. However, we will identify key aspects of different species life-history strategies as required to aid in our interpretation of findings. All six of 237 these species occur naturally in Lake Opinicon and spawn in the littoral zone throughout the lake (Keast et al. 1978). To facilitate frequent monitoring of nests and to avoid heavy angling pressure, we used a study site close to the Queen’s University Biological Station that included 4 km of shoreline. All experiments were approved by the University of Illinois Office of Laboratory Animal Research and the Queen’s University Animal Care Committee. All procedures were in accordance with the guidelines for animal research in Canada and the United States. Scientific Collection Permits were furnished by the Ontario Ministry of Natural Resources. Snorkeling and telemetry Snorkeling surveys, initiated when the water temperature reached 12C, were used to monitor the onset and progression of reproduction by all six centrarchid species. Snorkelers swam the study site every 2–3 days. When nests were found, numbered tiles were placed adjacent to nests, the male was identified to species and his total length estimated to the nearest cm by trained divers, and the offspring stage was recorded. We used externally attached radio transmitters for monitoring the parental care duration of nest guarding fish because this method is rapid, less invasive than internal implantation, and does not require anesthetizing the fish. Furthermore, a recent study by Cooke (2003) determined that the same transmitters and methods we used here did not result in any changes in behavior or reproductive success relative to control fish for rock bass. These methods were developed with veterinarian consultation and used our extensive experience in transmitter attachment on fish and other animals. For this study, all transmitter attachments were conducted on a research vessel equipped with a surgical table. We located nesting males that were attending eggs or newly-hatched larvae and then angled these parental males from their nests using rod and reel. Fish were landed within 10 s and immediately placed ventral side down on a wet sponge pad where they were measured and weighed (Table 1). A wet cloth covered the head and caudal peduncle region of the fish while an assistant held the fish in place for transmitter attachment. A neoprene backing plate was placed on two 22-gauge hypodermic needles mounted on 3-ml syringes that were pushed through the dorsal back musculature, ventral to the junction of the soft and spiny dorsal fins (Beaumont et al. 1996; Cooke 2003). From the opposite side, the transmitter attachment wires (surgical stainless steel, 20 gauge) that had already been threaded through the transmitter (Model BD-2G, Holohil Systems, Ontario; wgt in air, 2.1 g, 14·6·4 mm, 120 mm antenna wire for small fish and Model AVM G3, AVM Instruments, Calif., wgt in air, 3.6 g, 18·9·6 mm, 200 mm antenna wire for large fish) and a neoprene pad (2 mm) were inserted into the lumen of the needles. The wires were pulled out on the opposite side of the fish, and when the needles were removed, the neoprene backing plate was left in place to protect the body of the fish. The wires were twisted carefully and trimmed prior to releasing the fish above its nest. The fish were out of water for less than 90 s. A snorkeler protected and monitored the nest during the attachment procedure until the fish had resumed parental care duties. At the time of this study, we used the smallest commercially available radio transmitters. Nonetheless, we were forced to select larger individuals (especially for the smallest species) relative to the broader population of nesting fish. Consequently, the size of fish monitored with telemetry were larger than those monitored with videography (see below). During the early stages of offspring development, the presence/absence of the nest-guarding males equipped with transmitters was determined by a snorkeler. As fish approached the period at which they would normally terminate care or move from the nest with their offspring, we located each fish using telemetry. We used programmable radio telemetry receivers equipped with two-element H antennas. Initially, we used a combination of triangulation and pinpointing fish through gain reductions to locate the radio-tagged fish. When we were within 10 m of a radio-tagged fish, we dispatched one or more snorkelers to search visually for the fish. Simultaneously, we switched to an electric trolling motor to maneuver the boat. We continued reducing gain until we had a strong signal with a gain of ‘‘0’’. Marker buoys were deployed and the boat was moved from the immediate area. Upon visually locating the fish, the diver recorded information on the general condition of the male, the activity and behavior of the male, and determined the presence or absence of offspring. When a fish was located on two successive occasions without offspring or engaged in activity unrelated to parental care, we assumed that care had terminated on the last day that parental care was observed. During these snorkeling observations we observed no ill effects of the transmitters on the tagged males. Videography We used small underwater cameras (Atlantis, AU-401) and time-lapse recorders (Sanyo, SRT 7072) to record detailed information on multiple nests. Our videographic observations were restricted to fish that were not carrying transmitters. Video recording gear was located aboard a boat that was anchored at least 25 m from the nest sites. Each camera had a 50-m cable that connected it to the boat. Cameras were positioned 0.5 m from the nest by a diver and were on a 45 angle pointing down towards the nest (Cooke and Bunt 2004). Because we relied on ambient light to provide illumination for the camera, all of our video observations were diurnal. Several studies of parental care in centrarchids have determined that activity rates remain unchanged at night (e.g., Hinch and Collins 1991; Cooke et al. 2002), so we assumed that our diurnal observations were also representative of nocturnal activity. We recorded male parental behaviors for fish not equipped with radio transmitters between 1000 and 1400 hours for a 10-min period during both the egg and wriggler stages. The egg stage is the period after egg deposition and fertilization, but prior to hatching. The wriggler stage occurs after eggs had hatched, but prior to them becoming free swimming. We only recorded video footage from each individual once so that we did not have to control for individuals in analyses. At the conclusion of the video recordings, the snorkeler recorded the species, the size of the fish, and examined the stage of offspring development (Table 1). We transcribed video records using a professional editing suite (Mitsubishi BV-100) at playback speeds of 1/5 to 1/30 normal. Although we recorded at least 10 min of video, we excluded the first 5 min to eliminate any disturbances arising from camera placement. We quantified the caudal fin beat frequency of the fish while on the nest and used those values to calculate the inplace swimming speed of the male. We used methods outlined in Hinch and Collins (1991) to establish swimming speeds and to ascribe energetic costs to swimming speeds (Weihs 1977). We quantified swimming speed using videography because it can be used to quantify activity costs when incorporated into existing energetics models (Boisclair and Legett 1989; Trudel and Boisclair 1996) and because it has been determined to be a robust behavioral indicator of fish energetics (e.g., Trudel and Boisclair 1996; Rennie et al. 2005). We used the Wisconsin Bionenergetics model (Hanson et al. 1997), a widely used tool in fisheries science (Ney 1993) that has benefited from substantial ground truthing and refinement. To derive energetic costs from behavioral data, we replaced the activity multiplier of the Wisconsin Bioenergetics model with our empirically determined data. We also had to modify model parameters using species-specific criteria derived from the literature because the Wisconsin model had not been parameterized for all six species. Also, even where existing models were available, we modified them to incorporate the most thermally relevant swimming energetics information. For example, we used previously developed information on metabolic costs of swimming for largemouth bass (Beamish 1970), pumpkinseed (Brett and Sutherland 1965; Evans 1984), and bluegill (Wohlschlag and Juliano 1959). Data for smallmouth bass were derived from data collected at 18 C (S.J. Cooke, Unpublished data). Because no empirical data were available for rock bass and because smallmouth bass and rock bass have similar thermal preferences and tolerances (Wismer and Christie 1987), as well as Q10 rates (S.J. Cooke, Unpublished data), we used smallmouth bass model parameters for rock bass calculations. Data were also unavailable for black crappie, so we used values collected for white crappie (Parsons and Sylvester 1992) because these fish have very similar life histories and physiological characteristics (Scott and Crossman 1973; Wismer and Christie 1987). Use of metabolic values from similar species when data are unavailable is a recognized technique in fish bioenergetics (Ney 1993). Our specific modifications to the models are listed in Table 2. Because metabolic rates vary with water temperature (Beamish 1964; Brett and Groves 1979), and because water temperatures during parental care differ among species, we used water temperatures measured in Lake Opinicon during each species’ parental care period for estimating their respective parental care activity costs (see Tables 1 , 2, Fig. 1). Oxycalorific equivalents were calculated for oxygen consumption values and then expressed in joules (Brett and Groves 1979). Because these data are redundant with oxygen consumption data, we use them only for descriptive purposes. We adjusted all values for the size of the fish using the generalized mass scaling exponent of 0.8 where required (Schmidt-Nielsen 1984). Where appropriate we standardized values to 1 kg for comparative purposes. Egg measurements We collected newly fertilized eggs from 15 nests of each study species. Eggs were held in lake water for no more than 2 h, dried with a paper towel and weighed in groups of ten (one group per nest) using a Metler AE 100 scale (0.0001 g). We then used a dissecting microscope at 160· magnification equipped with an ocular micrometer to measure the diameter of ten eggs from each nest. Phylogenetic corrections Phylogenetic relatedness poses a potential constraint for comparative studies because closely related species can share a certain character state through common ancestry rather than through independent evolution (Felsenstein 1985; Harvey and Pagel 1991). As a result, we conducted our analysis with and without controlling for phylogenetic effects. To control for phylogenetic effects, data were converted to phylogenetically independent standardized contrasts using the PDTREE module of the Table 2 Phylogenetic Diversity Analysis Package (PDAP, v.6.0) (Garland et al. 1992; 1993), which is based on the methods of Felsenstein (1985). Branch lengths were transformed prior to analyses using Grafen’s (1989) method for arbitrary branch lengths. We used a phylogeny obtained from a molecular genetic analysis by Roe et al. (2002), which included all the species we examined except for pumpkinseed. Because morphological (Mabee 1993) and genetic evidence (Neff et al. 1999) suggest that pumpkinseed are closely related to bluegill, the other Lepomis species, we considered pumpkinseed to be a separate branch, originating at the Lepomis node. The branch length for pumpkinseed was arbitrarily assumed to be the same as bluegill. Branch lengths provided for several other Lepomis in Roe et al. (2002) were all of similar length. To provide phylogenetic consistency throughout the manuscript, species are listed in order of evolutionary history in all tables and figures. Data analysis One-way analysis of variance with the conservative Tukey HSD test (Day and Quinn 1989) was used to assess differences in fish size, egg size, and duration of parental care among the species we examined. Pairwise correlations were also used to test for relationships in duration of parental care and size of fish among species. T-tests were used to compare parental care activity and energetic parameters among stages of offspring development for each species. One-way ANOVA was also used to compare total respiration rates and energetic costs for each species corrected for duration of care. We used multiple stepwise regression analysis to determine which parental care factors explained significant variation in egg diameter and mass. We used phylogenetically independent contrasts (PICs; see above) to account for phylogenetic influences and have reported these data in addition to phylogenetically uncorrected data throughout the results. All analyses were conducted using JMPIN (v.4.1, SAS Institute) and all tests were considered significant at a=0.05.