This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
 Author's personal copy

Zoology 113 (2010) 131–139

Contents lists available at ScienceDirect

Zoology
journal homepage: www.elsevier.de/zool

Juvenile colour polymorphism in the red rock crab, Cancer productus: patterns,
causes, and possible adaptive significance
Jacqueline Krause-Nehring a,b,c,∗ , J. Matthias Starck a , A. Richard Palmer b,c
a
Department Biologie II, Ludwig-Maximilians-Universität München, Großhaderner Str. 2, D-82152 Planegg-Martinsried, Germany
b
Department of Biological Sciences, University of Alberta, Edmonton, AB T6G 2E9, Canada
c
Bamfield Marine Sciences Centre, 100 Pachena Road, Bamfield, BC V0R 1B0, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Juveniles of the common red rock crab of the Northeastern Pacific, Cancer productus, display a stun-
Received 1 April 2009 ning diversity of colours and patterns, while adults all have the same drab colouration. Although this is
Received in revised form 8 September 2009 widely known, key questions remain: (1) Does the frequency of different juvenile colours or patterns
Accepted 11 September 2009
vary among collection sites or seasonally? (2) Does juvenile colour polymorphism reflect genetic hetero-
geneity or phenotypic plasticity in response to variable environmental conditions? (3) Do juveniles of
Keywords:
different colours or patterns prefer substrata of different heterogeneity or brightness? We therefore: (i)
Crustacea
described the variation in colour and pattern of juvenile C. productus; (ii) tested for associations between
Cryptic colouration
Colour change
colour/pattern morphs and crab size, collection site, and season, in the field; (iii) conducted preliminary
Habitat preference tests for habitat preferences (background colour, substratum type, light level) of different colour/pattern
Predator avoidance morphs in laboratory experiments, and (iv) tested the effect of diet (mussels versus shrimp) and feeding
rate (high versus low) on juvenile colour/pattern. We describe 30 phenotypes that embrace a wide range
of colour and pattern variants. The proportions of these phenotypes did not vary significantly among four
collection sites, but they did vary significantly with season: over the summer and fall, juvenile colour
and pattern variation was gradually replaced by the uniform adult colouration. The number of crabs dis-
playing adult colouration also increased with crab size. Laboratory experiments suggest no significant
preferences of different juvenile morphs for different backgrounds, substrata, or light levels. Diet (mus-
sels versus shrimp) and feeding frequency had no effect on colour/pattern. Collectively, these results,
although limited in scope, are not consistent with two likely hypotheses that could explain the extensive
colour and pattern variation in juvenile C. productus: (i) selection for background matching by different
cryptic forms and (ii) direct effects of diet or feeding rate on colour or pattern. Most probably, the large
variety of different juvenile morphs is a result of frequency-dependent selection in which abundant vari-
ants are attacked disproportionately often and rarer forms are favoured. Juvenile colour polymorphism
in C. productus may reduce the vulnerability to visual predators, impede the formation of a search image,
and consequently decrease the risk of predation during the juvenile stages.
© 2010 Elsevier GmbH. All rights reserved.

1. Introduction By polymorphism, we mean merely a variety of forms within

Colour and pattern polymorphisms occur in many brachyuran
crabs (Palma et al., 2003), particularly in juveniles (Hogarth, 1978;
Bedini, 2002, 2006; Todd et al., 2006; Reuschel and Schubart, 2007)
or small-bodied species (Pessani and Tirelli, 2006). Despite being
rather widespread, surprisingly little is known about the causes
(genetic vs. environmental) and the adaptive significance of these
polymorphisms (but see Palma and Steneck, 2001).
∗ Corresponding author. Current address: Alfred-Wegener-Institut für Polar- und
Meeresforschung, Am Alten Hafen 26, D-27568 Bremerhaven, Germany.
E-mail address: jacqueline.krause-nehring@awi.de (J. Krause-Nehring).
a species (e.g., Wennersten and Forsman, 2009). Among crus-
taceans, different colours or colour patterns develop for different
reasons (Pessani and Tirelli, 2006). Polymorphism may be caused by
genetic differences between morphs or by environmental effects.
For example, the red and green forms of the hermit crab Ces-
topagurus timidus arise from a genetic polymorphism (Pessani and
Premoli, 1992), whereas diet (Jones et al., 1996; Davis et al., 2005;
Tlusty and Hyland, 2005) and colour of the growth environment
(Davis et al., 2005; Bedini, 2006) may induce different colours in
crayfish, lobsters, blue crabs, or pigmy crabs.
Although colour and pattern polymorphisms are presumably
adaptive, direct evidence of this is limited in crabs. Organisms avoid
detection or recognition by predators in various ways (Sherratt et
al., 2005). For example, camouflage is widespread among animal

0944-2006/$ – see front matter © 2010 Elsevier GmbH. All rights reserved.
doi:10.1016/j.zool.2009.09.002
 Author's personal copy
132 J. Krause-Nehring et al. / Zoology 113 (2010) 131–139
taxa (Thayer, 1909; Cott, 1940) and often involves body coloura-
tion (Stevens and Merilaita, 2009). Perhaps the most common form
of camouflage is background matching or crypsis where colours
and patterns of organisms match their backgrounds (Endler, 1981).
Another type of camouflage is disruptive colouration. Here, par-
ticular markings create false, or hide existing boundaries of the
organism to make detection of characteristic shapes more difficult
(Sherratt et al., 2005). Bold and high-contrast colours also break up
the prey’s body into apparently unconnected objects (Cuthill et al.,
2006).
Juveniles of Cancer productus (Randall, 1839), the common red
rock crab of the shallow-water, Northeastern Pacific (Alaska to Baja
California; Harbo, 2004), exhibit a particularly stunning diversity
of juvenile colour and pattern morphs, in sharp contrast to adults,
which have a nearly uniform red-brick colouration (Jensen, 1995).
As juveniles of this species adapt readily to laboratory conditions,
they offer an ideal opportunity to study the causes and potential
adaptive significance of juvenile crab polymorphism. First, we set
out to describe the remarkable diversity of juvenile colour/pattern
morphs in C. productus. Second, we described morph frequency
variation among several collection sites and as a function of crab
size and season. Third, we tested whether diet type or feeding fre-
quency in the laboratory had any effect on colour/pattern morph
changes from one moult to another. Finally, we tested whether
different juvenile morphs of C. productus exhibited different micro-
habitat preferences regarding background colouration, substratum
type, and light intensity under controlled laboratory conditions.
We hypothesised that: (i) darker juveniles would prefer a black
background while lighter crabs would prefer a white background,
(ii) uniformly coloured crabs would prefer sandy substrata while
those with distinct patterns would choose a heterogeneous gravel
background, and (iii) lighter crabs would be less attracted to lower
light intensities than darker-coloured crabs. These three experi-
ments therefore tested whether individual crabs could “recognise”
their own colour/pattern and chose substrata or lighting levels that
would increase camouflage against visual predators. If laboratory-
adapted juvenile crabs of different colours or patterns exhibited no
microhabitat preferences in laboratory tests, this would imply that
different juvenile morphs are not capable of choosing more cryptic
microhabitats in the field.
2. Materials and methods
2.1. Collection sites and scoring of crabs
A total of 163 juvenile C. productus were collected at low tide
in the intertidal zone from various sites near Bamfield, British
Columbia, Canada, on the west coast of Vancouver Island in May,
June, July, and August 2007. Sites were selected opportunisti-
cally depending on their apparent suitability for crabs. Among the
eight sites where juveniles were found, only four yielded sufficient
numbers for statistical analysis of among-site variation in morph
frequencies: Little Ohiat (latitude 48.852, longitude −125.182),
Diana Island (latitude 48.848, longitude −125.184), Sanford (lat-
itude 48.871, longitude −125.164), and Marble Cove (latitude
48.909, longitude −125.105).
Superficially, these four sites appeared to be similar, and juve-
niles were typically discovered buried in sandy or gravelly substrata
either underneath rocks or in eelgrass beds. The four sites were
situated on different islands in Barkley Sound and were 0.5–5 km
apart from one another. Crabs were exposed to different environ-
mental conditions at each of the four collection sites: Marble Cove
is an embayment of an extensive shallow bay; Little Ohiat is on
the protected side of a small island, less than 100 m from a deep,
open ocean channel; Diana Island is a rocky shoreline, also less than
100 m from the open ocean channel, but with a higher degree of
exposure than at the other three collection sites; and Sanford is an
open bay along the protected side of an island. Three out of the
four collection sites have a small sandy beach that is surrounded
by the rocky intertidal. On Sanford, crabs were mainly collected at
the fringe between the sandy beach and rocky shoreline. At Marble
Cove, crabs were sometimes sampled in the latter area, but mostly
in crevices within the rocky intertidal further away from the beach.
At Little Ohiat, no crabs were found near the beach, but exclusively
hiding in rocky crevices. No sandy beach was in close proximity
to the collection site on Diana Island. Here, crabs were sampled
from little tide-pools and crevices underneath eelgrass beds. No
large eelgrass beds were present at the other three collection sites.
Instead, rocky crevices and boulders were often covered with var-
ious species of seaweed (including crustose and coralline algae)
at Marble Cove and Little Ohiat. At Sanford, crabs were usually dis-
covered walking on sand in between submersed boulders and rocks
close to the rocky shoreline.
Upon collection, crabs were transported to the laboratory.
Each crab was assigned a number, measured for maximum
carapace width, and housed and fed separately in individu-
ally labelled, porous plastic containers (14 cm × 14 cm × 10 cm
or 12 cm × 12 cm × 12 cm). Containers were held in sea tables
(168 cm × 71 cm × 23 cm) with running, unfiltered seawater
(9–11 ◦ C; 29–31 ppt salinity) and oxygen supply, and were cleaned
once a week. To document the different morphs, each crab was
photographed (Canon Digital IXUS 65, 6.0 Mpix; Canon Inc., Tokyo,
Japan) under water. To achieve equal reflection from all sides, each
picture was taken in a circular white bucket (diameter: 27 cm;
height: 24 cm) filled with seawater. The bottom of the bucket
was covered with a blue piece of fabric to reduce reflection (by
providing a rough surface) and absorption (by a blue background).
Two strings of blue fabric were attached to the bottom, placed
between the legs and the carapace on each side of the crab and
used to restrain each crab while being photographed. In the few
cases where a crab moulted in the laboratory before a picture could
be taken, the first picture after moulting was used for analysis.
Pictures of all 163 individuals were grouped according to colour
and pattern into six colour and five pattern groups (Fig. 1) which
yielded a total of 30 morph categories plus one category called
“other”. To assess the reliability of these categorisations, five inde-
pendent people were asked to assign each of the 163 crab pictures
to one of the 31 morph categories. The five data sheets were then
combined into one sheet that listed for each crab the five categories
which it had been assigned to by the five observers. The category
chosen by at least two of the five observers was then used as the
morph category for each individual crab. Among the 163 crabs, 81
were assigned to the same morph category by at least four of the
five observers. Where crabs were assigned to different categories,
they were almost invariably assigned to an adjacent, rather similar
category (e.g., morphs 12 vs. 17, or 11 vs. 12; Fig. 1). Since simi-
lar morph categories (those most prone to scoring disagreements)
were pooled for statistical analyses (see next paragraph), scoring
disagreements likely had little effect on the results. The goal here
was to try to minimise the possible bias that might arise if scoring
was done only by a single individual.
To facilitate statistical analysis, and to test for more spe-
cific associations, the 31 different colour/pattern morphs were
further pooled into three overall brightness (greyscale) groups,
three colour groups, and three pattern groups. First, to obtain a
quantitative measure of overall brightness, colour images were
converted into eight-bit grey level images using Adobe Photo-
shop 7.0 and a mean greyscale value (MGSV) between 0 (pure
black) and 255 (pure white) was computed for each image. Three
greyscale groups were created by dividing this range into thirds:
“dark”: 0 < MGSV ≤ 83.33, “intermediate”: 83.33 < MGSV ≤ 166.66,
 Author's personal copy
J. Krause-Nehring et al. / Zoology 113 (2010) 131–139
Fig. 1. Representative images of all morph categories observed among 163 C. productus juveniles. The combination of six colour types (=rows) and five pattern types
(=columns) yielded 30 morph categories. Some boxes do not contain a picture either because no sampled crab fit into the specific category or because the category was too
diverse (=“other” as an extra category, see Section 2.1).
and “light”: 166.66 < MGSV ≤ 255. Second, crabs were pooled into
three colour groups as “red and wine” (morphs 11–20; Fig. 1),
“brown and orange” (morphs 1, 21–30; Fig. 1), or “whitish” (morphs
6–10; Fig. 1). Finally, crabs were pooled into three pattern groups
as “nearly solid or one large spot” (first and last columns; Fig. 1),
“irregular patches” (column 2; Fig. 1), or “parallel lines or maze”
(columns 3 and 4; Fig. 1).
2.2. Effects of crab size, collection site, and sampling season
Contingency table analysis was used to test for associa-
tions between juvenile morphs and crab size, collection site,
and season of collection. A frequency distribution of the sizes
of all sampled crabs was graphed using NCSS version 2007
(NCSS, Kaysville, UT, USA) and used to assign crabs to size
classes according to carapace width (CW) for subsequent anal-
ysis: CW ≤ 17 mm, 17 mm < CW ≤ 27 mm, 27 mm < CW ≤ 37 mm,
37 mm < CW ≤ 45 mm, 45 mm < CW ≤ 55 mm or CW > 55 mm. Sites
from which fewer than 10 individuals were collected were
excluded from analysis.
2.3. Effects of diet type and feeding rate on colour changes
In total, 144 crabs were held in the laboratory for 7 months (May
to November 2007) to document colour and pattern changes from
one moult to the next. Each crab was housed and fed in a sepa-
rate box to prevent interaction between replicate crabs. Ten days
after each moult, pictures were taken again. This period allowed for
hardening of the exoskeleton and recovery of the colours. Pre- and
post-moult pictures of 89 moulted crabs were visually compared
133
and colour/pattern changes assigned to one of six colour-change
categories (Fig. 2) for statistical analysis: (i) increase in red (legs
first), (ii) increase in colour intensity (same colour), (iii) change to
grey (or transparent), (iv) fading of colours (same colour), (v) same
colour, and (vi) colour change.
All 144 crabs were randomly assigned to one of two diet type
groups. Half of the crabs were fed mussels (Mytilus trossulus at
0.40 g wet weight per crab) while the other half was fed shrimp
(Pandalus sp. at 0.26 g wet weight per crab). Mussels and shrimp
were chosen because a sufficient and constant supply of both diets
could be maintained over several months. Food portions were cho-
sen to feed each crab approximately the same weight of flesh: 0.40 g
of mussel including the shell contains the same amount of flesh as
0.26 g of shrimp including the shell. The shell was not removed from
either food type so that the flesh would sink and be discovered by
the crab. Since the number of moulting events decreased after 5
months, food portions were doubled to ensure that these larger
crabs received enough food. Feeding frequencies were kept the
same. Each diet type group was separated into high- and low-food
subgroups, yielding a total of four feeding groups (=mussel once
a week (N = 36 crabs), mussel twice a week (N = 41 crabs), shrimp
once a week (N = 34 crabs), shrimp twice a week (N = 33 crabs)).
Contingency-table analysis was used to test for associations
between colour changes in moulted juveniles and type of diet or
feeding frequency.
2.4. Microhabitat choice experiments
One-hundred and forty out of 144 C. productus juveniles kept
in the laboratory were exposed to trials to test for morph-related
 Author's personal copy
134 J. Krause-Nehring et al. / Zoology 113 (2010) 131–139
Fig. 2. Six colour change categories observed among C. productus juveniles that
moulted in the laboratory over a 7-month time period. Photographs on the left
were taken prior to moulting; photographs on the right were taken after moulting
(number of moults between both pictures = 1). The six colour change categories
were (from top to bottom): “increase in red” (x = 10), “increase in colour intensity”
(x = 97), “change to grey” (x = 76), “fading of colours” (x = 25), “same colour” (x = 97),
and “colour change” (x = 12); x = number of days between left and right image.
background preferences. Some crabs died prior to running one of
the experiments and were therefore not included in the data set
(different overall N for experiments 1–3). Carapace width ranged
from 6.9 mm to 59.2 mm. Each crab was tested only once in each of
the three experiments, except for five crabs that were each tested
five times to assess the repeatability of their choices in each exper-
iment.
2.4.1. Experiment 1: white versus black background
The first experiment tested preferences of different juvenile
morphs for light or dark background. The experimental arena con-
sisted of an open box (43 cm × 25 cm × 18 cm) that was painted half
white and half black on the inside with non-toxic, edible, water-
proof acrylic paint and placed in the same location for all trials in
a sea table partially filled with running seawater. Standing seawa-
ter in the arena was changed prior to each trial. Trials were run
with the room lights on to maintain light conditions as consistent
as possible. The orientation of the box relative to the sea table was
varied at random among trials. Only one crab at a time was tested
in each arena to prevent crabs from influencing each other’s choice.
At the beginning of each trial the midline of the crab was placed on
the middle line of the box (i.e., legs on one side of the body were
on black while legs on the other side were on white). The crab’s
location (white or black side) was recorded after 0, 1, 2, 5, 10, 15,
20, 30, 40, 50, 65, 80, 95, 110, and 125 min.
To assess the consistency of the behaviour, five individual crabs
were tested five times. Repeat trials were run to ensure that the
experimental setup was appropriate to test for behavioural pref-
erences and to be able to exclude flaws in the setup as possible
causes for negative results. Trials were run once per day to min-
imise the likelihood that crabs might get habituated, or that other
factors might bias a crab’s behaviour.
To simplify analysis, only initial (=the first side a crab explored)
and final choices (=where a crab spent most of the time during
the last hour of the 2-h trial) were tested statistically. In some
cases crabs never moved from one side to the other side of the
box after their initial choice. If an individual never explored both
backgrounds, the final location of the crab could not be considered
a fully informed “final choice”. In such cases only initial, but not
final choices were used in the analysis (Ninitial is not necessarily
Nfinal ). Contingency-table analysis was used to test for associations
between behaviour and morphs (greyscale group, colour group, or
pattern group).
2.4.2. Experiment 2: sand versus gravel substrata
The second experiment tested for morph-related preferences
for a sandy or gravelly substratum. For these trials, the experi-
mental arena consisted of an open box covered half with sandy
and half with gravelly substratum (Fig. 3). Although these two
substrata likely differed in several ways (background colour, back-
ground heterogeneity, grain size, chemical cues, textural cues),
this experiment nonetheless allowed us to ask whether different
morphs responded at all differently to substratum type. Exper-
imental setup, procedures, and analysis, including repeat trials,
were the same as for habitat experiment 1.
For both experiments 1 and 2, three replicate arenas were held
in the same seawater table and used to run trials concurrently. As in
experiment 1, five individual crabs were tested five times to assess
the repeatability of their behaviour.
2.4.3. Experiment 3: light intensity
The third experiment tested for possible morph-related pref-
erences for different light intensities. The test arena consisted of a
Fig. 3. Experimental arena (43 cm × 25 cm × 18 cm) with half the bottom covered
with sand, the other half covered with gravel to test for substratum preferences
among different C. productus juvenile morphs.
 Author's personal copy

J. Krause-Nehring et al. / Zoology 113 (2010) 131–139 135

box (18.25 cm × 63.5 cm × 30.5 cm) divided into three equally large Table 1
Prevalence of juvenile vs. adult colouration as a function of carapace width in C.
compartments (18.25 cm × 21.2 cm × 30.5 cm) and was built from
productus. Larger crabs were more likely to exhibit adult colouration (P < 0.05, Chi-
corrugated plastic sheet. The three compartments were separated square = 22.5, DF = 5, N = 163). By approx. 40 mm carapace width, more than half the
by walls to minimise light flux between compartments, but small crabs exhibited adult colouration.
openings at the bottom of the walls allowed crabs to walk freely
Carapace width Morph category
from one compartment to another. The tops of two compartments
were covered with a white lid to reduce light intensity and one of Juvenile Adult
these was additionally covered with black plastic to reduce light CW ≤ 17 mm 7 0
conditions further. Finally, the walls were wrapped in black plastic 17 mm < CW ≤ 27 mm 27 8
to restrict light entry from the sides and to keep light conditions in 27 mm < CW ≤ 37 mm 39 11
37 mm < CW ≤ 45 mm 20 21
each division as consistent as possible. Little doors in the lids were 45 mm < CW ≤ 55 mm 10 15
used to check the location of each crab during the experiment. The CW > 55 mm 2 3
three light intensities, measured with a photometer in each com-
partment of the box, were on average: full light 232.50 lx (±44.56
SD, N = 8; =100%),
ˆ medium light 57.00 lx (±6.50 SD, N = 8; ≈25%), Chi-square = 18.0, DF = 6). The proportion of lined or maze-
and low light 4.00 lx (±0.53 SD, N = 8; ≈2%). About 7 cm of fresh patterned crabs tended to decrease over the summer, but those
seawater was filled into the sea table prior to each trial and sea- exhibiting a patchy pattern of colouration increased (P = 0.022, Chi-
water was fully replaced between trials. Trials were run with the square = 14.84, DF = 6).
room lights on to keep light conditions as consistent as possible.
Eight replicate arenas were constructed so eight independent tri-
3.2. Effects of diet type and feeding rate
als could be run in parallel. At the beginning of each trial a crab was
placed into the compartment with full light conditions (the only
The change from juvenile to adult colouration appeared to be
one without a lid). The crab’s location (full, medium, low light) was
gradual rather than by one moult. Furthermore, legs seemed to
recorded after 0, 10, 20, 30, 45, 60, 75, 90, 105, and 120 min. Crabs
turn red before the carapace did. However, the number of observed
that did not explore all three compartments of the box throughout
moults was not large enough for a rigorous statistical test. Other
the entire trial were excluded from analysis.
colour changes were also observed in crabs that moulted in the lab-
To simplify analysis, only the final choice was taken into account.
oratory (Fig. 2). In some cases colour intensity increased, in others
Also, because each trial was started in the compartment with full
it decreased, and sometimes it remained the same (Fig. 2).
light conditions, no initial choice could be observed. Contingency-
table analysis was used to test for associations between final
light-level choice and colour/pattern morphs (greyscale, colour, or
pattern group). As in experiments 1 and 2, five individual crabs were
tested five times to assess the repeatability of their behaviour. Con-
sistency was assessed with choices of full and intermediate light
conditions pooled as one outcome and low light conditions as the
second outcome because few crabs chose the intermediate or full
light level.

3. Results

Juvenile C. productus showed a wide variety of colour and pat-

tern phenotypes. These phenotypes could be grouped into six major
colour types and five major pattern types that, together, yielded 30
potential combinations of colour and pattern (Fig. 1). However, of
these 30 potential colour/pattern phenotypes, only 18 were actu-
ally observed (Fig. 1); additionally, we had to form an extra category
of “other” phenotypes (N = 1) which could not be assigned to one of
the latter 30 categories.

3.1. Variation with crab size, collection site, and sampling season

The number of crabs exhibiting adult colouration increased sig-

nificantly with increasing crab size (Table 1). The average size
at which at least 50% of field-collected juveniles displayed adult
colouration was 40.42 mm (±10.21) carapace width.
The frequencies of overall brightness (greyscale score), colour
type, and pattern type varied among the four collection sites (Fig. 4).
Fig. 4. Numbers of juvenile C. productus of three different brightness (A), colour
However, these differences were only significant for overall bright-
(B), and pattern (C) groups collected from four different study sites. Dark: 0 < mean
ness (P = 0.015, Chi-square = 15.7, DF = 6). Among brightness groups, greyscale value (MGSV) ≤ 83.33 (e.g., morph category 16 or 18); intermediate:
crabs with an intermediate greyscale value were proportionally 83.33 < MGSV ≤ 166.66 (e.g., morph category 1 or 27); light: 166.66 < MGSV ≤ 255
more common at Little Ohiat. (e.g., morph category 6 or 26). Red/wine: e.g., morph category 16 or 18;
Further analysis revealed significant correlations between brown/orange: e.g., morph category 27 or 28; whitish: e.g., morph category 6 or
9. Solid/spot: nearly solid carapace or one large spot (e.g., morph category 1 or 10);
sampling season and juvenile morph groups (Fig. 5). The pro-
patches: irregular patches (e.g., morph category 12 or 27); lines/maze: parallel lines
portions of whitish C. productus decreased over the summer but or maze (e.g., morph category 28 or 29). See Fig. 1 for photographs of each morph
the proportions of red/wine coloured crabs increased (P = 0.006, category.
 Author's personal copy

136 J. Krause-Nehring et al. / Zoology 113 (2010) 131–139

Fig. 6. Numbers of C. productus juveniles exhibiting six different types of colour

changes from one moult to the next after having been fed two different kinds of diet
(mussels or shrimp) and at two different rates of feeding (once or twice a week). See
Fig. 2 for representative photographs of each colour change category.

independent observations because most crabs thoroughly explored

Fig. 5. Numbers of juvenile C. productus of three different brightness (A), colour (B),
and pattern (C) groups collected at four different times during the spring and sum-
mer of 2007. Dark: 0 < mean greyscale value (MGSV) ≤ 83.33 (e.g., morph category
16 or 18); intermediate: 83.33 < MGSV ≤ 166.66 (e.g., morph category 1 or 27); light:
166.66 < MGSV ≤ 255 (e.g., morph category 6 or 26). Red/wine: e.g., morph category
16 or 18, brown/orange: e.g., morph category 27 or 28; whitish: e.g., morph category
6 or 9. Solid/spot: nearly solid carapace or one large spot (e.g., morph category 1 or
10); patches: irregular patches (e.g., morph category 12 or 27); lines/maze: parallel
lines or maze (e.g., morph category 28 or 29). See Fig. 1 for photographs of each
morph category.
Colour change was independent of both diet type (P = 0.91,
Chi-square = 1.56, DF = 5) and feeding frequency (P = 0.28, Chi-
square = 6.3, DF = 5) (Fig. 6). Similar ratios of crabs being fed mussels
to crabs being fed shrimp were observed across all colour change
groups. In general, crabs that were fed twice a week seemed to
moult more often than crabs that were fed only once a week.
This was observed for four out of the six different kinds of colour
changes.
3.3. Habitat choice experiments
the experimental arena. Those trials in which an individual moved
immediately to one region of an experimental arena and remained
there were excluded from the analysis of final choices; this explains
why final choice sample sizes are smaller than initial choice sample
sizes in Tables 2 and 3.
3.3.1. Experiment 1: white versus black background
Statistical analysis revealed that both initial and final back-
ground choices were independent of carapace brightness (Table 2),
colour type (initial choice: P = 0.41, Chi-square = 1.76, DF = 2; final
choice: P = 0.75, Chi-square = 0.58, DF = 2), and pattern type (initial
choice: P = 0.32, Chi-square = 2.28, DF = 2; final choice: P = 0.37, Chi-
square = 1.98, DF = 2; data for colour and pattern groups not shown).
However, all groups preferred the black background more often
than the white background.
3.3.2. Experiment 2: sand versus gravel substrata
Juvenile C. productus exhibited no significant preferences dur-
ing initial and final choices of substratum types depending on
their overall carapace brightness score (Table 3), colour type (ini-
tial choice: P = 0.091, Chi-square = 4.61, DF = 2; final choice: P = 0.17,

In order to determine whether individual crabs moved at Table 2

Numbers of C. productus juveniles of different carapace brightness levels choosing to
random or exhibited consistent behaviours under experimental
move to a light or dark background when first introduced into an experimental arena
conditions in the laboratory, we first analysed only the subset of and during the last hour of the experiment. Carapace brightness did not affect either
crabs that were tested multiple times in the same experiment. Anal- initial (P = 0.54, Chi-square = 1.25, DF = 2; overall N = 129) or final choice (P = 0.41,
ysis of all these repeat trials together suggests that individual crabs Chi-square = 0.20, DF = 2; overall N = 90) of white versus black background. Neither
did not select background or light levels at random. Among 25 indi- initial (=the first side a crab explored) nor final choices (=where a crab spent most
of the time during the last hour of the 2-h trial) differed significantly from random
vidual tests (five each for initial and final choices in experiments (P > 0.05; Fisher’s Exact Test).
1 and 2, and five final choices in experiment 3), three crabs made
the same choice in all five trials, 13 made the same choice four Carapace brightness (MGSV) Initial choice of Final choice of
background background
out of five trials, and nine made the same choice three out of five
trials (P < 0.025, Chi-square test versus the frequencies expected White Black White Black
from a binomial distribution, DF = 2). Therefore, tests for different Dark (0 < MGSV ≤ 83.33) 12 22 10 17
responses to experimental conditions by different morphs were Intermediate (83.33 < MGSV ≤ 166.66) 33 42 19 29
justified. Light (166.66 < MGSV ≤ 255) 10 10 5 10
Total 55 74 34 56
In the analysis of habitat choice experiments 1 and 2, initial and
final choices by an individual crab in a single trial were treated as MGSV, mean greyscale value.
 Author's personal copy

J. Krause-Nehring et al. / Zoology 113 (2010) 131–139 137

Table 3 width (Table 1), well below that of sexual maturity (Orensanz and
Numbers of C. productus juveniles of different carapace brightness levels choosing to
Gallucci, 1988). C. maenas exhibits a similar pattern: most juveniles
move to different substrata when first introduced into an experimental arena and
during the last hour of the experiment. Carapace brightness did not affect either transform to adult colour by about 25 mm carapace width, although
initial (P = 0.085, Chi-square = 4.93, DF = 2; overall N = 128) or final choice (P = 0.11, the size at which this happens varies among habitats (Hogarth,
Chi-square = 4.35, DF = 2; overall N = 92). Unlike final choices (=where a crab spent 1978).
most of the time during the last hour of the 2-h trial; P < 0.05; Fisher’s Exact Test),
initial choices (=the first side a crab explored) were not significantly different from
random (P > 0.05; Fisher’s Exact Test).
4.2. Stability of juvenile colour polymorphism

Carapace brightness (MGSV) Initial choice of Final choice of
substratum type substratum type
Sand Gravel Sand
Dark (0 < MGSV ≤ 83.33) 18 16 10
Intermediate (83.33 < MGSV ≤ 166.66) 44 32 15
Light (166.66 < MGSV ≤ 255) 15 3 1
Total 77 51 26
MGSV, mean greyscale value.
Chi-square = 3.49, DF = 2), and pattern type (initial choice: P = 0.28,
Chi-square = 2.57, DF = 2; final choice: P = 0.054, Chi-square = 5.83,
DF = 2; data for colour and pattern types not shown).
3.3.3. Experiment 3: light intensity
Preferences for different light intensities by juvenile C. productus
were also independent of overall brightness score (Fig. 7; P = 0.58,
Chi-square = 2.85, DF = 4), colour type (P = 0.72, Chi-square = 2.08,
DF = 4), and pattern type (P = 0.34, Chi-square = 4.49, DF = 4), but all
crabs in all morph groups preferred the lowest light condition.
4. Discussion
4.1. Diversity and ontogeny of variation
Juvenile C. productus exhibit a remarkable range of both colours
and patterns (Figs. 1 and 2). In addition to the great diversity we
report here, other unusual phenotypes have been observed as well,
including a solid black carapace with a white anterior margin and
white legs and chelae, as well as an entire body that is nearly uni-
form, pearly white (ARP, pers. obs.). So, other morphs undoubtedly
also exist. Other studies have reported a large diversity of morphs
among juvenile crabs, including Cancer irroratus (Cancridae) (Palma
and Steneck, 2001), Carcinus maenas (Portunidae) (Hogarth, 1978;
Todd et al., 2005, 2006), Xantho poressa (Xanthidae) (Bedini, 2002;
Reuschel and Schubart, 2007), and Sirpus zariquieyi (Pirimelidae)
(Bedini, 2006). Unfortunately, juvenile colour polymorphism has
only been studied systematically in a few species, thus it is hard to
say how unusual the diversity of C. productus is in this regard.
The transition from juvenile to adult colouration is gradual in
C. productus, and appears to require from two to four moults to
be complete. This transition takes place at around 40 mm carapace
In our experiments, diet type (mussels versus shrimp) had virtu-
ally no effect on C. productus juvenile colouration (Fig. 6), suggesting
Gravel that carapace pigment components are not derived directly from
18 prey tissues. Hogarth (1978) also found that juvenile C. maenas
34 morphs were generally “stable at ecdysis” when reared in the lab-
14 oratory. This contrasts with studies of other decapods where diet
66
may have a large effect on colouration in freshwater crayfish (Jones
et al., 1996) and American lobsters (Tlusty and Hyland, 2005). Too
few studies have been done to determine whether these differences
reflect a fundamental difference in how carapace pigments are pro-
duced in astacidean versus brachyuran decapods. As we tested for
the effects of diet on juvenile carapace colour using only two types
of food, we also cannot exclude the possibility that other food
might influence juvenile morph variety in C. productus. However,
filter-feeding molluscs and scavenging shrimp contain rather dif-
ferent complements of pigments (Fox, 1979). In particular, elevated
carotenoid levels in the shrimp diet would have been expected to
increase levels of red pigments, which we did not observe. Hence
the absence of any strong differences between juveniles fed such
different prey suggests that the effects of diet-derived pigments on
colour are likely minimal in C. productus.
Feeding rate did appear to have some effect on the colour
changes observed in C. productus as more individuals “changed
to grey” when fed twice per week (Fig. 6). But this effect may
have had more to do with habitat than diet. Other crabs appear to
modify carapace colour in response to different-coloured habitats.
Hogarth (1978) suggested that juvenile C. maenas might be able
to alter colour to suit their background. Callinectes sapidus do alter
brightness, hue, and saturation of the carapace after moulting in
experimental arenas of different background, but the effects were
rather small and did not significantly affect survivorship (Davis et
al., 2005). And in the pygmy crab Sirpus, individuals also appear to
be able to adjust the colour and distribution of chromatophores to
be more cryptic in the different habitats in which they live (Bedini,
2006). We did not specifically test the effects of habitat colouration
on carapace colouration in laboratory-reared C. productus. How-
ever, crabs that were fed more also moulted more often over the
course of the study. The average number of moults per crab for
each of the four treatments over 7 months was 0.62 moults/crab
(fed mussels once per week), 1.03 moults/crab (fed mussels twice
per week), 1.11 moults/crab (fed shrimp once per week), and
1.26 moults/crab (fed shrimp twice per week). Therefore, colour
changes due to habitat effects among laboratory-reared juveniles
may have been more pronounced in the more frequently fed group,
which may have nothing to do with diet composition.

4.3. Adaptive significance of juvenile colour polymorphism

Ontogenetic colour change occurs in many animal groups,

Fig. 7. Final choices of light intensity observed for C. productus juvenile crabs
of three different brightness groups during the last hour of each trial. Dark:
0 < mean greyscale value (MGSV) ≤ 83.33; intermediate: 83.33 < MGSV ≤ 166.66;
light: 166.66 < MGSV ≤ 255.
where it is commonly associated with differences in habitat or
predation regimes experienced by juveniles compared to adults
(Booth, 1990; Todd et al., 2006). Distributional differences are likely
due, at least in part, to differences in active habitat choice by smaller
and larger crabs (Richards, 1992). Furthermore, defences against
predation may especially be important early in life when mor-
tality is high (Palma and Steneck, 2001). Visual predators of C.
productus include dogfish, halibut, sculpins, wolf eels, and octopi
(Fisheries and Oceans Canada, 2007) as well as rockfish (Knudsen,
 Author's personal copy
138 J. Krause-Nehring et al. / Zoology 113 (2010) 131–139
1964) and mink (ARP, pers. obs.). Birds also prey on C. productus,
but their predation is restricted to intertidal juveniles (Orensanz
and Gallucci, 1988). Therefore, bird predation on C. productus juve-
niles should also be considered when studying habitat preferences
of this species.
The advantage of polymorphism in C. productus appears to be
strongly size-dependent, as the polymorphism all but disappears
in adults. Smaller crabs (<40 mm CW) are undoubtedly vulnera-
ble to a much greater diversity of predators, particularly visual
predators such as fish and birds. Significantly, it is in larger-bodied
crabs, including some Cancer species (Palma and Steneck, 2001; this
study) and C. maenas (Hogarth, 1978; Todd et al., 2006) where the
polymorphism is limited to juveniles. In two small-bodied crabs,
Xantho (CW < 15 mm) (Reuschel and Schubart, 2007) and Sirpus
(CW < 10 mm) (Bedini, 2006), polymorphism persists into the adult
stage, although, intriguingly, the polymorphism is less pronounced
in adults of the somewhat larger-bodied X. poressa (Reuschel and
Schubart, 2007).
We cannot say with much conviction whether or not
the observed juvenile colour polymorphism in C. productus is
maintained by selection for crypsis associated with different micro-
habitats, because we did not quantify microhabitat characteristics
where individual crabs were collected. In general, juveniles were
discovered in similar conditions: buried in sandy or gravelly sub-
strata either underneath rocks or in eelgrass beds. However, two
lines of evidence suggest that selection for crypsis/background
matching in heterogeneous environments is a less likely hypothe-
sis to explain this extensive polymorphism. First, the four primary
sites from which crabs were collected were on different islands and
likely were characterised by different environmental conditions, so
the morph frequencies (Fig. 4) seem much more similar than would
be expected given that the relative proportions of different micro-
habitats clearly varied among these sites (see Section 2.1). In other
crabs, such as Carcinus, the frequency of juvenile colour morphs
does vary among microhabitats: plain crabs occur more commonly
in macroalgae whereas patterned crabs are more abundant in mus-
sel beds (Hogarth, 1978; Todd et al., 2006). Thus, a more systematic
field survey might reveal a stronger association with habitat types
in C. productus. Second, in all three habitat choice experiments,
only substrata or light levels were varied in the testing arenas,
which were otherwise featureless. No refuges were provided so
that crabs would experience maximal risk (no cover) and therefore
be more likely to choose a risk-minimising substratum. Despite
these higher-risk testing conditions, laboratory-adapted juvenile
crabs of different colour or pattern morphs showed none of the
expected behavioural preferences in laboratory experiments (Fig. 7
and Tables 2 and 3). In addition, in the substratum choice experi-
ment 2, the two substratum types (Fig. 3) differed in several ways
(background colour, background heterogeneity, grain size, textural
cues, and possibly chemical cues), yet no differences in substra-
tum choice were apparent among different juvenile morphs. So,
if individual crabs are not “aware” of what colour or pattern they
are, it is hard to imagine how they could behave in such a way as
to choose microhabitats that ensure maximum crypticity. Future
habitat choice tests would be stronger if conducted shortly after
crabs were collected so that they had less time to accommodate
to laboratory conditions, and with better control of substratum
variables.
Selection for disruptive colouration might explain the juvenile
colour polymorphism in C. productus, because disruptive coloura-
tion would clearly be beneficial in a chromatically heterogeneous
environment (Cuthill et al., 2006; Reuschel and Schubart, 2007) that
varies in colour, pattern, and lightning, or for animals that move
around in their environment (Cott, 1940). Background matching
is rather unlikely to fully prevent predator detection, because the
outline of the animal may still be easily detected and reveal the
animal’s presence (Thayer, 1909). However, camouflage may be
achieved through disruptive colouration (Thayer, 1909). A combi-
nation of markings breaks the true outline of the animal and creates
false edges and boundaries (Stevens and Merilaita, 2009). Such
disruptive markings hinder detection or recognition of prey and
help decrease the risk of detection (Stevens and Merilaita, 2009),
and many of the morphs we observed were clearly disruptively
coloured (Figs. 1 and 2). Highly contrasting colours should be more
disruptive than less contrasting colours (Cuthill et al., 2005), as
bold and high contrast colours help break up the prey’s body into
apparently unconnected objects (Cuthill et al., 2006). However, a
significant fraction of all juveniles we collected exhibited a solid
or nearly solid colour pattern (Fig. 4), so disruptive colouration
seems unlikely to be the primary adaptive advantage of juvenile
polymorphism in C. productus.
Frequency-dependent selection (Allen, 1988) in which abun-
dant variants are attacked disproportionately often and rarer
forms are favoured remains the most appealing hypothesis to
account for juvenile colour polymorphism in C. productus. Nega-
tive frequency-dependent selection is one of the few mechanisms
that can promote substantial within-population polymorphism
(Poulton, 1890; Allen, 1988; Bond and Kamil, 2002; Wennersten
and Forsman, 2009). Visual predators such as fish and birds are
likely to develop a search image based on previous prey; thus, they
would be more likely to miss the other morphs within a poly-
morphic population that do not fit the search image (Krebs and
Davies, 1993). Frequency-dependent selection, therefore, offers the
simplest explanation for (i) the great diversity of juvenile colours
and patterns we observed in C. productus juveniles, (ii) the mini-
mal morph-frequency variation observed among habitats, and (iii)
the absence of ‘cryptic’ behaviours by individual crabs of different
colour or pattern.
Acknowledgments
We thank the German Academic Exchange Service (DAAD),
the Ludwig-Maximilians-Universität München, the University of
Alberta, and NSERC Canada (Discovery Grant A7245 to A.R.P.) for
funding this project. Special thanks to C. Gruman for help with
catching and caring for crabs, as well as to C. Bergstrom, R. Wyeth,
C. Neufeld, H. Parkis, Q. Wong, S. Tyne, K. Irmscher, T. Ridder, and
the staff of the Bamfield Marine Sciences Centre for advice and dis-
cussion. Additional thanks go to E., G., and C. Krause-Nehring. We
are also grateful to Chris Todd and an anonymous referee who both
provided thoughtful and detailed comments that greatly improved
the paper.
References
Allen, B., 1988. Frequency-dependent selection by predators. Philos. Trans. R. Soc.
London B. 319, 485–503.
Bedini, R., 2002. Colour change and mimicry from juvenile to adult: Xantho poressa
(Olivi, 1792) (Brachyura, Xanthidae) and Carcinus maenas (Linnaeus, 1758)
(Brachyura, Portunidae). Crustaceana 75, 703–710.
Bedini, R., 2006. Polymorphism of colour patterns related to environmental fac-
tors: correlation between habitat, sight, and mimicry in Sirpus zariquieyi Gordon,
1953. Crustaceana 79, 53–67.
Bond, A.B., Kamil, A.C., 2002. Visual predators select for crypticity and polymorphism
in virtual prey. Nature 415, 609–613.
Booth, C.L., 1990. Evolutionary significance of ontogenic color-change in animals.
Biol. J. Linn. Soc. 40, 125–163.
Cott, H.B., 1940. Adaptive Coloration in Animals. Methuen & Co. Ltd, London, UK.
Cuthill, I.C., Stevens, M., Sheppard, J., Maddocks, T., Parraga, C.A., Troscianko, T.S.,
2005. Disruptive coloration and background pattern matching. Nature 434,
72–74.
Cuthill, I.C., Stevens, M., Windsor, A.M.M., Walker, H.J., 2006. The effects of pat-
tern symmetry on detection of disruptive and background-matching coloration.
Behav. Ecol. 17, 828–832.
Davis, J.L.D., Eckert-Mills, M.G., Young-Williams, A.C., Hines, A.H., Zohar, Y.,
2005. Morphological conditioning of a hatchery-raised invertebrate, Callinectes
sapidus, to improve field survivorship after release. Aquaculture 243, 147–158.
 Author's personal copy
J. Krause-Nehring et al. / Zoology 113 (2010) 131–139
Endler, J.A., 1981. An overview of the relationship between mimicry and crypsis.
Biol. J. Linn. Soc. 16, 25–31.
Fisheries and Oceans Canada, 2007. Fact Sheet: Dungeness, Red Rock and Tanner
Crab. Fisheries and Oceans Canada, Vancouver, British Columbia.
Fox, D.L., 1979. Biochromy. Natural Coloration of Living Things. University of Cali-
fornia Press, Berkeley, CA.
Harbo, R.M., 2004. From Whelks to Whales. Harbour Publishing, Madeira Park.
Hogarth, P.J., 1978. Variation in carapace pattern of juvenile Carcinus maenas. Mar.
Biol. 44, 337–343.
Jensen, G.C., 1995. Pacific Coast Crabs and Shrimp. Sea Challengers, Monterey, CA.
Jones, P.L., DeSilva, S.S., Mitchell, B.D., 1996. The effect of dietary protein source
on growth and carcass composition in juvenile Australian freshwater crayfish.
Aquacult. Int. 4, 361–376.
Knudsen, J.W., 1964. Observations of the reproductive cycles and ecology of common
Brachyura and crab-like Anomura of Puget Sound. Pacific Sci. 18, 3–33.
Krebs, J.R., Davies, N.B., 1993. An Introduction to Behavioural Ecology. Blackwell
Science Ltd., Oxford.
Orensanz, J.M., Gallucci, V.F., 1988. Comparative study of postlarval life-history
schedules in four sympatric species of Cancer (Decapoda, Brachyura, Cancridae).
J. Crustac. Biol. 8, 187–220.
Palma, A.T., Steneck, R.S., 2001. Does variable coloration in juvenile marine crabs
reduce risk of visual predation? Ecology 82, 2961–2967.
Palma, A.T., Orrego, C., Arriagada, M., 2003. Crypsis in early benthic phases of
brachyuran decapod crustaceans in central Chile. Rev. Chil. Hist. Nat. 76,
149–156.
Pessani, D., Premoli, C., 1992. Chromatic varieties in a Mediterranean population of
the hermit crab Cestopagurus timidus (Roux, 1830) (Crustacea Decapoda Paguri-
dae). Boll. Mus. Reg. Sci. Nat. Torino 10, 173–178.
139
Pessani, D., Tirelli, T., 2006. Chromatic patterns of the hermit crab Calcinus tubularis
related to the occupied shell. Hydrobiologia 557, 107–112.
Poulton, E.B., 1890. The Colours of Animals. Appleton, New York.
Reuschel, S., Schubart, C.D., 2007. Contrasting genetic diversity with phenotypic
diversity in coloration and size in Xantho poressa (Brachyura: Xanthidae), with
new results on its ecology. Mar. Ecol. 28, 296–305.
Richards, R.A., 1992. Habitat selection and predator avoidance—ontogenic shifts in
habitat use by the Jonah crab Cancer borealis (Stimpson). J. Exp. Mar. Biol. Ecol.
156, 187–197.
Sherratt, T.N., Rashed, A., Beatty, C.D., 2005. Hiding in plain sight. Trends Ecol. Evol.
20, 414–416.
Stevens, M., Merilaita, S., 2009. Defining disruptive coloration and distinguishing its
functions. Philos. Trans. R. Soc. London B 364, 481–488.
Thayer, G.H., 1909. Concealing-Coloration in the Animal Kingdom: An Exposition of
the Laws of Disguise Through Color and Pattern: Being a Summary of Abbott H.
Thayer’s Discoveries. Macmillan, New York.
Tlusty, M., Hyland, C., 2005. Astaxanthin deposition in the cuticle of juvenile Amer-
ican lobster (Homarus americanus): implications for phenotypic and genotypic
coloration. Mar. Biol. 147, 113–119.
Todd, P.A., Ladle, R.J., Briers, R.A., Brunton, A., 2005. Quantifying two-dimensional
dichromatic patterns using a photographic technique: case study on the shore
crab (Carcinus maenas L.). Ecol. Res. 20, 497–501.
Todd, P.A., Briers, R.A., Ladle, R.J., Middleton, F., 2006. Phenotype–environment
matching in the shore crab (Carcinus maenas). Mar. Biol. 148, 1357–
1367.
Wennersten, L., Forsman, A., 2009. Does colour polymorphism enhance survival of
prey populations? Proc. R. Soc. B 276, 2187–2194.