Archaeofauna 6 (1997): 59 - 66
The impact of salting and drying on fish bones: Preliminary observations on four marine species from Parita Bay, Panama
IRIT ZOHAR1 & RICHARD COOKE2 (I) Departmem of Zoology. Tel Aviv University Ramat Aviv, 69978. Israel (2) Smithsonian Tropical Research Institute Panama, Unit 0948 APO AA 34002-0948 USA (Received 20 June 1996; accepted 18 November 1996)
ABSTRACT: Actualistic («middle range») studies of present day fishing communities facilitate the reconstruction of fishing methods and fish utilization from archaeofaunal data. Fish dehy- dration by salting and sun and wind drying is still practised around Parita Bay, central Pacific Panama. This report describes the ways in which two different butchering methods, dependent upon fish size, affect the fish skeleton. Each butchering method leads to different patterns of bone loss and breakage. These data will be useful to archaeologists who wish to distinguish bet- ween fish processing and consumption sites. KEYWORDS: BONE DAMAGE, BUTCHERING METHOD, DRYING, FISH REMAINS, MIDDLE RANGE THEORY, PANAMA, PRESERVATION, SALTING, TAPHONOMY RESUMEN: Las investigaciones actualfsticas en torno a comunidades de Pescadores facilitan sobremanera la inferencia de tecnicas de peso a y la interpretacion de los datos arqueofaum'sti- cos relatives a peces. La deshidrataci6n de peces a traves de la salazon y el secado at sol y al viento continua siendo practicada en el entorno de la Bahi'a de Parita en el Panama Pacifico- Central. Este estudio describe los modos en que dos diferentes tipos de procesado del pescado, dependientes de la talla, afectan al esqueleto de los peces. Cada uno de los metodos implica di ferentes patrones de perdida y fracturacion de los huesos, datos que sin duda resullaran de inte- res para aquellos arqueologos que deseen diferenciar lugares de procesado y de consume en el registro fosil. PALABRAS CLAVE: DANOS EN HUESO, METODOS DE DESCUARTIZAM1ENTO, SECADO, RES- TOS DE PECES. PANAMA, CONSERVACION, SALAZON, TAFONOMIA
INTRODUCTION skeleton. Such «middle-range» research (Trigger, 1989: 361-367) underlines differences in bone dis- Many coastal fisherpeople around the world tribution and modification at sites where fish are dehydrate fish in order to preserve them for future processed and consumed. dietary or commercial use (Firth, 1975; Stewart, 1982, 1989; Michael, 1984; Essuman & Diakite, 1990; Belcher, 1994). Dehydration methods, ho- wever, vary with regard to fish species and size, se- GEOGRAPHIC AND CULTURAL CONTEXT ason, climate and local cultural tradition. Hence, it is informative to undertake ethnographic studies at This research was conducted near Parita Bay, a different localities in order to gain insights into the small tidal embayment in the north-western corner impact of specific processing methods on the fish of Panama Bay (Figure 1). Contemporary coastal 60 IR1TZOHAR & RICHARD COOKE fisherpeople are of mixed Native American-Afri- 2. Which skeletal elements are discarded and/or can-European descent, but they probably retain damaged during butchering? some pre-contact food procurement practices. Pre-Spanish (pre-A.D. 1520) dietary faunal sam- ples from Parka Bay archaeological sites attest to the intensive utilization of estuarine fish, espe- MATERIAL AND METHODS cially marine catfish (Ariidae), croakers (Sciaeni- dae), grunts (Haemulidae) and thread-herrings We examined 573 fish, belonging to 34 species (Opisthonema), and their transport inland (Cooke, and 9 families, which we recorded at two salting 1992). The transport offish in this lowland tropical and drying stations. 431 fish were prepared by country implies some kind of preservation (Cooke Francisco ViUarreal at Boca de Parita and 142 fish & Tapia, 1994). by Lucia Almendas at Aguadulce. We took stan- Modern fishing also concentrates on littoral es- dard measurements for each fish before it was but- tuarine waters. Individuals or small groups employ chered (total length-TL, standard length-SL and throw and gill nets, and hand-lines from dug-out total weight-BM). We observed two different but- canoes, with and without outboard motors. Alt- chering techniques: hough most modern catches are sold fresh, a few Method 1 (442 fish): A longitudinal cut is made families in the coastal settlements of Aguadulce, with a steel knife along the ventral mid-line of the El Rompio and Boca de Parka still process consi- fish, starting at the anus. This passes through the derable amounts of fish by salting and wind and branchiocranial region and usually ends near the sun drying. articular symphysis. The entrails and gills are re- Two questions guided the field study: moved and discarded. Then a few longitudinal or 1. Which butchering methods precede fish dr- oblique cuts are made on both sides of the fish to ying and salting? facilitate salt penetration. One cut often descends obliquely from the dorsal fin, straightening out over the caudal vertebrae and ending near the cau- dal peduncle. Butchered fish are salted in a brine tank and then dried whole, strung on a line, with the head still appended (Figure 2).
♦ H
\ V* ^ Aguduke o "&/ "•"
Pania Bay
Bocasde Parit/o^ r-4$^ FIGURE 2 igf Small fish, bulchered by method I, stung on a line (note the ly intact skull). mn*
FIGURE 1 Method 2 (131 fish): A longitudinal cut is made dorsally, starting at the base of the caudal fin and Map of Panama and Parita-Bay, showing location of the study sites. extending to the anterior part of the skull. The first FISH SALTING AND DRYING 61 dorsal spine and predorsal plate of large marine We collected 252 fish of the eight commonest catfish (Ariidae sp.) are removed and discarded be- species that were butchered by the two informants: fore the cut is made. Splitting the skulls of these «Arius» kessleri, «Arius» seemani, Sciadeops tros- and other larger fish, such as crevalle jacks {Ca- chelii, «Cathorops» furthi, «Cathorops» multira- ranx caninus), is sometimes facilitated by hitting diatus (Ariidae), Haemulopsis nitidus (Haemuli- the back of the knife with a blunt machete (Figure dae), Caranx caninus, and Oligoplites altus 3). Thus, the fish is opened from the dorsal region. (Carangidae). Each species was represented by The exposed entrails are discarded along with the more than 25 fish (Table I). 117 fish, belonging to gills. Then the fish is turned over, skin or scales four species, were selected for detailed osteologi- uppermost, and additional transverse cuts are cal analysis. These were skeletonised by macera- made on the flesh. tion in warm water and digestive enzymes (Davis & Payne, 1992) at the Smithsonian Tropical Rese- arch Institute's archaeozoology laboratory. Butchered by method I: (1) 32 silver (or shi- ning) grunt Haemulopsis (formerly Pomadasys ni- tidus) (Allen & Robertson, 1994: 153; FAO, 1995: 1161), (2) 29 box (or many-rayed) catfish «Catho- rops» multiradiatus (Allen & Robertson, 1994: Plate V, 4; FAO, 1995:881). Butchered by method 2: (1) 28 Pacific crevalle jack Caranx caninus (Allen & Robertson, 1994: Plate VIII, 4; FAO, 1995: 954), (2) 28 sculpted (or Kessler's) catfish «Arius» kessleri (Allen & Ro- bertson, 1994: Plate V, 10; FAO, 1995: 869). We defined our intraspecific index for breakage of individual skeletal elements as the total number of undamaged bones divided by the total number FIGURE 3 of bones expected in an intact fish. A relative mea- A large fish (Arius kessleri) butchered by the method 2. sure of similarity in overall breakage pattern bet-
,S,.„i„„ Sputtius '-'"""— "—'-"""
„„..,.,.,..„ *> ,,,,, '*» —""-"""" », „.,» '
W ..«, '
'—"" ."""' - ,,.,,, ,.,
« ia.7» '
» n.s*. 2
■* ,.,.,» '
—— —'- (-""""'" .,'",'.,,.„,» - ,,.5« '
TABLE I Frequency distribution for the number of Fishes, collected for this study, at each locality. 62 IRIT ZOHAR & RICHARD COOKE
ween each of the two taxa is expressed as Eucli- dean distance based on breakage index. The matrix of Euclidean distances was used in Multidimensio- nal Scaling (MDS) to separate butchering me- 8uk thods. MDS is an ordination procedure that com- Ot«J> Mttblfl SuiU-irJ U-iigil: intm.
presses multidimensional space onto a simple *".«» ...... -... »*. v.... <•,. two-dimensional representation (Borg, 1981). The »,—«' .W.I u I Ki.O-MU.0 2"* 1 !\,,'l >I4 11
method has no underlying assumptions about the 4 iAmny-*M/rr> 17-7 27 u ii ■no 22" 1) 2* ;., ....
normality or linearity of the data. The fit to the two ,..„.*- -7.O : M2.O49&0 .Ml, 2 iiaa.ua n dimensional model is evaluated by a stress factor, r «„„„„,...», U7.1 ii 9&0.23&0 2.2-1 5» IttO 2S2JB which ideally should be lower than 0.1. (". (nnhn .17.7 ».. y ;.n-. ■ 2.4.1 « 170.0-510(1 Since our data are not normally distributed we 1 pint: ■ . ■ i IK" 1 MS.O-JM.U :iu J applied the following non-parametric tests: (I) S 1 **, isn 12 imam 2*2 7 14 . ' 11 : Chi-square test, for comparing the observed num- (_,W, 96U « «,.M40».7 ,562 « 2V2 II 1960 ber of bones in a butchered fish with the expected C lt\-fn
KW- 1077/. 1 .7.4. .750 RESULTS l .*,.;. Will ' «K'4« 2M" 4
'.' !:':;■;■ ■.,. 4V....O 1 WM no. ' " Our initial observations on 573 fish from 34 «.,.,„„„,„,,„,
More detailed anatomical observations on the "'""""","" ')• ran 1 171,) I84J) 1 l«0 silver grunt, box catfish, Pacific crevalle jack and r „„„„., •*• ' mo »" ' 217.1, sculpted catfish demonstrate the following diffe- rences between the two butchering methods: TABLE 2 I) The body size (body mass and standard length) distribution (Figure 5) offish butchered by Descriptive statistics regarding body mass and standard length (SL) based on 573 fish belonging to 34 species, method I (Haemulopsis niiidus, Calhorops imilii- butchered in two different techniques. radialus) differed statistically from that of the fish butchered by method 2, (Caranxcaninus and Arius kessleri) (Mann-Whitney test; p<0.001). FISH SALTING AND DRYING 63
3) Although fish processed by method I do not 626 650 MtU'Qrt-1 Method -2 601-625 - exhibit damage or loss of any neurocranial bones 576-600 551-575- (Figure 7) other cranial bones, i.e., the cleithrum, 526-550 coracoid and the basypterygia are often damaged
476-500 (Table 3).
- 426-150
■5 376 400 J 351-375 •g 326-350 ■| 301-325 CtJJiumi-McilwlI 5 276 300
201-225
5 0 5 Fluency (7<.)
I IGURE4 Standard length (SL) frequency distribution of 573 fish be- FIGURE 6 longing to 34 species butchered by two different techniques. Percentage of fish skeletal elements lost due to butchering methods I and 2.
KM.6B- MtUwLl (L Mrlhort 2
~». Q Cjnir-'rl.^."-'
: '.:_,.. J . 5»«0- —— Co Mc- ..i 501 y...... C . _y - . AV,. ■ - ...... rtwMMi 451-47!- ** 4:«-i«i. •MS-
SI .!'!■ toiKvumnmai I -.-! K ■...: :. <: -
*"■ d
«■»»- FIGURE 7 ^—,—,—i—f The percentage of fish skeletal elements damaged by butchering methods I and 2.
FIGURE 5 Standard length (SL) frequency distribution of four fish spe- cies used in this study butchered by two different 4) Fish processed by method 2 exhibit damage techniques. to neurocranial elements (Figure 8) and also to other, extra cranial bones (Tables 3 and 4). This pattern is significantly different from that observed 2) Loss of most branchial region elements is in fish butchered by method 1 (Kruskal-Wallis; observed in both butchering methods (Figure 6). p<0.01). The most frequently damaged neurocra- The observed distribution of bones grouped by nial bones of both the crevalle jack and sculpted anatomical region differ significantly from the ex- catfish are: ethmoid, vomer, frontal, exoccipital, pected distribution in an intact fish (c2 test; parasphenoid, and basioccipital (Table 4). The su- p<0.001). However, if the branchial elements are praoccipital of the sculpted catfish is damaged excluded, bone loss by anatomical region did not when the first dorsal spine and its base are sliced differ statistically between the two butchering me- off. thods. 64 IRIT ZOHAR & RICHARD COO"KE
5) In butchering method 2 cranial bones are da- Species C. canimis (n=28) A. kessleri (n=28) maged and lost less frequently than neurocranial bones, but more frequently than when fish are but- Neurocranial bones No* % No* % chered by method 1. Elements most often affected Prefrontal 17 30.3% - - are: the cleithrum, coracoid, premaxilla, quadrate, metapterygoid, epihyal, preopercular, interopercu- Frontal 35 62.5% 21 37.5% lar, and urohyal (Table 3). According to MDS Epiotic 28 50% 2 3.57% analysis, morphologically distinct species butche- Exoccipital 48 86% 26 46.4% red by the same method display a similar bone bre- akage pattern (Figure 9). The MDS two-dimensio- Prootic 36 64.3% 27 48.2% nal representation explained 99% of the variance in Elhmoid 26 92.9% 28 100% the original data, giving a stress factor of 0.00088. Vomer 22 78.6% 28 100%
Tripus - - 18 32.1%
Malhod-1 Mithod-2 Supraoccipital 27 96.4% 28 100%
Sp« C. mIMdiatus H nitidus C. caninus A kesskli Parasphenoid 26 92.9% 28 100% [n=29| (n=32) ln=2«) |n=28) Basioccipital 22 78.6% 28 100% Riflion Cranial bones No' %" No' % No' % No' %
Appendicular Cleithfum 22 379% 33 51.6% 30 53.6% 51 96.4%
region Coracoid 24 41 4% 0 » 20 35.7% 35 62 5% ' The values are calculated for both sides (left and right).
Basypteiygia 7 12.1% 12 18.7% 3 5.35% 0 0
Oromandibular Premaxilla 0 0 , 1.56% 17 30.3% 12 21.4% TABLE 4 Oenlary 0 0 0 0 6 10.7% 18 32.14% The most frequent neurocranial bones damaged due
Quadrate 0 0 1 1.56% 8 14.28% 30 53.6% to butchering by method 2.
Metapterygoid 1 1.73% 1 1.56% 13 23.2% 17 30.3%
Qpercular Preopercular 0 0 2 3.12% 13 23.2% 32 57.1% studied, promise to enhance the universality of our, Inleropercular 0 0 0 15 26.8% 16 28.6% » as yet, preliminary observations. Opercular 0 0 0 0 2 3.45% 8 14,5% The two butchering methods employed by our 18 1, 19.6% Hyoid rag ion Ep,hV3l 0 0 0 0 32.1% informants for preparing fish for salting and drying Urohyal 1 3.4% 1 3.1% 4 14.3% | 16 57.1% show that bones from fishes that weigh less than ' The values are calculated for bo Hi sides (left and right). 400 g, and measure less than 325 mm SL, suffer less damage than bones of larger fish.
TABLE 3 Since all fish are gutted, both butchering methods The most frequent cranial bones damaged due lead to considerable loss of branchial arch bones. In to butchering by method 1 and 2. large fish, several neurocranial and some othe*r cra- nial elements are damaged. In small fish, the bones of the neurocranium are not damaged, but some ap- pendicular bones are considerably damaged. DISCUSSION AND CONCLUSIONS These data infer that bone and body part distri- butions should be very different at processing and consuming sites. Assuming that similar butchering Our data from Parka Bay, Panama, represent methods were employed, and providing no ani- the developmental stage of an empirical model ap- mals were around to devour discarded pieces of propriate for identifying certain butchering met- fish, bone surviving at processing sites would con- hods practiced at archaeological sites. We cannot sist predominantly of branchial arch elements, and demonstrate cultural continuity between our infor- in the case of marine catfish, whole and damaged mants' butchering methods and those of pre-His- panic fisherpeople in this region. Our fieldwork predorsal plates, dorsal spines and supraoccipitals. was restricted to two sites. Despite these limita- At consuming sites there would be minimal quan- tions, however, the world-wide occurrence of simi- tities of branchial arch bones. Neurocranial, cra- lar salting and drying methods, and the ample dis- nial and postcranial bones of large fish would tribution of the fish genera and families we show relatively increased physical damage compa- FISH SALTING AND DRYING 65
Caranx caninus Ventral view Dorsal view
Arius kessleri Ventral view Dorsal view
FIGURE 8 Caranx caninus and Arius kessleri skulls after butchering by method 2 (notice the breakage along the skull mid-line).
l.UU -
A.kessleti - method II M 0.50- • C.multiradiatus - method I C _o • q 0.00- H.nitidus - method I a C.caninus - method II • • 0.50 -
1 1 1 I 1 ■1.50 -1.00 -0.50 0.00 0.50 1.00 1.50 Dimension 1
FIGURE 9 Interspecific Euclidean distances based on pattern of bone loss and breakage were used for this Multidimentional Scaling (MDS) plot of four fish species. Patterns of butchering are indicated. 66 I KIT ZOHAR & RICHARD COOKE red to those of small fish. This suggest that survi- BORG, I. 1981: Multidimensional Data Represen- val rate of head bones of small fish in archaeologi- tations: When and Why? Mathesis Press, Ann cal deposits is greater than some specialist expect Arbor. (Butler, 1993;Lubinski, 1996). BUTLER, V. L. 1993: Natural versus cultural sal- Lastly, MDS analysis has demonstrated our abi- monid remains: origin of the Dalles Roadcut lity to distinguish between fish butchering bones, Columbia River, Oregon, USA. Journal methods based on patterns of bone loss and brea- of Archaeological Science 20: 1-24. kage. It may be possible to use a similar approach COOKE, R. G. 1992: Prehistoric nearshore and lit- in order to identify processing methods practiced by past populations. toral fishing in the eastern tropical Pacific: an ichthyological evaluation. World Archaeology In future communications, we plan to include 6: 1-49. data from other fish species, describe a typology of damage observed on individual bones and body COOKE, R. G. & TAPIA, G. R. 1994: Stationary in- parts, and compare the results of this experimental tertidal fish traps in estuarine inlets on the Paci- research with specific archaeofaunal assemblages. fic coast of Panama: descriptions, evaluations of early dry season catches and relevance to the in- terpretation of dietary archaeofaunas. Archaeo- Ichthyological Studies, Offa 51: 287-298. ACKNOWLEDGEMENTS DAVIS, S. J. M. & PAYNE, S. 1992: 101 Ways to deal with a dead hedgehog: notes on the prepa- This research was funded by the Smithsonian ration of disarticulated skeletons for zoo-archa- Tropical Research Institute (Panama), the Aharon eological use. Circaea 8: 1-6. Katzir Centre-Weitzman Institute, the Irene Levi Sala CARE Archaeological foundation and the ESSUMAN, K. & DIAKITE, B. 1990: L'utilisation du Morris M. Polver Scholarship Funds of Israel. Re- batistes (Batistes carolinensis) au Ghana et au search facilities were provided by the Smithsonian Senegal. FAO circ. Fishes 828: 1-23. Tropical Research Institute-Panama, the Cleveland FAO 1995: Guia FAO para la Identificacion de Es- Museum of Natural History, the Department of pecies para los Fines de la Pesca. Paciftco Cen- Zoology and the Department of Anatomy and tro-Oriental, Vols. 2, 3. Food and Agriculture Anthropology of Tel Aviv University. Organization of the United Nations, Rome. We gratefully acknowledge assistance in the FIRTH. R. 1975: Malay Fishermen. Their Peasant field from Conrado Tapia, Jose Tapia, Gonzalo Ta- Economy. The Norton Library. New York. pia, and the various fishfolk from Aguadulce and Parka who kindly let us study their work. LUBINSKI, P. M. 1996: Fish heads, fish heads: An experiment on differential bone preservation in We thank Prof. Israel Hershkovitz, Dr. Tamar Dayan and Dr. Eli Geffen for their helpful com- a salmonid fish. Journal of Archaeological ments and encouragement. Science 23 (2): 175-181. MICHAEL, S. M. 1984: The fishermen of India. In: Gunda, B. (ed.): The Fishing Culture of The World. Studies in Ethnology, cultural Ecology REFERENCES and Folklore, Vol. II: 651-676. Akademia Kiado, Budapest. ALLEN, G. R. & ROBERTSON, D. R. 1994: The Com- STEWART, H. 1982: Indian Fishing: Early Methods plete Divers' and Fishermen*' Guide to Fishes on the Northwest Coast. University of Washing- of the Tropical Eastern Pacific. Crawford ton Press, Vancouver. House Press, Bathurst. STEWART, K. M 1989: Fishing Sites of North and BELCHER, W. R. 1994: Butchery practices and the East Africa in the Late Pleistocene and Holo- ethnoarchaeology of South Asian fisherfolk. In: Van Neer, W. (ed.): Fish Exploitation in the cene. BAR. (International Series) 521, Oxford. Past. Annales clu Musee Royalde I'Afrique Cen- TRIGGER, B. G. 1989:A History of Archaeological trale, Sciences Zoologiques no 274: 169-176, Thought. Cambridge University Press, Cam- Tervuren. bridge.