This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
The diets of fish are routinely examined by management and research agencies to address a wide array of topics. The objectives of these investigations can be as simple as establishing an inventory of what prey items are consumed or as complex as measuring dietary responses to changes in the environment (Bowen 1992). Historically, the standard protocol for describing the diet involves visually identifying the contents of fish stomachs and summarizing prey items by number, frequency of occurrence, and either volume or weight (Hyslop 1980). Despite widespread use, there are limitations inherent to visual techniques that can bias results. For example, the rate at which different types of prey digest in the stomach varies, and as a result some prey items will be detectable for longer time periods than others (Gannon 1976; Berens and Murie 2008). Also, physical structures of the fish can pulverize certain prey items, making identification difficult if not impossible.
Alternatives to visual analysis of stomach contents are available, and Polymerase Chain Reaction (PCR) techniques represent an appealing alternative. PCR uses a series of chemical reactions to produce millions of copies of targeted DNA fragments that can uniquely identify to individual species (Hebert et al. 2003). Only minutes amounts of tissue recovered from prey would be needed for PCR due to the exponential nature of the reaction. PCR is a widespread tool used in a diverse array of research fields and as result databases have been created that contain hundreds of thousands of DNA sequences from known individuals (e.g. GenBank, http://www.ncbi.nlm.nih.gov/Genbank/). The utility of these expansive databases is that they allow for used for comparison and subsequent identification of unknown samples through tests of sequence similarity. Finally, DNA based approaches have been used to identify individual prey items with missing or degraded morphological characteristics in birds, marine mammals, and fish (Jarman et al. 2004; Blankenship and Yayanos 2005; Smith et al. 2005; Deagle et al. 2005b; Deagle 2006)
The overall goal of this study was to compare the results from two techniques, visual and molecular analysis, applied to the same set of prey items recovered from the stomach contents of French grunt (Haemulon flavolineatum). The resulting prey identification determined by each method alone and combined was compared. This research examined the potential for molecular analysis to identify samples that would be unidentifiable by visual analysis alone.
Visual analysis of stomach contents was detailed in Chapter 2, and tests were conducted to look for changes in diet by sampling event, location and fish size. These goals were addressed through the quantification of prey items and comparison of niche breadth and dietary overlap (Chapter 2). Molecular identification of prey items was performed to determine if prey type, prey state of digestion, or marker selection affected the ability to sequence extracted DNA (Chapter 3). Finally, critical factors influencing the success of both visual and molecular analysis from the same samples will be discussed (Chapter 4).
Food habits of French grunt
French grunt occur in waters less than sixty meters deep in the western Atlantic from South Carolina to Brazil, including parts of the Gulf of Mexico. French grunt are common throughout much of their range and in some locations are the most abundant fish species observed on coral reefs (Randall 1967) .
French grunt undergo ontogenetic shifts in habitat use, shifting from nursery areas to coral reefs (Cocheret de la Morinière et al. 2003a; Cocheret de la Morinière et al. 2003b; Nagelkerken and van der Velde 2004; Nagelkerken and Velde 2004). Newly settled larvae inhabit interstitial spaces in seagrass beds and occur as solitary individuals. Juveniles aggregate into schools and utilize mangrove roots, patch reefs, and structures within seagrass beds as nursery habitat (Cocheret de la Morinière et al. 2003a; Cocheret de la Morinière et al. 2003b). As individuals reach sexual maturity, they form resting schools on coral reefs and eventually move to offshore habitats (Meyer and Schultz 1985).
French grunt feed via a winnowing behavior, whereby potential prey and non-nutritive debris are separated within the oropharyngeal cavity. Prey items that are retained are macerated by pharyngeal teeth before entering the stomach. Several studies have documented the diet throughout their range, including the Netherland Antilles (Cocheret de la Morinière et al. 2003a; Cocheret de la Morinière et al. 2003b; Nagelkerken and van der Velde 2004; Nagelkerken and Velde 2004), Puerto Rico(Austin and Austin 1971; Dennis 1992), Haiti(Beebe and Tee-Van 1928), Florida (Davis 1967; Hein 1996) and the United States Virgin Islands (Randall 1967). Diets change as fish transition through habitats, and major foraging guilds have been determined by size: pre-juveniles (<50 mm FL), juveniles (51-150 mm FL) and adults (>150 mm FL) (Hein 1999). Pre-juvenile fish feed during the day on planktonic copepods within the nursery habitat (Gaut and Munro 1983; Cocheret de la Morinière et al. 2003a). Juveniles begin making migrations to seagrass beds to forage at night on benthic invertebrates which includes tanaids, decapod crabs and shrimp, and polychaete worms (Dennis 1992). Adults form resting schools on coral reefs and make crepuscular migrations to grassbeds and sand patches where they consume primarily polychaete and sipunculid worms, gastropods and decapod crustaceans (Randall 1967; Estrada 1986; Dennis 1992).
The relative importance of prey items varied some between studies; however, unidentifiable prey items were routinely encountered. Dennis (1992) recorded at least one unidentified prey item in 263 out of 330 (79%) stomachs examined, and Cocheret de la Moriniere et al. (2003a) found unidentified material constituted up to 58% of total stomach contents by volume for particular sized fish.
French grunt were selected as the study species for two reasons: 1) their ecological importance within the seagrass and coral reef habitats as predators of benthic invertebrates and transporters of nutrients between habitat types, and 2) in order to qualitatively enhance diet results that have frequently contained large numbers of unidentified prey.
The objectives of this chapter were to visually quantify the diet of French grunt, and test for changes in diet by sampling event and habitat type. Frequency of occurrence data for various prey types from this chapter will be used for direct comparison with the results of molecular analysis of stomach contents in chapter 4.
French Grunt Collections
French grunt were collected from the southern shore of St. John Island, US Virgin Islands, USA, in June of 2008 and May/June of 2009. For each individual, the time of day, date, GPS location, fork length (FL), depth (meters), and gear type was recorded. In 2008, collections were made throughout the day to determine periods of peak foraging and once established sampling was targeted towards times with greatest chance of prey filled stomachs. In total, fish were collected from two reef habitats (Tektite and Fish Bay) and one seagrass habitat (Viers Dock).
Grunts were collected using multiple gears including hook and line, trap, pole spear, and hand net. Hook and line sampling consisted of sabiki rigs with multiple hooks baited with squid. Fish traps (1m long, 1m wide, and .5m in height) with mesh size of ~7.5cm2 were set on sand patches near reef habitats, baited using cat food (Kozy Kitten), and set overnight. Both spear fishing and hand net collections were conducted by divers using scuba gear. Collections by hand net utilized a modified cast net (brail and hand lines removed) to capture French grunt.
Fish captured alive were brought to the surface and their and stomach contents were obtained via gastric lavage (Foster 1977; Light et al. 1983; Hartleb and Moring 1995). A 3 mm diameter tube attached to a pressurized sprayer tank was inserted into the fish's esophagus and water was pumped in pulses as the wand was moved back and forth to loosen prey items. Regurgitated contents were deposited onto a 200 micron sieve and stomach contents were collected and put into individually labeled sterile sampling bag. Sampling bags were kept on ice in the field ice prior to freezing in the lab. For fish less than 75 mm FL a 10 mL syringes rigged with an inflating needle were substituted in the place of the sprayer tank. Fish harvested using a pole spear were placed directly into individual plastic bags underwater and placed on ice at the surface prior to transportation to VIERS laboratory. Speared grunts were dissected, and contents from the esophagus to upper intestine were removed for visual analysis.
Stomach Content Analysis
Stomach contents were transported to University of Florida FAS facilities where they were thawed and examined using a Leica MZ 12.5 stereomicroscope. Diet items were identified to the lowest taxonomic level using relevant identification keys (Manning 1969; Fauchald 1977; Abele and Kim 1989; Kensley and Shotte 1989; Thomas 1993; Cutler 1994; Hendler et al. 1995; Heard et al. 2003). Individual prey items were catalogued by fish number, given a distinct identification number and photographed using a digital imaging system (Motic). Once sorted, individual items were rinsed with deionized water and placed into individual 1.5 mL Eppendorf vials containing 100% non-denatured ethanol.
Recovered food items were catalogued by percent occurrence (%O), and percent by number (%N). Percent occurrence is defined as the total number of stomach containing a particular prey type divided by the total number of stomachs containing food. Numerical abundance is calculated as the number of each prey type from all stomachs divided by the number of all prey items from all stomachs. Volumetric measurements were not taken because prey types were diverse and subject to volumetric distortion as a result of digestion. Gravimetric measurements were not taken for two reasons: 1) due to the error associated with wet weights and the inability to remove water equally from all prey types and 2) to minimize processing time between thawing and immersion into DNA preservative.
Cumulative prey curves, which count the number of new prey item per each individual stomach, were generated to assess the adequacy of sample size (Ferry and Caillet 1996). A random number generator was used to determine the order in which stomach were analyzed and the number of novel prey items per new stomach was recorded. This process was iterated ten times to establish averages and standard deviations for each stomach number. A graph showing the total number of stomachs versus the number of new prey items in each stomach was generated. If the curve for this graph reaches an asymptote then the diet is considered to be well described (Ferry and Caillet 1996).
Niche Breadth and Diet Overlap
Comparisons between sampling events and habitat types were calculated using %N. Niche breadth was calculated using Levin's standardized index (BA) to determine if differences in sampling locations and collection event influenced the niches exploited (Krebs 1999a). This index returns ranges on a scale from 0 (narrow niche) to 1 (broad niche).
Diet overlap for sampling event, habitat type and fish size was calculated using Morisita's index of similarity. Morisita's index was selected because of low bias with varied samples sizes and large numbers of resource states (Smith and Zaret 1982). Both sets of calculations were performed using Ecological Methodology software package (Krebs 1999b).
French Grunt Collections
A total of 99 fish were collected during two sampling trips in June of 2008 (69 specimens) and May/June of 2009 (30 specimens). Multiple gear types were employed in both years with the majority of fish collected using a hand net (Table 2-1). Overall, sampled French grunt ranged in size from 57-188 mm FL (Figure 2-1) with a mean length of 119 mm (SD = 40.3). Fish collected in 2008 spanned a larger range of sizes and displayed a bimodal distribution with peaks around 70 cm and 170 cm. Samples from 2009 were significantly larger on average than 2008 (Welch's t-test=-2.92, df=76.5, p=0.004 ) and were distributed evenly across sizes from 80 mm to 180 mm. Multiple habitat types were sampled with 69 (70%) fish collected from coral reefs and 30 (30%) from seagrass beds. Fish collected on coral reefs were larger in general compared to seagrass beds and encompassed a larger range of sizes.
Samples collected at different times during the day in 2008 indicated morning hours yield more stomachs filled with prey and subsequently all collections in 2009 occurred prior to 09:00 AM (GMT- 4:00) (Figure 2-2).
Stomach Content Analysis
In 2008 a total of 69 fish were collected and 26 (38%) of stomachs contained prey items. From 2009, a total of 30 fish were collected and 25 (83%) had stomach contents. Regardless of digestion code, all prey items recovered were included in analysis.
French grunt collected 2009 contained more prey items on average (2.6 in 2008 vs. 7 in 2009) and a greater array of prey types. Unidentified crustaceans were commonly encountered in both years by occurrence (46% and 48%) and number (13% and 10%) for 2008 and 2009 respectively (Table 2-2). Polychaete worms were abundant by occurrence (19% and 20%) as were harpacticoid copepods (23% and 16%) again for 2008 and 2009 respectively. Numerically unidentified prey items (6% and 13%) and polychaete worms (5% and 3%) were commonly counted prey items. Sipunculid worms showed the largest difference between years by occurrence (15% and 96%) and number (8% and 35%) for 2008 and 2009 samples respectively. Unidentified prey items were less common by occurrence in samples from 2008 (23%) relative to 2009 (68%). Stomatopods and taniad crustaceans were absent from stomachs collected in 2008 but present in 2009 samples (36% and 16% by occurrence and 7% and 2% numerically).
Diet by Sampling Location
Stomach contents of French grunt were divided by sampling location to examine potential trends. In total, two coral reef habitat types were sampled, Fish bay (n=9 stomachs) and Tektite reef (n=27 stomachs), and one seagrass bed, Viers Dock (n=15 stomachs). Prey types were lumped into broader taxonomic categories (Amphipoda, Copepoda, etc.) to simplify analysis. Several similarities were observed between diets collected at the three sites. Unidentified crustaceans were abundant by %O (53%, 56%, and 37%) and %N (17%, 15%, 10%) for Viers Dock, Fish Bay, and Tektite Reef respectively (Table 2-3). Unidentified prey items were also abundant at all spots (33% VD, 56% FB, and 48% TR) by occurrence.
Distinct differences were revealed by location with fish only being consumed at the Viers Dock location (13% and 17%) by %O and %N. Foraminifera (4% and 3%) and polyplacophora (4% and 1%) by %O and %N were only observed in stomachs from Tektite reef. Stomatopods (11% for both locations by %O) and tanaid crustacean (11% FB and 30% TR) were identified only from reef samples. Sipunculid worms were consumed at all three locations, with more individuals consumed by occurrence and number in Fish Bay (89%O and 32%N) and Tektite reef (67%O and 35%N) relative to Viers Dock (13%O and 13%N).
Diet by Size
The diet of French grunt was divided into two size groups (A = <90mm standard length (SL) and B = >90mm) to examine potential patterns in the abundance and frequency of different prey items. These size groups were selected because previous work has suggested that nocturnal foraging begins during a mid-juvenile phase (Hein 1999). Fork lengths of French grunt collected in this study were converted to standard length (SL) using the regression FL = 1.04 * SL + 4.04 (Hein 1992). Prey types were lumped into broader taxonomic categories (Amphipoda, Copepoda, etc.) to examine general trends. In total, size group A contained 25 individuals and group B had 26 individuals.
For guild A the most important prey items numerically were unidentified crustaceans (18%), sipunculid worms (17%), copepods (10%), fish (9%) and polychaetes (8%) (Table 2-4). By occurrence, unidentified crustaceans also dominated (52%) followed by copepods (28%), polychaete worms (24%), shrimps (24%), and sipunculids (24%). For guild B, sipunculid worms were most important by number (36%), unidentified prey items (15%), unidentified crustaceans (10%) and tanaid crustaceans (6%). By occurrence, sipunculid worms were consumed by almost all individuals (85%), followed by unidentified prey types (69%), unidentified crustaceans (42%) and ophiuroids (27%). Both shrimp and crabs were more commonly consumed by smaller individuals than the larger groups (44%O and 23%O).
Cumulative Prey Curves
The cumulative prey curve constructed based on all diet items consumed by all fish collected did not reach an asymptote, indicating that the diets of French grunt were not well characterized.
Niche Breadth and Diet Overlap
Niche breadth for the two sampling trips was considerably different (0.19 and 0.52 for 2008 and 2009 respectively). The number of resource states utilized was higher in 2009 (38) compared to 2008 (18), however fish from 2009 used only three resources frequently. Niche overlap as calculated using Morisita's simplified index for the two sampling trips was moderate (0.62).
When analyzed by location, niche breadth was greatest at Viers dock (0.64) followed by Fish Bay (0.41) and Tektite Reef (0.31). The number of resource states was similar amongst sites (VD=13, FB=14, TR =17) with 8 resources commonly used at Viers Dock and Fish Bay and only 5 resources on Tektite Reef. Niche overlap was higher between Fish Bay and Tektite Reef (0.91) compared with Fish Bay to Viers Dock (0.70) or Tektite Reef to Viers Dock (0.63).
By size, the niche breadth was greater for smaller individuals (0.64) relative to larger ones (0.27) as calculated by Levin's standardized measure. Both guilds utilized an array of prey types (Guild A = 14, B=18), however much fewer prey categories were routinely utilized by larger individuals (4 vs. 8 for A and B respectively). A moderate amount of diet overlap between guilds was observed according to Morisita's simplified index of overlap (0.73).
Sipunculid worms were both numerically and by occurrence the most commonly consumed prey item by French grunts collected in this study. The majority of sipunculid worms counted in this analysis were partial organisms and consisted of little more than the distal portion of the sipunculid introvert, suggesting that numerically their importance might be overestimated. Both unidentified crustaceans and unidentified prey items were found in almost half of all individuals stomachs analyzed and represented important prey types numerically. Prey items coded as unidentified crustaceans or unidentified prey encompassed a wide range of sizes and without gravimetric or volumetric measurements minimal inference can be made regarding the dietary importance of theses prey types. Harpacticoid copepods were found in a large portion of the stomachs however likely contributed little nutritional value to the diet of French grunt given their small size (1mm total length).
When stomachs were divided by sampling location sipunculid worms were found to be important numerically and by occurrence in both reef habitat types, but were consumed much less frequently from the seagrass habitats. There is no clear explanation why sipunculid worms might be consumed more commonly on coral reefs without knowing something about the relative abundance of prey items. Stomatopods and tanaid crustaceans were only observed in stomach collected on reefs and again this might be related to prey abundance.
Stomach contents of individuals categorized by size revealed that fish less than 90mm SL was the only size class to consume fish as prey and can potentially be explained by the large schools of silversides (Atheriniformes) present in the sampling area. Fish greater than 90mm SL consumed the only ophiuroids recovered from stomach contents however none of these individuals represented whole individuals but distal portions of legs.
When calculated for all individuals, niche breadth suggests that French grunt exploit neither a narrow nor broad niche as evidenced by moderate Levin's index of niche breadth (0.38). Niche breadth appeared to decrease with fish size suggesting that either fewer prey types are exploited or that diets are heavily influenced by specific prey types consumed in large quantities. Similarity between diets was greatest between the two reef sampling sites and to a lesser extent seagrass habitat according to Morisita's index of similarity which is what one would expect.
The results of this study are generally consistent with previous diet studies conducted in the region, however the ability to draw direct comparisons to previous studies was limited due to different indices used to catalog prey items. Dennis (1992) analyzed the diets of 330 French grunt from Puerto Rico and concluded that polychaete worms, sipunculids, gastropods and shrimp were the four most important prey types by biomass. Randall (1967) found that for French grunts collected in the US Virgin Islands and Puerto Rico, polychaetes were the most important prey type volumetrically followed by crabs and sipunculid worms. Estrada (1986) examined French grunt diets from Columbia and concluded that crustaceans, mollusks and polychaetes were most commonly encountered by frequency and consumed more worms (sipunculids and polychaetes) relative to other grunt species. Interestingly enough, multiple studies conducted in the Netherland Antilles concluded that neither sipunculid worms nor polychaetes were ingested at any level (Cocheret de la Morinière et al. 2003a; Cocheret de la Morinière et al. 2003b; Nagelkerken and van der Velde 2004). Diets from fish collected on coral reefs were dominated volumetrically by decapods crabs and prey fishes while fish from nursery habitats consumed primarily tanaid crustaceans, copepods and decapods crustaceans (Cocheret de la Morinière et al. 2003a; Cocheret de la Morinière et al. 2003b).
The high rate of stomach contents with unidentified prey items and the fact that previous studies produced similar results provides justification for exploring alternative means of identification, specifically molecular analysis which does not rely on morphological characteristics to generate identifications.
Table 2-1. French grunt collections by year and gear type.
Table 2-2. Occurrence (%O) and numerical abundance (%N) for prey sampled from French grunt stomach contents collected from St John, US Virgin Islands in May/June of 2008 and June of 2009.
Table 2-3. Occurrence (%O) and numerical abundance (%N) for French grunt stomach contents divided by sampling site.
Table 2-4. Diet of French grunt by number (N%) and (O%) based on size category.
Figure 2-1. Size distribution of French collected from St. John, US Virgin Islands in May/June of 2008 and June 2009.
Figure 2-2. Collection times and number of individuals containing stomach contents by time for French grunts.
Figure 2-3. Cumulative prey curve for all prey items recovered from French grunts collected in 2008 and 2009.