2012 NORTHEASTERN NATURALIST 19(3):527–532
Crow Pellets from Winter Roosts in Lancaster,
Pennsylvania
Mary Annala1, Eric A. Tillman2, Gregory Backus3, Kandy L. Keacher2,
and Michael L. Avery2,*
Abstract - Although crows cast pellets, there is little quantitative information on pellets
from Corvus brachyrhynchos (American Crow), and none from C. ossifragus (fish
Crow). During a study of crow roost dispersal in Lancaster, PA, we collected samples of
pellets from several locations. By mass, pellets consisted mostly of grit and other fi ne
inorganic material, various seeds (principally Toxicodendron radicans [Poison Ivy] and
Celtis occidentalis [Common Hackberry]), and vegetation remnants. Six pellets contained
small-mammal bones. Because the Lancaster winter crow population included
many fish Crows, the source of the pellets was not certain. To clarify this, we compared
the size of the Lancaster crow pellets to those produced by captive fish Crows, and we
provide the fi rst quantitative description of pellets for either species. Our size comparisons
suggest that >90% of the pellets in our sample from Lancaster were produced by
American Crows.
Introduction
Pellet-casting has been reported within at least 18 orders of birds (Below
1979). Pellets usually contain indigestible parts of prey items, so pellet analysis
is a standard means for studying avian food habits (Coleman and Fraser 1987,
Errington 1932, Glading et al. 1943). Among passerines, Corvus spp. (crows)
regularly produce pellets (Berrow et al. 1992, Dean and Milton 2000, Kurosawa
et al. 2003).
Barrows and Schwarz (1895) described pellet production in Corvus brachyrhynchos
Brehm (American Crow). They provided a sketch of a typical crow
pellet, but included no measurements, nor did they characterize crow pellet
contents except in a general way. We are aware of just two additional studies
of crow pellets in the USA. Black (1941) collected 1214 pellets from 12 winter
American Crow roosts in Illinois, but he combined all pellets from each site
for analysis. Major food items identified by Black (1941) included corn, various
common weed seeds, and small mammals (bones or teeth occurred in 178
pellets). He estimated that grit comprised 17.6% of the pellets, by volume.
Platt (1956) collected 617 American Crow pellets year-round in south-central
Kansas and examined them singly, although he quantified only the food
items, “excluding sand and other extraneous material”, and he provided no
1Department of Wildlife Ecology, Michigan Technological University, Houghton, MI
49931. 2United States Department of Agriculture, Wildlife Services, National Wildlife
Research Center, 2820 East University Avenue, Gainesville, fl32641. 3Department
of Biology, Bard College, Annandale-on-Hudson, NY 12504. *Corresponding author -
michael.l.avery@aphis.usda.gov.
528 Northeastern Naturalist Vol. 19, No. 3
measurements of pellet size. field crops (wheat, sorghum, oats, sunflower,
and corn) comprised 59% of the identifiable food residues, invertebrates comprised
26.5%, and mammal remains 2.6% (Platt 1956). Apparently, published
information on pellets from Corvus ossifragus Wilson (fish Crow) is lacking
(McGowan 2001).
While conducting a study of winter crow roost dispersal, we noticed large
numbers of pellets at several locations in the Lancaster, PA area. This roosting
crow population consisted of 30,000–40,000 birds, including both fish and
American crows (Avery et al. 2008). Because quantitative information on crow
pellets is scarce, particularly regarding the fish Crow, we collected and analyzed
pellets to understand more about this phenomenon. Further, to separate
pellets according to species origin, we compared the Pennsylvania field sample
to pellets produced by captive fish Crows in Florida.
Methods
We opportunistically collected 113 crow pellets from locations throughout
Lancaster, PA during November–January 2005 and 2006. The locations included
a parking garage, various roads and lawns, and a vacant warehouse. For each
pellet, we recorded location and date of collection, placed them individually in
plastic bags, and froze them for later analysis.
We measured length and width of intact pellets using digital calipers.
We placed the pellets in a drying oven for 2–4 hours and then we weighed
each pellet on a digital balance. To determine the composition of the
pellets, each pellet was crushed and sifted through a series of sieves (US
Standard Sieve Series with ASTM specifications). We used level 4 with a
4.75-mm (0.19-in) mesh, level 6 with a 3.35-mm (0.13-in) mesh, level 12
with a 1.68-mm (0.07-in) mesh, and level 18 with a 1.00-mm (0.04-in) mesh.
For each pellet, we separated material caught in each sieve into categories of
Grit, Seeds, Vegetation, and Other. For sieve levels 4, 6, and 12, we counted
the individual pieces in each category and then weighed each category in aggregate.
The materials in level 18 and the bottom collection pan were too fine
to be counted individually, so only the total mass was recorded. We assessed
differences among collection sites in mass, length, width, and composition of
the pellets using one-way analyses of variance.
In Gainesville, FL, we maintained 7 fish Crows in 2 outdoor group pens (3.1 x
9.3 x 1.7 m). Mean body mass of these crows (281 g, SE = 7, n = 7) did not differ
(F1,60 = 0.53, P = 0.468) from that of the fish Crows we trapped in Lancaster, PA
(288 g, SE = 3, n = 55). The birds received a diet of cracked corn, chopped fruit,
and commercial dry dog food, as well as grit and sand. During April–June 2011,
we collected intact pellets opportunistically and measured their length and width,
but did not deconstruct them. We compared the measurements of fish Crow pellets
with the fi eld sample using one-way analyses of variance.
2012 M. Annala, E.A. Tillman, G.Backus, K.L. Keacher, and M.L. Avery 529
Results
Pellet composition
Among the Pennsylvania pellets, grit comprised almost half of the mass of all
pellets, while 35% consisted of fi ne materials too small to identify in the bottomlayer
collection pan. Vegetation remnants including corn seed fragments, bits
of stems, and ground-up leaf material comprised 6% of the pellet mass. Weed
seeds were very abundant in pellets from two sites and comprised 8% of the pellet
composition overall. The most common seed was Toxicodendron radicans L.
Kuntze (Poison Ivy). These seeds averaged 10 mg, with mean length of 4.5 mm
and mean width of 2.9 mm. The only other identifi ed seed was Celtis occidentalis
L. (Common Hackberry), which averaged 6.1 mm in diameter with mean mass of
0.1 g. The “other” category included teeth and bone fragments from small mammals
(n = 6 pellets) and one piece of rubber (n = 1 pellet).
Pellet size
The mean width of pellets from the Pennsylvania field sample (16.9 mm,
SE = 0.2, n = 113) exceeded (F1,158 = 347.51, P < 0.001) that of the captive fish
Crows (11.6 mm, SE = 0.2, n = 46). The mean length of pellets from the field
sample (32.6 mm, SE = 0.5, n = 113) also exceeded (F1,158 = 135.25, P < 0.001)
that of pellets produced by the captive fish Crows (21.5 mm, SE = 0.7, n =
46). There was very little overlap in the width distributions of the two samples
(fig. 1). Thus, based on the width dimension, it appears that only 11 of the 113
(9.7%) pellets collected in Lancaster were small enough to be within the size
range of those produced by fish Crows. The composition of these 11 pellets
did not differ from the 102 larger pellets in percent grit (P = 0.754), seeds (P =
figure 1. Distribution of crow pellet widths collected from Lancaster, PA (n = 113) and
produced by captive fish Crows (fiCR) in Gainesville, fl(n = 46).
530 Northeastern Naturalist Vol. 19, No. 3
0.661), or vegetation remnants (P = 0.412). Except for their respective sizes,
the pellets from Pennsylvania and Florida were very similar in appearance
(fig. 2).
Discussion
The paucity of quantitative information on crow pellets can be ascribed to
seasonality and fragility. Foods taken during the non-breeding season, particularly
waste corn, are often hard and diffi cult to digest, necessitating intake of grit
and sand, which comprise the bulk of the pellets. During other months, soft fruit,
insects, and other more readily digestible items dominate the diet, presumably
lessening the need for digestive aids. Platt (1956) noted that the availability of
crow pellets diminished outside the winter season.
In Lancaster, crows often staged or roosted on lawns, parking lots, and roofs
(Avery et al. 2008). Thus, at these sites disgorged pellets did not fall far and
therefore many remained intact. The chance of pellets produced by crows in tree
roosts surviving intact is greatly diminished (Barrows and Schwarz 1895). Furthermore,
crow pellets are not weather-resistant. Unless collected the day after
being deposited, the likelihood of fi nding a pellet intact is greatly reduced.
Direct comparisons of the composition of pellets we collected in Pennsylvania
with those reported previously by Black (1941) in Illinois and Platt (1956)
in Kansas are difficult because of differences in collecting, analyzing, and
reporting. We can say that grit was a predominant constituent in each sample,
that the bulk of the food items were vegetation, and that weed seeds (including
Poison Ivy and Common Hackberry) were common. Bones and teeth of small
mammals occurred in 5% of the pellets in our sample compared to 15% in Illinois
and 12% in Kansas. Unlike the samples from Illinois and Pennsylvania,
which contained corn remnants, the Kansas pellet sample included a wide
variety of crop seeds, reflecting differences in what was locally available to
wintering crows.
In comparing dietary information derived from pellets versus stomach contents,
Black (1941) noted consistent similarity. The usefulness of pellets will
likely depend on the questions being investigated. For food-habit studies, crow
figure 2. Crow pellet collected in Lancaster, PA (left; 31.1 mm long x 16.9 mm wide)
and pellet produced by captive fish Crow in Gainesville, fl(right; 25.6 mm long x 12.8
mm wide).
2012 M. Annala, E.A. Tillman, G.Backus, K.L. Keacher, and M.L. Avery 531
pellets have an obvious advantage in that large samples can be obtained without
killing any birds, and repeated sampling at a given roost site can provide a temporal
view of the food habits of the population (e.g., Platt 1956). There are several
limitations, however. For example, the specifi c foraging site cannot be known, so
it is not possible to relate the pellet contents to food availability; pellets largely
consist of grit and other non-food material; and as in our case, the species that
produced the pellet is not always known.
The winter crow population in Lancaster, PA consists of fish and American
Crows. The percent species composition is not known and several lines of
evidence produce disparate estimates. During our study, we captured 601 fish
Crows in modifi ed Australian crow traps compared to 46 American Crows (Avery
et al. 2008). Cannon-netting yielded 10 fish Crows and 4 American Crows.
Conversely, counts of calls from crows exiting Lancaster roosts in the morning
suggested approximately 90% were American Crows (M. Brittingham, Pennsylvania
State University, State College, PA, unpubl. data). Recent Christmas Bird
Count results indicate >90% of the Lancaster area crows were American Crows
(http://audubon2.org/cbchist/count_table.html).
Based on the size distribution of the pellets produced by the captive fish
Crows relative to our fi eld sample, we conclude that the vast majority of the pellets
we collected from Lancaster were produced by American Crows. Whether or
not this reflects the actual species composition of the winter crow population in
the Lancaster area is unknown.
Acknowledgments
Support was provided by the Berryman Institute, Mississippi State University (M. Annala
and G. Backus), and by the National Wildlife Research Center (E.A. Tillman, K.L.
Keacher, and M.L. Avery). W.E. Bruce assisted with animal care.
Literature Cited
Avery, M.L., E.A. Tillman, and J.S. Humphrey. 2008. Effi gies for dispersing urban crow
roosts. Vertebrate Pest Conference 23:84–87.
Barrows, W.B., and E.A. Schwarz. 1895. The Common Crow of the United States. Bulletin
6, Division of Ornithology and Mammalogy, US Department of Agriculture.
Washington, DC.
Below, T.H. 1979. first reports of pellet ejection in 11 species. Wilson Bulletin
91:626–628.
Berrow, S.D., T.C. Kelly, and A.A. Myers. 1992. The diet of coastal breeding Hooded
Crows, Corvus corone comix. Ecography 15:337–346.
Black, C.T. 1941. Ecological and economic relations of the crow, with special reference
to Illinois. Ph.D. Thesis, University of Illinois, Urbana, IL.
Coleman, J.S., and J.D. Fraser. 1987. Food habits of Black and Turkey Vultures in Pennsylvania
and Maryland. Journal of Wildlife Management 51:733–739.
Dean, W.R.J., and S.J. Milton. 2000. Directed dispersal of Opuntia species in the Karoo,
South Africa: Are crows the responsible agents? Journal of Arid Environments
45:305–314.
532 Northeastern Naturalist Vol. 19, No. 3
Errington, P.L. 1932. Technique of raptor food habits study. Condor 34:75–86.
Glading, B., D.F. Tillotson, and D.M. Selleck. 1943. Raptor pellets as indicators of food
habits. California fish and Game 29:92–121.
Kurosawa, R., R. Kono, T. Kondo, and Y. Kanai. 2003. Diet of Jungle Crows in an urban
landscape. Global Environmental Research 7:193–198.
McGowan, K.J. 2001. fish Crow (Corvus ossifragus). In A. Poole and F. Gill (Eds.). The
Birds of North America. No. 589. The Birds of North America, Philadelphia, PA.
Platt, D. 1956. Food of the crow, Corvus brachyrhynchos Brehm, in south-central Kansas.
University of Kansas Publications, Museum of Natural History 8:477–498.