Florida State University Libraries
Faculty Publications FSU Coastal and Marine Laboratory
2011 Cownose Ray (Rhinoptera Bonasus) Predation Relative to Bivalve Ontogeny Robert Fisher, Garrett Call, and Dean Grubbs
Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected] Journal of S/11'1/jish Research. Vol. 30. o. I. 187 196.2011.
COWNOSE RAY (RHINOPTERA BONASUS) PR EDATION RELATIVE TO BIVALVE 0 TOGENY
ROBERT A. FISHER,' * GARRETT C. CA LL 1 AND R. DEAN GRUBBS2 1 Virginia Institllfe of Marine Science. College of William and Mary, PO Box 1346. Gloucester Point, VA 23062: 2Fiorida State University Coastal and Marine Labor(ltory . 36 18 Hll')' 98. St. Teresa. FL 32358
ABSTRACT The purpose of this ｾｉｕ､Ｉ＠ was to determine the ability of the cownose ray. Rhiuoptera houasus (Mitchill. 1815). 10 manipulate oysters and ｣ｬ｡ｭｾＮ＠ to test for relative prey preference. and to investigate whclher susceptibility to cownosc ray predation changes wi1h bivalve omogeny. We investigated patterns of preda1ion forcaplivc aduh and young-of-yearcownose rays on 4 species of bivalves, including Crassostrea 1•irgiuica (Gmelin, 1791 ). Cm.uostrea ariakeusis (Fuji ta. 1913). ;\-lercenaria mercenaria (Linnaeus. 1758). and My a areuaria Linnaeus, 1758. In oyster (C. ｬ ｩｲｾＺｩ ｵｩ ｣ｴｴＩ＠ irials, predation probabilities by adull rays were highest at shel l heights of30 70 rnrn and shell depth s of8 22 mm. The ra tes of preda ti on by adult rays in trials in which sa me• size oysters were used were higher than rates in most comingled trials. Adult rays showed no differences in predation between nati ve oysters (C. virginim) and nonnative oysters (C. ariakensis; P > 0.05). Adult rays selected hard-and soft-shell clams (Manly• Chcsson index M. merrenaria. a = 0.736 ± 0.002. elcctivity = 0.473 ± 0.007: M. arenaria. a = 0.742 ± 0.003. elcctivity = 0.485 ± 0.013) over oysters (C. 1•irginira. a = 0.263 ± 0.002. clcctivity = - 0.473 ± 0.007: a = 0.257 ± 0.003. elcctivit)' = 0.485 ± 0.003). In young-of-year feeding trials. oysters with a shell height of 10-35 mm and a shell depth of3- 12 mm had the highest probabili1y of predation. a ative oyster and hard clam peak force or load crush tests resulted in forces of200-1.500 and 400 1.400 1 across shell depths of 10-35 mm and 21 34 mm. respectively. before valve failure. The results of th is study indicate that cownose ray predation on shellfish is limited by shell size and is likely related to ray jaw gape and bite force.
KEl' WORDS: Durophagy. cownose ray. prey selection. predation risk. Rhinnptera. bivalve mollusc
I TROOUCI'IO has been documented (Gray ct al. 1997). and studies have shown the general absence of oysters in the diets of rhinoptcrid The cownosc ray. Rhinoptem honasus (Mitchill. 1815), is and myli obatid rays (Smith & Mcrrincr 1985. Collins et al. a member of the order Myliobatiformcs. which includes 10 2007) even when associated with oyster beds (Gray et al. 1997). families of stingra ys. Cownosc rays (Rhinoptcridae) include at The ability of cownose rays to feed on large oysters is also least 7 species (Compagno 2005) of coastal pelagic rays that questionable a a result of the gape limitations of these rays often travel in large schools. R. /)onttsus is the onl y species that (Summers 2000. Sasko et al. 2006) and the force required to occurs along the cast coast of the United States. and is crush Eastern oysters (Bishop & Peterson 2006). However. distributed from southern cw England to Brazil and through• Peterson ct a!. (2001 ). report that cownose ra ys in orth out the Gulf of Mexico. Cownosc rays undergo long seasonal Carolina arc capable of depleting dense patches of weaker migrations similar to those exhibited by most coastal sharks shelled bay scallops (Argopecten irradians (Lamarck. 1819)). (Smith & Merriner 1987, Grusha 2005). In spring. they migrate Oyster restoration and commercial grow-out efforts in Virgi nia north. reaching the Outer Banks of North Carolina by April have undoubtedly experienced setbacks because of cownosc and subsequently the Chesapeake Bay in early May (Smith & rays consuming deployed oyster on experimental reefs and Mcrriner 1987). Cownose rays arc abundant in the Chesapeake commercial grounds. In 2004 and 2006. 1.2 million and 775.500 Bay and its tributaries throughout the summer. occurring at oysters were seeded for reef restoration in Virginia. respectively. salinities as low as 8 (practical salinity scale) and temperatures Wesson (2009) reported that 95% were c;llen by cownose rays. from 15-29°C (Smith & Merriner 1987). By early October. most Of the 9 species of batoids that inhabit the Chesapeake Bay cownosc rays have vacated the Chesapeake Bay to begin their during summer months, only 2 species the cownose ray and the southerly migration to wintering areas primarily oli the Atlan• bullnose ray (M_rliobatisj'remim•illii Lcscur. 1824)- ha ve grinding tic coast of Florida (Grusha 2005). plates and jaw musculature potentiall y capable of manipulating Cownosc rays are durophagous (feeding on hard-shelled and crushing oysters and hard clams (Mercenaria mercenaria prey) predators. feeding on molluscs as well as crustaceans and (Linnaeus. 1758)). Although the bullnose rays may be capable of benthic polychaetes. Collins ct al. (2007) reported that cownose manipulating and crushing adult oysters and hard clams. they arc rays from Charlotte Harbor on the Gulf coast of Florida fed relatively uncommon in Virginia waters. are generally solitary, primarily on small crustaceans (cumacca ns) and sedentary and arc therefore unlikely to be major predators on bivalves in polychactcs, but most earlier studies reported that the dominant this region. Cownose rays. in contrast. arc ex tremely abundant in prey arc small. weak-shelled bivalves (e.g .. Smith & Mcrriner the Chesapeake Bay. The reports of cownose ray predation on 1985. Blaylock 1993). Concern over predation on commercial commercial bivalves coupled wi th questionable claims of dra• bivalve resources have been raised by fishe ry and aquaculture matic increa es in the cownose ray population coastwide (Myer operation for many years and in several regions of the world. et al. 2007) have spurred interest in developing a commercial However. little evidence of actual predation on these re ources fishery for cownosc rays or at least identifying nonlethal de• terrents for keeping cownosc rays from commercial beds. •corresponding author. E-mail: rlishcr(a vims.cdu Cownose rays use several behaviors in feeding on bcnlhic DOl : 10.2983/035.030.0 126 prey. Cown ose rays excavate invertebrate prey from the substrate
187 188 FtSII ER ET 1\1 .. by using vigorou o cillati ons of the pectoral fins and ｢ｾ＠ jelling Iarine Science in Gloucester Point. VA. We held adult rays and water taken in through the spiracles during ｲ･ｾｰｩｮｴｬｩｯｮ＠ from the performed predation trials in an aboveground. oblong fiber• mouth to separate prey funher from sediment (Schwal'li' 1967. ｧｬ｡ｳｾ＠ tank (3 X 4.2 m) with sand filter recircula 1ion . Water depth Sasko 2000). Inertial suction feeding moves prey from the was maintained SH and YOY comi ngled trials to examine the effect of each SH group, SO. and time period (when appropriate) on predation probability where a binary response- 0. alive: I . dead- is related to one or more predictor variables. A logistic regression model predicted the probability of predation of 3 different bivalves in the comingled trials- C. 1•irginica. C. ariakensis. and M. mercenaria- by capti ve cownose rays. The model can be ex pressed as Log it {p(x)} = log {p{.r)/ 1- {.r)} = bo +b1x+b2x2 Figure I. Side view of an oyster (C. 1"ir;:i11ica). S O. shell depth: S H. shell height. where p(x) is the probability that a bivalve will be preyed on as a function or a variable x. and b0• h1• b2 are the regression parameters. The equation can be rearranged to define estimated end of the holding tank (opposite end from where prey were probabi lity p(x) as introduced) using a fence constructed of PVC that extended the width of the tank. The rays were corralled there until shell fish eiJo+hox+lll.r and crushed shell renwants were collected from the tank p(x) = { 1+e!Jo +l>or1 1>1.t' } bottom. Collection was performed by compiling the shell from the tank bottom using a 1-m long rubber queegce. followed by Factors (x) contributing to the probability of predation scooping shell from the pile wi th a 2-gal capacity funnel (p(x)) included SO and SH groups, and. in one instance. time attached with a 1-mm-mesh filter bag. The final removal of period for C. 1•irginica. For analysis of C. virginica. the SH small pieces was conducted using a 6-gal wet-dry shop vacuum. groups were 15- 25, 30- 40. 45- 55, 60- 70, 75- 85, and 90- 100 Whole bivalves recovered after each trial were sorted from shell mm. For C. ariakensis the SH groups were 45- 55. 60- 70. a nd remnants, grouped to size or species classification. counted, and 75- 85 mm. The groups for M. mercenaria were 30- 35. 40--45, remeasured (S H or SW and SO). and 50- 55 mm. We applied this model to each trial for time periods of 7.5. 15. 30. 45. 60. 120. 240 min for C. 1•irginica: 30 Comi11gled Oyster Susceptibility Trials min for C. ariakensis: and 15 min for M. mercenaria. Time (x) To evaluate size preferences, we comingled multiple shellfish was added as a factor to the model for C. 1•irginica to ge nerate size groups together and introduced them simultaneously to the a predicted probabi lity across multiple time peri ods. Parameter rays. In comingled trials with adu lt cownose ra ys, 25 single estimates for each predictor va riable were generated and oysters or clams per SH group (for a total of !50 oysters or 75 evaluated for significa nce (P < 0.05). Model fi t was evaluated clams) were mi xed and dumped into the holding tank approx• using Hosmer and Lemeshow tests. imately I m from the tank's vertical end wa ll. resulting in a mound of randomly mixed bivalves of various sizes covering 2 £1·aluatio11 of Peak Load of C. \irginica a11d M . rnercenaria approximately 0.5 m • For C. l'irginica. feeding trials were conducted in triplicate for time periods of 7.5. 15. 30. and Forty oysters (C. virginica: Sl-1. 24-95 mm; SO, 12- 35 mm) 45 min. and duplicate for 60-. 120-. 240-min periods. For C. and 36 hard clams (M. mercenaria: SH. 33- 54 mm: SO. 21-31 ariakensis. we only tested 3 SH groups (because of availability) mm) we re used to evaluate the force (load) required to crush in triplicate 30-min trials. Preliminary investigations feeding each species. We used a 100 Ki p Enerpac manual hyd raulic rays M. mercenaria demonstrated that exceeding 15 min was pump and jack system. connected to a 5.500-lb (25-k ) MTS likely to exhaust the 25 clams in the 30-40-rnm SH size class: Systems Corporation (Eden Prairie. M ) load cell (model therefore. clam selecti vi ty trials were only conducted at 15-min 661.20 8-0 I). The load cell was connected to a voltmeter durations. For comingled trials with YOY ra ys, 25 oysters per through an AC-powered Bridge sensor (model DMD 465WB) SH group (SH 10- 20 mm, 20- 30 mm. and 30--40 mm) were for taking load measurements. A standard 0- 2-i n range de• comingled in a 2-gal bucket. then dumped into the holding tank. flecti on dial gauge (wi th a least count of 0.00 I in) was used to 2 resulting in a mound - 400 cm . Triplicate 18-h feeding trials record deformation/deflecti ons of the shellfish specimen. Cou• were conducted. pling the MTS load cell with the Bridge sensor inerea ed the Data analyses were conducted using SPSS for Windows resolution of the load readings greatly. and the manual (version 16.0.0. 1BM . Somers. Y). Adult comingled trials were hydraulic pump gave precise control over the load incre• initially evaluated using chi -square tests and G tests to test the ments/intervals. The least applicable load was 0.7 lb. or 3 N, null hypothesis that predation success was eq ual for bivalves of wi th this configuration. all SH. In trials in which predation success was uneq ual, we used The load cell was calibrated under the MTS load frame the Manly-Chesson alpha index of selectivity for va riable prey system before testing shellfish. The calibration involved the abundance and normalized it to get electi vit y ( I is complete applica ti on of a known load to the load cell assembly in avoidance and + I is complete preference) to evaluate prey increments and the corresponding vo ltage output recorded. preferences. Actual count data were standardized to display This process establishes the voltage-to-load calibration rela• the proportion of predation based on SH and SO measurements ti onship for the load cell. We weighed and measured SH and SO before and after comingled trials. The mortality data collected for all li ve bivalve samples. Specimens were placed on a soli d from these trials were also used to ge nerate proportions of steel platform under the load cell, a nd load testing commenced. predati on. We also used binary logistic regression for both adult With all shellfish samples, the load cell was gently brought in 190 ｆｴ ｓｾ ｉ ｉＺ ｒ＠ ET t\ L. contact with the specimen. and the dcAcction dial gauge was set RES LTS to 0. A small increment of load was then applied using the hydraulic pump. and corresponding deformation of the speci• ｃ｡ｭｩｬｬｾｴｨ ｊ＠ Trials men was recorded from the mechanical dial gauge. Thi process In comingled trials with adult cownose rays. the proportion continued until the specimen failed by crushing. One of the 2 valves ofs pecimens would fail first. at which point load reading of oyster successfully eaten increased for all SHs te ted as time increased. except for the largest SH class (90 100 mm: Fig. 2A). were recorded. indicating initial valve failure. or. for the purpose of this study. mortality. Load readings were made at point of first failure (cracking of one va lve) and again at point of 1 A .....7.5 (min) second valve failure. Load was measured in kilo- ewtons (from _ ,5 0.9 - 30 the load cell) versu vertical deformation in millimeters (ba ed ....-45 0.8 - 60 on the dial gauge readings). Compres ivc load readings were in -120 -+-240 pound force (lbl) with I lbf = 4.4482 c: 0.7 .Q ｾ＠- 0.6 Comparatil-e Predation Trials ｾ＠ ｾ＠ 0.5 Adult predation trials were conducted comparing C. ''irgin• 0 8 0.4 im and C. ariakensis. C. 1•irginica and M . llll!rc·c•Jwria. and C. t: 8. 0.3 1•irginim and M . arenaria. In comparative trials. 25 ｓ ｊｘＧ＼Ｚｩｭ･ｮ ｾ＠ of e both species from the same SH group wi th similar SO (Table I) Q. 0.2 were comingled and imultaneously introduced into the holding 0.1 tank with 4 adult rays. Trial time was held constant at 15 min 0.0 L--:-::-::-::-..._-=-:--:-::--'----,.,=-=:,...... --:,-:-::.,.....-'--=:-=-=-...._-=-...,..,.,,... and performed in triplicate. for oyster/soft-shell clam trials. we 15·25 30-40 45-55 60-70 75-85 90-100 visuall y counted mortalities at 3. 5. and 15 min for triplicate Shell Heights (mm) trials. For comparati ve experiment testing. preference chi• 1 B - 7.5 (m•nl square test or G tests were performed and combined to test -- 15 0.9 -...30 .. 45 for significant (o. = 0.05) differences in the number of each - so specie preyed on. Independent tests of significa nce were 0.8 - 120 240 combined using Fisher's (1954) method. We calculated the -.§ 0.7 Manly-Chesson alpha index of selectivity for va riable prey ｾ＠ "' 0.6 abu ndance and norma li zed it to get electi vi ty ( I is complete 0: 0 0.5 avoidance and + I is complete preference) (Chesson 1978) to ｾ＠ 0.4 determine prey preference when appropriate. ::0 e"' o.3 Rat!! Trials Q. 0.2 We evaluated size-mediated predation rates by adult rays 0.1 through predation trials grouping I00 C. l'l·rginim oysters from 0 ｾＭＭｾＭＭＭＭｾＭＭＭＭｾＭＭＭＭｾＭＭＭＭｾｾｾ＠ a given SH size over a 15-min period. Rate is defined as the mean 15-25 30-40 45-55 60-70 75-85 90-100 number of oyster mortalities per minute per ray within each Shell Heights (mm) individual time trial. OupliC Shell heights of 30-40, 45-55, and 60-70 111111 were the most Predation declined with increasing SD. The highest proportion heavil y selected for all time trial peri ods (Table 2). Lowest of predati on was observed in oysters with SOs between 8 mm predation success was observed on 15- 25-, 75- 85-. and 90- and 22 mm, whereas the lowest predation success was recorded 100-mm SH oysters. in oysters with SOs greater than 32 mm (Fig. 3). The probability of predation increased for all shell heights The hi ghest probability of predation among bivalves tested tested as time increased, except for the 75- 85- and 90-100-mm was for C. virginica in the 8-22-mm SD range, with predation oysters in the 15-min time period, and the 90-1 00-mm oysters in declining with increasing SO (Fig. 4). Probability of predation the 240-min time period (Fig. 28, C). Overall, oysters in the on C. ariakensis was highest for SOs of 14-20 mm. Simi larly, smallest and largest SH categories had the lowest selectivity. predation declined as SD increased above 22 mm. The hi ghest Mean SO of oysters within each SH group increased 2- 3 mm probability of predation in M. mercenaria was observed on SOs between pretrial and posttrial in the 60-70-, 75- 85-, and 90- between 21 111111 and 26 mm. A steep decline in predation was 100-mm oysters, suggesting selection for those oysters with observed as SD increased above 26 mm. A logistic regression smaller SO in larger oysters (Table 3). No difference in mean SD equation predicted the probabilities of predation for C. virginica was found in the 15-25-, 30-40-, and 45-55-mm SH oysters. based on the 8 variables tested (Fig. 4): TABLE 2. Combined predation (success or failure) on oysters (C. ••irgi11ica) for adult comingled trials. Time Trial Shell Height (mm) Success Failure Ot Elcctivity 7.5 min 15- 25 12 63 0.0504 :!: 0.0002 -0.5835 :!: 0.0097 11=3 Ｓ ｾ Ｐ＠ 47 27 0.3193 :!: 0.0057 0.3887 :!: 0.0232 45-55 38 35 0.2272 :!: 0.0008 0. 1871 :!: 0.0064 60-70 34 40 0.1853 :!: 0.0008 0.0605 :!: 0.0078 Chi-square = 73.42 75-85 32 44 0.1945 :!: 0.0156 0.0139 :!: 0. 1880 p < 0.05 90- 100 7 67 0.0233 :!: 0.0016 - 0.8180 :!: 0.0994 15 min 15- 25 18 57 0.0463 :!: 0.000 I - 0.6109 :!: 0.0044 11 =3 Ｓ ｾ Ｐ＠ 58 17 0.2637 ± 0.0072 0.2584 :!: 0.0513 45-55 64 I I 0.3373 :!: 0.0065 0.4217 :!: 0.0239 60-70 52 24 0. 1944 :!: 0.0007 0.0904 :!: 0.0071 Chi-sq uare = 146.7 75-85 42 33 0. 1461 ± 0.0039 -0.1072 :!: 0.0781 p < 0.05 90- 100 5 70 0.0121 :!: 0.0004 - 0.8945 :!: 0.0334 30 min 15-25 27 48 0.0526 :!: 0.0008 -0.5770 :!: 0.0345 11=3 Ｓｾ Ｐ＠ 64 I I 0.2283 :!: 0.0048 0.175 I :!: 0.0361 45-55 69 6 0.2853 :!: 0.0078 0.3117 :!: 0.0356 60-70 67 8 0.2741 :!: 0.0076 0.2852 :!: 0.0404 Chi-sq uare = 11 3.8 75-85 42 32 0.0997 :!: 0.0017 - 0.3038 :!: 0.0422 p < 0.05 90-100 28 47 0 06 16 :!: 0.0034 - 0.5559 :!: 0.1637 45 min 15- 25 29 46 0.0498 :!: 0.0005 -0.5915 :!: 0.()208 n = 3 ＳｾＰ＠ 65 10 0.2 188 :!: 0.0038 0.1524 :!: 0.0270 45-55 72 4 0.3076 :!: 0.0093 0.3576 :!: 0.0350 60-70 68 7 0.2374 :!: 0.0013 0.2 130 :!: 0.0087 G = 111.9 75-85 51 22 0.1213 ± 0.0005 -0.1875 ± 0.0103 p < 0.05 90-100 30 39 0.0652 :!: 0.0035 -0.53 13 :!: 0. 1522 60 min 15-25 26 24 0.0595 ± 0.000 I -0.5199 :!: 0.0008 11=2 Ｓｾ Ｐ＠ 49 0.2607 :!: 0.0003 0. 2755 :!: 0.0020 45-55 49 0.2607 ± 0.0003 0.2755 :!: 0.0020 60-70 48 2 0.23 13 :!: 0.0005 0.1998 ± 0.0039 G= IOI.I 75-85 38 12 0.1361 :!: 0.0068 - 0. 1562 :!: 0.1245 p < 0.05 90- 100 19 31 0.05 17 :!: 0.004 1 -0.6194 :!: 0.2023 120 min 15- 25 27 ?'-J 0.06 15 :!: 0.0005 - 0.5117 :!: 0.0204 11 = 2 ＳｾＰ＠ 50 0 0.2482 :!: 0.00 I 0 0.2430 :!: 0.0065 45-55 50 0 0.2482 :!: 0.00 I0 0.2430 :1: 0.0065 60-70 48 2 0.2191 :!: 0.0001 0.1673 :!: 0.0007 G = 78.40 75- 85 41 9 0. 1344 ± 0.0004 -0.1284 :!: 0.0073 p < 0.05 90- 100 32 18 0.0886 :!: 0.0032 -0.3716 :!: 0.0970 240 min 15-25 41 9 0.128 1 ± 0.0035 -0.1734 :!: 0.0679 11 = 2 Ｓｾ Ｐ＠ 25 0 0.2174 :!: 0.0001 0.1629 :!: 0.000 I 45-55 50 0 0.2174 ± 0.0001 0.1629 ± 0.000 I 60-70 50 0 0.2174 ± 0.000 I 0.1629 ± 0.0001 G = 123.3 75-85 48 2 0.1941 :!: 0.0120 0.0885 :!: 0.0 120 p < 0.05 90-100 15 35 0.0255 :!: 0.0260 -0.7740 :1: 0.0260 Shell heights in bold type indicate preferred prey items. 192 FISHER ET ;\ L. T;\81.E 3. 1 ----.._ ····-·-·. ... - C. vitginlc8 \ . •. - C. ariakensis Range of shell depths and mean shell depth of C. •·irgini<'a 0.9 ·, - ... M mereenana before and after ad ult ray comi ngled predation trials. ' 0.8 '• • .§ 0.7 ｾ ｬ ･｡ ｮ＠ Shell ｾ ｬ ･｡ ｮ＠ Shell • -<0 ! "Q 0.6 • Shell llciJ!hl Range or Shell Oeplh Beron• Ocplh Aflcr ｾ＠ a. •. (mm) Oeplh (mm) Feedi ng (mm) Feeding (mm) - 0.5 \ 0 • 15 25 4 9 6 6 ｾ＠ 0.4 :0 \ \ 30 40 8 18 11 12 <0 D 0.3 45 55 11 22 16 16 e \ 60 70 13 33 21 24 a. 0.2 75 85 IS 35 27 29 0.1 aw 90 100 18 40 30 32 0.0 ｾＭｾｾＭ］ＭＢＺＭＧＭＺＭＢＺＧＭＺＭＺＭＭＭＭＭＭＭＺ ＭｌＮＮＮＺＭＢＺＧＺＮｌＭＮＭｬＺＺＮＺＮＮｾＮＮＮＮｬﾣＮＮＭ 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 Shell Depth (mm) Figure 4. Mean prediclcd probability or adull ray predalion rrom logislic:• regression ｭ ｯ､･ ｬ ｾ＠ ror C. ｴ ﾷ ｩｲｾＺｩｮｩｃＢ｡Ｎ＠ C. ariakt•nsis, and M. mercenuriu as funcrions or shell d('pl h. Vcrlicallines ｲ ＨＧ ｰｲＨＧ ｾ ｮｬｬｨ ･＠ range in maximum ja11 gape for adul l rays ｵ ｾ ､＠ in pr('dalion. whe re p(.\') is Prob (0. I) . .\' 1 is time.. \· 2 is SO . .\') is an SH of 15- Ｎ｜Ｇｾ ｩ ｳ＠ 25 mm . an SH of 30 40 mm . .\'5 is an Sl-1 of 45 55 mm. x6 is an Sl-1 of 60- 70 mm. and .\'7 is and SH of 75- 85 mm. All p(x) ＭＭＭＺＭＺＭＺＺＭＺＭＺＭＺ｣ＭＭＭＬｾＭＭＭＮＬＮＭＺＺＭＺＺＭＭ Ｍ ｾ］ＭＭ variables were significant at the 0.05 leve l. and the Hosmer and = 1+e tJO ..l55 • o'I'H ,, • 13 ＼ＩＮＱＮＱ ＬＺｾ＠ - 0.93-1, , 1 · Lemcshow Test was nonsignificant ( P > 0.108). suggesting the model adequately fit the data. Individual analy is of each time The Hosmer and Lemeshow test wa nonsignificant ( P > 0.394). trial period resulted in nonsignificant Hosmer and Lemeshow uggesti ng a better model fit and. in additi on. 2 (intercept and tests for all time periods except for the 15-min period. Between SO: P < 0. 12 and P < 0.05. re pectively) of 4 parameter estimates 3 5 of 7 parameter estimates were significant for each period. were significant. Predicted probabilities from the model arc shown (Fig. 4). but the parameter estimates for SO and the smallest SH group ( 15 25 mm) were significant for all time trials (Table 4). In comingled trials wi th YOY rays, the probability of We ge newted a logistic rcgrcs. ion equation for('. ariakensis. predation declined as SH and SO increased (Fig. 5). The equation generated for YOY predation is I p(x) = l+e Ｑ Ｑｾ＠ 3:!9 • u S5<>,,· ＱＶＭｭ ＬＬｾ＠ ｾ ＰＸＱＹ ＬＬＱ＠ • where the intercept and SO pam meter estimates were significant (P > 0.0 I) and the SH parameter es1imatcs were nonsignificant Paramc1er estimales for intercept and SH were nonsignifi• (P > 0.998 and P > 0.472. rcspecli ve ly). However. the Hosmer cant ( P > 0.997. P > 0.850. and P > 0.285). whereas the estimate and Lcmeshow test was signifi cant (P < 0.026). suggesting the for SO was significant ( P < 0.05). The Hosmer and Lemeshow test model did not adequately fit these data. For hard clams (i'v/. suggesred the model did not adequately fit the data (P < 0.049). mrrcenaria). the logistic regression equation is The force needed to cause failure in one or both va lves in C. rirgini('(l and M . mercenaria increased as SO increased (Fi!!S. 6 and 7). The plot of the log-tntnsformed SO and peak -load 1 - C. wgmtea displays that the load scales isometrically wi th SO. 0.9 For M. mercenaria. linear peak load is lowest at a 21-mm SO and ｩｮ｣ｲ･｡ｾ･ ｳ＠ to nea rl y 1.400 Nat a 33-mm SO (Fig. 7A). Adult 0.8 probability of preda tion and peak load inrerseet at 30 mm for c: 0.7 .2 M. mercenaria. Peak load for C. •·irxinica is lowest at an SO of -<0 "Q 0.6 10 mm and increa es to more than lm500 at a 35-mm SO 41 ｾ＠ Q. 0.5 (Fig. 78). Adult probability of predation and peak load in• -0 C. rirf(inica. YOY c: te rsect at the 29-rnm SO for predation and 0 0.4 t: linear penk load (C. l'irginica) intersect at a 17-mm SO. 8. 0.3 e a. 0.2 Ratt• Trial,, 0.1 The rate of predation for all oyster SH groups decreased 0.0 ＧＭＺＺｾｾ ｾｾ ｾＭＭＭＭ］ＭＭＺＭＺＭＺＺＭＺＭＺＭＺＭＭＭＭＮｬ ＭＭＭＭＭＭＭＺＭｬｾＮＮＮＮＮｬＮ］Ｍ Ｍ with increasing trial time. In the 7.5-min time rrial . 30-40-rnm 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 SH oyster· were preyed on quickest. followed by the 45- 55·. 60- Shell Deplh (mm) 70-. then 75 85-mm SH oysters (Fig. 8A). Cownose ray pre• Figur(' 3. l'ro porlion or C. ｯ ﾷ ｩｲ ｾＺｩｮｩ｣ﾷ ｡＠ ('31('n as a funclion of shell deplh in dation rates on oysters were onl y sli ghtl y higher on same-size c:-orninJ!ICd I rials. Vcrlica llines n•pres<'nl I he range in maximum jaw ga pe oysters compared with cominglcd oysters of varying sizes. for adull ruys ｵｾ､＠ in preda1ion I ri als. except in the 75 85-mm SH (Fig. 88 ). CowNosE R AY PREDATION 193 TABLE 4. Parameter estimates j30 ... j36 corresponding to the intercept, sheU depth, 5 sheU height categories (15-25, 30-40, 45-55, 60-70, and 75-85 mm) for indi vidual adult ray comingled trials. Time Period (min) Intercept (13o) 13./p SO I32/P 15-25 133/P 3Q-40 134/P 45-55 13s/P 60-70 136/P 75-85 H L 7.5 4.608 -0.206 - 5.004 - 1.653 - 1.1 12 - 0.205 0.723 0. 107 0.001 <0.05 <0.05 0.076 0.139 0.705 0.088 15 5. 73 1 -0.279 - 5.162 - 1.085 0.690 1.069 1.990 0.048 <0.05 <0.05 <0.05 0.287 0.41 5 0. 103 <0.05 30 9.302 -0.340 - 7.744 - 3.644 - 1.292 1.208 0.217 0.230 <0.05 <0.05 <0.05 <0.05 0.121 0.016 0.574 45 9. 144 -0.325 - 7.546 - 3.531 -0.458 0.594 0.492 0.202 <0.05 <0.05 <0.05 0.001 0.614 0.288 0.225 60 I 0.653 - 0.384 - 8. 106 - 2.329 - 0.337 0.573 0.830 0. 104 <0.05 <0.05 <0.05 0. 163 0.807 0.551 0.119 120 I 0.234 -0.328 - 7.982 14.704 16.1 74 0.295 0.51 8 0.981 <0.05 <0.05 <0.05 0.998 0.998 0. 75 1 0.320 240 16.638 -0.554 - 11.569 11.780 14.131 18. 154 1.927 0.997 <0.05 <0.05 0.002 0.999 0.998 0.997 0.031 P va lues of each variable are shown below parameter estimates, and significance level for the Hosmer and Lerneshow tests (HL) of model fi t are displayed. SO. shell depth. Comparatil•e Trials Between Bi1•ah•e Species mm) had the highest probability of being eaten by adult rays whereas predation probability on smaller and larger oysters was No significant difference in predation was observed between C. significantly lower. Adult rays appeared unable to detect virginica and C. ariakensis in both SH groups (SH 45-55 mm, SH shorter ( 15- 25 mm SH) oysters, and ingestion of those sizes 75-85 mm; P> 0.222, 0.186, respectively) tested. Predation success was a result of collateral feeding only on smaller oysters in close was highest (90-96% eaten) in 45- 55-mm oysters of both species. proximity to larger target oysters. The tallest oysters (>75 mm Predation success was significantly higher (P < 0.000 I) and the SH) were eaten in fewer numbers because they were too big (SH rays selected hard clams (M. mercenaria, = 0.736 ± 0.002, a and SO) to be easil y manipulated and required more handling electivity = 0.473 ± 0.007) over oysters (C. virginica, a = 0.263 ± time to consume than oysters of smaller SH and typically 0.002, electivity = -0.473 ± 0.007). Rays also selected so ft clams. shallower SD. Thus, midsize oysters (30- 70 mm SH) fit more M. arenaria at 5 min into a 15-min trial (a = 0.742 ± 0.003, easily between the rays' jaws, resulting in higher predation. electivity = 0.485 ± 0.0 13) over oysters (C. virginica, a = 0.257 ± Given longer time to forage , however, successful predation on 0.003, electivity = -0.485 ± 0.003) initially, then selection was larger oysters increased. However, predation rates of the largest more equal at the end of 15-min trial (M. arenaria. a = 0.570 ± 2 size classes remained lower than the 3 intermediate size classes 0.014, elcctivity = 0.141 ± 0.059: C. virginica, a = 0.429 ± 0.014, regardless of time allowed, further indicating that physical electivity = -0.141 ± 0.059). Although SH was greater for C. constraints, such as jaw gape, limited predation success. virginica in oyster- hard clam trials. mean SO was si mi lar for both In eomingled trials with YOY rays, the smallest oysters were species (mean SO of clams, 24.9 mm; mean SD of oysters, 22.9 mm). most susceptible to predati on. YOY rays attempted to feed on DISCUSSION the largest oysters offered (30-40 mm SH, 15- 19 mm SO), but were unsuccessful because of gape limitations. Observations of cownose rays feeding throughout this study The logistic regression model was used to determine the showed that bivalves were viewed as a general food source. and effect of SH and SO on predation. Although direct application initial selection of potential prey was not based on a prey size. of this model might oot reflect predation in a natural setting Cownose rays would indiscriminately suck shellfish toward with unlimited time. the model does support the generalization their mouth and, if the shellfish fit between the ray's jaws and that adult cownose rays do not primarily prey upon ve ry small adequate crushing force was applied, the shellfish was eaten. If or very large, deeper bivalves. the prey was too large to fit between the biting plates, it was At nearly all SDs, there was a direct relationship between discarded and escaped predation, at least initially. trial duration and mortality for C. virginica. Given more time, Shellfish mortality caused by cownose ray predation of partic• rays would continue to manipulate larger oysters that had been ular SH and SO supports the idea that cownose ray jaw attempted earlier in the trial by one or more rays without morphology has a quantitative gape limitation related to prey size. success. This aggregate crushing effect. combined with increases In general, adult cownose rays in this study were unable to consume in feeding time, contributed to the hi gher amount of predation. bivalves larger than 31- 32 mm SD regardless ofSH, and YOY rays Regardless of time, greatest predation success in comingled were not able to consume those larger than 15- 16 mm SO. trials were on oysters 30-70 mm SH and 14-20 mm SO. This Data suggest that rays select oysters of intermedia te SH or suggests the rays activel y selected oysters of this size range SD. During comingled trials. 3 SH groups (30-40, 45- 55. 60-70 because they are within ray gape limitations. The Manly-Chesson 194 fiSHER ET AL. 100 1 2000 A --- .... 90 A 0.9 ' \ 1800 \ c 80 0.8 -- M. me""'nana Prob8bl¢y Of I e 1600 0 P 1 1 ••• ••••• 2000 ｾ＠ .. - C. ｾｴｲｪｮｫ｡＠ ·. 0.9 8 0 .9 8 \ 1800 0.8 c 0.8 ••• 1600 c •• .2 •• ·B 0.1 0.7 •• 1400 co • z ｾ＠ '0 • ｾ＠ •• ｾ＠ 0.6 • 1200 '0 £0.6 • a. ｾｾ＠ Cll •• 0 0.5 0 0.5 • 1000 - - •• .3 • ?:- >- •• .:zt. ｾ＠ 0.4 ｾ＠ 0.4 • 800 •• :ll 2l Cll ' \ a. !:! 0.3 0.3 • 6. 600 a. -e Ｖｾ＠ ' .... 0.2 Q. 0.2 tP /),. 6. 400 0.1 0.1 200 YOY Jaw Gapo •• •• Adult Jaw Gapo Jaw Gape 0.0 ｾ｟Ｎ｟｟ｾ ｾｾ ＭＮ｟｟ＮＮＮ｟｟ＮＮＮｾ｟｟｟ＮｾｾｾＭＧＭＧＭｾＭＭＭＧＬＮＮＮＭ ＮｌＮＮＮＮＮＮｊ＠ 0 0 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 Shell Depth (mm) Shell Depth (mm) Figure S. (A) ｾ ｬ ･｡ｮ＠ predicted probability of young-of-year (\'OY) pre• Figure 7. (A) ) lean predicted probability, adult ｲ｡ｾ ﾷ＠ predation. and dation from the logistic regression model for C. l'irJ:inica as related to shell nontran:.i'ormed peak load of M. mtrCI'naria as related to shell depth. (B) height. (B) Mean predicted probability of YO\' predation from the logistic ｾ ｬ ･｡ ｮ＠ predicted probability of young-of-year (YOY) adult ray preda tion rrgrcssion model of C. l'irJ:inica as related to shell depth. Vertical lines and nontransformcd peak load of C. Ｑ ﾷｩｲｾＺ ｩｮｩ｣｡＠ as related to shell depth. -C. rrprcscnt maximum jaw gape rant:r for YO\' rays used in predation trials. r·irJ:ini<·a Probability of Predation (adult), . .. C. r•irgi11ica Probability of Prcd11tion (YOY). 6 C. ｬＧｩｲﾷｾＺｩｬｲｩ｣｡＠ Pcn k Load (N). index further supports the preference for the aforemen tioned oyster SHs. These preferences may be further explained by rorce requirements. The force required to crush bivalves (peak load) with jaw ga pe and bite force limits ma y work in concert to lower was posi ti vely correlated with SO and scales isometrically. The ontogenetically the susceptibility of bivalves to predation. rise in force needed to crush a bivalve at increa ed SO along Comparing results from comingled versus same-size trials. slightly higher rates of predation were observed in same-size trials except for the 75-85-mm-SH oysters. The difference in the rate of 4 predation may be the result of the greater time required to sort (passively or actively) through oysters of various sizes. including 0 M. morcenaria large oysters that cannot be successfully preyed upon at first - M. morconana y = 3.62181n(x) • 1.5744 attempt (75-85 mm). However. the differences in predation rates R' =0.719 between trial types may be explained by passive foraging. Adult '0 Cll rays were observed manipulating ;md preying upon shelllish as 0 :;2 they were encountered. regardless of the proximity of more Cll Q) susceptible prey. This passive foraging on oysters was also ＮＬ［ｾｮｩ｣｡＠ a. 6 C. ob ervcd in YOY rays that indiscriminately initiated prey manip• C. Ｂｾｮｫ｡＠ ulation on the first oyster encountered. regardless of oyster size. y =1. 883Sin(x) • 2.3054 Oyster predation rate in comingled trials declined for each R'=0.7441 SH category as time increased (Fig. 8A). Rays initially depleted more susceptible prey. resulting in fewer available prey as time 1 progressed. Fewer available prey. a larger proportion of prey 1 2 Shell Depth (mm) approaching or exceeding the ga pe or bite force limitations Figure 6. J>eak load of C. l'irJ:illica and M. merce11aria as rrlatcd to shell (i ncreasing handling time). and satiation resulted in decreasing depth plotted on logarithmic axes. rates of predation over time. COWNOSE RAY PREDATION 195 A Our data suggest cownose rays are ga pe limited and unable 1 to produce the force needed to crush larger oysters. Therefore, Shell Height oyster growers and those attempting to seed reefs with mature ｾ＠ 0.9 Groups(mm) Q) oysters (broodstock) should consider some measure of pro• • go -100 1/) ->- 0.8 tecti on for shellfish until they reach a shell depth of 22- 24 mm 0 • 75 -85 and/or breed shellfish able to withstand forces greater than 0) c: 0.7 1,400 N. We demonstrated that YOY rays can successfull y prey .c: ro 0.6 on seed oysters up to 40 mm SH. In most aquaculture settings, E Q) 0 45 -55 oyster seed is protected throughout grow-out by vari ous con• 0:: 0.5 tainment methods (bags. floats, racks). However, cultchless 0 30 -40 0 oysters are produced for restoration efforts where small oysters -c: 0.4 .o 0 15 -25 are used to seed constructed reefs. In this applica tion, thought t:: 0 should be given to habitat structure, with reefs providing refuge a. 0.3 e (crevices) for small oysters to settle a nd be less susceptible to ray a.. 0.2 predation. Cultched or spat-on-shell oysters have been proposed . and used in restoration efforts, and could limit susceptibility of • 0.1 oysters to cownose ray predation. Future work on cownose ray 0.0 and cultched oyster in teracti on is needed to evaluate any benefi ts. 0 7.5 15 30 45 60 120 240 The results also indicate that oyster restoration effo rts might Time (min) not benefit from introducing different oyster species. Our data 1.8 B indicate cownose ra ys prey on C. ariakensis no differentl y than • Single Size Trials on C. virginicct. Although the introduction of the fast-growing 1.6 a Comingled Trials C. ariakensis has been suggested as a possible solution, the 1.4 results of comparative predati on trials indicate tha t rays do not c 0 discriminate between C. ariakensis and C. virginica, and there• :p <11 1.2 fo re the introduction of C. ariakensis to the Chesapeake Bay to ¥ 1.0 restore oyster reefs or to rev itali ze the commercial industry may a.. not be an adequate solution. 0 0.8 Our data suggest cownose rays prefer the clams M . merce• 2 <11 naria and M . arenaria over the oyster C. virginica. Preference for 0:: 0.6 soft clams (M. arenaria) was expected as a result of their hi gh 0.4 SH-to-SD rati o and relati vely weak va lves. This species was historica ll y the dominant natural prey of cownose rays in the 0.2 Chesapeake Bay (Smith & Merriner I 985); however. natural 0 disaster (Tropical Storm Agnes in 1972), disease, and over• 30-40 45-55 60-70 75-85 exploitation have led to the coll apse of soft-shell clam stock in Shell Height (mm) the estuary. Given the signi fica nt influence of SD on predation Figure 8. (A) Proportion of remaining oysters after each time period (C. in the comingled trials of C. virginica and the similarity of SD in 1•irginica) for corninglcd trial periods. (B) Mean number of oysters (C. oyster- dam trials, higher predati on on hard clams was un• 1•irginica) consumed per minute per ad ult ray for same-size (rate) trial.s expected. A ray must crush the clam at or near its deepest point compared with comingled trials. Corni ngled trial times of IS min were u.sed (SD), whereas in oysters, rays ca n '·nibble" the fla ttened, for compa rison. posterior edge of the shell s. The ability to handle oysters and apply fo rce along the edges of oysters negates some of the effects Managemelll Implications of the gape limitation. Further investiga tion into the amount of nutrition gained by clams over oysters or shell composition and Cownose ray predation on commercial bi valves has been structure could explain the preference. a concern for shellfish industries fo r more than40 y (Merriner & Smith 1979). These concerns are acute in the seeding and grow• ACK OWLEDGME 'TS out operations that are part of restoration effo rts. Considering that cownose rays have among the slowest reproductive rates of This research was funded by the NOAA Chesapeake Bay Office any vertebrate, usuall y producing a single pup each yea r (Smith and Virginia Sea Grant College Program. A special thanks to Dr. & Merriner 1986), eradication programs are not a viable Jim Kirkley and David Rudders for discussions on statistical solution. However, there may be o ther means to protect analysis, and to Janet Krenn for reviewing drafts of this pa per. commercial and restored shell fish beds. This is VIMS Contribution #3146. LITERATURE CITED Bishop. M. J. & C. H. Peterson. 2006. When r-selection may not predict Blaylock. R. A. 1993. Distribution and abundance of cownosc rays, introduced-species proliferation: predation of a nonnative oyster. Rhi110ptera bo11asus (Rhinopteridae), in lower Chesapeake Bay, £col. Appl. 16:7 18- 730. Virgi ni a. Estuaries 16:255- 263. 196 fiSHER IH t\L. ｃｨ｣ｾｾｯｮＮ＠ J. 1978. Measuring prefercncc in selective predation. Ecolo[(y J>ctC"-On. C. II.. F. J. Fodric. II. C. Summerson & S. J>. Powers. 2001. 59:211 215. ｓｩｴ｣ＭｾＱＩｃｃｩｬｩ｣＠ and density-dependent extinction of prey by schooling Colli ns. A. ll .. M. R. Heupel. R. E. I !cuter & P. J. Mott;t. 2007. liard pre) rays: generation of a population ｾＱＱＱ ｫ＠ in top-quality habitat for bay SJ>I.'Cialist or opportunistic gcncrali>b'! An examination of the diet of ｾ｣｡ｬｬｯｰｳＮ＠ Oecnlogia 129:349 356. the cownose ray. Rhinoptem hmiii.I"IIS. Mar. Freshll". Res. 58:135 144. Sasko. D. 2000. The prey c