See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/286213483

Species abundance models as potential discriminators of natural

and anthropogenic paleoichthyocenoses with a case example
from African sites

Article · January 1994

CITATIONS READS
7 45

2 authors, including:

Wim Van Neer

Royal Belgian Institute of Natural Sciences
301 PUBLICATIONS   5,600 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Weird Fish: fish pathologies from archaeological and modern skeletons View project

TDASP - Tübingen-Damascus Excavation and Survey Project View project

All content following this page was uploaded by Wim Van Neer on 18 December 2015.

The user has requested enhancement of the downloaded file.

Journal of Archaeological Science 1992,19,683-695

“Fish Middens”: Anthropogenic Accumulations

of Fish Remains and Their Bearing on
Archaeoichthyological Analysis
Wim Van Neeraand Arturo Morales Muf&

(Received 9 October 1991, revised manuscript accepted 7 February 1992)

Extensive accumulations of animal remains, in the absence of anthropogenic

contextual evidence, tend to be interpreted in strictly palaeobiological terms. The fact
that anthropogenic thanatocoenoses may, to a great extent mimic natural ones has so
far not been documented in detail. This paper gives an example of a recent man-made
accumulation of fish remains, some characteristics of which might lead researchersto
incorrect thinking. Should the assemblagebe covered by sediment and later excavated,
it would probably be misinterpreted as a product of mass mortality of isolated
populations in seasonalpools. In particular, the large number of specimensretrieved,
the undisturbed articulation of most skeletal elements and the small sizeof the fishes,
together with an almost complete absenceof signs of human activity could lead to such
a misinterpretation. A palaeontological example of accumulated fish remains, which
has very similar characteristics to the present-day one, is also discussed.

Keywords: FISH, TAPHONOMY, AFRICA, ARCHAEOZOOLOGY.

1. Introduction
The importance of man as a major taphonomic agent has long been recognized. While
occasionally misunderstood and overrated, this importance runs, to a large extent,
parallel to the existence of activity signs (artefacts, chopmarks, burned bones etc.) which
provide clues to the interpretation of a particular assemblage’s formation (Noe-Nygaard,
1967: Brain, 1981). When no signs of human activity appear, however, there might be a
tendency to disregard anthropogenic influences as causative agents and to concentrate
instead on “natural causes”. One should always be careful when attempting to dismiss
human activity, for abundant data indicate that humans can work in most subtle ways, as
the present paper intends to illustrate.
Studies describing distinguishing features between naturally deposited and man-made
bone assemblages are increasing in the literature, especially those concerned with
terrestrial faunas (Koch, 1989). For fishes, however, few data are available (Schafer, 1972;
Butler, 1987; Colley, 1987,199O; Richter, 1987; Yerkes, 1987; Wheeler&Jones, 1989) and,
particularly in Africa almost no work has been done. Robbins (1980: 130-l 35) conducted
an ethnoarchaeological study of a fisherman’s camp at Lake Turkana (Kenya), which
“Royal Museum of Central Africa, B-3080 Tervuren, Belgium.
Wniversidad Autonoma de Madrid, E-28049 Madrid, Spain.

683
03054lO3/92/060683 + 13 $08.00/O 0 1992 Academic Press Limited
684 W. VAN NEER AND A. M. MUfiIZ

Figure 1. Location of Dakar-Bango. The dam is indicated by the black line that
croses the river Djeuss east of the village.

involved the excavation of the refuse areas of a camp occupied for about 2 years but
abandoned I7 months before investigation. In an attempt to find criteria to distinguish
naturally derived lacustrine bone scatters from anthropogenic ones, Stewart (1989: 78-98)
analysed a natural fish bone assemblage from a beach at the western shore of Lake
Turkana and then compared it with the bones found at a site used by Turkana fishermen
for the roasting and consumption of fish. In a later paper, Stewart (1991) also includes
channel-associated contexts in the comparison of natural and anthropogenic deposits.
The assemblages described here represent a somewhat different type of approach to the
problem.

2. The Fish Middens

2.1. Location and history offormation
The middens were discovered in June 1990 at Dakar-Bango, Senegal, during fieldwork in
the region east of Saint-Louis (Figure I). The village of Dakar-Bango lies on the left bank
of the Djeuss river, close to its outlet to the Senegal river. At the eastern extremity of the
village a dam crosses the Djeuss, allowing regulation of the drainage and preventing
incursion of saltwater. Just outside Dakar-Bango, at the west side of the dam five heaps of
decaying fish were found (Figure 2). Inquiries among the inhabitants of the village,
including fishermen, made it clear that all heaps had been “recently” deposited. Our
informants could not give a precise chronological reconstruction of the events that
resulted in the formation of the heaps. One single fisherman was said to be responsible for
the accumulation of fish, which occurred over a time period of “several months”. This
person had left the village “a few months ago”. The man used to empty his nets at the
landing place and selected marketable from unmarketable fish in function of their size.
The discarded fishes were those which were difficult to sell because of their small size.
According to one of our informants the small fish could originally have been left at the
surface to sun-dry. This idea was finally abandoned after speaking with other people and
 “FISH MIDDENS” 685

Figure 2. View of the fish middens at Dakar-Bango, taken from the north. The fish
midden in the front is heap 5.

also due to the fact that some of the heaps were rather thick with many layers of fish on
top of each other. The fish had been caught with drifting monofilament gill nets with a
mesh size of 20 mm (knot to knot), the same type used by the remaining fishermen at
Dakar-Bango.

2.2. Analysis of thejsh middens

2.2.1. Size and overall features. Before studying the content of the heaps, measurements
were taken and a sketch of the area was made (Figure 3). The greatest length of the
middens ranged from 2.2 to 3.25 m and the greatest width ranged from 1.6 to 2.0 m. The
height of the heaps varied from about 10 cm (for heap 1) to 50 cm (for heap 5). All heaps
were lying on firm ground but middens 3 and 4 were partially in contact with the water.
Some other occasional material was mixed with the fishes. This was especially true for
heap 1 which could be considered as a more general waste heap; it included lots of broken
branches of trees, some stones, some Arca shells and even a sandal. This waste was lying
below a layer of fishes. Part of this material, especially the branches, may also have come
from the nets. The other heaps also contained such plant remains.
Each heap contained numerous dermestid beetles and their larvae. They were identified
as Dermestes maculatus. The degree to which these beetles had attacked the fish bodies
varied from one area to another for any given heap. The most striking differences were
observed in heap 2. At the eastern edge of that pile, the skin and the flesh of the fishes was
almost completely eaten away by the dermestids, leaving perfectly cleaned skeletons
(Figure 4). It was noted that this was the best-aerated part of the midden. During the
course of the investigation of the heaps there was a strong and constant wind blowing
from the east. It is known from the behavior of the dermestids that they only start feeding
on carcasses which have lost a great part of their humidity (Waterman, 1976: 26). This
explains why the fishes lying in or close to the water had not yet been attacked and also
why the more humid parts inside the heap contained largely undefleshed individuals
686 W. VAN NEER AND A. M. MUSIZ

hl
Irn
l-l DAM

Figure 3. Sketch of the landing place and the fish mounds.

Figure 4. Eastern edge of heap 2 with well-defleshed specimens of mainly

Chrysichthys (scale bar is 5 cm).

(Figure 5). The defleshed skeletons were still articulated and had no tendency to fall apart,
except for an occasional disconnection of the head, or breakage of the vertebral column. It
was also noted that in certain fishes, such as Elops and Liza, isolated heads and bodies
were more frequent than in other species. Strong winds, such as the prevailing easterly
 “FISH MIDDENS” 687

Figure 5. Heap 3 with Erhmalosa not well-defleshed (scale bar is 5 cm).

one, are believed not to be responsible for dispersal of corpses or isolated portions from
them.
Further analysis of the fish middens involved the establishment of a species list, an
estimation of the relative abundance of the taxa in each heap and an overall description of
the size distribution of the main species.

2.2.2. Species diversity. A total of 23 different fish taxa, both marine and freshwater have
been identified from the heaps (Table 1). All marine species represented are known to enter
rivers and small specimens from these taxa constitute a substantial portion of the annual
catch in estuaries (Daget & Iltis, 1965). Among the freshwater species several taxa
(e.g. Chrysichthys and Tilapiini) are reputed for their high salinity tolerance. Others,
however, are less tolerant of brackish water (e.g. Mormyridae). This indicates that the
heaps accumulated not only during the dry season, when the incursion of saltwater into
the Djeuss was high, but also during the wet season when the freshwater influx was
important. Our visit to the area took place at the very end of the dry season (June) when
salinity levels were high downstream from the Dakar-Bango dam. Fishing with hand-
thrown cast nets in the downstream section of the river, close to the midden site, yielded
Mugilidae, Sarotherodon melanotheron, Sarotherodon galileus and Citharichthys stamp&
which are all euryhaline species (persona1 data). However, due to the aforementioned
strong winds blowing during our stay at the site, no gill-net fishing was practicable and
hence the regular catches could not be verified.

2.2.3. Relative importance of the taxa. The fish middens included accumulations ranging
from several hundreds (for heap 1) to several thousands of individuals (all remaining
heaps). In order to get an impression of the importance of each taxon, counts of samples
were made (Table 2; Figure 6). Specimens were counted in a 10 cm wide strip along the
greatest length and at the surface of each heap. In each case approximately 100 individuals
were identified to species. The species composition within each heap seemed to be more or
688 W. VAN NEER AND A. M. MURIIZ

Table 1. List offishes identiJiedfrom theJive heaps ai Dakar-Bango

Marine fishes Freshwater fishes

Elopidae Mormyridae
Elops sp. Small species
(cf. Marcusenius)
Clupeidae Cyprinidae
Ethmalosafimbriata Labeo sp.
Ilisha africana
Carangidae Citharinidae
Selene dorsalis Citharinus sp.
Lichia amia
Haemulidae Characidae
Brachydeuterus auritus AlesteslBrycinus
Monodactylidae Bagridae
Psettias sebae Bagrus bajad
Chrysichthys maurus
Mugihdae Schilbeidae
Liza sp. Schilbe mystus
Polynemidae Clariidae
Galeoides decadactylus Clarias anguillaris
Bothidae Mochokidae
Citharichthys stampflii Synodonris schall
Cynoglossidae Centropomidae
Cynoglossus sp. Lates niloticus
Cichlidae
Oreochromis niloticus
Tilapiini indet.

Table 2. Species composition of the heaps at Dakar-Bango, expressed as the number

of individuals found in samples of about 10afishes. Minor fish taxa have been lumped
as “others”

Hla Hlb H2 H3 H4 H5 Total

Elops 22 6 1 I - 1 31
Ethmalosa 14 20 25 76 73 53 261
Liza 29 12 - 1 17 4 63
Chrysichthys 15 63 52 12 - 18 160
Synodonris 13 50 3 5 - 15 86
Tilapiini 6 22 4 4 25 3 64
Others 7 19 19 4 4 2 55

Total 106 192 104 103 119 96 720

less homogeneous in each area without apparent concentrations of single species at any
place. An exception to this was a concentration of Chrysichthys occurring in a small spot
at the eastern edge of heap 2 (see also Figure 4). It was also noticed that heap 1, which
contained the lowest number of specimens, would give a different spectrum depending
upon the area chosen for a subsample. Therefore, two different counts were made on this
heap, one along the greatest length (heap la in Figure 6) and one including all specimens
lying in the northwestern “subheap” (heap 1b).
 “FISH MIDDENS” 689

Others

60 Synodonlis

1
z I Chrysichthys
L:
40
Ethmalosa
1 --I

20
a Hops
0 -L- d
heap la heap lb heap 2 heap 3 heap 4 heap 5
(N= 106) (N=192) (N= 104) (N= 103) (Nzll9) (N=96)

Figure 6. Relative importance of the main fish taxa in each heap. Percentages have
been calculated from number of individuals.

2.2.4. Size distribution of the d&@rent species. In order to evaluate the sizes of the discarded
fishes, standard lengths (SL) were measured for the most common species. For statistical
relevance, we tried to measure about 25 specimens for each species. This was straight-
forward for Ethmalosa, Synodontis, Chrysichthys and Tilapiini since they were common
and well preserved. Analysis of the graph (Figure 7) indicates that animals with a standard
length between 9 and 11 cm are most common in the samples. Data for other taxa are
incomplete, mainly because they were not so well represented, but also because whole
specimens were rare. Only seven complete specimens from Clarias gariepinus were
measured (between 14 and 20 cm SL). The Bagrus bajad specimens are in the same size
range (15.5-20 cm SL). In Psettiassebae the range is from 6.5 to 8.5 cm SL (four specimens
only). Z
The observed minimum length varies from species to species. This phenomenon has
long been known from fishery studies where the effectiveness of nets of a given mesh size
has been investigated as a function of the length and height of the captured fishes (Fryer &
Iles, 1972: 417-421; Jensen, 1990). In our samples it is obvious that deep-bodied species
(Psettias and tilapia) are represented by smaller individuals than more fusiform fishes such
as Elops. According to our informants the fishes were unmarketable specimens because of
their small size. Despite this, however, occasional larger individuals have been found:
Ethmalosa of more than 20 cm, as well as Clarias and Bagrus of approximately 20 cm. In
the case of Ethmalosa, those larger individuals may have fallen among the refuse fishes
during sorting of the catch or they may have been considered unmarketable for another
reason (damaged or spoiled). Initially it might seem surprising that Clarias and Bagrus
between 14 and 20 cm are found among the unmarketable fish, but they do represent the
smallest individuals of the catch and may therefore have been considered not worth
taking.

3. Discussion
Wheeler & Jones (1989: 77) give an overview of the processes involved in the formation of
archaeological fish assemblages. In their scheme they assume that undesired fishes are
690 W. VAN NEER AND A. M. MUr;;rIZ

‘thmoloso

SLkm)

6
Chrysichlhys

0
SL(cm)

4
z

0
8 9 10 II 12 SL(cm)

0
5 6 7 6 9 IO I2 I3 SL(cm)
Figure 7. Length distribution of the most common fishes.
 “FISH MIDDENS” 691

discarded at the place of capture and hence all such animals are rarely landed (Wheeler &
Jones, 1989: 64). During our stay in Senegal we have noticed that this is true when the fish
are captured far offshore. However, when fishing is practised close to the shore-in
coastal, riverine as well as lacustrine environments-fish selection usually occurs at the
landing place. This was the case at Dakar-Bango.
Hence, in this study we are dealing with an accumulation of fish which underwent only
very limited handling by man. Excluding the catch, the selection by size and the accumu-
lation in heaps, man did not manipulate the fishes. One is tempted to speculate how these
mounds would be interpreted if they were to be covered by soil and later excavated.
Naturally, we cannot predict possible future disturbances, nor can we gauge the
taphonomic loss that might occur during further exposure at the surface, or during and
after burial. Neither can we predict how accurate the methods would be during future
excavation. For discussion’s sake we will assume that most of the observable features
described thus far would also be apparent, to some extent, during future excavation.
Certainly, the contents of the mounds give almost no indication of their anthropogenic
origin. Only heap 1 contains abundant waste that is easily attributable to man and which
would give sufficient grounds for interpreting the heap as a culturally-derived assemblage.
Conversely, in the other mounds the only objects included in the deposits are broken
branches, and these are likely to have come from the nets. It is highly improbable that
these wood pieces, if preserved, would be considered as an indication of an anthropogenic
origin for the heaps, except perhaps if all mounds were totally excavated. If this were done
then, based on the premiss that heap 1 is of anthropogenic origin (see above), the others
could possibly be considered to have had a similar taphonomic history.
Nevertheless, the fishes do not show any sign of manipulation (butchery marks, traces
of fire, etc.). Moreover, the majority of the skeletons are still articulated. When these facts,
combined with the small size of the individuals, are considered on their own, then they
imply, albeit by convergent means, that the deposit resembles a natural death assemblage.
Circumstances under which such natural deposits can originate are numerous in Africa
along waters with a seasonally fluctuating level.

3.1. Distinguishing natural and anthropogenic thanatocoenosis

The level of rivers and lakes in many regions in Africa fluctuates seasonally as a result of
flood waters. During such periods when lake margins and floodplains are inundated, most
fishes migrate to the flooded areas to spawn. Their fry grow fast in this nutrient-rich
environment. When the water level drops, the larger fish migrate back towards the main
water body but the smaller individuals delay their return and may eventually become
trapped in residual pools at the lake margin or on the floodplain (Welcomme, 1975). The
species composition of such water bodies is not well studied but the work of Holden (1963)
indicates there is much variation even between pools lying close to each other. The
seasonal pools studied by Holden (1963) were relatively large and the number of genera
was high. When the pools dry up further, they become smaller and less oxygenated,
making it difficult for most species to survive. The genera with the highest tolerance to low
oxygen levels are Clarias (with its accessory breathing organs) and tilapia (Fish, 1956).
This also explains why these two genera are present in many Saharan waterbodies isolated
since the end of the last humid period of the Holocene (Le Berre, 1989).
A well-documented fossil assemblage of African freshwater fishes that died as a result of
seasonal drying-up of a pool comes from Bir Tarfawi (Eastern Sahara, Egypt). The fauna
is dated to the Last Interglacial and corresponds to the contents of a small pool which
was seasonally connected with a lake. Geomorphological observations and the lack of
artefacts indicate that the assemblage is a naturally derived one. It is also believed
that most of the fish remains represent individuals that died as the pool dried up, but
692 W. VAN NEER AND A. M. MUAIZ

Table 3. List qfJi.sh identified,from the natural-death assemblages at Bir Tarfan?.

expressed as percentages of number qf,fragments (NF) and of minimum number of
individuals ( MNI)

MN1
(N,N:47) (N=355)

Mormyridae 0.3% 0.3%

Alestes/Brycinus 0.3% 0.3%
Bagrus 0.1% 0.6%
Clarias 38.7% 32.7%
Synodontis 8.1% 39.1%
Lales 3.5% 4.2%
Tilapiini 49.0% 22.8%
indet. 1 + +
indet. 2 + +
indet. 3 + +

the abundant remnants of reed cormorant Phalacrocorax africanus (including many

juveniles) indicate that this species has probably also contributed to the deposit (Kowalski
et al., 1989). The ichthyofauna of Bir Tarfawi contains 10 different taxa, but comprises
mainly Clarias, tilapia and Synodontis (Table 3). Size reconstructions of the different
species indicate a preponderance of small individuals. Most specimens are between 10 and
20 cm SL for Clarias, between 5 and 15 cm SL for tilapia and less than 7 cm SL for
Synodontis (Van Neer, in press).
Differences between the assemblages of Dakar-Bango and Bir Tarfawi lie not in the size
distribution of the species, but rather in the species diversity itself. At Dakar-Bango a
much wider spectrum of the ichthyofauna is present. This probably relates to the fact
that this assemblage comes from a much larger catchment area which includes different
habitats exploited over different seasons, whereas in the natural assemblage of Bir Tarfawi
only species that are tolerant of the adverse conditions in the pool occur. A particular
feature of this natural assemblage, besides its low specific diversity, is the abundance of
Clarias and tilapia. These taxa have the highest chances of survival in poorly oxygenated
biotopes. This last criterion, however, will not always allow discrimination between
naturally derived and anthropogenic accumulations. Should people discard small fish
after exploitation of seasonal pools dominated by Clarias and tilapia, then it will be
difficult to discriminate such an assemblage from a naturally derived one. In such cases,
one can only hope to find circumstantial evidence to distinguish between the two.

3.2. Contextually derived evidence: the Wadi Howar case study

A specific case, recently studied by one of us (Van Neer, 1988), illustrates how geomorpho-
logical and archaeological observations can occasionally help to distinguish between
natural and anthropogenic assemblages. The Holocene site BOS 84/ 13 in the Middle Wadi
Howar, Sudan, incorporates eight pits which have been excavated on a dune of the
southern bank of the wadi. All pits are refuse dumps containing pottery, lithic material
and bones, mainly from cattle. However, one pit is different in that it yielded a large
number of small fish. So far only a small part of the 3 mm sieve fraction has been investi-
gated, the 1 mm fraction not being available for analysis at the time of this study. A
preliminary analysis of the sample indicates that at least 17 different taxa are present and
that Clarias and tilapia are the most abundant fishes. While size reconstructions are
unavoidably biased, since the 1 mm fraction is not included, they nevertheless indicate an
 “FISH MIDDENS” 693

abundance of small individuals. It seems clear from the relative frequencies of the taxa
(mostly Clurius and tilapia) and from the small sizes of the specimens, that the fishes must
have come from a residual pool along the wadi. During identification of the material, and
reconstruction of the sizes of the corresponding fishes, the overwhelming similarities with
the natural death assemblage of Bir Tarfawi, inclined us towards a similar explanation for
the history of deposition at the Middle Wadi Howar. Despite recognizing that people
readily eat small fishes, we still initially tried to explain the coexistence of artefacts and fish
bones as a result of natural causes. The scenario we first devised involved high floods
covering that part of the dune. The anthropogenic pit structure of the abandoned site was
believed to have acted as a trap in which small fish were retained after the water level
dropped. However, a geomorphological survey of the site and its surroundings made it
clear that floods could have never reached such a high level on the dune (Kropelin, pers.
comm.). Furthermore, the archaeological analysis showed the pit to be probably a multi-
functional unit, viz, a storage place during the initial phase of filling and a refuse dump
later on (Keding, 1986, in press). Thus, the initial archaeozoological interpretation of
the fish assemblage as a natural thanatocoenosis was actually a misguided automatic
response, resulting from the presence of large numbers of small fish combined with the
absence of manipulative marks on the bones. The results from the work at Dakar-Bango
clearly indicate that anthropogenic accumulations of this kind are indeed perfectly
possible.

4. Conclusions
The processes whereby fishes become deposited in heaps similar to the middens described
above may be governed by several factors. Inefficiencies associated with food processing
are responsible for many such factors. For example, rapid decomposition of these fishes
caused by weather changes, insect pests, etc. will render them unacceptable for human
consumption and hence they will be discarded. Other, related processes of product rejec-
tion include the disposal of caught fishes which are too small for profitable marketing.
There are, of course, alternative solutions to overcome such waste of resources (e.g. fishes
may be used for soup, or dried and used as fodder for domestic stock). Nevertheless,
rejective decisions of the type described above will remain as important factors in
the formation of fish middens and they are always complicated phenomena to evaluate
a posteriori. There can be other factors, including epiphenomena such as ritualistic
offerings, foods, status symbols etc. which may be regarded as potential causative
agents of non-consumable and consumable fish accumulations. However, any detailed
discussion of these lies beyond the scope of this paper.
The most significant point is the degree of convergence occasionally produced between
some man-made accumulations and natural ones (Rose110 & Morales, 1990; Morales &
Rosello, in press). Consequently, in order to describe accurately the history of any
particular accumulation, the researcher is required to make comparative analyses of all
complementary information relevant to the assemblage. We hope that future investi-
gations along such lines will eventually provide a database against which other hypotheses
may be tested.

Acknowledgements
The field work in Senegal would not have been possible without the help of Marie-Amy
Mbow (I.F.A.N.-Ch.A.Diop; Dakar) and Marie-Isaac Diop who acted as interpreters
and arranged several practical matters. The drawings for this article were made by Alain
Reygel (Tervuren). J. Decelle (Tervuren) identified the dermestid beetles. Earlier drafts of
694 W. VAN NEER AND A. M. MUNIZ

this paper were read and commented upon by Dick Brinkhuizen (Groningen), Inge
BDdker-Enghoff (Copenhague), Leif Jonsson (Gijteborg) and Ian Harrison (Tervuren).
To all these colleagues and friends we express our sincere thanks.

References
Brain, C. K. (1981). Hunters or the Hunted? An Introduction to African Cave Taphonomy. Chicago:
Chicago University Press.
Butler, V. L. (1987). Distinguishing natural from cultural salmonid deposits in the Pacific
northwest of North America. BAR International Series 352, 131-149.
Colley, S. (1987). Fishing or facts. Can we reconstruct fishing methods from archaeological
evidence?Australian Archaeology 24, 1626.
Colley, S. (1990). The analysis and interpretation of archaeological fish remains. In (M. Schiffer,
Ed) Archaeological Method and Theory. Tucson: University of Arizona Press.
Daget, J. & Iltis, A. (1965). Poissonsde Cote d’Ivoire. Memoires de f’lnstitut Francais de 1’Afrique
Noire 74, l-385.
Fish, G. R. (1956). Some aspectsof the respiration of six speciesof fish from Uganda. Journal of
Experimental Biology 33,186195.
Fryer, G. & Iles, T. D. (1972). The Cichlid Fishes ofthe Great Lakes of Africa: Their Biology and
Evolution. Edinburgh: Oliver and Boyd.
Holden, M. J. (1963). The populations of fish in dry seasonpools of the River Sokoto. Fishing
Publications Colonial Ofice 19, l-58.
Jensen, W. (1990). Comparing fish catches taken with gill nets of different combinations of mesh
sizes.Journal of Fish Biology 37,99-104.
Keding, B. (1986). Gabarona 84113. Siedlungsfunde des 2. und 3. vorchristlichen Jahrtausends im
Wadi Howar (Ostsahara). Magisterarbeit, Universitat Koln, Germany.
Keding, B. (in press). Leiterbandsites in the Wadi Howar. In (L. Krzyzaniak & M. Kobusiewicz,
Eds) Environmental Change and Human Culture in the Nile Basin and Northern Africa Until 2nd
Millenium BC. Poznan: Poznan Archaeological Museum.
Koch, C. P. (1989). Taphonomy: A Bibliographic Guide to the Literature. Peopling of the Americas
Publications. Bibliographic Series.Orono, Maine: Center for the Study of the First Americans,
University of Maine. .
Kowalski, K., Van Neer, W., Bochenski, Z., Mlynarski, M., Rzebik-Kowalska. B., Szyndlar, Z.,
Gautier, A., Schild, R., Close A. E. & Wendorf, F. (1989). A Last Interglacial fauna from the
Eastern Sahara. Quarternary Research 32,335-341.
Le Berre, M. (1989). Faune du Sahara. 1. Poissons-amphibiens-reptiles. Paris: Chabaud-
Lechevalier.
Morales, A. & Rosello, E. (in press).Casual or intentional? Comments on fish taxa skeletal
representation from Iberian archaeological settlements. Proceedings Fourth Meeting of the
ZCAZ Fish Working Group, September 1987, University of York. In (A. K. G. Jones & R. A.
Nicholson, Eds) Fishes and Man.
Noe-Nygaard, N. (1967). Present-day “kitchen middens” in Ghana. Geogrufisk Tidsskrift 66,
179-197.
Richter, J. (1987). Evidence for a natural deposition of fish in the middle neolithic site,
Kainsbakke, East Jutland. Danish Journal Archaeology 5,11&l 24.
Robbins, L. H. (1980). Lopoy: a Late Stone-Age fishing and pastoralist settlement in the Lake
Turkana Basin, Kenya. Publications of the Museum, Michigan State University, Anthropological
Series 3, l-140.
Rosello, E. & Morales, A. (1990). Global patterns of skeletal abundance in Spanish
archaeoichthic assemblages.In (S. Fernandez-Lopez, Ed) Communicaciones Reunion de
Tafonomia y Fosilizacion. Madrid: Universidad Complutense de Madrid pp. 319-325.
Schafer, W. (1972). Ecology and Paleoecology ofMarine Environments. Edinburgh: Oliver and
Boyd.
Stewart, K. M. (1989). Fishing sitesof North and East Africa in the Late Pleistoceneand
Holocene. Environmental change and human adaptation. BAR International Series 521, l-273.
 “FISH MIDDENS” 695

Stewart, K. M. (1991). Modern fishbone assemblages at Lake Turkana, Kenya: a methodology to

aid in recognition of hominid fish utilisation. Journal qf Archaeological Science l&579-603.
Van Neer. W. (1988). Fish remains from a holocene site (84/13-9) in Wadi Howar, Sudan
Archaeozoologia 2,339-348.
Van Neer, W. (in press). Fish remains from the Last Interglacial at Bir Tarfawi (Eastern Sahara,
Egypt).
Waterman, J. J. (1976). The production of dried fish. FAO Fisheries Technical Paper 16, l-52.
Welcomme, R. L. (1975). The fisheries ecology of African floodplains. FAO CIFA Technical Paper
3, l-51.
Wheeler, A. & Jones, A. K. G. (1989). Fishes. Cambridge Manuals in Archaeology. Cambridge:
Cambridge University Press.
Yerkes, R. W. (1987). Seasonal patterns in late prehistoric fishing practices in the North American
Midwest. Archaeozoologia 1, 137-148.

View publication stats