Environmental Biology of Fishes Vol. 17, No. 4, p. 291-297. 1986
0 Dr W. Junk Publishers, Dordrccht.

Depth distribution of Campostoma grazing scars in an Ozark stream

William J. Matthews, Mary E. Power & Arthur J. Stewart
University of Oklahoma Biological Station, Kingston, OK 73439, U.S.A.

Keywords: Herbivory, Attached algae, Periphyton, Minnows, Cyprinidae, Stream fishes, Feeding behavior

Synopsis

Campostoma spp., widespread and abundant herbivorous minnows of eastern North America, produce
distinctive `grazing scars' when feeding on algae attached to natural substrates in streams. These scars are
particularly prominent upon the low growth forms of blue-green algae that dominate the attached algal flora
of many upland streams. In one stream pool in the Ozark uplands of Oklahoma, numbers and sizes of grazing
scars coincided with numbers and sizes of individual Campostoma that occurred across a depth gradient,
demonstrating that the information contained in the scars can provide quantification of microhabitat use and
grazing intensity of these important herbivores. The results also support the hypothesis that in environments
free of aquatic predators, larger fish use deeper parts of available stream habitats, particularly if threats from
terrestrial or avian predators exist.

Introduction

Grazing fishes can have profound effects on pri-
mary productivity, standing crops, taxa, or growth
forms of attached algae in marine systems (Ogden
& Lobel 1978, Montgomery 1980, Montgomery et
al. 1980, Lobel 1980, Sammarco 1983, Miller 1982,
Lubchenco 1982, Weinstein et al. 1982) or tropical
freshwater streams (Power 1983, 1984). Grazing
fish in temperate streams may also have important
consequences for distribution, standing crops, or
taxonomic composition of the attached algal flora.
In small to medium-sized streams in Oklahoma and
Arkansas, herbivory by stoneroller minnows,
Campostoma anomalum, can control pool-to-pool
distribution of attached algae, or maintain algae at
low standing crops despite high primary pro-
ductivity (Power & Matthews 1983, Power et al.
1985, A. J. Stewart, personal observation). Precise

information on microsites and intensity of Camp-
ostoma activity is thus of basic interest to stream
ecologists.
Campostoma make distinctive `grazing scars'
when they feed upon low growth forms (`felts') of
attached algae on substrates in streams of the
Ozark Mountains (Oklahoma-Arkansas). Viewing
underwater via mask and snorkel we have actually
observed individual Campostoma producing these
scars; they also make similar scars on spinach-agar
coated tiles in the laboratory. After a large school
of Campostoma has grazed for several minutes in a
localized area, large numbers of fresh grazing scars
are conspicuous, contrasting with older, `healing'
grazing scars from earlier feeding bouts. [Kraatz
(1923) noted that grazing by Campostoma pro-
duced trials on algae-coated aquarium walls, but
these differed in form from the scars we find on
natural substrates.]

292

Fig. 1. Feeding scars left by Campostoma on slate from a southwestcrn Ozark strcam. Dark scars indicate recent feeding; lighter scars arc
older and pai-tiallv `healed' by algal rcgrowth and coloniration.

The algal felts in our study streams comprise
mostly blue-green (Cyanobacteria) taxa, e.g. Cal-
othrix, Phormidium, and Oscillutoria spp., which
produce a slippery, blackish surface 1-2 mm
(= `felt') thick on rock or wood substrates. When
feeding Cumpostornu remove attached algae from
such surfaces, they usually do so with behaviors we
describe as `swiping' (the head is thrust downward
and sideways, with the lower jaw scraping algae
from the rock in the sideways motion): `shoveling'
(pushing the lower jaw against the surface and
swimming vigorously forward); or `nipping' (small,
rapid bites with the body at 45" angle to the sub-
strate). The swiping and shoveling modes are most
commonly observed when Campostoma graze on
algal felts, and appear to most often produce iden-
tifiable grazing scars. The grazing scars (Fig. 1) are
typically rectangular, 3-4 mm wide and 5-20 mm
long, and are readily distinguishable from scars left
on rocks by other common grazers. Snails (e.g.
Oxtrema) are common grazers in Ozark streams,
but produce long, convoluted scars or trails. Her-
bivorous chironomids of these streams, e.g.
Orthocladius, graze circular or oval areas around

their attached cases. Notropis nubilis is another
common algae-eating minnow in Ozark streams,
but lacks the reinforced scraping lower jaw of
Cumpostornu, and nips at substrates rather than
swiping or shoveling. Consequently, N. nubilis
does not make grazing scars like those of Camp-
ostoma (W. J. Matthews, personal observations by
snorkeling).
The grazing scars of Cumpostornu are of particu-
lar interest in that they provide a record of activity
by these important herbivores. Grazing scars can
provide a convenient tool for quantifying feeding
activities of Campostoma and other grazing fishes,
if density and size of scars correspond with num-
bers or size of fish grazing a given area. Miller
(1982) and Hixon & Brostoff (1983) have used graz-
ing scars to document feeding intensity of marine
reef fishes, and Bowen (1979) used feeding scars to
quantify feeding activity of fish in a tropical lake.
Here we describe a situation that allowed us to
quantify distribution of grazing scars and of Camp-
ostoma under virtually ideal circumstances for ob-
servation. We asked specifically if the distribution
and size of grazing scars, and distribution and size

of Campostoma coincided along a depth gradient in
a pool of an Ozark stream.

Methods

The study site was a single pool, 65.8 m long and an
average of 10.4 m wide in Tyner Creek, a tributary
of Baron Fork of the Illinois River, Adair County,
Oklahoma. Much of the pool was of relatively uni-
form width, ranging 7.1-12.9 m wide in 8 transects.
The head and tail of the pool ended in swift riffles.
The long axis of the pool was north-south; no tree
canopy existed. The west side of the pool consisted
of prominent benches of black or dark gray shale,
which produced `stair-step' surfaces beginning im-
mediately beneath the surface of the water and
leading down to the bedrock and gravel bottom of
the pool, with greatest depths of -55 cm near the
west side. From the zone of maximum depth, a
gently sloping gravel and cobble substrate ex-
tended to the east shore. Water was very clear, and
the two days of study (8-9 May 1984) were cloud-
less, providing nearly perfect `bank-to-bank' view-
ing underwater. Surface flow through the pool
averaged 59 cm sec-' (timing a floating object three
times through the length of pool). Midday water
temperature was 14.5" C.
Several hundred Campostoma and several hun-
dred Notropis pilsbryi occupied the pool. A de-
tailed snorkeling survey (WJM) of -0.5 km down-
stream (a series of 5 pools and riffles), and a search
upstream approximately the same distance re-
vealed no potential piscine predators (e.g. small-
mouth bass, Micropterus dolomieui, which are
common in many Ozark streams).

This reach of Tyner Creek was dry (at the sur-
face; subsurface flow persisted) during part of the
year before our study (Mr. R. Whitmire, property
owner, personal communication). We assume that
since rewatering occurred, minnows had moved
back into this reach of stream, but larger predators
had not yet arrived. Thus, the system provided an
opportunity to not only compare distribution of
Campostoma and grazing scars, but to do so in an
environment free of piscine predators.

In this paper, as in all our work in Ozark streams,

we refer to the study animals as `Campostoma'.
The most abundant species is Campostoma anoma-
lum, but Campostoma oligolepis also occurs in
these streams. The two are similar in feeding habifs
and general ecology (Pflieger 1975), and only tu-
berculate adult males can be distinguished with
confidence in underwater observations. Small indi-
viduals cannot be distinguished, even in laboratory
examination (Robert C. Cashner, personal com-
munication).
The slate substrates in this pool were covered by
a bluegreen algal felt, against which Campostoma
grazing scars were easily observed (Fig. 1). To
quantify the numbers of grazing scars per unit area
at different depths on the shale benches of the west
side of the pool, we used a 50m tape to lay three
longitudinal transects parallel to the west side of
the pool, and along the benches. At every meter on
each transect, we recorded water depth and
counted scars within 10 cm x 10 cm quadrats on the
substrate. Depths of points on the transects used to
quantify numbers of Campostoma grazing scars
ranged 9-59cm deep. Only 5 points were in the
50-59cm interval, and thus were pooled with
40-49 cm deep points for analysis (Fig. 2). To quan-
tify size of grazing scars in various depth intervals,
we selected areas with prominent new scars at deep
(47-51 cm), medium (35-42) cm, and shallow (15-
22 cm) locations. In each of these areas, one of us
with mask and snorkel (1) `blindly' touched the
rock platform with a ruler and (2) measured the
width and length of the nearest scar (viewing un-
derwater with mask and snorkel) to the nearest
millimeter. This procedure was repeated 10 times
at each position. The observer then moved to an-
other position and repeated the procedure until a
total of 40 scars was measured in each depth zone.
On 9 May we made direct observations of habitat
use by Campostoma in the area of the shale
benches where we counted and measured grazing
scars. We each lay motionless in positions on the
east side of the pool, from which we could readily
observe by mask and snorkel the movement and
activities of Campostoma on the shale benches of
the west side of the pool. Our entry and departure
from the water was as inconspicuous as possible,
and Campostoma seemed undisturbed by our pre-

294

5c

40

N

`E

u
0
2 30

v) L

0
U

w? 2c

0

:

c

in

: IC

E

o
0

(

n

0-19 20-29 30-39 40-59

Depth (cm)

Ffg. 2. Mean (horizontal bar) plus or minus one standard devia-
tion (vertical bars) ol number of C'arnpo.sromu grazing scars 100
em ?. in three tranaccts in a pool of Tyner Creek, 8 May 1984. At
thc 2&29 cm depth interval one outlier is omitted. as discussed
in text.

sence: they fed, moved about freely, and in every
way seemed to behave normally. The observation
areas ranged from shale surfaces <20 cm below the
water's surface to surfaces at or near maximum
depth of the pool (>SOcm). As an aid to estimating
size of Campostoma, elongate pieces of cobble we
called large (= 150-161 mm), medium (110-
115 mm), and small (all 75 mm), were placed in the
observation area as references. From -1045 to
1330 h we each made scans every 2 min of predeter-

mined areas of the shale bench at different depth
zones. In each scan we counted Campostoma by
size classes (`small', `medium', or `large') corre-
sponding to our `reference rocks' nearest matching
their total length.

Results

Campostoma grazing scars were more numerous
on deeper (>30cm) than on shallow (<30cm)
substrates (Fig. 2). At the depth interval 20-29 cm,
we found the following variates (numbers of scars
100cm-'): 0, 0, 0, 0, 0, 0, 3, 4, 14, 17, 38, and 123.
The variate 123 was approximately 1.5 times as
large as any other variate at any depth in the trans-
ects, and is a distinct outlier in the 20-29 cm depth
treatment. A Kruskal-Wallis test (Sokal & Rohlf
1981) indicated significant differences of mean
grazing scars per unit area across the four depth
intervals (Fig. 2), whether the outlier at the 20-29
cm depth treatment is (HI,,, = 24.604; p<O.O01) or
is not (H,,,, = 27.912; p<O.OOl) included.
Grazing scars were largest in the deepest areas.
Those at depths of 47-51cm were on average
longer, wider, and had greater surface area than
those at shallow or medium depths (Table 1).
Underwater observations of numbers and size of
Campostoma occupying different depths (Fig. 3)
were consistent with our observations of more and
larger grazing scars in deeper parts of the pool.
During the approximately 3 h that all three obser-
vers recorded data, a total of 1402 observations of
individual Campostoma were made at depths of
38-S6cm (Zones I11 and IV, Fig. 3), 556 observa-
tions were made at 22-30cm, and only 2 observa-

7hblc 1. Mean and standard deviation (in parentheses) of dimensions of grazing scars in shnllow. medium, and deep parts of the Tyner
Crcck pool. F, is lrom One-way ANOVA for cach dimcnsion, across depths as trcatments. N  40 grazing scars at each dcpth interval.

Dimension of grazing scar Depth

Shallow

( 15-22 cm)

Medium       Deep
(35-42 cm)      (47-5 1 cm)

Width (mni)
Length (mm)
Area (md)

3.50 ( 0.60)     3.65 ( 0.83)      4.15 ( 0.67)       9.005          <0.001
13.75 ( 3.17)     13.08 ( 3.22)      16.3X ( 3.67)       10.746           <0.00 1
48.78 (16.76)     48.55 (18.87)     (18.90 (22.44)      14.360           t0.00 1

295

401 30

Small

n (N 2555)

- . . . . , . .

IIIIILX

8-14 22-30 38-47 50-56

Depth (cm)

Fi,y. 3. Size distrihution of Cnrnposroniu (size classes `small',
`medium` and `large' defined in text) across depth intervals in
one pool of Tyner Creek, 9 May 1984. N = actual numbers of
individuals observed hy snorkeling: vertical axis = numbers of
individuals observed, adjusted for area of substrate under ob-
servation at each depth interval. Data presented represent
cumulative counts for -3 h of scans at 2min intervals at fixcd
sites. Because occurrences of individual fish in successive time
intcrvals are not likely independent. we prcsent no statistical
analysis of these data, and view them only as descriptive.

tions were made at 8-14cm. Adjusted to account
for differences in the substrate surface areas ob-
served in depth zones I-IV (Fig. 3), there were, in
successive zones from shallowest to deepest, 0.74,
65.4, 61.8, and 94.4 counts of Campostornu m-? of
area under observation. These results show that
Campostoma are scarce in areas less than 20cm
deep, and that in environments free of predatory
fishes Campostorna preferentially occupy deep
parts of pools.
Size of Campostoma observed per depth zone
and size of grazing scars were related. Grazing
scars in deepeqt areas were on average larger than
those in shallow areas (Table 1), as were Cump-
ostoma. Small Campostorna were most common in
relatively shallow water (22-30 cm), medium sized
Cumpostoma were abundant both at 38-47 and

50-56cm, and large Campostornu were observed
almost exclusively in the deepest parts of the pool
(5G56cm) (Fig. 3).

Discussion

In addition to documenting concordance of grazing
scars and Campostornu depth distribution in Tyner
Creek, the present findings support our earlier ob-
servations (Power et al. 1985) that in environments
free of predatory fishes, Carnpostoma tend to use
deep parts of pools, and that in such conditions
their grazing effects are more evident at deep than
at shallow depths (Fig. 2 of Power & Matthews
1983). Herons and kingfishers fish most commonly
and effectively in water less than 20 cm deep (Sal-
yer & Lagler 1949, Whitfield & Blaber 1978,
Kushlan 1978, Whitfield & Cyrus 1978, Boag 1982,
Kaiser 1982, Kramer et al. 1983, Power 1984).
Larger Carnpostoma, which occupy deeper areas
than smaller individuals (Fig. 3), may be more
vulnerable to avian predators. Preference by fish-
ing birds for larger fish has been documented in
several freshwater habitats (Eipper 1956, Kushlan
1978, Alexander 1979). In a New York stream,
Carnpostoma anornalurn dominated fish remains in
nests of belted kingfishers (Ceryle alcyon), despite
the fact that Campostoma were less common than
smaller dace, Rhinichthys spp. (Eipper 1956).
Kingfishers (C. alcyon) and herons (Ardea hero-
dim, Egretta caerulea, Butoroides virescens) are
common along Ozark streams. A little blue heron
(E. caerulea) was in the immediate vicinity of the
study pool during our observations. Further evi-
dence that large Campostornu in Tyner Creek may
be subject to attacks from birds are scars and
wounds we have seen on the backs of Carnpostoma
in streams of the area (M. E. Power &L W. J.
Matthews, personal observations).

Use of grazing scars to assess microdistribution
of Campostornu activities is not entirely free of
difficulty. Dense groups of grazers may denude
large patches making counts of individual scars
impossible. However, even intensely grazed sub-
strates show individual scars characteristic of
Campostoma at the edges of patches. Thus, al-

though counts of individual scars may be impossi-
ble, one can infer recent and intense grazing by
Campostoma from such grazed patches. A second
difficulty is that Campostoma grazing scars are ill-
defined on very rugose substrates. However, such
substrates are often interspersed in Ozark streams
with flat or smooth-surfaced cobble, upon which
well-formed scars are found. Finally, information
on recovery rates of algae in grazing scars is needed
if one is to know precisely how much time is repre-
sented in the record the scars provide. Rates of
algal regrowth vary widely, depending upon tem-
perature, light, available nutrients, and other fac-
tors. So far, we have used only relatively fresh scars
in estimating Campostomu activity, but the distinc-
tion is admittedly subjective.
Even with the qualifiers above, grazing scars
have much potential as tools for study of Camp-
ostoma microhabitat use or grazing intensity. In
addition to Campostoma, approximately 40 other
North American stream fish species may be impor-
tant grazers of attached algae (Lee et al. 1980).
Several, such as Acrocheilus alutaceus (chisel-
mouth) and several Catostomus species, are speci-
fically known to use the lower jaw to scrape algae
from substrates. Acrocheilus. like Cainpostoma,
leave distinct scars on substrates where they graze
(Moodie & Lindsey 1972). Many other freshwater
fishes might produce characteristic grazing scars
useful in quantitative ecological studies.

Acknowledgements

This study was funded under a National Science
Foundation grant BSR - 8307014.

References cited

Alexander. G.R. 1979. Prcdators of fish in coldwater streams.
pp. 153-170 ZH: H. Clepper (cd.) Predator-prey systems in
fishcrics management, Sport Fishing Institute. Washington,
D. C.

Boag. D. IS82 The kingfisher. Blandord Press. Dorset.

Bowen. S.tI. 1979. A nutritional constraint on detrivory in
lishcs: the stuntcd populations of Sarothcrotlori n~os.\rrmhictrs
in Lake Sibava, South Africa. Ecol. Monogr. 49: 17-31.

Eipper, A.W. 1956. Differences in vulnerability of the prey of
nesting kingfishers. J. Wildl. Manag. 20: 177-183.
Kaiser. M.S. 1982. Foraging ecology of the green heron on
Ozark streams. M.S. Thesis. University of Missouri, Colum-
bia. 112 pp.
Kraatz, W.C. 1923. A study of the food of the minnow, Carnp-
ostomri tinomalum. Ohio J. Sci. 23: 265-283.
Kramcr. D.L.. D. Manley & R. Bourgeois. 1983. The effect of
respiratory mode and oxygen concentration on the ribk of
aerial predation in fishes. Can. J. Zool. 61: 65.3-665.
Kushlan. J.A. 1078. Feeding ecology of wading birds. pp. 239-
297 /fi: A. Sprunt IV. J.C. Ogden & S. Winckler (ed.) Wad-
ing Birds, Research Report No. 7. National Audubon So-
ciety. New YorL.
Lee, D.S., C.R. Gilbert, C.H. Hocutt, R.E. Jenkins, D.E.
McAllister & J.R. Stauffcr, Jr. 1980. Atlas of North Amcr-
ican freshwater fishes. North C:irolina State Mus. Nat. Hist.,
Raleigh. 854 pp.

Lobel, P.S. 1980. Herbivory by damselfishes and their role in
coral reef community ecology. Bull. Mar. Sci. 30: 273-2239,
Lubchenco, J. 1082. Effects of grazers and algae competitors on

fucoid colonization in tide pools. J. Phycol. 18: 534550.

Miller, A.C. 1982. Effects of differential fish grazing on the
community structure of an intertidal recf flat at Enewctak
Atoll, Marshall Islands. Pacific Sciencc 36: 467-482.
Montgomery, W.L. 1980. The impact of non-selective grazing
by the giant blue damsclfish. Microspalhodon dorsulis. on
algal communities in the Gulf of California, Mexico. Bull.
Mar. Sci. 30: 29(&303.
Montgomery. W.L., T. Gerrodette & L.D. Marshall. 1980.
Effects of grazing by the yellowtail surgeonfish. Prionurus
punctutus, on algal communities in the Gulf of California.
Mexico. Bull. Mar. Sci. 30: 901-908.
Moodic, G.E.E. & C.C. Lindsey. 1972. Life history of a unique
cyprinid fish. the chiselmouth (Acrocheilus riluluceus). in
British Columbia. Syesis 5: SS-61.

Ogden. J.C. & P.S. Lobel. 1978. The role of herbivorous fishes
and urchins in coral reef communities. Env. Biol. Fish. 3:
39-63.

Power. M.E. 1983. Grazing responses of tropical freshwater
fishes to different scales of variation in their food. Env. Biol.
Fish. 9: 103-115.

Power. M.E. 1984. Depth distributions of armorcd catfish: pre-
dator-induced resource avoidance? Ecology 65: 523-528.
Power. M.E. & W.1. Matthews. 1983. Algae-grazing minnows

( Campostomu anomalum). piscivorous bass (Micropterns
spp.), and the distribution of attached algae in a small prairie-
margin stream. Oecologia 60: 328-332.

Power. M.E.. W.J. Matthews & A.J. Stewart. 1985. Grazing
minnows. piscivorous bass and stream algae: dynamics of a
strong interaction. Ecology 66: 144s1456.
Salycr, J.C. & K.F. Lagler. 1949. The castern belted kingfisher.
Megacervle alcyon alcyon (Linnaeus) in relation to fish man-
agement. Trans. Amer. Fish. Soc. 76: 97-117.
Sammarco. P.W. 1983. Effects of fish grazing and damselfish
territoriality on coral reef algae. 1. Algal community struc-

297

turc. Mar. Ecol. Prog. Series 13: 1-14.

Sokal. R.R. & F.J. Rohlf. 19x1. Biometry- 'The principles md
practice of statistics in biological research. W.H. Freeman.
San Francisco. 859 pp,
Weinstein, M.P., K.L. Heck, Jr., P.E. Giehcl & J.E. Gatcs.
1082. The role of herbivory in pinfish (Lrrgodon rhomhoides):
a preliminary investigation. Bull. Mar. Sci. 32: 791-795.

Whitfield. A.K. & S.J.M. Blabcr. 1978. Feeding ecology 01

piscivorous birds at Lake St. Lucia. Part 1: Diving birds.
Ostrich 49: 185-198.

Whitfield, A.K. & D.P. Cyrus. 1978. Fceding succession and
zoiiation of aquatic birds at False Ray. Lake St. Lucia. Ost-
rich 49: 8-1.5.

Received 2.7.198.5 Acccpied 19.11.1985