www.scencemag.org/cg/content/full/315/5820/1846/dc1 Supportng Onlne Materal for Cascadng Effects of the Loss of Apex Predatory Sharks from a Coastal Ocean Ransom A. Myers, Jula K. Baum,* Travs D. Shepherd, Sean P. Powers, Charles H. Peterson* *To whom correspondence should be addressed. E-mal: baum@mscs.dal.ca (J.K.B.); cpeters@emal.unc.edu (C.H.P.) Ths PDF fle ncludes: Materals and Methods SOM Text Fgs. S1 to S3 Tables S1 to S5 References Publshed 30 March 2007, Scence 315, 1846 (2007) DOI: 10.1126/scence.1138657
Supportng Onlne Materal Cascadng Effects of the Loss of Apex Predatory Sharks from a Coastal Ocean Ransom A. Myers, Jula K. Baum, Travs D. Shepherd, Sean P. Powers, Charles H. Peterson Materals and Methods Speces Great sharks Large shark speces n the northwest Atlantc were consdered for ncluson n ths category based on ther sze and occurrence of elasmobranchs n ther det. Eleven speces met these crtera (Table S1). These sharks are among the largest (notable exclusons beng baskng and whale sharks, whch feed at much lower trophc levels), reachng maxmum lengths rangng from ~2.0m n blacktp and sandbar sharks up to 5-6m n great hammerhead and great whte sharks (S1-S3). Bull, blacktp, sandbar, and scalloped hammerhead reach sexual maturty below or close to 2m, but all others mature at a greater length (S1-S3). These large fshes are all tertary consumers (trophc level!4) wth catholc dets. Fve speces (bull, great hammerhead, tger, sand tger, and great whte sharks) are true apex predators, whle the remanng sx speces feed at and near the top of the food web. Smaller elasmobranchs form a key component of the det of large sharks (S1, S2, S4, S5), and conversely, sharks are the most common predators of other elasmobranchs (S6, S7). Among the large sharks, however, there s consderable varaton n the proporton of elasmobranchs consumed n ther det. Bull, great hammerhead, sand tger, and great whte sharks are each consdered to be mportant predators on other elasmobranchs, wth about 30-40% of ther det comprsed of these fshes (S5). For the other speces, the proporton of elasmobranchs n ther det has ranged n dfferent studes between approxmately 1 and 15% (see Table S1 references; S5). We compled data on elasmobranch consumpton by each of the large sharks, wth partcular consderaton of the speces ncluded n the elasmobranch mesopredator category (see below). At the speces level there s evdence that large sharks are predators of seven of the elasmobranch 1
mesopredator speces, lttle and clearnose skates, bullnose eagle ray, spotted eagle ray, cownose ray, bonnethead and Atlantc sharpnose sharks (Table S1). Notably, two speces, blacktp and sandbar sharks, are known to eat the cownose ray, and four of the other great sharks speces are known to consume speces wthn the cownose ray genus (Rhnoptera). In general, there s a dearth of speces-specfc prey nformaton for most sharks (most nformaton n the lterature s reported at hgher taxonomc levels (usually famly or genus)), and a lack of bologcal nformaton n general for several of the lttle known mesopredator speces. For example, there s no nformaton n the lterature (that we are aware of) on predators of fve of the mesopredators (rosette skate, spny and smooth butterfly ray, lesser devl ray, and chan catshark). However, there s evdence that large sharks consume speces n ten out of the 11 mesopredator genera, and on the famly (Scylorhndae) of the only other genus, that of the chan catshark (Table S1). We assessed trends n relatve abundance for ten of these 11 great shark speces, based on the crteron that for each speces there had to be at least one source of longlne data (the most effectve gear type for samplng these speces) avalable for analyss. Evdence from a sharktargeted longlne research survey n Chesapeake Bay (not avalable for analyss) suggests that the eleventh speces, sand tger shark (Carcharas taurus), has experenced declnes smlar to those of the other great sharks (S8). The sand tger shark has been consdered for proposed lstng under the U.S. Endangered Speces Act, and s currently lsted as a Speces of Specal Concern and a prohbted speces (to land) by the U.S. Natonal Marne Fsheres Servce (S9). Mesopredatory elasmobranchs We ntally consdered all small shark, and all skate and ray speces from wthn the geographc range of our study that are preyed upon by larger shark speces, and for whch there were suffcent data avalable to assess ther trends n relatve abundance. However, because both low ntrnsc rates of populaton ncrease and heavy fshng pressure lmt the potental responses of elasmobranch mesopredator populatons followng a loss of ther predators, we restrcted our analyses to a subset of these speces based on the followng crtera: () of the mesopredators subject to fshng pressure (whether as drect targets or as bycatch), we ncluded only those speces wth female age at maturty <4 years and thus relatvely hgh potental rates of populaton ncrease. These crtera were relevant to all mesopredatory shark speces, and excluded speces lke spny dogfsh (Squalus acanthas) and smooth dogfsh (Mustelus cans), whch mature late, 2
and Atlantc angel shark (Squatna dumerl), whch s presumed to mature late based on ts sze and the age at maturty of other speces n ts genus. For skates, ths meant that three northern speces (thorny (Amblyraja radata), wnter (Leucoraja ocellata), and barndoor (Dpturus laevs) skate) were excluded because of both late age at maturty and hgh rate of explotaton; () for those mesopredators not subject to hgh explotaton rates we ncluded speces wth female age at maturty up to 7 years; () fnally, we excluded stngrays from our analyss because they are subject to hgh rates of post-dscard mortalty (S10), presumably as a consequence of ther thn body type relatve to thcker boded skates and ther mstreatment by fshermen fearng ther venomous spnes. Fourteen elasmobranch mesopredator speces met our crtera (Table S1). They comprse 11 dfferent genera from 7 famles, and range from farly well known speces (e.g. Atlantc sharpnose, blacknose, and fnetooth sharks, cownose rays) to very poorly known speces (bullnose eagle ray, lesser devl ray, chan catshark, smooth and spny butterfly ray). Bvalves We examned all northwest Atlantc bvalve speces that are components of the cownose ray det (S11, S12), for whch suffcent data were avalable. These ncluded only commercally fshed speces: the eastern oyster (Crassostrea vrgnca), hard clam (Mercenara mercenara), softshell clam (Mya arenara), and bay scallop (Argopecten rradans). We suspect that cownose ray predaton also now nfluences surf clam (Spsula soldssma) populatons of the New Jersey and Delmarva Pennsula coasts but were unable to locate suffcent nformaton to nclude ths nteracton n our analyss. Data sources Research survey data We analyzed 17 scentfc research surveys (Tables S2, S3) from U.S. coastal waters (Fg. S1) that recorded elasmobranch speces, began pror to 1990, and were conducted usng a consstent methodology over at least 12 years. Two of the surveys used longlnes and were carred out specfcally to sample sharks, the UNC survey (detaled below) and the SC survey (see S13, S14 for detals). Ffteen other surveys used ether bottom trawls or senes, and were desgned to sample a varety of fnfsh and nvertebrate speces. In total, all 17 surveys caught elasmobranch 3
mesopredator speces; 12 caught large sharks (Table S3). Survey start years ranged from 1959 to 1989, wth a medan start year of 1976 (Table S3). The long-term UNC-IMS research survey of sharks has been conducted each year snce 1972 by Dr. F.J. Schwartz of the Unversty of North Carolna at Chapel Hll Insttute of Marne Scences n Onslow Bay off the central coast of North Carolna near Cape Lookout. The UNC-IMS data set that we analyzed comprsed a total of 760 longlne sets from 1972-2003. Survey methods (S15) have remaned dentcal over ths 32-year perod. Unanchored longlnes have been set bweekly from about Aprl 15 to November 1 each year usng a desgn employng the same gear at two fxed statons. Pror to settng out the longlne, fresh fsh were collected by trawlng and used as whole fsh to bat the hooks. Two successve sets of bated hooks consttuted the samplng for every date (except less than one quarter of days when bad weather prevented establshment of the second set). Samplng was carred out durng the day between the hours of 0800 and 1500hr. The East-West set was establshed frst, near shore and approxmately parallel to the beach of Shackleford Banks n 13 m depth, runnng up to 4.8 km eastward from 34º 38.029' N, 76º 37.835' W. Sets employed between 27 and 483 hooks (mean = 151), wth one plastc foam nternatonal orange buoy of 1.3-m dameter attached for every 10 hooks and hooks spaced every 4.5 m. Case-hardened steel 9/0 Mustad tuna hooks were attached to 1.8-m drop lnes of No. 2 (95 kg) porch swng chan, whch were snapped onto the 7.6-cm braded nylon man lne. Soak tme after settng was 1 hr. Durng the "45 mn requred to pull n the lne, the speces, sex, and fork length of each hooked shark was recorded and all lve sharks were tagged and returned to the sea. After 35-40 mnutes travel tme, the North-South set was establshed further offshore n Onslow Bay n 22 m depth, runnng southwards from 34º 33.071' N, 76º 37.422' W. The procedures followed were dentcal to those of the East-West set. Trawlng for addtonal bat was occasonally requred between sets. Fsheres data For the large sharks, we also examned logbook (1986-2000) and observer (1992-2005) data from the U.S. pelagc longlne fshery. Fsheres-dependent data are the only type that covers a substantal proporton of the geographc range of these shark populatons, and pelagc longlne gear s partcularly sutable for catchng these speces. The U.S. pelagc longlne fleet fshes offshore of the Grand Banks (50ºN), along the U.S. eastern coast and wthn the Gulf of Mexco, 4
and as far south as the equator. The broad geographc coverage of these data therefore serves to complement the long temporal coverage of the research surveys. These data also nclude two speces, shortfn mako and the great whte shark, that consume elasmobranchs but were almost never caught n the research surveys (for great whte n=1 n the UNC survey and n=0 for all other surveys; n=0 for shortfn mako n all surveys). In the fsheres data analyses, speces wthn the same genera that could not relably be dstngushed from one another were grouped. Ths ncludes a groupng of hammerhead sharks, genus Sphyrna (scalloped, smooth, and great hammerheads), mako sharks, genus Isurus (prmarly shortfn mako), and large coastal sharks of the genus Carcharhnus (blacktp, bull, dusky, sandbar, bgnose, nght, slky, spnner; the frst four of whch consume elasmobranchs). Commercal landngs data Data on U.S. landngs were obtaned from the NMFS commercal landngs database, whle those for Canada came from the Unted Natons Food and Agrculture Organzaton (FAO). Data on eastern U.S. landngs were avalable by state from Mane to Texas and are an aggregate of both fshery and aquaculture producton. For the purposes of testng for effects of the ncreasng east coast populaton of cownose rays on bvalves, only shellfsh landngs from states where these cownose rays would be expected to nteract wth bvalves were ncluded n the meta-analyss (New Jersey, south to eastern Florda). Landngs from outsde the regon nhabted by ths cownose ray populaton (e.g. Canada, Rhode Island, and Texas) are presented n Fgure S2 for comparson. For one speces, the hard clam, aquaculture makes up a large porton of producton (up to 82%) snce the md-1980s (based on comparson of FAO aquaculture producton data and NMFS commercal landngs data for the U.S. east coast). Wthout a reasonable method of parttonng these two producton sources by state, we were requred to obtan fshery landngs data for hard clams n U.S. states from other sources. Data were avalable only for Vrgna and Rhode Island, from the Vrgna publc fshery hard clam producton database (1973-1999) and the Rhode Island shellfsh management plan (S16) respectvely. Trends n relatve abundance models Research survey data Trends n relatve abundance of each speces, from each fshery-ndependent survey, were analyzed usng generalzed lnear models (GLMs) wth a negatve bnomal error structure and a 5
log lnk. The negatve bnomal s an approprate probablty dstrbuton for dscrete, overdspersed data lke the survey data, whch contan a large number of zero (no catch) observatons and are more varable than expected n a Posson dstrbuton. Accordng to the negatve bnomal dstrbuton, the probablty of catchng C ndvduals of a gven speces n survey tow has mean!, ( 1 % * & C ) #! k, ",, for C = 0, 1, 2,, k ' k $ p( C ; k; ) + ( 1 % ( 1 % C )& # *( C ) *& ' k $ 1) (1 ) k, ) ' k # $ where * s the gamma functon and k s the negatve bnomal dsperson parameter. Usng a log lnk n the GLMs means that the log of the mean catch s assumed to be a lnear combnaton of predctor varables. The expected mean catch of a gven speces s then,!, " + -! log! offset" log ) x where x- s a vector of explanatory covarates for observaton,! s a vector of unknown coeffcents for the explanatory varables and offset s the offset term. All analyses were conducted usng SAS v9.1 (S17). The breadth of ancllary data whch could be used as explanatory covarates n estmatng trends n relatve abundance vared among surveys. For all surveys and speces, we employed the general strategy of usng the followng covarates n the generalzed lnear models as the vector of explanatory varables ( x- ): year, the second order polynomal of depth, the second order polynomal of bottom temperature and q (the seasonal cycle) (Table S4). The seasonal cycle, q, was characterzed by a seres of sne and cosne terms, wth perods, j, of ½ and 1 year as, q ( d ) + where d 2. j+ 4 ( 25jd % ( 25jd % 1 27 j cos& # ) 6 j sn& #/ 1 3 ' 365.25 $ ' 365. 25 $ 0 s the sequental day of the year that observaton occurred n, and 7 and 6 are estmated parameters. Modelng the seasonal cycle, q, also allowed us to generate common estmates for surveys conducted durng multple, dstnct tme-perods each year (NMFS offshore, NMFS nshore and SEAMAP surveys). The NMFS surveys and the SEAMAP surveys covered relatvely large lattudnal ranges (Fg. 6
S1) and there was some nter-annual varaton n the tmng of these surveys. For speces that do not undertake seasonal mgratons out of each survey area, ths was not a concern. However, changes n the tmng of the survey could have sgnfcant effects on estmates for those speces that do mgrate out of the area surveyed. To account for ths effect, we used the addtonal term of lattude when modelng the NMFS and SEAMAP survey data. Furthermore, for these surveys we allowed the seasonal cycle, q, to vary by lattude by ncludng the nteracton term between lattude and q. There were exceptons to our general strategy of parameter selecton and error structure used for the generalzed lnear models (Table S4). Data from the CTDEP and GSO trawl surveys were avalable only n the form of mean annual estmates so only year could be ncluded n the model. For these two surveys, the negatve bnomal error structure was not approprate snce t s used for dscrete data only, and nstead we used a gamma error structure and a log lnk for the generalzed lnear models. The probablty of a mean catch C of a gven speces n year was assumed to follow a gamma dstrbuton wth the mean!, 9 1 ( C % ( % 9 C & # & 8 9 p ( C + # ; 9 ;, ) exp, for 0 < C < #, *( 9 ) C ', $ ', $ where * s the gamma functon and " s the gamma dstrbuton scale parameter. The expected mean catch s then, log!, " + # x - where x- s the year n whch observaton occurred, and # s the coeffcent for year. When surveys followed a fxed staton desgn (DNREC trawl survey, UNC longlne survey, GSO trawl survey), we ncluded a unque staton dentfer as a model factor. In some cases, covarates other than those n our standard lst were avalable, ncludng rver basn for the VIMS sene survey, and the second order polynomal of salnty for the Maryland sene survey (MDNR). Fsheres data Methods of the logbook data analyss are reported n S18 and ts Supplementary Materal, Trends n relatve abundance for large shark speces were estmated from the observer data usng generalzed lnear mxed models wth a negatve bnomal error structure and log lnk n the 7
GLIMMIX procedure of SAS v.9.1 (S17, S19 ). In these models, to account for nonndependence of longlne fshng sets made by the same vessel and on the same trp, we specfed vessel as a G-sde random effect and fshng trp as an R-sde random effect wth an autoregressve one (AR1) correlaton structure. Addtonal detals of the logbook and observer data analyses are found n Table S4. Calculatng change n abundance Changes n abundance reported n the man text for ndvdual speces were computed by applyng the estmated rate of change over the tme perod from the speces frst appearance n the data set untl the end of the data set. Thus, for example, n the UNC data set the rate of change for most speces was calculated for the entre tme perod (1972-2003), but for sandbar sharks was calculated only from 1976 to 2003. Meta-analyss of trends n relatve abundance We summarzed trend estmates from multple surveys for each speces usng meta-analytc technques. The nstantaneous rate of change of speces s from survey s estmated as :ˆ. The unts of the nstantaneous rate of change wll have the same unts for all surveys (estmated usng generalzed lnear models wth a log lnk as descrbed n the prevous secton) and can be thought of as slopes on a log scale. The estmate of ˆ ~ (, 2 s, N : s, v s, : ). : s, wll be approxmately normal, that s, we assume We plotted the log-lkelhood profle to check the normalty assumpton. As sample szes were large for most surveys, ths assumpton was reasonable n most cases. s, For the random effects meta-analyss of the nstantaneous rate of change for a gven speces, we assumed that the true rates of change came from a normal dstrbuton,.e. 2 : s, ~ N( : s, 6 s ). Maxmum lkelhood estmaton of the random effects meta-analyss was carred out n SAS usng Proc Mxed (S17). Testng for heterogenety s equvalent to testng H 0 : $ 2 = 0 aganst H 1 : $ 2 > 0. The standard lkelhood rato must be modfed n ths case because the null hypothess s on the boundary of the parameter space (.e. the varance cannot be less than zero), whch n ths case means the p-value of the nave lkelhood rato test must be dvded by two (S20). 8
It s common n meta-analyss to use a fxed effect meta-analyss f there s not statstcally sgnfcant heterogenety among studes; however, the power of ths test s low for small numbers of studes and a fxed effect meta-analyss wll underestmate the standard errors of the estmate f heterogenety s present. We thus used the random-effects meta-analyss n all cases. In the results secton of the supplement, we examne cases where there may be dfferences among surveys and over dfferent tme perods. Trends n large shark length Trends n large shark length were analyzed usng the UNC longlne survey data wth generalzed lnear models wth a gamma error structure and a log lnk, after removng any bologcally mplausble length values (e.g, fve lengths for dusky sharks were smaller than the mnmum sze of neonates). Month of catch was ncluded as a covarate for each speces length model. Results are shown n Fg. S3 and dscussed n the man text. Cownose ray absolute abundance & nvertebrate consumpton estmates To estmate the number of cownose rays that currently pass through Chesapeake Bay durng ther fall mgraton, we combned our survey-based meta-analytc estmated rate of ncrease wth Blaylock s populaton estmate of 9.3 mllon (s.e. = 1.8 mllon) for Chesapeake Bay, whch was based on aeral surveys between 1986 and 1989 (S11). Then, to estmate the total food demand for benthc bvalve mollusks by cownose rays n the Chesapeake Bay area annually, we combned our cownose ray abundance estmate wth ts annual occupancy tme n Chesapeake Bay of 100 days (S11) and ts ndvdual consumpton rate. Blaylock (S11) estmated an ndvdual daly consumpton rate of 210g for the cownose ray. Schwartz (S21) estmated that ndvdual cownose rays consume up to 1.5L of bvalve mollusks per day. Ths equates to 250g/day based on the converson of 1.5L to 1kg bvalve mollusk, wth about 25% meat, whch s comparable to Blaylock s estmate. To be conservatve, we used the lower estmate of 210g, yeldng a total estmate of 840,000 metrc tons (wet flesh) per annum. Quantfyng cownose ray mpacts on bay scallops Trends n bay scallop commercal landngs data To llustrate changes over tme n the magntude of North Carolna bay scallop landngs (Fg. 1), 9
we ftted these data wth a generalzed addtve model (GAM) n R v2.2.1 (S22), usng a gamma error structure and log lnk, wth a loess curve of span 1, degree 2. The gamma s an approprate probablty dstrbuton for these data, whch are contnuous, postve, and have non-constant varance. Bay scallop densty To evaluate the populaton-level mpacts of fall mgratng rays across the full geographc range of tradtonal scallopng grounds n North Carolna, scallop denstes were measured b-weekly at sx seagrass beds located wthn Core (Cedar Island, Yellow Shoal ), Back (Oscar Shoal, Straghts) and Bogue (Marker 34 and 40) Sounds from August through October n 2002, 2003, and 2004. Bay scallop densty was measured wthn each seagrass bed n early August and agan n md-october because ths perod brackets the fall mgraton of the cownose rays. At each ste 5 replcate 1-m 2 quadrats were haphazardly thrown near the edge and at the center of each seagrass bed (10 quadrats total per bed). All bay scallops wthn each quadrat were counted, measured and returned to ther orgnal locaton. Physcal parameters (% cover of seagrass, salnty, temperature, sedment type) were also recorded durng samplng. In the Back Sound porton of our study area, the North Carolna Dvson of Marne Fsheres (NCDMF) allowed a lmted hand harvest of scallops concdent wth the expected tmng of fall mmgraton by cownose rays. Sx harvest days were permtted between md August and early September wth a daly harvest rate of 10 bushels/fsherman. Few fshermen partcpated and fshng mpacts were trval compared to estmated losses from ray predaton. Nevertheless, to prevent our densty estmates from beng confounded by ths addtonal treatment and to quantfy the relatve mpact of ths harvest, the NCDMF establshed and we conducted our samplng n two 25 m 2 shellfsh sanctuary areas wthn all seagrass beds. A substantally longer data base exsts for one of the stes, Oscar Shoals. Although some small dfferences exst among years n methodologes, adult bay scallop densty was measured n late July or early-md August and agan n September or October n 1992, 1993, 1994, 1996, 1998, 1999 and 2000 (detaled methods are reported n S23). For all years, bay scallop survval was calculated by dvdng denstes measured on the last samplng date by the densty measured on the ntal samplng date. Expermental assessment of cownose ray predaton To determne to what extent any decrease n scallop densty s attrbutable to ray predaton, we 10
establshed four 2-m 2 exclosures at the center and four at the edge of the 6 seagrass beds where NCDMF shellfsh sanctuary areas were establshed. The exclosures, short (50 cm) PVC poles arranged as a stockade, exclude cownose rays whle allowng other predators (crabs and whelks) nto the matrx of poles (S23, S24). The number of scallops survvng wthn the stockade s compared to areas of free ray access (controls). The experment was performed durng the fall of 2002, 2003, and 2004. The stockades were constructed n stu and bay scallops allowed to move freely nto and out of the exclosure. Exclosures were erected n md August of each year and bay scallop densty measured wthn the exclosure and n the controls at that tme and agan n late September. A smlar set of experments had been performed at the Oscar Shoal ste n 1996 and 1998 (S23). As n the later experments, naturally occurrng bay scallops were allowed free access to the exclosure, but n addton ten marked and tethered bay scallops were placed wthn and outsde the stockades. Mortalty wthn the stockade should be substantally less than n the control areas f large moble consumers are the chef predator on bay scallops durng ths tme perod. Bay scallop mortalty wthn the stockades was calculated as 1 mnus survval, computed by dvdng denstes measured on the last samplng date wthn the stockade by the densty measured on the ntal samplng date pror to constructon of the enclosure. The dfference between scallop survval nsde and survval outsde the stockades greatly underestmates the proporton of natural mortalty attrbutable to large moble consumers (of whch cownose rays were the only ones observed) because bay scallops ntally nsde stockades emgrate throughout ths perod of tme and thereby become susceptble to consumpton by rays. Earler tetherng experments (S23) ndcate that emgraton explans a large majorty of the apparent mortalty of bay scallops nsde the stockades. Supportng Text: Results and Dscusson Trends n relatve abundance Overall, the trend estmates from the 17 research surveys and the 2 fsheres data sets gve broadly consstent estmates of populaton declnes of great sharks and populaton ncreases n elasmobranch mesopredators (Table S5). Earler trend estmates for great sharks from logbook reports (S18) have been crtczed for usng fsheres-dependent data reported by fshers and for relyng on only one data source (S25, but see S26). Here, we have analyzed the complementary scentfc observer data set from the same fshery, and shown smlar results for each speces (group) except tger sharks (dscussed below). We have also analyzed all avalable, long-term 11
scentfc research surveys (n=12) for great sharks. Importantly, the longlne research surveys desgned to catch sharks (UNC, SC) suggest declnes for every great shark speces, and show large, statstcally sgnfcant declnes for each great shark that was caught n suffcent numbers to estmate trends. In a few cases, there are qualtatve dfferences (.e. ncreasng vs. decreasng trends) amongst elasmobranch speces trend estmates from dfferent data sets. Such dfferences could arse for several reasons, ncludng dfferences n the years (early vs. recent) or areas (e.g. north vs. south, nshore vs. offshore) sampled (Fg. S1, Table S5). Here we dscuss the detals of each of these cases. Only 2 of the 30 trend estmates for great sharks are statstcally sgnfcant ncreases. The frst s for juvenle hammerhead sharks caught n the recent (1989-2005) SEAMAP survey. We also note that juvenle blacktp sharks n the SEAMAP survey and juvenle sandbar shark n the NMFS surveys show nonsgnfcant changes (Table S5). These data suggest that declnes n the juvenles may have ceased for these speces, and that juvenle survval could have ncreased because of declnes of larger sharks. The second ncreasng trend occurs n the observer data set for the tger shark: ts abundance has apparently begun to ncrease n the past couple of years. Ths fshery catches manly juvenle tger sharks, and of all the shark speces caught, tger shark has the hghest survval rate (S27). Thus, ths change from a declnng to an ncreasng trend also may represent an ncrease n juvenle survval assocated wth a declne n predaton by large sharks. For elasmobranch mesopredators, most dfferences n trend estmates wthn speces appear to be caused by small sample sze and/or hgh samplng varablty at the edge of the speces range. For example, there were 7 surveys for cownose ray, and 6 of these have estmates of nstantaneous rates of ncrease between 0.044 and 0.17 per year (Table S5). Cownose rays n the NMFS-Offshore survey data have a non-sgnfcant trend estmate of -0.26 (95%CI: -0.54 0.01). NMFS-Offshore surveys caught cownose rays n only 3 years because of the survey locaton at the edge of the geographc regon consstently nhabted by ths speces, and we thus do not regard ths estmate as beng ndcatve of temporal change n the populaton. A smlar problem probably exsts for the trend estmate derved from the NMFS-Inshore data for Atlantc 12
sharpnose shark. Ths s the only non-sgnfcant estmate for ths speces, and these surveys occur at the extreme northern lmt of ths speces range. For some of the smaller skates, dfferences n trend estmates may represent real dfferences among populatons. For example, lttle skate shows statstcally sgnfcant ncreases n three surveys, but does not appear to be ncreasng n Long Island Sound (CTDEP survey, Table S5). We suspect that ths may represent a real pattern because there s an ntense fshery for lobster bat n ths regon that catches lttle skate (S28). The only two exceptons to the general pattern of ncreasng abundance amongst the mesopredators are the blacknose shark, whch has been decreasng accordng to the UNC survey, and the spotted eagle ray, whch has been decreasng accordng to the recent SEAMAP survey (Table S5). Lke the sharpnose shark, the blacknose shark s caught n several recreatonal and commercal fsheres (S29); however, ts age at maturty s greater (3.8 years on average) than the Atlantc sharpnose shark (2.3 years) (S30-S33). Thus, t may be more susceptble to fshng than the Atlantc sharpnose, whch s clearly ncreasng. Insuffcent nformaton about the fshng pressure on spotted eagle ray (whch matures between the age of 4 and 6 years (S34)) lmts our nterpretaton of abundance trends for ths speces. Inferences from lfe hstory theory about the cownose ray rate of ncrease Females n the U.S. Atlantc cownose ray populaton reach sexual maturty between age 7 and 8 (S12) and have one pup per year (S35). Lke most other elasmobranch speces, there are no drect estmates of natural mortalty for the cownose ray. However, usng the meta-analytc mean ncrease, 0.087 (95% CI: 0.034 0.14), as the rate of populaton ncrease (r), we can solve the Euler-Lotka equaton to estmate the mortalty that the cownose ray populaton must be subject to. The mortalty rate was calculated as 0.076 (95% CI: 0.021-0.127), whch s much lower than a speces of fsh wth ths populaton growth rate would be expected to have (compared to smlarly szed speces) under natural condtons. For example, usng the observed growth rate and the asymptotc length (and an assumed temperature of 20ºC), Pauly s equaton (S36) gves a natural mortalty estmate of 0.26, whle Hoeng s equaton (S37), based on longevty, gves an estmate of 0.33. Thus, a natural mortalty rate as low as that calculated n the Euler-Lotka 13
equaton mples that the cownose ray populaton has experenced substantally reduced natural mortalty. In addton, because the mortalty rate of the cownose ray populaton also must nclude some bycatch mortalty, the natural mortalty must actually be somewhat less then the estmate of 0.076. We conclude that gven the lfe-hstory of cownose rays and the observed rate of ncrease that the populaton must now have extraordnarly low natural mortalty rate compared to what t would experence under normal levels of predaton. We nfer that the loss of naturally more ntense predaton by the great sharks explans why the cownose ray now devates so greatly n mortalty rate from what s expected on the bass of lfe hstory relatonshps (S36, S37). Comparson of shrmp fshng effort between the southeast U.S. & northern Gulf of Mexco Whereas the cownose ray populaton on the east coast of the U.S. has ncreased substantally, n the Gulf of Mexco, where shrmp trawl fshng effort s enormously greater, ncdental catches have apparently reduced that cownose ray populaton (S38), and oyster landngs have ncreased (Fg S2a). We compared shrmp fshng effort between the southeast U.S. (North Carolna to eastern Florda) and Gulf of Mexco wthn equvalent tme perods for whch data were avalable (1991-1993). Along the southeast U.S., the annual average number of shrmp fshng trps durng that perod was 55,878 (S39). Ths ncludes ocean waters, sound waters and some areas possbly unsutable for cownose rays, such as rvers. We were not able to exclude unsutable areas due to the resoluton of the data. For the same tme perod, the Gulf of Mexco shrmp trawl fleet fshed an annual average of 306,910 24-hour shrmpng days or 7,365,829 fshng hours (S40). Typcally, a southeast U.S. shrmp fshng trp n the early 1990s was approxmately 5 hours long (S39). Thus the southeast U.S. shrmp fshng effort equaled 279,390 fshng hours (55,878 trps x 5 hours per trp), or 3.8% of the Gulf of Mexco effort. As cownose rays mgrate northward n the sprng and southward n the late summer and fall along the coast of the southeast U.S., they wll be exposed to local shrmp fsheres over restrcted tme perods. In North Carolna for example, the majorty of the shrmp fshng occurs n July and August (S41) whle cownose ray abundance does not peak n the regon untl September (S23), mssng the tme of most ntense effort. In contrast, the northern Gulf of Mexco shrmp fshery mantans a very hgh ntensty from May nto December (NMFS 14
commercal landngs data), whch ncludes the tme at whch cownose rays nhabt the area (S38). Thus, we conclude that the dfference n shrmp fshng effort, and spatal and temporal overlap between cownose rays and fshng effort, could explan the dfferent trends n abundance for these two cownose ray populatons. 15
Fgure S1. Map of the U.S. Atlantc coast showng the locaton of each of the 17 research surveys, wth 200m, 500m, and 1000m sobaths (dotted lnes) gven for reference.
Fgure S2a. Changes n landngs (metrc tons) by ndvdual states of the U.S.A. plus east coast of Canada for oysters. Regons enclosed by red lnes are those n whch the east coast populaton of cownose rays s expected to nteract wth bvalves.
Fgure S2b. Changes n landngs (metrc tons) by ndvdual states of the U.S.A. for bay scallops. Regons enclosed by red lnes are those n whch the east coast populaton of cownose rays s expected to nteract wth bvalves.
Fgure S2c. Changes n landngs (metrc tons) by ndvdual states of the U.S.A. plus east coast of Canada for hard clams. The regon enclosed by red lnes s that n whch the east coast populaton of cownose rays s expected to nteract wth bvalves.
Fgure S2d. Changes n landngs (metrc tons) by ndvdual states of the U.S.A. plus east coast of Canada for soft-shell clams. Regons enclosed by red lnes are those n whch the east coast populaton of cownose rays s expected to nteract wth bvalves.
a b Fgure S3. Change n length of great sharks between 1972 and 2003 from the Unversty of North Carolna shark-targeted longlne research survey (UNC): a) nstantaneous rates of change (± 95% confdence ntervals); b) overall trend (sold lne) and ndvdual year estmates ( ). Speces wth length samples nmore than three years were modeled n a) and b); only raw data are shown for great and smooth hammerheads.
Table S1. Taxa of elasmobranchs (sharks, skates, rays) consumed by the apex (or nearly apex) shark speces ncluded n the large shark group. Prey are lsted by speces level and at the genus and/or famly level because of the paucty of speces-specfc det data avalable n the lterature. Numbers correspond to references n the Table S1 reference lst. Large Sharks Elasmobranch Mesopredators Famly Genus Common name, Scentfc name Bull shark, Carcharhnus leucas Blacktp shark, C. lmbatus Dusky shark, C. obscurus Sandbar shark, C. plumbeus Tger shark, Galeocerdo cuver Great whte shark, Carcharodon carcharas Shortfn mako, Isurus oxyrnchus Great hammerhead, Sphyrna mokarran Scalloped hammerhead, S. lewn Smooth hammerhead, S. zygaena Sand tger, Carcharas taurus 8 8 8,11,14 12,2 Rajdae (Skates) 8 7 7,19 8 9 8 7,14,18 Lttle skate, Leucoraja ernacea Rosette skate, L. garman Clearnose skate, Raja eglantera Gymnurdae (Butterfly rays) Gymnura speces Smooth butterfly ray, Gymnura altavela Spny butterfly ray, G. mcrura Mylobatdae (Mantas and eagle rays) 12,2 12 8 8,10,1 8,11 8 6 9 10 11 6 9 5,8 8 11 12 8,17 3,7 4,7,19 2,6 13 7,14 Aetobatus speces (eagle rays) 8 7,19 2 13 7,14 Spotted eagle ray, A. narnar 8 7 2 13 14 Mobula speces (devl rays) 5,8 11 Lesser devl ray, M. hypostoma Mylobats speces 7 7,14,18 Bullnose eagle ray, M. fremnvll 14 Rhnopterdae (Cownose rays) 15 1 11 8,12 7,19 8 Rhnoptera speces (cownose rays) 15 1 11 8,12 19 8 Cownose ray, R. bonasus 1 12 Scylorhndae (catsharks) 5 10 8,11 7 4 6 9 18 Chan catshark, Scylorhnus retfer Sphyrndae (hammerhead sharks) 5,8 1,10 11 8 8 3,7 4,7 Sphyrna speces 1 8 3,7 4,7 Bonnethead shark, S. tburo 1 8 Carcharhndae (requem sharks) 5,8 8,10 8,11 12 8 3,7 4,7,19 6 8,9 7,14,18 Carcharhnus speces 5,8 8,10 8,11 12 8 3,7 4,7 6 8 14,18 Blacknose shark, C. acronotus Fnetooth shark, C. sodon Rhzopronodon speces 5,8 1,8,10 11,21 8,20 8 3,7 4,7 6 8,9 18 Atlantc sharpnose shark, R. terraenovae 1 22
Table S1 References: 1. J. Castro, Bull. Mar. Sc. 59, 508 (1996). 2. D. D. Chapman, S. H. Gruber, Bull. Mar. Sc. 70, 947 (2002). 3. G. Clff, S. F. J. Dudley, B. Davs, S. Afr. J. Mar. Sc. 8, 131 (1989). 4. G. Clff, S. F. J. Dudley, B. Davs, S. Afr. J. Mar. Sc. 9, 115 (1990). 5. G. Clff, S. F. J. Dudley S. Afr. J. Mar. Sc. 10, 253 (1991). 6. G. Clff, S. Afr. J. Mar. Sc. 15, 105 (1995). 7. L. J. V. Compagno, Sharks of the world: an annotated and llustrated catalogue of shark speces known to date. Vol. 2 (Food and Agrculture Organzaton of the Unted Natons, Rome, 2001). 8. L. J. V. Compagno, Sharks of the world: an annotated and llustrated catalogue of shark speces known to date. Vol 4, Parts 1 & 2. (Food and Agrculture Organzaton of the Unted Natons, Rome, 1984). 9. P. de Bruyn, S. F. J. Dudley, G. Clff, M. J. Smale, Afr. J. Mar. Sc. 27, 517 (2005). 10. S. F. J. Dudley, G. Clff, S. Afr. J. Mar. Sc. 13, 237 (1993). 11. S. F. J. Dudley, G. Clff, M. P. Zungu, M. J. Smale, Afr. J. Mar. Sc. 27, 107 (2005). 12. J. K. Ells, thess, The College of Wllam and Mary n Vrgna (2003). 13. F. Galván-Magaña, H. J. Nenhus, Calf. Fsh Game. 75, 74 (1989). 14. J. Gelslechter, J. A. Musck, S. Nchols, Envron. Bol. Fshes. 54, 205 (1999). 15. R. E. Hueter, C. A. Manre, Bycatch and catch-release mortalty of small sharks n the Gulf coast nursery grounds of Tampa Bay and Charlotte Harbor (Mote Marne Techncal Report No. 368, Fnal report to NOAA/NMFS, MARFIN Project NA17FF0378-01, 1994). 16. E. R. Hoffmayer, G. R. Parsons, 2, 271 (2003). 17. C. A. Smpfendorfer, A. B. Goodred, R. B. McAuley, Envron. Bol.Fshes. 61, 37 (2001). 18. M. J. Smale, Afr. J. Mar. Sc. 27, 331 (2005). 19. C.E. Stllwell, N.E. Kohler, Can. J. Fsh. Aquat. Sc. 39, 407 (1982). 20. C.E. Stllwell, N.E. Kohler, Fsh. Bull. 91,138 (1993). 21. R. P. van der Elst, Envron. Bol. Fshes 4, 349 (1979). 23
Table S2. Data sources. Data type Data Acronym Source Reference or web access Survey CTDEP Connectcut Department of Envronmental Protecton, Fsheres Dvson DNREC Delaware Department of Natural Resources and Envronmental Control, Dvson of Fsh & Wldlfe http://dep.state.ct.us/burnatr/fshng/fdhome.htm http://www.fw.delaware.gov/. GSO Unversty of Rhode Island, Graduate School of Oceanography http://www.gso.ur.edu/ MDNR Maryland Department of Natural Resources, Fsheres Servce http://www.dnr.state.md.us/fsheres/ NCDMF NMFS-Off & NMFS-In North Carolna Department of Envronment and Natural Resources, Dvson of Marne Fsheres Natonal Oceanc & Atmospherc Admnstraton (NOAA), Natonal Marne Fsheres Servce (NMFS), Northeast Fshery Scence Center SC South Carolna Department of Natural Resources S13, S14 http://www.ncfsheres.net/ http://www.nefsc.noaa.gov/ SEAMAP Southeast Area Montorng and Assessment Program, South Atlantc http://www.dnr.sc.gov/marne/mrr/seamap/seamap.html UNC Unversty of North Carolna - Insttute of Marne Scences, Longlne shark montorng survey http://www.marne.unc.edu/research VIMS Vrgna Insttute of Marne Scence http://www.fsheres.vms.edu/trawlsene/sbman.htm Fsheres Logbook NOAA, NMFS, Southeast Fshery Scence Center http://www.sefsc.noaa.gov/fls.jsp Observer NOAA, NMFS, Southeast Fshery Scence Center http://www.sefsc.noaa.gov/pop.jsp Landngs Landngs NOAA, NMFS, Offce of Scence & Technology http://www.st.nmfs.gov/st1/commercal Landngs UN Food and Agrculture Organzaton, Fsheres Department, Fshery Informaton, Data and Statstcs Unt http://www.fao.org/f/statst/statst.asp 24
Table S3. Survey, fsheres and landngs data set descrptons, ncludng area, gear type, season and years sampled, and total sample sze. Speces sampled n each data set: great shark speces (L), elasmobranch mesopredator speces (M), bvalve speces (S). Data type Acronym Area Gear Season Years Samples Speces Survey CTDEP Long Island Sound Trawl Fall/Sprng 1984 2004 - M DNREC Delaware Bay Trawl Year round 1966 2004 1874 L,M GSO Narragansett Bay, Rhode Island Trawl Year round 1959 2002 - M MDNR Chesapeake Bay Sene Summer 1960 2005 8022 M NCDMF Pamlco Sound, North Carolna Trawl Summer/Fall 1987 2004 1889 M NMFS-Off Northeast U.S. Offshore Trawl Sprng 1968 2005 10185 L,M NMFS-Off Northeast U.S. Offshore Trawl Fall 1963 2005 8829 L,M NMFS-Off Northeast U.S. Offshore Trawl Summer 1963-1995 1758 L,M NMFS-In Northeast U.S. Inshore Trawl Sprng 1976 2005 2084 L,M NMFS-In Northeast U.S. Inshore Trawl Fall 1974 2005 2228 L,M NMFS-In Northeast U.S. Inshore Trawl Summer 1977 1981 351 L,M SC Coastal South Carolna Bottom longlne Year round 1983-84, 1993-95 131 L,M SEAMAP Coastal Southeast U.S. Trawl Sprng 1989 2005 1441 L,M SEAMAP Coastal Southeast U.S. Trawl Fall 1989 2005 1389 L,M SEAMAP Coastal Southeast U.S. Trawl Summer 1989 2005 1393 L,M UNC Coastal North Carolna Longlne Aprl - November 1972 2003 760 L,M VIMS Chesapeake Bay Sene Summer 1968 2003 3166 M Fsheres Logbook Northwest Atlantc Pelagc longlne Year round 1986 2000 214234 L Observer Northwest Atlantc Pelagc longlne Year round 1992 2005 6967 L Landngs NMFS Landngs Coastal Eastern U.S. Varous Year round 1950 2003 - S FAO Landngs Atlantc Canada Varous Year round 1950 2003 - S 25
Table S4. Summary of generalzed lnear models used to estmate trends n abundance for large sharks and elasmobranch mesopredators. All data were modeled usng generalzed lnear models, except for the observer data, whch was modeled usng generalzed lnear mxed models. All models ncluded year as a covarate; q represents a seasonal term composed of a seres of sne and cosne terms wth perods of one year and one half year. Data source acronyms as n Table S3. Data Source Covarates Error dstrbuton Lnk Offset CTDEP no covarates avalable Gamma Log None DNREC depth, depth 2, staton, q Negatve bnomal Log Swept area GSO no covarates avalable Gamma Log None NCDMF no covarates avalable Gamma n.a. n.a. NMFS-Off depth, depth 2, temperature, temperature 2, lattude, q, lattude*q nteracton Negatve bnomal Log Swept area NMFS-In depth, depth 2, temperature, temperature 2, lattude, q, lattude*q nteracton Negatve bnomal Log Swept area SEAMAP depth, depth 2, temperature, temperature 2, lattude, q, lattude*q nteracton Negatve bnomal Log Swept area SC depth, depth 2, q, tme of set, soak tme, Negatve bnomal Log Number of hooks MDNR month, temperature, temperature 2, salnty, salnty 2 Negatve bnomal Log None UNC staton, q Negatve bnomal Log Number of hooks VIMS rver basn Negatve bnomal Log None Logbook area, season, temperature, use of lght stcks, area*season, area*lght stcks Truncated negatve Log Number of hooks Observer area, q, depth, depth 2, temperature, tme of set, number of lght stcks, hook depth, hook type, soaktme, target speces, bat, area*q nteracton, fshng trp, vessel Negatve bnomal Log Number of hooks 26
Table S5. Model results for each speces of great shark and elasmobranch mesopredator, from each of the research survey and fsheres data sources used n the meta-analyss shown n Fgure 2, ncludng the frst and last year of capture n the data set, the number of years caught, the total number of the speces caught, the model estmate (± 95% confdence ntervals (CI)) of the nstantaneous rate of change, for all years of data (All) and for only those years onwards from the baselne of 1970 (1970-). Statstcal sgnfcance levels for model estmates are * = <0.05; **=<0.01; ***=<0.001; ****=<0.0001, otherwse non-sgnfcant. Speces Data Frst Last n n Years Instantaneous rate of change Common name source year year years caught n model Estmate upper CI lower CI Great Sharks Blacktp UNC 1972 2003 32 905 All -0.084**** -0.065-0.103 SEAMAP 1990 2005 12 29 All 0.040 0.133-0.053 Bull UNC 1973 1995 10 23 All -0.181**** -0.093-0.270 Dusky UNC 1972 2003 29 1036 All -0.149**** -0.120-0.179 NMFS-Off 1967 1999 11 38 All -0.075* -0.005-0.144 1972 1999 8 26 1970 - -0.068 0.004-0.140 NMFS-In 1974 2005 17 100 All -0.092** -0.035-0.149 SEAMAP 1990 2004 6 24 All -0.199* -0.032-0.365 Great hammerhead UNC 1975 1997 4 5 All -0.080 0.037-0.197 Sandbar DNREC 1966 2004 26 242 All -0.048**** -0.035-0.062 1970 2004 22 159 1970 - -0.041**** -0.023-0.060 UNC 1976 2000 23 310 All -0.077**** -0.039-0.114 NMFS-Off 1967 2002 22 73 All 0.014 0.057-0.028 1973 2002 20 68 1970-0.009 0.054-0.036 NMFS-In 1974 2005 27 107 All 0.019 0.048-0.010 SC 1983 1995 5 196 All -0.281**** -0.225-0.337 SEAMAP 1990 2005 13 71 All -0.029 0.070-0.128 Scalloped hammerhead UNC 1972 2003 29 495 All -0.127**** -0.104-0.149 NMFS- In 1980 1995 3 4 All -0.110 0.065-0.285 SEAMAP 1989 2005 17 126 All 0.094** 0.155 0.033 Smooth hammerhead UNC 1973 1989 4 5 All -0.172* -0.010-0.334 Hammerhead speces SC 1983 1994 3 11 All -0.110 0.089-0.308 Logbook 1986 2000 15 60,402 All -0.158**** -0.143-0.172 Observer 1992 2005 14 1,292 All -0.110**** -0.062-0.157 Large coastal speces Logbook 1986 2000 15 80,480 All -0.118**** -0.103-0.133 Observer 1992 2005 14 8,186 All -0.084**** -0.048-0.121 Mako speces Logbook 1986 2000 15 65,795 All -0.037**** -0.025-0.050 Observer 1992 2005 14 3,433 All -0.032* -0.001-0.063 Tger UNC 1973 2002 18 39 All -0.117**** -0.064-0.169 SC 1983 1995 5 142 All -0.027 0.029-0.083 Logbook 1986 2000 15 16,030 All -0.076**** -0.061-0.091 Observer 1992 2005 14 1,190 All 0.037* 0.071 0.002 Great whte Logbook 1986 2000 15 6,087 All -0.117**** -0.074-0.146 Elasmobranch Mesopredators Atlantc sharpnose UNC 1973 2003 31 2239 All 0.084**** 0.098 0.071 NMFS-Off 1974 2003 15 39 All 0.084** 0.138 0.031 1974 2003 15 39 1970-0.084** 0.138 0.030 NMFS-In 1974 2005 26 331 All -0.025 0.002-0.053 SEAMAP 1989 2005 17 13187 All 0.065**** 0.079 0.051 SC 1983 1995 5 135 All 0.103*** 0.159 0.047 Blacknose SEAMAP 1989 2005 17 156 All 0.043 0.091-0.004 UNC 1972 2003 32 1304 All -0.090**** -0.073-0.107 Bonnethead shark SEAMAP 1989 2005 17 4925 All 0.028** 0.045 0.010 27
Bullnose eagle ray DNREC 1966 2004 28 3701 All 0.008* 0.015 0.001 1970 2004 24 3153 1970-0.010* 0.019 0.001 NMFS-Off 1967 2005 23 297 All 0.056* 0.104 0.009 1973 2005 21 279 1970-0.053* 0.102 0.003 NMFS-In 1974 2005 32 2230 All -0.003 0.014-0.020 SEAMAP 1989 2005 17 5300 All 0.041*** 0.065 0.018 Chan catshark NMFS-Off 1963 2005 43 778 All 0.052**** 0.065 0.038 1970 2005 36 715 1970-0.070**** 0.087 0.053 Clearnose skate CTDEP 1984 2004 21 - All 0.199**** 0.281 0.118 DNREC 1966 2004 28 5778 All -0.049**** -0.042-0.057 1970 2004 24 3359 1970- -0.008 0.001-0.018 NCDMF 1988 2004 7 9 All 0.053 0.206-0.099 NMFS-Off 1967 2005 39 1053 All 0.034**** 0.047 0.022 1970 2005 36 1029 1970-0.029**** 0.042 0.016 NMFS-In 1974 2005 32 2678 All 0.047**** 0.057 0.036 SEAMAP 1989 2005 17 6991 All 0.014 0.034-0.007 Cownose ray DNREC 1979 2003 14 76 All 0.117**** 0.168 0.065 1979 2003 14 76 1970-0.111**** 0.168 0.054 MDNR 1976 2003 12 26 All 0.063** 0.102 0.024 NCDMF 1987 2004 17 230 All 0.175**** 0.219 0.132 NMFS-Off 1972 1976 3 23 All -0.265 0.011-0.541 1972 1976 3 23 1970 - -0.432 0.063-0.928 NMFS-In 1974 2005 27 544 All 0.044* 0.081 0.006 SEAMAP 1989 2005 17 4817 All 0.059** 0.105 0.014 VIMS 1992 2003 7 11 All 0.104* 0.201 0.008 1992 2003 7 11 1970-0.101* 0.200 0.002 Fnetooth shark UNC 1977 1997 14 93 All 0.039 0.114-0.037 SEAMAP 1990 2005 7 23 All 0.092 0.261-0.078 Lesser devl ray SEAMAP 1990 2005 15 347 All 0.105** 0.173 0.037 Lttle skate CTDEP 1984 2004 21 - All -0.008 0.010-0.025 DNREC 1966 2004 25 2499 All 0.048**** 0.058 0.039 1970 2004 21 2378 1970-0.082**** 0.096 0.068 NMFS-Off 1963 2005 43 161330 All 0.018**** 0.022 0.015 1970 2005 36 151031 1970-0.015**** 0.019 0.011 NMFS-In 1974 2005 32 142760 All 0.076**** 0.084 0.067 GSO 1959 2002 44 - All 0.054**** 0.064 0.044 1970 2002 33-1970- 0.056**** 0.071 0.041 Rosette skate NMFS- Off 1963 2005 43 1014 All 0.039**** 0.052 0.025 1970 2005 36 939 1970-0.037**** 0.053 0.022 Smooth butterfly ray DNREC 1967 1999 4 11 All -0.108* -0.025-0.191 1971 1999 2 3 1970 - n.a. n.a. n.a. NCDMF 1989 2004 6 44 All 0.344**** 0.474 0.215 SEAMAP 1989 2005 17 5247 All 0.131**** 0.148 0.114 Spny butterfly ray DNREC 1966 2002 22 55 All -0.029* -0.004-0.053 1970 2002 19 45 1970 - -0.030 0.002-0.062 NMFS-Off 1973 2005 19 52 All 0.041 0.085-0.003 1973 2005 19 52 1970-0.030 0.073-0.014 NMFS- In 1974 2005 32 589 All 0.016 0.033-0.001 SEAMAP 1989 2005 17 317 All 0.102**** 0.152 0.053 Spotted eagle ray SEAMAP 1990 2005 16 159 All -0.070** -0.017-0.122 28
References S1. L. J. V. Compagno, Sharks of the world: an annotated and llustrated catalogue of shark speces known to date. Vol 4, Parts 1 & 2. (Food and Agrculture Organzaton of the Unted Natons, Rome, 1984). S2. L. J. V. Compagno, Sharks of the world: an annotated and llustrated catalogue of shark speces known to date. Vol. 2 (Food and Agrculture Organzaton of the Unted Natons, Rome, 2001). S3. E. Cortés, Conserv. Bol. 16, 1048-1062 (2002). S4. E. Clark, K. von Schmdt, Bull. Mar. Sc. 15, 13 (1965). S5. E. Cortés, ICES J. Mar. Sc. 56, 707 (1999). S6. S. Sprnger, n Sharks, skates and rays, P. W. Glbert, R. W. Mathewson, R. P. Rall eds. (John Hopkns Press, Baltmore, MD, 1967). S7. M. R. Hethaus, n Bology of sharks and ther relatves, J. C. Carrer, J. A. Musck, M. R. Hethaus, eds. (CRC Press, Boca Raton, 2004), pp. 487-521. S8. D. H. Ha, thess, The College of Wllam and Mary n Vrgna (2006). S9. NOAA Fsheres, Offce of Protected Resources, Speces of concern and canddate speces (www.nmfs.noaa.gov/pr/speces/concern/#lst). S10. I. C. Stobutzk, M. J. Mller, D. S. Heales, D. T. Brewer, Fsh. Bull. 100, 800 (2002). S11. R. A. Blaylock, Estuares, 16, 255 (1993). S12. J. W. Smth, J. V. Merrner, Estuares, 10, 153 (1987). S13. R. A. Low, G. F. Ulrch, Survey of the shark resource n shelf waters off South Carolna (South Carolna Marne Resources Center Tech. Rep. No. 61, 1984). S14. G. F. Ulrch, Fshery ndependent montorng of large coastal sharks n South Carolna (1993-1995), fnal report, (U.S. NOAA and Interjursdctonal Fsheres Act NA47FI0347-01, 1996). S15. F.J. Schwartz, Northeast Gulf Sc. 7, 29 (1984). S16. Management Plan for the Shellfsh Fshery Sector, (State of Rhode Island & Provdence plantatons, Department of Envronmental Management, Dvson of Fsh and Wldlfe, Marne Fsheres, 2006). S17. SAS/STAT 9.1 User s Gude. (SAS Insttute Inc., Cary, NC, 2004). S18. J. K. Baum et al., Scence 299, 389 (2003). S19. SAS Insttute Inc., The Glmmx Procedure, Nov. 2005. (http://support.sas/com/rnd/app/papers/glmmx.pdf) S20. H. C. van Houwelngen, L. R. Arends, T. Stjnen, Stat. Med. 21, 589 (2002). 29