This paper describes an experimental
approach to the interpretation of
archaeological fish assemblages
excavated from the Richardson Island site,
Haida Gwaii,1 British Columbia. This
early Holocene archaeological site has a
well-defined, artifact-rich, and highresolution
stratigraphic sequence and has
produced one of very few faunal
assemblages from coastal British Columbia
dating to earlier than 9,000 years ago (14C
YBP).2 The Richardson Island faunal
assemblage consists entirely of burnt fish
remains concentrated within hearth
features. In the sample analyzed thus far,
the fish taxa appear to be represented
predominately by relatively small
individuals. The research presented here
investigates the possible taphonomic
reasons for the lack of large fish in the
Richardson Island hearth assemblages.
Archaeological and Geological
Context
Richardson Island is located in
southeastern Haida Gwaii, near the
northern boundary of the Gwaii Haanas
National Park Reserve/Haida Heritage Site
(Figure 1). The archaeological site3 is
situated on the west side of Richardson
Island and includes both inter-tidal and
raised beach components. Archaeological
investigations at the site have focused on
the highly stratified raised beach deposits
that are positioned approximately between
15 and 20 m above current sea level. Test
1 By request of the Haida First Nation, we refer to the
Queen Charlotte Islands as Haida Gwaii.
2 All dates are in uncalibrated radiocarbon years before
present (BP).
3 Parks Canada archaeological site designation 1127T,
found within Borden block FeTw, but not assigned a
Borden number.
excavations by Parks Canada in 1995 and
1997 established that the site had been
occupied between 9,300 and 8,500 years
before present4 (Fedje and Christensen
1999; Fedje et al. 2005c); additional
excavations were conducted by the
University of Victoria in 2001 and 2002.
The site consists of over 50 distinguishable
layers which can be grouped into 20
separate, well-sealed depositional units of
analysis. These layers have been dated by
sixteen AMS radiocarbon age estimates
(Fedje 2003:33), all but one of which fall in
sequence consistent with stratigraphic
ordering. In addition to numerous hearth
complexes, the site contains features such
as ash lenses and post-moulds. The lithic
assemblage from the site consists of
approximately 3,600 tools and tens of
thousands of pieces of debitage. Several
small, calcined bone tool fragments were
recovered from the hearths. The lower
Richardson component (pre-8,750 BP) is
assigned to the Kinggi Complex,
characterized by large core and flake tools
and bifacial technology. The upper
Richardson component (8,750 to 8,500 BP)
is assigned to the Early Moresby Tradition,
characterized by the addition of microblade
technology to the Kinggi Complex toolkit
(Fedje and Christensen 1999; Fedje and
Mackie 2005). The hearth assemblages
discussed in this paper are from the Kinggi
Complex component of the site.
From approximately 12,000 to 8,900 BP,
the sea rose by about 165 m at Richardson
Island, from 150 m lower to 15 m higher
than modern levels (Fedje 1993; Josenhans
4 More precisely, the youngest radiocarbon date derived
from cultural layers at the Richardson site is 8490 ± 70
14C age BP; this is a 9,440-9,530 calibrated age range.
The oldest date from cultural layers is 9290 ± 50 14C age
BP; this is a 10,640-10,260 calibrated age range.et al. 1995, 1997; Fedje et al. 2005a).
Consequently, the timing of human
occupation at the Richardson Island site
coincided with the final centuries of sea
level rise and the first several centuries of
sea level stability. Sea levels remained
quite stable at this high stand until
approximately 5,000 years ago, before
slowly receding to their present position.
Rising sea levels contributed to rapid site
formation, resulting in very deep and
highly stratified deposits. The matrix at
this site is composed largely of well-sorted
beach gravels, presumably aggregated by
long-shore drift from the prevailing southeasterly
winds that push through Darwin
Sound.
Subsequent storm toss and supra-tidal berm
development created an ever-rising flat
platform of well-drained and lightly
vegetated terrain along the otherwise steep
slope of Richardson Island, attracting
repeated human settlement. The complex
stratigraphic profile of numerous sealed
layers is the result of sea-level
transgression, supra-tidal berm building,
and occasional down-slope silt mudflows
mixing with upslope gravel storm tosses,
especially in the low, wet swale
immediately inland of the berm. Humans
revisiting the site would have occasionally
found a “refreshed” gravel surface capping
earlier deposits. All this has resulted in an
unusually high-resolution stratigraphic
profile spanning almost 4.5 vertical meters
of deposit, with each depositional unit
representing at most a few decades.
Despite these rapid depositional episodes,
there is evidence that site formation
processes have not dramatically disturbed
the integrity of spatial patterning at the
Richardson Island site. This evidence
includes many clearly defined, intact hearth
and post-mould features, a paucity of
water-worn artifacts, the association of
artifacts with occupation surfaces and
features, some lithics in close proximity
that refit with one another, the consistent
ordering of the radiocarbon age estimates,
and the presence of numerous A and B soil
horizon couplets.
The Richardson Island Faunal
Assemblage
The Richardson Island faunal assemblage
consists entirely of calcined bone, mainly
from the contents of sixteen hearths. The
hearth features span a relatively short
period of less than 200 radiocarbon years,
from approximately 9,290 BP to 9,120 BP,
with each hearth representing one or
several burning events. Hearths were
excavated following their composite
morphology, which typically included a
central area rich with calcined bone
surrounded by a charcoal halo and firealtered
sediments. The different hearth
components were classified as follows: “a”
for the calcined bone-rich central areas, “b”
for peripheral charcoal-rich areas, and “c”
for fire-altered sediments . All
identifiable bone elements and fragments
were removed from the hearth matrices
with the aid of magnification, with some
fragments smaller than 1 mm in size. This
study focuses on three hearth samples from
unit EU-13 (Q12-F1a, S22-F1a, and K26-
F1a), the analysis of which has been
completed. Full analysis of all hearth
fauna from the site is currently in progress
(Steffen 2006).
Table 1: Fish Remains Recovered from Three Richardson Island Hearths
Figure 2: Cross-sectional model of hearth component structure
Richardson K26-F1A Skeletal Element NISP MNI
greenling (Hexagrammos sp.) vertebra (caudal) 2 1
Irish lord (Hemilepidotus sp.) scute 1 1
lingcod (Ophiodon elongatus) vertebra 1 1
lingcod/arrowtooth flounder/hake/cabezon tooth, tooth row 7 -
sand lance* (Ammodytes hexapterus ) vertebra 3 1
prickleback (Stichaeidae) vertebra 2 1
rockfish (Sebastes sp.) (see Appendix A) 80 5
Richardson S22-F1A
dogfish (Squalus acanthias) tooth 10 1
flatfish* (Pleuronectiformes) scute 2 1
Pacific herring (Clupea pallasi) vertebra 1 1
Irish lord (Hemilepidotus sp.) gill raker, pterygiophore 8 1
great-type sculpin (Myoxocephalus sp.) vertebra (abdominal) 1 1
lingcod/arrowtooth flounder/hake/cabezon tooth, tooth row 22 -
rockfish (Sebastes sp.) (see Appendix A) 150 2
Richardson Q12-F1A
dogfish (Squalus acanthias ) tooth, vertebra 16 1
halibut* (Hippoglossus stenolepis) vertebra 1 1
lingcod/arrowtooth flounder/hake/cabezon tooth, tooth row 48 -
Pacific herring (Clupea pallasi) prootic 1 1
salmon (Oncorhynchus sp.) gill raker, parapophyses, vertebra 41 1
starry flounder* (Platichthys stellatus) scutes 5 1
rockfish (Sebastes sp.) (see Appendix A) 163 2
*tentative identification
A list of fish species and elements
recovered from the three hearths is
presented in Table 1. At least 13 taxa are
represented in the assemblages. In each
hearth, rockfish (Sebastes sp.) is most
abundant, both in terms of number of
identified specimens (NISP) and minimum
number of individuals (MNI). All fish
bones were assigned to a size class when
possible. This was done by visual
comparison with comparative specimens at
the University of Victoria. Rockfish
elements were divided into categories from
very small (total fish length of <15 cm),
small (15-30 cm), medium (30-50 cm),
large (50-70 cm), to very large (>70 cm).
Upon initial observation, the rockfish
specimens from the Richardson Island
hearths appeared to be noticeably smaller
than those recovered from the early
Holocene site of Kilgii Gwaay, 90 km to
the southeast. Subsequent examination
confirmed this assessment.
At Kilgii Gwaay (Parks Canada site
designation 1325T), which dates to 9,450
BP (Fedje et al. 2001, 2005b), the faunal
assemblage is dominated by fish remains,
which represent 72% of the total
assemblage NISP (Fedje et al. 2005b). As
at Richardson Island, rockfish remains
dominate the Kilgii Gwaay fauna,
comprising 83% of the fish NISP. When
assigned to size classes, the fish bones
from Kilgii Gwaay exhibit a different
pattern than Richardson Island, with larger
individuals represented (Table 2). The two
sites represent very different depositional
contexts. Kilgii Gwaay is an inter-tidal,
single component wet site with excellent
organic preservation,5 while Richardson is
5 There are also differences in faunal recovery
methods between these two sites. At Richardson,
a highly stratified raised-beach site with
fauna preserved within hearth feature
contexts. Despite their different
taphonomic histories, the two sites may
represent human exploitation of a similar
ecological niche during the early Holocene.
Both sites are located within protected
areas of coastline with immediate access to
near-shore fishes, although Kilgii Gwaay is
closer to the exposed west coast where
there may have been greater opportunity
for deep-water fishing for large rockfishes.
This difference in access may have resulted
in a greater number of larger rockfishes
being present at Kilgii Gwaay. Deep-water
fishing would nonetheless have been
possible from the Richardson Island site as
it is situated at the northern end of Darwin
Sound with good access to Hecate Strait.
Table 2: Size Comparison of Sebastes Elements
Richardson
Island
Kilgii
Gwaay
n 224 601
very small 1.3% 0%
small 73.7% 20.5%
medium 25.0% 67.0%
large 0% 11.5%
very large 0% 1.5%
Before the size difference in rockfish
between the two sites may be attributed to
ecology or human behavior, it is necessary
hearth samples were sieved through mosquito
netting to maximize recovery of small elements,
while 1/8 screen was used for faunal recovery at
Kilgii Gwaay. This may account for the relative
lack of small and very small individual specimens
identified at Kilgii Gwaay, but it does not account
for the absence of large individuals at Richardson.
to examine how taphonomic processes – in
particular, the burning of the fish bones in
the Richardson Island assemblage – may
have affected the size, preservation, and
subsequent quantification of the fish bone
specimens. Two experiments were
designed to address this issue. In the first,
a laboratory-based controlled burning of
fish bones was conducted to determine how
the size of fish elements is affected by
exposure to high temperatures. In the
second, a series of experimental hearths
was created to simulate those found at the
Richardson Island excavation. The hearths
were used to burn fish of various species,
abandoned, and then subsequently
excavated. An analysis of the experimental
hearth contents was conducted to determine
the influence of this depositional
environment on the survival, recovery, and
quantification of fish bones. In this paper,
we focus on fish remains because these
dominate the Richardson Island
assemblage and because most of the
literature on burnt bone addresses mammal
bone. (For studies of the burning of
mammal bone, see Andrews 1995; Binford
1981; Bonnichsen 1989; Coard and
Dennell 1995; Lyman 1994; Noe-Nygaard
1983; Outram et al. 2005; Rabinovich et al.
1996; Stiner et al. 1995; for fish bone
studies see Butler 1993, 1996; Lubinski
1996; Nicholson 1991, 1993, 1996; Richter
1986; Stewart 1991; Van Neer et al. 1992).
Experimental Program 1: Fish
bone size reduction from
exposure to high temperatures
As noted above, many of the rockfish in the
Richardson Island hearths are small in size.
This laboratory-based examination of the
effects of high temperatures on fish bone
was designed to investigate one aspect of
the size of burnt bone. Has burning
reduced the size of the fish bone in the
Richardson Island assemblage, and if so,
by how much? Does burning cause fish
bone shrinkage to a degree that may
significantly affect our size estimates of
archaeological samples? In studies of
mammal bone, Shipman et al. (1984) found
a mean percent shrinkage of about 15%,
while Gilchrist and Mytum (1986)
documented a range of shrinkage of 5% to
30% for bovine and sheep bones. Size
reduction of such magnitude would lead to
inaccurate assessment of live body size,
possibly biasing our understanding of fish
procurement and utilization. This study
looks at Sebastes fish bone size reduction
due to burning.
Methodology
This experiment was conducted in the
archaeology labs at the University of
Victoria. Four rockfish (Sebastes sp.) were
purchased, weighed, measured, and gutted.
To facilitate flesh removal, the fish were
poached slightly by placing them into a pan
of shallow water above a hot plate emitting
only enough heat to loosen the bone from
surrounding flesh. A variety of different
bone elements were selected for
measurement. Some were chosen based on
their frequency in the Richardson Island
assemblage and others for their inclusion in
the size regression formulae developed by
Orchard (2003). The bones included in this
experiment are the atlas, vomer, dentary,
pre-maxilla, epihyal, maxillary, second
vertebra, and basioccipital. Both left and
right sides of paired bones were removed
and measured. Measurements of the first
vertebrae (atlas), vomer, dentary, premaxilla,
and epihyal elements followed
Orchard (2003). These and measurements taken of the maxillary, second vertebra,6
basioccipital, and otolith elements are
listed in Table 3.
After measurement to the nearest 0.01 mm,
the Sebastes bones were placed on a flat
ceramic tile in a Fisher Scientific Isotemp
Muffle furnace, Model 182A kiln. The
bones were burned in two identical kiln
episodes with two rockfish individuals
each. In both episodes, kiln temperature
was brought to 900°C (1650°F) over the
duration of 35 minutes. This temperature
was chosen as it was the maximum
temperature recorded during the field-based
burning experiments, which will be
described later. It also approximates
maximum temperatures recorded in
previous experimental research (see
Shipman et al. 1984 for a review).
Subsequent to reaching maximum
temperature, the kiln was left to cool
overnight. The cooling rate was measured
the first half hour after the kiln was turned
off, during which the temperature fell
approximately 290°C (550°F). Identical
measurements of each element were taken
before and after kiln burning.
Results
Size Reduction After Burning
All elements showed some degree of size
change after burning. For those that
shrank, the average size reduction was
9.0%. There was some variability in
percent shrinkage between elements, with
parasphenoids displaying the least amount
of shrinkage at 7.6% and vertebrae
displaying the most at 10.9%. There was
also considerable variability in the size
6 The anterior measurements of the second vertebra were
taken as a proxy for the posterior measurement of one
first vertebra (atlas) that was lost during processing. All
other vertebral measurements are posterior.
reduction within each element type (Table
4).
Table 3: Measurements Taken Before and After
Burning
Vomer
1. Maximum width of toothed surface
2. Maximum anterior-posterior diameter of
toothed surface
Dentary
1. Maximum anterior-posterior diameter of
the body (superior margin)
2. Maximum height of the symphysis
3. Maximum anterior-posterior diameter
from the symphysis to the external
posterior incision
Premaxilla
1. Maximum antero-posterior diameter of the
body
2. Maximum height of the ascending process
3. Maximum height of the articular process
Epihyal
1. Maximum length of the ventral margin
2. Height of the posterior process
3. Maximum height perpendicular to the
ventral margin
First vertebra (atlas) and second vertebra
1. Maximum height of the centrum
2. Maximum width of the centrum
3. Maximum anterior-posterior diameter of
the centrum
Maxillary
1. Maximum length
2. Maximum width of the posterior margin
3. Maximum depth of the articular end
Basioccipital
1. Maximum height of the centrum
2. Maximum width of the centrum
Otolith
1. Maximum length
2. Maximum width
Table 4: Percent Shrinkage and Standard
Deviation (SD) After Burning
Skeletal Element Average %
Shrinkage SD
vomer 8.2 2.1
dentary 8.1 2.0
premaxilla 9.1 1.7
epihyal 8.8 4.1
first vertebra (atlas) 10.2 2.8
second vertebra 10.9 2.6
basioccipital 8.4 3.2
maxilla 9.4 2.6
parasphenoid 7.6 2.3
otolith - 4.5 1.8
Average 9.0 2.8
The six otoliths in the sample exhibited an
average of 4.5% size increase. Each
otolith was measured the day after burning,
and each specimen exhibited a number of
small fracture lines. This process of
fragmentation appeared to continue
unaided, and within a week of the burning
episode, all otoliths had disintegrated into
ash (Plate 4). This is an interesting
observation because, as noted below, no
otolith fragments have been recovered from
the Richardson Island hearths thus far.
Variability and Measurement
In addition to investigating the overall
shrinkage of fish bones due to burning, this
study was also interested in examining the
variability in shrinkage between elements
(Table 4). Some measurements were more
skewed after burning than others. For
example, even though the premaxillae have
the lowest SD, it was evident that the curve
of the premaxilla anterior-posterior length
had changed during burning. This would
introduce a slight skew into not only the
measurement of its length but also into the
maximum height of the extending
processes, with the original curvature
height being slightly modified through
warping. The maxillary and parasphenoids
have thin and fragile posterior edges, which
curled slightly during burning. This
change in shape could have increased
apparent shrinkage. Given that only two or
three measurements were taken for each of
a small sample of element types, the actual
range of variation in shrinkage between
elements may not be clearly defined here,
but it is important to note the existence of
such variation if burned elements are used
to reconstruct the live size of individual
fish.
Estimates of Richardson Island Fish Size
Size regression formulae have been
developed to estimate the size of different
species of animals, including fish, based on
measurements of various elements (e.g.,
Crockford 1997; Orchard 2003; Casteel
1974). We used Orchard’s (2003)
regression formulae to examine how an
approximate 9% reduction in the size of
elements due to burning would affect the
estimates of the length and weight of whole
fish at the Richardson Island site.
Calculations based on a vomer and epihyal
from hearth Q12-F1a and on five left
epihyals from hearth K26-F1a determined a
range of live fish lengths from 247 mm to
365 mm and live fish weight from 187 g to
870 g . If the percent size
reduction due to burning is incorporated
into these calculations, the estimated live
fish lengths would increase by
approximately 4-8% and the estimated live
weights by approximately 20-31%. The
adjusted weight estimates indicate that the
fish were small (mostly from 400-600g),
but qualitatively speaking, large enough to
be worth eating.
This experiment was designed to determine
if burning causes Sebastes fish bone to
shrink enough to bias our interpretations of
archaeological fish remains. The degree of
shrinkage documented is less than that
observed in the studies of mammal bones
noted above but may significantly affect
size (primarily weight) estimates of fish
individuals. It does not appear to explain
the size difference in Sebastes from the
Richardson Island and Kilgii Gwaay sites.
Table 5: Comparison of Size Estimates Based on
Measurements of Burnt and Unburnt Bone
Vomer Epihyal
1 1 2 3 4 5 6
A. Measurement of burnt specimen (mm)
8.5 1.2 2.7 1.8 2.1 1.9 2.2
B. Projected measurement (mm) if not burnt
9.3 1.3 2.9 2.0 2.3 2.1 2.4
Live length (mm, based on A)
286 239 346 283 305 292 309
Live length (mm, based on B)
307 247 365 296 320 306 324
% Difference
7.3 3.3 5.5 4.6 4.9 4.8 4.9
Live weight (g, based on A)
366 156 725 342 462 388 483
Live weight (g, based on B)
468 186 867 409 552 464 578
% Difference
27.9 19.2 19.6 19.6 19.5 19.6 19.7
To obtain “projected” measurements (B), values
for vomer and epihyal were increased by 8.2% and
8.8%, respectively (see Table 4).
Length = alpha + (beta*bone_measurement)
Weight = alpha * (bone_measurment^beta)
Experimental Program 2: Shortterm
use hearth replications
A field-based replication of short-term
single-use and multiple-use hearths was
conducted to examine additional effects of
high temperatures on fish bones and their
quantification. Two short-term use hearth
features were created to better understand
the formation processes affecting the
Richardson Island faunal assemblage.
While archaeological hearth features are
exposed to taphonomic factors not
replicable in this type of experiment (e.g.,
millennia of compression under four metres
of gravel overburden), this approach can
help us address a number of issues
concerning the interpretation of
archaeological fish bone assemblages. Our
objectives were to determine 1) which
skeletal elements of the fish species
observed within the Richardson Island
assemblage are most identifiable after
burning within hearths (see also Nicholson
1995) and 2) how burning and deposition
in hearth contexts affect the subsequent
quantification of fish remains (e.g., the
calculation of number of identified
specimens [NISP] and minimum number of
individuals [MNI]).
Methodology
“Short-term use hearth” is defined here as
the repeated use of a single hearth for fewer
than 10 burning episodes. The fish placed
in these hearths were caught at various
locations in southern Juan Perez Sound,
Haida Gwaii, near the field campsite on
southeast Wanderer Island where the
experiments were conducted. This camp is
approximately 50 km south of the
Richardson Island site. Two fires, each approximately 50 cm in
diameter, were assembled with wood
placed directly on beach gravel. No pit
was dug and no hearth lining used. A
variety of wood from the surrounding
beach was used, including alder, red cedar,
and yellow cedar. Fire temperature was
measured at regular intervals with a
thermocouple pyrometer, the probe tip of
which was placed in the centre of the fire
and at points along its periphery.
The first fire was a single-use hearth that
was lit only once and within which was
placed a single, filleted rockfish (Table 6).
This hearth was created as a general
reference for what one might expect from a
single burning event in terms of charcoal,
ash, and other fire alterations, as well as
quantity and condition of calcined bone;
clearly, there may be much variation
between such single-use events.
The second hearth was lit eight times.
Fauna were introduced during all but the
last of these burnings. Fish specimens
included eight rockfish (Sebastes sp.), two
lingcod (Ophiodon elongatus), one rock
sole (Lepidopsetta bilineata), one dogfish
(Squalus acanthias), and one kelp
greenling (Hexagrammos decagrammus)
(Table 6). These individuals were filleted
but not gutted before being placed
skeletally whole into the fire. Because of
the large size of halibut (Hippoglossus
stenolepis), only the ultimate seven caudal
vertebrae and the tail assemblage from one
individual was included in this experiment.
Each burning episode lasted from between
45 to 135 minutes, during which time the
fire was fed before being allowed to
extinguish naturally overnight.
Table 6: Fish Used in Hearth Experiments
SINGLE USE HEARTH Length Weight
copper rockfish (Sebastes
caurinus) 23.0 cm 140 g
MULTIPLE USE
HEARTH
rockfish (Sebastes sp.) 29.4 cm 365 g
rockfish (Sebastes sp.) 34.1 cm 725 g
rockfish (Sebastes sp.) 42.0 cm 1050 g
rockfish (Sebastes sp.) 49.0 cm 1800 g
rockfish (Sebastes sp.) 32.5 cm 500 g
rockfish (Sebastes sp.) 26.6 cm 225 g
rockfish (Sebastes sp.) 30.5 cm 320 g
rockfish (Sebastes sp.) 34.5 cm 635 g
lingcod (Ophiodon
elongatus) 67.0 cm 2,315 g
lingcod* (Ophiodon
elongatus) 62.0 cm no data
rock sole (Lepidopsetta
bilineata) 28.5 cm 225 g
canary rockfish (Sebastes
pinniger) 31.0 cm 450 g
greenling (Hexagrammos
decagrammus) 38.0 cm 590 g
dogfish (Squalus
acanthias) 76.0 cm 2,360 g
Pacific halibut**
(Hippoglossus stenolepis) 98.0 cm ~12,220 g
* Stomach contents included two small fish which
were also placed in hearth.
** Only 7 caudal vertebrae and tail were placed
into fire. Weight is estimated from length.
Temperature may be highly variable in
small campfires, with localized temperature
fluctuating considerably at any given
moment. Most hearths had a maximum
temperature around 760°C (1400°F), with
the highest temperature of 900°C (1650°F)
recorded during one episode that took place
on a particularly blustery evening. In
general, fire temperature increased quickly
after lighting. For example, in Episode 2
the first temperature reading was 260°C
(500°F), taken less than five minutes after
the fire was started. Four minutes later, it had risen to 730°C (1350°F). The hearths were excavated twenty days after the first
burning episode.
The methods used to excavate the
experimental hearths were identical to
those used in the archaeological excavation
of the hearth features at the Richardson
Island site, following the component
morphology of each hearth (Figure 2).
This morphology developed quite quickly.
After just two burnings of the multiple-use
hearth, the charcoal rich “b” component
had developed around the “a” component
which was centrally concentrated with less
charcoal. This pattern is comparable to that
observed in the Richardson Island
archaeological hearths and suggests that
they are not necessarily the result of
frequent re-use. Subsequent burning
episodes did seem to increase the visual
definition of the hearth structure.
Interestingly, portions of fish from burning
episodes sometimes remained charred, with
blackened fleshy components still visible at
the periphery of the hearth, while other
portions of the same fish placed in the
middle of the fire became completely
calcined. Also notable was the lack of
evidence or observation of any disturbance
of the experimental hearths by scavengers
such as eagles, ravens, raccoon, or bear, all
of which were present on the Wanderer
Island shoreline at this time.
Results
Skeletal Element Representation
Burning renders bones more susceptible to
fragmentation, resulting in a corresponding
reduction in identifiability (Stiner et al.
1995). It has been noted that, for
mammals, small dense bones of smaller
animals are more likely than other bones to
survive in identifiable condition in highly
fragmented assemblages (Klein and Cruz-
Uribe 1984). Our findings suggest that this
observation also applies to fish bones, but
on a much smaller size scale than for
mammals.
In order to better understand if smaller
bones survive more often in identifiable
condition in burnt fish bone assemblages,
we examined which skeletal elements from
burnt assemblages are commonly damaged
beyond recognition during burning (see
also Nicholson 1995). This is a primary
methodological consideration relevant to
the Richardson Island archaeological hearth
assemblages, keeping in mind that both
human behavior (e.g., species choice and
butchery practices) and taphonomic
processes (e.g., differential durability, soil
chemistry, coarseness of matrix, and
compression from overburden) have
influenced the assemblage composition.
This study obviously cannot replicate all of
the relevant, complex site-specific and
time-dependent taphonomic processes. It
attempts primarily to assess pre-burial
hearth formation processes to gain insight
into how fish bone elements are affected in
archaeological hearth contexts. Because
the “b” component of the experimental
hearths remains to be analyzed, only the
“a” components of the experimental and
archaeological hearths are used in this
comparison.
Sebastes: Appendix A lists the number of
Sebastes bone elements identified after the
two hearth experiments and three
Richardson Island hearth assemblages.
Elements from all regions of the skeleton
are present in both the experimental hearths
– in which whole fish were deposited – and
in archaeological hearths. There is no
complete absence of any skeletal region in
the archaeological examples, suggesting that all portions of at least some fish were
deposited in this context. In general, skull
elements, the suspensorium, and gill rakers
are fairly well-represented in all five
hearths. When assessing the presence of
specific bones, elements that appear in all
five hearth contexts include the nasal,
dentary, gill rakers, vertebra, and
pterygiophores. In many cases, only part
of the bone survived, but diagnostic
portions were present for identification.
The diagnostic attributes of bone elements
sometimes survived in unusual patterns.
For example, in hearth K26-F1a a total of
seven epihyal bones, five left and two right,
were recovered, resulting in a rockfish
MNI of five within a hearth containing
relatively few identifiable rockfish
elements overall (n=80).
Rockfish exhibit considerable discrepancy
in the relative representation of elements
between the experimental and the
archaeological contexts. Specifically, there
are 166 rockfish vertebrae in the multipleuse
and 16 in the single-use hearth,
compared to the three archaeological hearth
contexts, which produced only 4, 5, and 2
rockfish vertebral elements in total. There
has been considerable interest in the
variable frequencies of fish cranial and
vertebral elements in the study of food
processing and storage on the Northwest
Coast (e.g., Calvert 1973; Huelsbeck 1983;
Moss 1989). The paucity of identifiable
vertebral elements in the Richardson Island
hearths may be a result of cultural factors,
such as differential processing of fish
carcasses (Chatters 1984; Butler and
Chatters 1994). Butchery practices that are
dependent upon the size of fish are known
on the Northwest Coast and elsewhere
(Zohar et al. 2001). A particularly salient
example of this has been observed for large
Pacific halibut, which are often butchered
on the beach (Stewart 1977). The edible
fleshy parts of the halibut are then hauled
up to living areas, while the guts and the
remainder of the carcass, likely including
most bones, are left on the beach. Smaller
fish were more likely brought into camps
whole for processing and cooking (e.g.,
Binford 1981). Accordingly, it is likely
that the many smaller rockfish within the
Richardson hearths were brought into camp
whole, and the relative lack of rockfish
vertebrae may be indicative of a type of
fish processing or consumption that was
practiced within the vicinity of hearths.
The presence of gill structures (gill rakers)
and many bony head elements indicates
that the initial processing of small fish taxa,
including rockfish, may have taken place
here. Conversely, the lack of vertebrae
suggests that these elements, possibly
along with the fish fillets, may have been
used and deposited elsewhere.
Differential preservation of skeletal
elements can also contribute to
disproportionate representation of fish
vertebrae. Butler and Chatters (1994)
found that the density of salmon vertebrae
far exceeded that of most of their cranial
elements and are thus more likely to
survive over time, potentially skewing their
relative abundance in archaeological
contexts. In addition, experimental
modeling of the effects of cooking and soil
pH on various fish bones demonstrates that
the vertebrae of some fish are better able
than cranial elements to withstand postdepositional
degradation after cooking
(Lubinski 1996). It may be that Sebastes
vertebrae are also relatively dense,
although density values have not been
derived for this taxon. Other intrinsic
factors, such as bone shape and size, may have influenced element survival in the
archaeological deposits (Lyman 1994).
Overall, the paucity of rockfish vertebrae in
the Richardson Island hearths may not be
entirely attributable to differential
preservation. Any fragmentation of these
vertebrae renders them extremely difficult
to identify to species, and it is possible that
rockfish vertebrae were present but too
fragmentary to be counted.
Other Fish Taxa: Other than Sebastes, the
fish deliberately placed in the multiple-use
experimental hearth include dogfish,
lingcod, greenling, rock sole, and the
partial halibut skeleton. Fish elements
identified from the multiple-use
experimental context are compared to those
identified in the three Richardson hearths
in Appendix B. There are few identifiable
elements of taxa other than Sebastes within
the three archaeological hearths, making it
difficult to compare patterns in element
representation. However, the distribution
of elements of several species invites
comment despite their small numbers.
Dogfish occurred in two of the
archaeological hearths, represented by both
teeth and vertebrae. Because dogfish are
predominately cartilaginous, their dorsal
spines are the only other element that one
might expect to find (see Rick et al. 2002).
The absence of these dorsal spines appears
to be as a result of the peripheral location
of the dogfish within the experimental
hearth. The dogfish was burned near the
fire periphery during the last burning
episodes. Charred portions of the fish
remained apparent around the outside of
the hearth feature during excavation and
constituted much of the “b” component of
the experimental hearth, which was not
included in this study. This does not
explain the absence of dogfish spines
within archaeological hearths, but
placement of ancient fish remains within
different locations in the hearth may also
have biased the appearance of specific
skeletal elements within archaeological
hearths in general. Herring were
represented by vertebral and prootic bones
in two archaeological hearths and the
multiple-use experimental hearth,
suggesting that these elements were most
likely to survive. There was only one
herring element in each of the two
archaeological hearths, presumably because
their small bones do not survive well once
burnt (but see Nicholson 1995). Given the
paucity of herring elements, this taxon may
have been introduced as the stomach
contents of other fish, as was the case in
the experimental context. Vertebral
elements occurred for several taxa
including greenling, sculpin, halibut, and
salmon. A number of teeth and tooth row
fragments were identified as lingcod, arrow
tooth flounder, hake, or cabezon, as they
display a diagnostic “arrow”-tip tooth and
similar tooth row patterning. Halibut also
have a similar tooth row structure. It is
difficult to distinguish between these
species simply on the basis of teeth or very
small fragments of the tooth row alone. In
contrast, scutes, which are modified scales
or skin spines, can be relatively diagnostic.
Irish lord, flatfish, and starry flounder were
all identified by their scutes. This is
interesting for the archaeological samples
because it suggests that the skin of these
fish was deposited into the hearths, likely
as result of the clean-up of fish processing
debris.
Element Representation Summary: In the
current study, the survival of fish bone
after burning appears to have been
influenced by shape, size, and, perhaps, bone density. Flat and less “sculptural”
diagnostic elements tend to survive burning
and subsequent fragmentation less
frequently than diagnostic bones with a
more spherical or simple, robust shape.
Bone size is also a factor in survivability,
with smaller, more diagnostic bones such
as gill rakers and pterigiophores
fragmenting less than larger flat bones.
They are thus more likely to remain
identifiable despite – and perhaps because
of – their small size. In fact, very small
fish bones and fragments were identifiable.
In addition, smaller fish are more likely to
be brought into camps whole than large
fish, contributing more skeletal elements
per individual.
In archaeological fish bone assemblages
that have not been burned, the higher
density bones, including vertebrae and
robust head bones such as the angular and
maxillary, may be expected to survive after
burial more readily than less dense bones,
such as the ceratohyal (Butler and Chatters
1994). This is likely to be true of burnt fish
bone assemblages as well, although enough
of the specific diagnostic regions of bones
must also be present for elements to be
identified to a specific taxon. The relative
lack of rockfish vertebrae in the three
Richardson hearth samples may be due to
the high level of fragmentation, as very
small fragments of these vertebrae are not
identifiable to taxon. It has been noted that
burning causes some loss in the mechanical
strength of bone (Knight 1985 in Lyman
1994; Stiner et al. 1995). In the absence of
density data for these fish species, size and
shape may present a reliable but coarse
indicator of the potential of fish bone
survival within some contexts.
What is of particular interest is the fact that
no otolith fragments have yet been
recovered from the Richardson Island
hearths. Only a few, very fragile otolith
remnants were identified in the
experimental hearths. It appears that,
despite being extremely dense (Butler and
Chatters 1994), otoliths do not survive well
in burnt contexts (or in other archaeological
contexts [Wigen and Stucki 1988]).
Fragmentation, Identification, and the Size
of Skeletal Elements
In their analysis of how fragmentation
affects the identification of mammal bones,
Lyman and O’Brien (1987) concluded that
there is a minimum identifiable size of
bone fragment which varies between taxa
and skeletal elements. Beyond a certain
variable size threshold, the proportion of
identifiable mammal bone fragments will
decrease dramatically (Watson 1972; Hesse
and Wapnish 1985). Elements that occur
in an assemblage may not be identified and
quantified due to high levels of
fragmentation. Here we are interested in
examining the concept of a minimum
identifiable size for various fish bones.
It is apparent that there are few identified
elements in each of the three Richardson
hearths compared to the experimental
contexts (Appendices A and B), which is
likely evidence of significant taphonomic
attrition. The heavy overburden at this site,
consisting of over four meters of gravelrich
sediments, may alone have been
sufficient to cause loss of identifiable
specimens through high levels of
fragmentation. The level of fragmentation
and the paucity of element types identified
at Richardson Island suggest that the
number of identifiable elements has
decreased since deposition. Virtually all the identified elements from the
archaeological hearths are entirely calcined
and hence would be expected to have
turned to powder due to soil compaction
and, in some contexts, trampling and other
disturbances (Stiner et al. 1995). The
calcined fragments recovered from the
Richardson Island hearths may have
survived in the gaps between individual
gravels.
The relative difference in fragment size
between the experimental and the
archaeological hearths offers a coarse
comparison of degree of fragmentation (see
also Grayson 1984; Klein and Cruz-Uribe
1984). Three sub-samples of two hundred
fragments of unidentifiable bone were
randomly selected from both experimental
and archaeological contexts and then
weighed. The average experimental hearth
sub-sample weighed 2.9 g (0.015 g/
specimen), while the same number of
elements from Richardson Island hearth
K26-F1a weighed 1.3 g (0.007 g/specimen)
and those from hearth Q12-F1a weighed
0.9 g (0.005 g/specimen), showing the
average unidentified fragment in the
experimental context to be much larger. A
similar pattern holds true for identified
elements. During the identification process
the amount of each bone element present
was assessed and recorded on the following
scale of completeness: 1 (0-20%), 2 (20-
40%), 3 (40-60%), 4 (60-80%), and 5 (80-
100% whole) (Figure 3).
Statistical analysis shows that identified
skeletal elements were significantly more
complete in the experimental hearths than
in the archaeological hearths (Mann-
Whitney U=201105, p=0.0001). The
significantly higher level of fragmentation
in the archaeological samples has likely resulted in an overall reduction of the
number of identifiable skeletal elements. Quantification of Hearth Assemblages
There has been a longstanding debate about
the relative merits of number of identifiable
specimens (NISP) and the minimum
number individuals (MNI) as measures for
quantification of taxonomic abundance in
faunal assemblages (e.g., Casteel 1977;
Grayson 1973, 1984; Lyman 1979;
Marshall and Pilgram 1993; White 1953).
Highly fragmented bone assemblages
introduce additional interpretive challenges
(for reviews, see Grayson 1984; Klein and
Cruz-Uribe 1984; Ringrose 1993).
Considering the problems that arise from
the use of either NISP or MNI, it is
generally agreed that neither figure should
be used in isolation.
This portion of the analysis examines how
accurately the numbers of individual
specimens put into the fire are detected or quantified after burning. MNI was derived
through visual assessment of skeletal
elements, incorporating size comparisons
and the siding of paired elements. NISP is
the number of whole or fragmentary
specimens identified, not including
fragments that were unidentifiable beyond
the classification “fish”.
Table 7: Fish Introduced Into and Recovered
from Multiple-Use Experimental Hearth
Taxon
Number
placed in
fire
MNI
after
burning
MNI derived
from:
NISP
after
burning
rockfish 8 8 first vertebra
(atlas) 923
dogfish 1 1 vertebra, teeth 109
halibut 1 1 caudal
vertebra 10
herring 0 *1 vertebra,
prootic 41
flatfish 0 *1 vertebra 33
rock sole 1 1 posttemporal 8
lingcod 2 2 basioccipital,
quadrate 295
starry
flounder 0 *1 basioccipital 1
greenling 1 *2 basioccipital,
2nd vertebra 66
crab 0 *1 claw fragment 1
TOTAL 14 19 1490
* possible stomach contents of other fish
Table 7 shows the number of fish
individuals introduced into the multiple-use
experimental hearth as well as the MNI and
NISP of all species identified during
analysis. This analysis produced a number
of interesting results. Of note are the
elements from which MNI was derived for
each taxon. These elements were
considered to be the most reliable measures
of MNI because they were often the least
fragmentary. Other elements, such as the
dentary and premaxilla, which are
commonly used in calculating MNI in
unburnt assemblages, were in some cases
very fragmentary, making accurate visual
matching of fragments difficult. It has
been suggested by some researchers that
only a limited range of specific elements
need to be identified within fish bone
assemblages for the calculation of
measures such as MNI (Leach 1997). In
contrast, this study indicates that, when
calculating MNI within highly fragmented
fish bone assemblages, a wide range of
elements should be examined.
Some of the individuals identified in the
experimental assemblage had not been
documented as part of the original
experiment. A herring and a small flatfish
were observed to be part of the stomach
contents of the lingcod carcass. Because
taxonomic determination is often very
difficult when using Pleuronectiformes
vertebrae, the flatfish vertebrae that were
recovered were not identified to species
(Table 7). While both rock sole and starry
flounder were identified in the calcined
assemblage, the presence of starry flounder
is based on a single basioccipital element
which is morphologically quite similar to
rock sole. Elements of one greenling and
one crab were also introduced
unintentionally, either through stomach
contents of another fish or through
environmental contamination. This second
greenling was much smaller than the one
that was intentionally placed in the fire, and
it was most likely introduced as the
stomach contents of another fish. Butchery
practices that involve the gutting of small
fish at fires may result in the deposition of
non-food refuse into hearth features. In quantifying Sebastes remains recovered
from the multiple-use hearth, it was noted
that many bones were fragmented, resulting
in high NISP counts for specific elements
(Appendix A). Elements with long
diagnostic components, such as the
parasphenoid, the tooth row regions of the
premaxilla, and the dentary, are most
affected because a large proportion of these
bones are distinctive, resulting in the
identification of more fragments. At the
same time, other elements have become
unidentifiable through fragmentation and
thus are not counted toward NISP. The
NISP for Sebastes in the multiple-use
hearth is 923, which may seem fairly high
for the number of individuals (8) placed
into the fire. One might assume that the
relatively high NISP is a result of elements
being fragmented and counted multiple
times, but in this case, the major cause of
the high NISP was the identification of gill
rakers, which contribute 479 of the 923
identified specimens. The identification of
gill rakers within the hearths is a result of
our methodology, a process that
incorporated the assessment of very small
diagnostic elements. Sebastes have a large
number of gill rakers that seem to be more
durable than those of other species (Susan
Crockford, pers. comm.), and the
abundance of these elements may increase
the relative abundance of Sebastes NISP
when compared to other taxa.
Each individual fish that was intentionally
introduced into the multiple-use hearth was
represented in the recovered assemblage.
Therefore, while fragmentation of elements
in the experimental hearth may have
resulted in increased NISP for some taxa, it
had no effect on the quantification of the
MNI of taxa that were intentionally placed
into the fire. The implications of this for
the Richardson Island hearth assemblages
are unclear, partly because the sample size
of each hearth is so small. The use of MNI
with small sample sizes may result in the
exaggeration of the dietary significance of
less important species (Payne 1972).
While we cannot know the number of
individuals originally deposited in the
archaeological hearths, our experiments do
suggest that MNI values may not be greatly
affected by burning.
Species Representation
Species representation may offer insight
into economic activity when dealing with
highly fragmentary assemblages that are
from short-term contexts and contain few
identifiable elements, as is the case at
Richardson Island.7 Consideration should
be given to the fact that not all species
present will have direct economic
importance and may simply represent
discard of offal or other unintentional
occurrence. The experimental hearth study
presented here affirms the possibility that
some species or individuals recovered from
the archaeological hearth contexts may
constitute refuse (such as fish stomach
contents) and thus do not represent human
dietary items. Thus, it is not only
important to identify species but also to
understand their relationships within the
specific ecological niche being exploited
by people. In addition, for large faunal
assemblages, it may be possible to
investigate whether site occupants had
fished out the larger individuals from the
near-shore environment so that only small
individuals remained.
7 For example, one way of presenting presence and
absence data within a number of different contexts is
through the development of a ubiquity index of taxa (see
Dean 2005, McKechnie 2005). The Richardson Island assemblage
included small individuals representing
several near-shore fish species. This shows
that its prehistoric occupants were
exploiting the near-shore environment,
catching rockfish one day and greenling the
next, then processing and cooking the fish
and discarding the refuse from their catch
in their campfires.
Conclusions
The effects of human activity may be
difficult to distinguish from those of
natural taphonomic processes within
archaeological contexts. This presents
complications for the interpretation of
faunal assemblages, challenges that are
compounded by the added taphonomic
complexity typical of hearth contexts. Few
studies have focused on fragmentary burnt
fish bone within hearth features (but see
Hanson 1998), primarily because highly
fragmented bones are not easily identifiable
to skeletal element or taxon. The
importance of the Richardson Island site –
given its early Holocene age, its unusually
high-resolution stratigraphy, and its rich
lithic assemblage – and its lack of any
other faunal evidence led us to focus on the
study of the calcined fish assemblages.
Preliminary investigation of the Richardson
Island fauna has found that rockfish
(Sebastes) are by far the most abundantly
represented fish and that the rockfish
individuals from this site are small
compared to those from nearby Kilgii
Gwaay. This study investigated potential
reasons for the lack of large fish within the
Richardson hearth assemblages. Controlled
burning and hearth replication experiments
were conducted not to replicate the entire,
complex taphonomic history of the
Richardson assemblages, but to provide
insight into the specific characteristics of
the archaeological hearth features.
Our controlled burning experiment
demonstrated that burning causes a size
reduction of these fish bones that may
result in significant underestimates of live
fish weight. An average bone shrinkage of
about 9% was observed for the rockfish
elements in this study. Researchers
studying calcined fish bone may wish to
conduct similar burning experiments to
determine the degree of size reduction for
other taxa of interest. Our experiment also
showed that otoliths turn to ash after being
exposed to high temperatures (900ºC),
providing a viable explanation for the lack
of otoliths in the Richardson hearths.
The hearth replication experiments
demonstrated the complex taphonomy of
burnt fish bone assemblages. For example,
fragmentation affects NISP in two ways:
when specific skeletal elements (for
example the parasphenoid) are broken,
pieces of the same bone can be counted
more than once, resulting in an increased
NISP for those elements. More commonly,
fish skeletal elements were broken beyond
recognition, reducing NISP. This study
also observed that high levels of
fragmentation may result in an inverse
relationship between the body size of an
animal and the identifiability of its remains.
Large mammal and bird bones fragmented
into small pieces may be less identifiable
than fish bones broken into similarly sized
fragments. At the Richardson Island site,
not only were there very few identified
mammal and bird specimens, but there
were also very small burnt fish bone
fragments that were identifiable. In examining how MNI counts may be
affected by burning in hearths, we found
that all of the individual fish that were
placed into the experimental hearth were
accounted for in the recovered assemblage.
Of particular note was the fact that several
additional fish had been introduced into the
assemblage, likely as stomach contents of
other fish. Small fish in archaeological
assemblages may thus represent discard
that did not contribute directly to human
subsistence. This does not appear to be
entirely the case at Richardson Island.
Preliminary analysis of Sebastes skeletal
elements from two hearth contexts suggests
that the individuals recovered were large
enough to represent a food resource. It is
unlikely that all these fish were brought to
the hearth to be discarded in the fire or that
they only represented stomach contents of
larger fish. In addition, the large
proportion of Sebastes of all sizes in a
variety of contexts at the Kilgii Gwaay site
supports the conclusion that this taxon was
caught to be eaten.
The Richardson Island hearth assemblages
derive from a very localized and specific
hearth context. As a result, they represent
the material remnants of activities
conducted at or near the hearth. If
complete butchery and discard of bones
preceded arrival at the hearth, those
remains would not have been introduced
into these features. This is especially
relevant given that larger animals, such as
bear, albatross, halibut or seal, which
represented a major portion of the Kilgii
Gwaay faunal assemblage, are more likely
to be butchered prior to transport to
residential sites (Binford 1981).
While taphonomic factors may complicate
interpretations concerning the species that
contributed to the diet of the prehistoric
inhabitants of Richardson Island, the hearth
assemblages still provide behavioural
information about the activities of the site
occupants. The experimental hearths
suggest that their archaeological counterparts
represent a series of short-term events
– specific activities, such as logistical
forays into the environment. The hearth
contents thus illustrate the niches that were
exploited by humans across short time
spans in the distant past. Despite the
difficulties in working with such highly
fragmented faunal assemblages, they
provide a rich source of information that
may prove critical in our understanding of
the subsistence practices of peoples in the
past.
Nessun commento:
Posta un commento