the use of scales to estimate the prey size of nearctic river otters (lontra canadensis) and other...

BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors,nonprofit publishers, academic institutions, research libraries, and research funders in the common goal ofmaximizing access to critical research.

The Use of Scales to Estimate the Prey Size ofNearctic River Otters (Lontra canadensis) and OtherPiscivoresAuthor(s): Cory R. Stearns, Dorothy M. Fecske, and Thomas L.SerfassSource: The American Midland Naturalist, 166(1):163-176. 2011.Published By: University of Notre DameDOI: http://dx.doi.org/10.1674/0003-0031-166.1.163URL: http://www.bioone.org/doi/full/10.1674/0003-0031-166.1.163

BioOne (www.bioone.org) is a nonprofit, online aggregation of coreresearch in the biological, ecological, and environmental sciences. BioOneprovides a sustainable online platform for over 170 journals and bookspublished by nonprofit societies, associations, museums, institutions, andpresses.

Your use of this PDF, the BioOne Web site, and all posted and associatedcontent indicates your acceptance of BioOne’s Terms of Use, available atwww.bioone.org/page/terms_of_use.

Usage of BioOne content is strictly limited to personal, educational, andnon-commercial use. Commercial inquiries or rights and permissionsrequests should be directed to the individual publisher as copyright holder.

http://dx.doi.org/10.1674/0003-0031-166.1.163

http://www.bioone.org/doi/full/10.1674/0003-0031-166.1.163

http://www.bioone.org

http://www.bioone.org/page/terms_of_use

The Use of Scales to Estimate the Prey Size of Nearctic RiverOtters (Lontra canadensis) and Other Piscivores

CORY R. STEARNS, DOROTHY M. FECSKE AND THOMAS L. SERFASS1

Department of Biology and Natural Resources, Frostburg State University, Frostburg, Maryland 21532

ABSTRACT.—Estimating the size of fish prey requires the use of relationships between thesize of non-digestible fish remains (recovered as prey in scats or digestive tracts) and fishlength. The applicability of scales for estimating the size of fish prey eaten by river otters(Lontra canadensis) or other piscivores was evaluated by conducting a linear regression analysisof scale size and fish length for 22 species and six multi-species groups from the Red River ofthe North tributaries of eastern North Dakota. Analyses included six scale measurements, andseparate models were constructed for lateral line and non-lateral line scales. Positiverelationships existed between scale size and fish length in most (42 of 44) single speciesmodels, with lateral line scales (r2 ranging 0.317–0.994) typically producing betterrelationships than non-lateral line scales (r2 ranging 0.136–0.959). Body-scale relationshipsalso existed when including multiple species in a model (r2 ranging 0.471–0.984 and 0.305–0.953 for lateral line and non-lateral line scales, respectively). Scales, and particularly lateralline scales, are highly useful in estimating the size of fish prey, but there are limitationsincluding: the lack of scales on some fish species, potential degradation in passage throughthe digestive system and the absence of lateral line scales from some prey remains.

INTRODUCTION

Numerous studies have assessed the types of fish most frequently preyed on by river otters(Lontra canadensis) (e.g., Greer, 1955; Melquist and Hornocker, 1983; Serfass et al., 1990).However, none of these investigations have quantitatively examined the size of fish includedin the river otter’s diet. Information about the size of fish consumed by river otters andother piscivores is fundamental for developing insights about their feeding behavior andtheir ecological role in aquatic systems.

Relationships between the size of non-digestible fish remains (recovered from scats ordigestive tracts) and fish length are required to estimate the size of fish consumed by riverotters or other piscivores. These positive relationships result from concurrent andproportional growth of the structure in relation to the growth of the fish in length (Lagler,1956; Daniels, 1996). For instance, previous studies of the European otter (Lutra lutra) haveused relationships between fish length and the dimensions of vertebrae to estimate preylength (e.g., Wise, 1980; Adrian and Delibes, 1987; Carss et al., 1990; Kemenes and Nechay,1990). Scales also have been used to estimate the prey size of the European otter (i.e.,Kozena et al., 1992) but in general have received little evaluation or application for thispurpose. Other structures that have been suggested for use in estimating the size of fish preyinclude: cleithra, jaw bones (i.e., dentary, premaxillary, maxilla), pectoral spines, pharyngealteeth, opercula and otoliths (Hamilton, 1961; Hansel et al., 1988; Prenda and Granado-Lorencio, 1992; Dellinger and Trillmich, 1999; Granadeiro and Silva, 2000; Noordhuis,2002; Copp and Kovac, 2003; Hajkova et al., 2003; Ross et al., 2005).

The use of scales instead of vertebrae or other structures for estimating the size of fishprey has several advantages. One advantage of scales is that they can be collected without

1 Corresponding author present address: Department of Biology and Natural Resources, FrostburgState University, Frostburg, Maryland 21532; Telephone: (301)687-4171; e-mail: [email protected]

Am. Midl. Nat. (2011) 166:163–176

163

dissection, and thus there is no need to kill fish to obtain samples. In contrast, fish must besacrificed to obtain vertebrae, otoliths and other bony structures. In cases where bonystructures for size estimation can be obtained from previously sampled fish (e.g., those incollections), scales retain the advantage of being quicker and easier to obtain (i.e., nodissection is required). Also, using hard structures that occur in the head or anterior part ofthe body (the origin of many of the aforementioned structures) could bias outcomes againstlarger fish because the anterior portions of larger fish may not be consumed (Erlinge, 1968;Rowe-Rowe, 1977). Difficulty in identification and the potential for breakage or otherdegradation in passage through the digestive system are additional disadvantages of usingvertebrae or other bones in size estimation (Carss and Nelson, 1998). Generally, even afterpassing through the digestive system scales are easily identified to the family level throughthe use of keys (e.g., Daniels, 1996) and reference collections. When an ecosystem containsrelatively few species within a family, identification can occasionally be done to the specieslevel.

Although vertebrae are the most frequently used structure for estimating the size of fishprey, their use is further complicated (in addition to the previously mentioned challengesassociated with the use of hard structures) by variation in size among regions of the vertebralcolumn (Wise, 1980). Consequently, predictive models for estimating fish length fromvertebrae typically have been developed using vertebrae from the same vertebral region(Wise, 1980). Therefore, their use for estimating prey size involves identifying vertebraefrom the region from which the model was constructed, which is complicated and timeconsuming. Scales also vary by shape and size over the body of an individual fish, which hasresulted in criticism of their use for size estimation (Phillips, 1948; Joeris, 1956; Scarnecchia,1979; Wise, 1980; Daniels, 1996; Miranda and Escala, 2007; Roberts et al., 2007). However,lateral line scales are easily distinguished from other scales by a pore or line on the scale andcan, therefore, be identified among prey remains (Daniels, 1996; Roberts et al., 2007).Although lateral line scales also vary in size over the length of a fish, constructing modelsusing only lateral line scales reduces the area from which scales are taken for buildingmodels. Therefore, the amount of variation in predictive models could be reduced,resulting in better models for prey size estimation.

Our study was designed to assess the relationship between fish length and variousdimensions of lateral line and non-lateral line scales. We were primarily interested indetermining if variation in regression models would be reduced when using lateral linescales, which would demonstrate their utility for use in estimating the size of fish prey.

STUDY AREA

We sampled fish from the Red River of the North (hereafter the Red River) drainage ofeastern North Dakota. The Red River forms at the convergence of the Bois de Sioux River,and the Ottertail River at Wahpeton, North Dakota and Brackenridge, Minnesota (Fig. 1).The river flows north forming the boundary between North Dakota and Minnesota fornearly 640 km before entering Manitoba, Canada (Koel and Peterka, 1998). The landscapeof the Red River drainage has low relief, and mostly occurs within the former lake bed ofLake Agassiz (Eddy et al., 1972; Stoner et al., 1993). The Red River has 10 major tributaries inNorth Dakota which are all similar in appearance, typically having low gradients, frequentmeanders and high turbidity (Copes and Tubb, 1966; Stoner et al., 1993). The majority ofsamples were obtained from the Forest and Turtle rivers, but samples also were obtainedfrom other rivers throughout the Red River drainage (Fig. 1).

164 THE AMERICAN MIDLAND NATURALIST 166(1)

METHODS

Scale samples were taken from fish collected 12 Jun.–16 Aug., and 28 Sept.–2 Nov. 2007using a backpack electrofishing unit, a seine, cloverleaf traps, minnow traps and fyke nets.We measured and recorded the total length (from snout to the tip of the tail) to the nearestmm of each fish captured, and collected a randomly selected lateral line scale and a non-lateral line scale. The randomization process involved partitioning the fish into 10 verticalregions (dorsum to ventrum) from the posterior end of the gill cover to the base of thecaudal fin (Fig. 2), with a lateral line scale being randomly selected from one of the 10regions. To select a non-lateral line scale, we visually imposed a 10 3 10 grid on the side of

FIG. 1.—Map of the Red River of the North drainage of eastern North Dakota. The stars indicate riversfrom which samples were taken but are not indicative of sampling locations

2011 STEARNS ET AL.: USE OF SCALES TO ESTIMATE PREY SIZE OF NEARCTIC RIVER OTTERS 165

the fish’s body and randomly selected a cell from which the scale was removed (Fig. 2). Atleast one individual was sampled from each species that was captured at a particularsampling location, with the intent of obtaining at least 30 total scale samples (regardless ofcollection site) for each species. Thereafter, sampling was mostly restricted to species withlow sample size (,30) or lengths that were not well represented from previous sampling.

For each scale sample, six scale measurements (length, height, diagonal, anterior radius,posterior radius and antero- or posterolateral radius; Fig. 3) were taken using digital calipers

FIG. 2.—Randomization of scale selection for the assessment of body-scale relationships of 22 fishspecies and six multi-species groups of eastern North Dakota. In the selection of lateral line scales, fishwere visually divided into 10 equal sized regions (the numbers), and a lateral line scale was taken from arandomly selected region. For non-lateral line scales a 10 3 10 grid was visually imposed on the fish, anda non-lateral line scale was taken from a randomly selected cell

FIG. 3.—Scale measurements used in assessing body-scale relationships of 22 fish species and six multi-species groups of fish of the Red River of the North tributaries of eastern North Dakota. A 5 anteriorradius, B 5 anterolateral radius, D 5 diagonal, H 5 height, L 5 length, P 5 posterior radius


accurate to 0.01 mm. An anterolateral radius was measured for most species, but aposterolateral radius was measured for cyprinids other than carp (Cyprinus carpio). For eachspecies and scale type, we independently performed linear regression analyses to assess therelationships between scale dimensions and fish length using Minitab Version 14 (MinitabInc., State College, Pennsylvania, USA). Multiple regression analyses were initiallyconducted; but the variables were highly correlated, so further analysis was restricted tosingle variable regressions.

The analyses included 22 species for which $10 lateral line and random scale samples(most contained 20–40) were obtained. The best models for each species were determinedby maximizing the coefficient of determination (r2). For species that could not reliably bedistinguished by the morphology of their scales, we developed and assessed multi-speciesmodels. Cyprinids other than carp and dace [longnose dace (Rhinichthys cataractae) andwestern blacknose dace (R. obtusus)], which could be differentiated from other cyprinids,were divided into the ‘‘small cyprinid group’’ and the ‘‘large cyprinid group’’ [speciesattaining total lengths .10 cm; in our study represented by common shiner (Luxiluscornutus), creek chub (Semotilus atromaculatus) and hornyhead chub (Nocomis biguttatus)].Doing so allowed us to identify the best scale measurement for estimating fish length forboth groups and determine a scale size that could be used to differentiate between them foruse in subsequent models to estimate the prey size of river otters (for an example of theapplication see Stearns and Serfass, 2011). The small cyprinid group contained species thatattain maximum length ,10 cm, and small individuals (,10 cm) from the species thatattain larger sizes. For the small cyprinid models, 15 samples were included from eachspecies (when available). When more than 15 samples were available for a particular species,we conducted stratified (by length) random subsampling to ensure samples fromthroughout the length range of a species were included in the group model. Thestandardized sample size for the number of lateral line and non-lateral line scales from eachspecies included in the other multi-species models were: 18 (lateral line) and 16 (non-lateralline) for Centrarchidae [bluegill (Lepomis macrochirus) and black crappie (Pomoxisnigromaculatus)], 18 (lateral line) and 17 (non-lateral line) for dace, 24 for both the lateraland non-lateral line for large cyprinids, 36 (lateral line) and 43 (non-lateral line) for darters[johnny darter (Etheostoma nigrum) and blackside darter (Percina maculata)], and 10 (lateralline) and 12 (non-lateral line) for the model with white bass (Morone chrysops) andfreshwater drum (Aplodinotus grunniens). Largescale stonerollers (Campostoma oligolepis) werenot included in the group analyses because of low sample size.

RESULTS

For most species, positive relationships existed between scale size and fish total lengthusing lateral line and non-lateral line scales (Table 1; refer to Appendix for selectedexamples of the relationships and Stearns (2008) for the relationships of all speciesstudied). Generally, relationships had high r2 values, with 28 of 44 of the best (the lateralline and non-lateral line scale model with the highest r2 for each species) models having r2

$ 0.70 (Table 1). Using lateral line scales, the best model for all 22 species was significant,with r2 ranging from 0.317 for sand shiners (Notropis stramineus) to 0.994 for bluegill(Table 1). The best non-lateral line models were significant for 20 of 22 species, with r2

ranging from 0.136 to 0.954 for sand shiners and freshwater drum, respectively (Table 1).Lateral line models usually were better estimators of fish total length than non-lateral linemodels, with 20 of 22 species having a higher r2 using lateral line scales (Table 1).Additionally, significant positive relationships existed for all 12 multi-species models, with r2


TA

BL

E1.

—T

he

coef

fici

ent

of

det

erm

inat

ion

(r2),

regr

essi

on

coef

fici

ents

and

sign

ific

ance

of

the

rela

tio

nsh

ip(P

)o

fb

od

y-sc

ale

regr

essi

on

sfo

r22

fish

spec

ies

of

the

Red

Riv

ero

fth

eN

ort

htr

ibu

tari

eso

fea

ster

nN

ort

hD

ako

tau

sin

gsa

mp

les

coll

ecte

dJu

n.–

No

v.20

07,

by

scal

ety

pe.

Var

iab

les

incl

ud

e:A

5

ante

rio

rra

diu

s,B

5an

tero

late

ral

rad

ius,

C5

po

ster

ola

tera

lra

diu

s,D

5d

iago

nal

,H

5h

eigh

t,L

5le

ngt

h,

P5

po

ster

ior

rad

ius.

*In

dic

ates

anin

sign

ific

ant

rela

tio

nsh

ip(P

.0.

05)

Spec

ies

To

tal

len

gth

ran

ge(c

m)

Lat

eral

lin

eN

on

-Lat

eral

lin

e

nV

ar.

Slo

pe

Inte

rcep

tr2

Pn

Var

.Sl

op

eIn

terc

ept

r2P

Cat

ost

om

idae

Sho

rth

ead

Red

ho

rse

(Mox

osto

ma

mac

role

pido

tum

)7.

1–24

.616

P4.

909

2.64

50.

850

,0.

001

15P

4.37

94.

549

0.88

2,

0.00

1W

hit

eSu

cker

(Cat

osto

mu

sco

mm

erso

ni)

6.4–

44.9

82D

3.29

14.

473

0.90

3,

0.00

184

H4.

833

4.09

50.

817

,0.

001

Cen

trar

chid

ae

Bla

ckC

rap

pie

(Pom

oxis

nig

rom

acu

latu

s)4.

4–21

.029

H3.

239

1.24

90.

980

,0.

001

28L

3.46

01.

673

0.93

5,

0.00

1B

lueg

ill

(Lep

omis

mac

roch

iru

s)2.

9–18

.318

D2.

752

1.29

80.

994

,0.

001

16L

3.23

20.

594

0.95

7,

0.00

1

Cyp

rin

idae

Big

mo

uth

Shin

er(N

otro

pis

dors

alis

)4.

7–7.

322

P2.

280

3.45

10.

563

,0.

001

26C

3.23

43.

069

0.55

0,

0.00

1B

lun

tno

seM

inn

ow

(Pim

epha

les

not

atu

s)2.

7–8.

635

L2.

874

1.99

70.

792

,0.

001

37C

3.09

53.

096

0.65

2,

0.00

1C

arp

(Cyp

rin

us

carp

io)

3.0–

68.9

125

L2.

455

0.57

50.

963

,0.

001

128

L2.

382

2.06

80.

915

,0.

001

Co

mm

on

Shin

er(L

uxi

lus

corn

utu

s)3.

3–16

.864

L3.

450

1.71

00.

958

,0.

001

56P

4.14

53.

594

0.61

5,

0.00

1C

reek

Ch

ub

(Sem

otil

us

atro

mac

ula

tus)

3.5–

23.7

59L

4.54

91.

035

0.94

2,

0.00

164

P5.

910

3.31

80.

781

,0.

001

Fat

hea

dM

inn

ow

(Pim

epha

les

prom

elas

)3.

4–7.

413

C4.

847

1.98

80.

938

,0.

001

21D

2.52

33.

271

0.61

5,

0.00

1H

orn

yhea

dC

hu

b(N

ocom

isbi

gutt

atu

s)3.

2–15

.843

L3.

186

1.08

90.

965

,0.

001

48H

3.67

01.

206

0.86

0,

0.00

1L

arge

-sca

leSt

on

ero

ller

(Cam

post

oma

olig

olep

is)

8.9–

12.9

13H

2.07

34.

404

0.71

1,

0.00

113

H0.

849

9.02

80.

139

0.20

9*L

on

gno

seD

ace

(Rhi

nic

hthy

sca

tara

ctae

)5.

8–10

.918

H5.

845

1.72

40.

749

,0.

001

17C

2.92

55.

595

0.39

50.

007

San

dSh

iner

(Not

ropi

sst

ram

ineu

s)4.

8–7.

515

P1.

879

3.74

80.

317

0.02

919

C1.

161

4.78

70.

136

0.12

0*Sp

otf

inSh

iner

(Cyp

rin

ella

spil

opte

ra)

5.3–

8.6

17L

3.19

50.

621

0.80

0,

0.00

114

L1.

836

3.47

60.

597

,0.

001

Wes

tern

Bla

ckn

ose

Dac

e(R

hin

icht

hys

obtu

sus)

3.5–

9.4

21H

4.08

22.

810

0.63

0,

0.00

126

H4.

855

2.71

60.

659

,0.

001


Spec

ies

To

tal

len

gth

ran

ge(c

m)

Lat

eral

lin

eN

on

-Lat

eral

lin

e

nV

ar.

Slo

pe

Inte

rcep

tr2

Pn

Var

.Sl

op

eIn

terc

ept

r2P

Eso

cid

ae

No

rth

ern

Pik

e(E

sox

luci

us)

13.2

–55.

039

H8.

932

2.40

30.

952

,0.

001

44L

6.01

96.

623

0.77

9,

0.00

1

Mo

ron

idae

Wh

ite

Bas

s(M

oron

ech

ryso

ps)

5.8–

15.3

25H

4.73

91.

139

0.93

5,

0.00

130

H3.

732

2.99

10.

823

,0.

001

Per

cid

ae

Bla

cksi

de

Dar

ter

(Per

cin

am

acu

lata

)3.

4–9.

044

L5.

512

0.95

00.

871

,0.

001

43H

3.37

03.

482

0.47

6,

0.00

1Jo

hn

ny

Dar

ter

(Eth

eost

oma

nig

rum

)2.

6–7.

336

L4.

254

1.25

20.

734

,0.

001

55H

2.56

22.

464

0.44

0,

0.00

1Ye

llo

wP

erch

(Per

cafl

aves

cen

s)6.

5–21

.943

L3.

813

2.97

60.

797

,0.

001

44L

2.89

45.

382

0.68

6,

0.00

1

Sci

aen

idae

Fre

shw

ater

Dru

m(A

plod

inot

us

gru

nn

ien

s)3.

6–43

.510

P12

.772

4.19

90.

989

,0.

001

12H

5.04

20.

536

0.95

9,

0.00

1

TA

BL

E1.

—C

on

tin

ued


ranging from 0.471 (large cyprinids) to 0.984 (centrarchids and white bass/freshwaterdrum) and 0.305 (large cyprinids) to 0.953 (white bass/freshwater drum) for lateral lineand non-lateral line scales, respectively. For all six multi-species groups, lateral line scalesproduced better models than non-lateral line scales.

Scale length provided the best fit for nine species using lateral line scales, the most of anymeasurement (Table 1). However, scale height provided the best fit for the most species (n 5

8) using non-lateral line scales (Table 1). The majority of multi-species groups attained theirbest fit using scale length for lateral line models, whereas maximum r2 was attained using thelength, height or posterior radius measurement using non-lateral line scales (Table 2).

DISCUSSION

The fish body-scale relationship is well known in fisheries literature, so it is not surprisingthat there are positive relationships (in most cases) between scale size and fish total length(Whitney and Carlander, 1956; Hile, 1970; Carlander, 1982; Francis, 1990; Pierce et al.,1996). Other studies (e.g., Pierce et al., 1996; Giordano, 2005; Miranda and Escala, 2007)have typically reported r2 . 0.85, but for many species in our study values were lower(particularly using non-lateral line scales). But, in previous studies scales typically were takenfrom a specific location (e.g., at the tip of the pectoral fin when it is flattened against thebody) on the fish to minimize variation, and thereby maximizing r2 (Regier, 1962;Scarnecchia, 1979; Carlander, 1982; Pierce et al., 1996). However, because scales vary in sizeand shape over the body of the fish, and the exact location on the body where the scaleoriginated can’t be determined from the scale itself, this method would inaccuratelyestimate the size of fish prey (Phillips, 1948; Joeris, 1956; Scarnecchia, 1979; Daniels, 1996;Roberts et al., 2007). Therefore, predictive relationships need to be established using scalesamples from the entire body. Our study documented that there are positive relationshipsbetween scale size (using both lateral line and non-lateral line scales) and the total length offish, which demonstrates the utility of using these relationships for estimating the size ofsome fish species. Our study also demonstrates that variation in scale size can be reducedsubstantially by focusing model development on lateral line scales.

Constructing models using only lateral line scales restricts the area from which scales areselected, resulting in models typically having r2 higher than models built with non-lateralline scales (Tables 1 and 2). There were considerable differences in r2 values (i.e., modelsconstructed with lateral line scales were better fitting than models constructed using non-lateral line scales) for several species [e.g., common shiner, northern pike (Esox lucius), andpercids], but for other species (e.g., carp, centrarchids and freshwater drum) r2 valuesdiffered only slightly between models (Table 1). Small sample size (e.g., limited size range,having clumps of fish of similar sizes and having unequal numbers of scales selected fromeach scale region) likely contributed to the observed small differences between models forsome species [e.g., shorthead redhorse (Moxostoma macrolepidotum), bluegill and freshwaterdrum] and poor relationships for others (e.g., sand shiners and largescale stonerollers)(Table 1). However, the degree of uniformity in scale size among regions of the bodyappears to be a species-specific characteristic, with some species possessing non-lateral linescales that are similar in shape across the entire body, leading to an r2 close to that of lateralline models. For example, we had a large sample size (nLL 5 125, nR 5 128) and anadequate distribution of selected scales for carp; but the non-lateral line model stillpossessed a very high r2 (0.915, compared with 0.963 for lateral line scales). For such species,the use of non-lateral line scales instead of lateral line scales for obtaining prey size estimateswould have little effect on the precision of estimates.


TA

BL

E2.

—T

he

coef

fici

ent

of

det

erm

inat

ion

(r2),

regr

essi

on

coef

fici

ents

and

sign

ific

ance

of

the

rela

tio

nsh

ip(P

)o

fb

od

y-sc

ale

regr

essi

on

sfo

rsi

xm

ult

i-sp

ecie

sfi

shgr

ou

ps

of

the

Red

Riv

ero

fth

eN

ort

htr

ibu

tari

eso

fea

ster

nN

ort

hD

ako

tau

sin

gsa

mp

les

coll

ecte

dJu

n.–

No

v.20

07,

by

scal

ety

pe.

Var

iab

les

incl

ud

e:A

5an

teri

or

rad

ius,

B5

ante

rola

tera

lra

diu

s,C

5p

ost

ero

late

ral

rad

ius,

D5

dia

gon

al,

H5

hei

ght,

L5

len

gth

,P

5p

ost

erio

rra

diu

s

Gro

up

To

tal

len

gth

ran

ge(c

m)

Lat

eral

lin

eN

on

-Lat

eral

lin

e

nV

ar.

Slo

pe

Inte

rcep

tr2

PN

Var

.Sl

op

eIn

terc

ept

r2P

Cen

trar

chid

ae

Bla

ckC

rap

pie

(Pom

oxis

nig

rom

acu

latu

s)o

rB

lueg

ill

(Lep

omis

mac

roch

iru

s)2.

9–21

.036

H3.

231.

110.

984

,0.

001

32L

3.34

1.14

0.92

3,

0.00

1

Cyp

rin

idae

Dac

e(R

hin

icht

hys

cata

ract

aean

dR

.ob

tusu

s)3.

5–10

.936

D3.

962.

890.

654

,0.

001

34P

3.83

4.25

0.57

6,

0.00

1L

arge

Cyp

rin

ids

(.10

cm)1

10.1

–23.

772

L2.

983.

920.

471

,0.

001

72L

2.11

8.15

0.30

5,

0.00

1Sm

all

Cyp

rin

ids

(#10

cm)2

2.7–

10.0

117

L2.

402.

840.

623

,0.

001

119

L2.

023.

660.

445

,0.

001

Per

cid

ae

Dar

ters

(Eth

eost

oma

nig

rum

and

Per

cin

am

acu

lata

)2.

6–9.

072

L5.

880.

200.

792

,0.

001

86P

10.5

53.

080.

402

,0.

001

Mo

ron

idae

or

Sci

aen

idae

Wh

ite

Bas

s(M

oron

ech

ryso

ps)

or

Fre

shw

ater

Dru

m(A

plod

inot

us

gru

nn

ien

s)3.

6–43

.520

L4.

552.

050.

984

,0.

001

24H

4.98

0.30

0.95

3,

0.00

1

1In

clu

des

:co

mm

on

shin

er(L

uxi

lus

corn

utu

s),

cree

kch

ub

(Sem

otil

us

atro

mac

ula

tus)

and

ho

rnyh

ead

chu

b(N

ocom

isbi

gutt

atu

s)2

Incl

ud

es:

big

mo

uth

shin

er(N

otro

pis

dors

alis

),b

lun

tno

sem

inn

ow

(Pim

epha

les

not

atu

s),

com

mo

nsh

iner

,cr

eek

chu

b,

fath

ead

min

no

w(P

imep

hale

spr

omel

as),

ho

rnyh

ead

chu

b,

san

dsh

iner

(Not

ropi

sst

ram

ineu

s)an

dsp

otf

insh

iner

(Cyp

rin

ella

spil

opte

ra)


Generally, the best scale measurement for estimating fish size was scale length.Accordingly, scale length has been demonstrated as the preferred dimension for moststudies where the relationship between scale size and fish length has been assessed (Daniels,1996; Pierce et al., 1996; Giordano, 2005; Miranda and Escala, 2007). Scale height was a goodestimator for some species, and has previously been shown to be a better predictor of fishsize than scale length for white suckers (Catostomus commersoni) (Giordano, 2005). Althoughthere are exceptions, scale length and height are usually the measurements that mostaccurately estimate the length of fish prey and, therefore, should be the focus of futurestudies.

Despite advantages, using scales for estimating the size of fish prey does have limitations.Perhaps the largest limitation is that not all fish possess scales. Ictalurids lack scales andbecause of their abundance are commonly preyed upon by river otters (Field, 1970; Serfasset al., 1990; Noordhuis, 2002; Giordano, 2005; Stearns and Serfass, 2011). Therefore, by onlyusing scales for size estimation, the size of some fish in the diet (occasionally a largeportion) can’t be evaluated. Therefore, to completely assess the length of fish in the diet,other non-digestible structures would need to be used in size estimation. Pectoral spines aredistinctive features of ictalurids, previously have been reported to have a relationship withfish length (Klaasen and Townsend, 1973) and, therefore, warrant further investigation foruse in food studies designed to estimate prey size. Other limitations of using scales for preysize estimation include difficulty in identifying the actual number of fish of similar sizewithin a particular scat and possible degradation through the digestive system. Degradationof a scale in passage would result in an underestimation of the size of the fish it originatedfrom. However, a bias towards larger fish could result if smaller scales become moredegraded than larger scales, especially if they are degraded completely.

This study demonstrates the utility of scales for estimating the prey size of piscivores, andaccordingly the models and techniques described here were used to estimate the prey size ofriver otters in eastern North Dakota (Stearns and Serfass, 2011). This study also documentsthat using only lateral line scales for model construction results in stronger predictiverelationships than using non-lateral line scales and, therefore, should be preferred forestimating prey size. Nonetheless, fish prey remains (e.g., in scats or digestive tracts) may notalways contain a lateral line scale, which requires a decision whether or not to use non-lateralline scales. Our study suggests that regressions of non-lateral line scales and fish length oftenhave significant linear relationships and r2 values acceptable for estimating the length of manyspecies of fish included in this investigation (Table 1). Because of their abundance in scats,the existence of identification keys to the family level, the positive relationships between scalesize and fish length and the noninvasive method of establishing predictive relationships, scalesare an ideal structure for use in estimating the size of fish prey.

Acknowledgments.—We thank the North Dakota Game and Fish Department for providing funding(through wildlife and sport fish restoration funds under the state wildlife grant program) and assistancethroughout the project. Thank you to South Dakota State University researchers Lucas Borgstrom andCari-Ann Hayer for allowing us to join them during their fisheries research, so that we could obtain scalesamples. We also thank the staff at Turtle River State Park (Arvilla, ND) for providing housing and ourfield and lab crews—Zach Olson, Brenda Newton, Sarah Negium, Lou Allard, Melissa Smith, StevenLoughery, Thomas Baden and Melissa Brannon.

LITERATURE CITED

ADRIAN, M. I. AND M. DELIBES. 1987. Food habits of the otter (Lutra lutra) in two habitats of the DonanaNational Park, SW Spain. J. Zool. (Lond.), 212:399–406.


CARLANDER, K. D. 1982. Standard intercepts for calculating lengths from scale measurements for somecentrarchid and percid fishes. Trans. Am. Fish. Soc., 111:332–336.

CARSS, D. N., H. KRUUK AND J. W. H. CONROY. 1990. Predation on adult Atlantic salmon, Salmo salar L., byotters, Lutra lutra L., within the River Dee system, Aberdeenshire, Scotland. J. Fish Biol.,37:935–944.

——— AND K. C. NELSON. 1998. Cyprinid prey remains in otter Lutra lutra faeces: some words of caution.J. Zool. (Lond.), 245:238–244.

COPES, F. A. AND R. A. TUBB. 1966. Fishes of the Red River tributaries in North Dakota. Contrib. Inst. Ecol.Stud.. No. 1.

COPP, G. H. AND V. KOVAC. 2003. Biometric relationships between body size and bone lengths in fish preyof the Eurasian otter Lutra lutra: chub Leuciscus cephalus and perch Perca fluviatilis. Folia Zool.,52:109–112.

DANIELS, R. A. 1996. Guide to the identification of scales of inland fishes of northeastern North America.New York State Mus. Bull. No. 488. The University of New York, Albany, New York. 97 p.

DELLINGER, T. AND F. TRILLMICH. 1999. Fish prey of the sympatric Galapagos fur seals and sea lions:seasonal variation and niche separation. Can. J. Zool., 77:1204–1216.

EDDY, S., R. C. TASKER AND J. C. UNDERHILL. 1972. Fishes of the Red River, Rainy River, and Lake of theWoods, Minnesota. Occas. Pap. Bell Mus. Nat. Hist. No. 11.

ERLINGE, S. 1968. Food studies on captive otters Lutra lutra. Oikos, 19:259–270.FIELD, R. J. 1970. Winter habits of the river otter (Lutra canadensis) in Michigan. Mich. Academ., 3:49–58.FRANCIS, R. I. C. C. 1990. Back-calculation of fish length: a critical review. J. Fish Biol., 36:883–902.GIORDANO, A. J. 2005. Feeding ecology of a reintroduced river otter (Lontra canadensis) population in

northcentral Pennsylvania. Thesis, Frostburg State University, Frostburg, Maryland.GRANADEIRO, J. P. AND M. A. SILVA. 2000. The use of otoliths and vertebrae in the identification and size-

estimation of fish in predator-prey studies. Cybium, 24:383–393.GREER, K. R. 1955. Yearly food habits of the river otter in the Thompson Lakes Region, northwest

Montana, as indicated by scat analysis. Am. Midl. Nat., 54:299–313.HAJKOVA, P., K. ROCHE AND L. KOCIAN. 2003. On the use of diagnostic bones of brown trout, Salmo trutta m.

fario, grayling, Thymallus thymallus, and Carpathian sculpin, Cottus poecilopus in Eurasian otter,Lutra lutra diet analysis. Folia Zool., 52:389–398.

HAMILTON, W. J., JR. 1961. Late fall, winter, and early spring foods of 141 otters from New York. New YorkFish Game J., 8:106–109.

HANSEL, H. C., S. D. DUKE, P. T. LOFY AND G. A. GRAY. 1988. Use of diagnostic bones to identify andestimate original lengths of ingested prey fishes. Trans. Am. Fish. Soc., 117:55–62.

HILE, R. 1970. Body-scale relation and calculation of growth in fishes. Trans. Am. Fish. Soc., 99:468–474.JOERIS, L. S. 1956. Structure and growth of scales of yellow perch of Green Bay. Trans. Am. Fish. Soc.,

86:169–194.KEMENES, I. AND G. NECHAY. 1990. The food of otters Lutra lutra in different habitats in Hungary. Acta

Ther., 35:17–24.KLAASEN, H. E. AND A. H. TOWNSEND. 1973. Age and growth of channel catfish in Tuttle Creek Reservoir,

Kansas. Trans. Kans. Acad. Sci., 76:248–253.KOEL, T. M. AND J. J. PETERKA. 1998. Stream fishes of the Red River of the North basin, United States: a

comprehensive review. Can. Field Nat., 112:631–646.KOZENA, I., P. URBAN AND I. STOURACOVA. 1992. The diet of the otter (Lutra lutra Linn.) in the Polana

protected landscape region. Folia Zool., 41:107–122.LAGLER, K. F. 1956. Freshwater fishery biology, 2nd ed. W. C. Brown Publishing, Dubuque, Iowa. 418 p.MELQUIST, W. E. AND M. G. HORNOCKER. 1983. Ecology of river otters in west central Idaho. Wildl. Monog.,

83:1–60.MIRANDA, R. AND M. C. ESCALA. 2007. On the use of cyprinid scales in the diet analysis of piscivorous

species: how much information is hidden in a fish scale? Trans. Am. Fish. Soc., 136:1008–1017.NOORDHUIS, R. 2002. The river otter (Lontra canadensis) in Clarke County (Georgia, USA): survey, food

habits, and environmental factors. IUCN Otter Special. Group Bull., 19:75–86.


PHILLIPS, J. B. 1948. Comparison of calculated fish lengths based on scales from different body areas ofthe sardine, Sardinops caerulea. Copeia, 1948:99–106.

PIERCE, C. L., J. B. RASMUSSEN AND W. C. LEGGETT. 1996. Back-calculation of fish length from scales:empirical comparison of proportional methods. Trans. Am. Fish. Soc., 125:889–898.

PRENDA, J. AND C. GRANADO-LORENCIO. 1992. Biometric analysis of some cyprinid bones of prey fishes toestimate the original lengths and weights. Folia Zool., 41:175–185.

REGIER, H. A. 1962. Validation of the scale method for estimation age and growth of bluegills. Trans. Am.Fish. Soc., 91:362–374.

ROBERTS, N. M., C. F. RABENI AND J. S. STANOVICK. 2007. Distinguishing centrarchid genera by use of lateralline scales. N. Am. J. Fish. Manage., 27:215–219.

ROSS, R. M., J. H. JOHNSON AND C. M. ADAMS. 2005. Use of fish-otolith-length regressions to infer size ofdouble-crested cormorant prey fish from recovered otoliths in Lake Ontario. Northeast. Nat.,12:133–140.

ROWE-ROWE, D. T. 1977. Variations in the predatory behavior of the clawless otter. Lammergeyer, 23:22–27.SCARNECCHIA, D. L. 1979. Variation of scale characteristics of coho salmon with sampling location on the

body. Progress. Fish Cult., 41:132–135.SERFASS, T. L., L. M. RYMON AND R. P. BROOKS. 1990. Feeding relationships of river otters in northeastern

Pennsylvania. Trans. Northeast Sect. Wildl. Soc., 47:43–53.STEARNS, C. R. 2008. Food habits of river otters (Lontra canadensis) in the Red River of the North drainage

of eastern North Dakota. Thesis, Frostburg State University, Frostburg, Maryland.——— AND T. L. SERFASS. 2011. Food habits and fish prey size selection of a newly colonizing population

of river otters (Lontra canadensis) in eastern North Dakota. Am. Midl. Nat., 165:169–182.STONER, J. D., D. L. LORENZ, G. J. WICHE AND R. M. GOLDSTEIN. 1993. Red River of the North basin,

Minnesota, North Dakota, and South Dakota. Wat. Res. Bull., 29:575–615.WHITNEY, R. R. AND K. D. CARLANDER. 1956. Interpretation of body-scale regression for computing body

length of fish. J. Wildl. Manage., 20:21–27.WISE, M. H. 1980. The use of fish vertebrae in scats for estimating prey size of otters and mink. J. Zool.

(Lond.), 192:25–31.

SUBMITTED 31 AUGUST 2010 ACCEPTED 3 JANUARY 2011

APPENDIX A

Select examples of species and multi-species group models of lateral line (left) and non-lateral line (right) body-scale relationships established from samples collected from the RedRiver of the North drainage of eastern North Dakota, Jun.–Oct. 2007. The relationshipspresented were constructed using the scale measurement that resulted in the best model foreach species (or group) of six measurements that were considered. Refer to Stearns (2008)for all species and group-models assessed in this study.


the use of scales to estimate the prey size of nearctic river otters (lontra canadensis) and other...

Documents