Volume 103 1994 > Volume 103, No. 3 > Skeletal evidence of kava use in prehistoric Fiji, by Edward P. Visser, p 299-317
SKELETAL EVIDENCE OF KAVA USE IN PREHISTORIC FIJI
The aim of this study is to identify the cultural behaviour which led to temporomandibular joint (TMJ) degeneration in a skeletal population. The underlying changes in the masticatory apparatus will also be examined, to establish how these changes led to TMJ degeneration.
Although degeneration of the TMJ is a problem in contemporary populations (Hansson and Nordstrom l977), it is not a recent phenomenon. Studies of prehistoric and historic skeletal populations have shown that degeneration of the joint afflicted many people (Filce-Leek 1973; Griffen et al. 1979; Hodges 1991; Moffett 1974; Owen et al. 1992; Richards 1990; Richards and Brown 1981; Sheridan et al. 1991; Webb 1989; Whittaker et al. 1990). Although there is not a great deal of detailed information about TMJ degeneration in skeletal populations, studies do show that differences between populations exist. Furthermore, the frequency of TMJ degeneration is more common in females than males, although it is rare for more than 40% of either sex to be afflicted (Hodges 1991; Oberg et al. 1971; Richards 1988; Richards and Brown 1981; Sheridan et al. 1991; Webb 1989). The high incidence of TMJ degeneration in some prehistoric populations is suspected to result from repetitive overuse of the joint caused by industrial use of the teeth (Webb 1989).
The principal factors leading to TMJ degeneration differ from population to population. Most of the studies have found a strong correlation between tooth wear and TMJ degeneration, and a low association with tooth loss (Griffen et al. 1979; Hodges 1991; Richards 1988; Richards and Brown 1981; Owen et al. 1992; Sheridan et al. 1991), although two studies have found contrary associations, with a strong correlation between posterior tooth loss (Sheridan et al. 1991), or anterior tooth loss (Richards 1990) and TMJ degeneration.
A number of Australian studies have compared the incidence of TMJ degeneration among different skeletal Aborigine populations (Richards 1988; Richards 1990; Webb 1989). The strongest associations were found with aspects of the orofacial morphology, such as facial angles and ramal breadth, and TMJ degeneration. What these studies suggest is that, in populations with robust orofacial characteristics, tooth retention is important for survival and is therefore biologically selected for. Once teeth are lost, the masticatory - 300 apparatus is disrupted and TMJ degeneration follows, consequently chewing becomes inefficient.
What can be made of these conflicting results? Clinical and skeletal evidence suggests that biomechanical forces and occlusal dysfunction through tooth loss and/or tooth wear play important roles in TMJ dysfunction (Brown 1965; Griffen et al. 1979; Hansson and Nordstrom 1977; Oberg et al. 1971; Whittaker et al. 1990). Recent studies have maintained that tooth loss and tooth wear are not the only variables involved, and have argued that TMJ degeneration has a multifactorial basis (Parker 1990;Richards 1990). Consideration should therefore be given to the physiological adaptability of the entire masticatory system.
In the present study a number of factors that may have caused TMJ degeneration are considered. (1), the association between tooth wear, tooth loss and TMJ changes; (2), the correlation between condyle skewing and temporal eminence degeneration; (3), an assessment of sex differences in three facial diameters, and TMJ changes. Lastly, consideration will be given to the cultural activity leading to the biological responses causing TMJ degeneration.
Figure 1. The temporomandibular joint. Representation of the structures of the temporomandibular joint in the open and closed positions.
The functional biomechanics involved in chewing are complex. Briefly, the TMJ is a synovial joint between the temporal mandibular fossæ and mandibular condyles (Fig. 1). The joint space between the articulating condyle and temporal bone contains a nonrobust fibrocartilagenous disc, which is thickest at its posterior end (Ryan 1989:1422). The convex shape of the condyles and concave mandibular fossæ allow the main movement of the mandible to be a hinged upward and downward movement, but gliding protrusive and retractive movements are also possible. Gliding actions can be accomplished by pivoting on one condyle so that the other condyle glides from the fossa in a scything movement (Shipman et al. 1985). Five main muscles are responsible for effecting these movements: the digastric, medial and lateral pterygoids, temporalis and masseter (Fig. 2.).
Molar mastication involves an asymmetrical chewing cycle because most people chew on one side of the mouth at a time and frequently have a preferred chewing side (Shipman et al. 1985:243). The way in which the chewing cycle works is that, during the opening movement, the lateral pterygoid relaxes on
Figure 2. The main muscles involved in mastication, The arrows indicate the approximate direction of muscle actions
the chewing side, which keeps the condyle in the mandibular fossa and slightly skews the mandible to that side. On the other side, the contracting lateral pterygoid pulls and holds the condyle on the eminence. During the chewing power-stroke, the medial pterygoid brings the mandible onto the midline, while the masseter and temporalis muscles contract to move the mandible simultaneously upward and medially to crush the food. The balancing condyle remains on the temporal eminence and absorbs the chewing force. During the closing stroke, the lateral pterygoid relaxes on the balancing side, returning the condyle to the mandibular fossa. The masseter and medial pterygoid on the working side relax, and the mandible returns to the resting position.
The greatest amount of shear force acts upon the cartilage on the balancing side as the condyle is pulled and slightly rotated out of the mandibular fossa. Pathological problems first arise on the subchondral bone, and thereafter the fibrocartilage begins to degenerate (Ryan 1989:1429). The pathology of TMJ bony degeneration involves two types of remodelling. Progressive remodelling is the result of articular cartilage being converted into bone by osteoblastic activity. This remodelling involves gradual changes to the temporal eminence, from a convex shape to an increasingly concave shape, but these changes do not involve osteoarthritic changes. The other type of remodelling is regressive, which involves the loss of cortical bone by osteoclastic resorption. Basically this is a degenerative process which leads to osteoarthritis. In severe cases, the temporal eminences and condylar heads are eroded and the efficiency of the joint is compromised (Cherrick 1979). Under normal physiological conditions progressive and regressive remodelling occur simultaneously, so that the contours of the TMJ articular surfaces are maintained but, when the joint experiences excessive stress, this relationship is disrupted, so that one remodelling process becomes dominant.
MATERIALS AND METHODS
The skeletal remains of a discrete population of 58 individuals were excavated from Sigatoka, Fiji, in 1987 and 1988 (Fig. 3). The burial site has been radiocarbon dated to A.D. 80 ± 70 years (Best 1989). The sample size used in this study was restricted because many individuals were represented only by fragmented bones, while, on a number of the better-preserved skeletons, the bony articular surfaces of the TMJ had decayed. Consequently, only 18 adult male and 14 adult female TMJ's could be studied. The skeletal material is currently held in the Department of Anatomy and Structural Biology, University of Otago.
Sex assessment was based on a number of indicators. The primary method used was based on the angle of the greater sciatic notch, following Genoves- 303
(1959) and Krogman (1962). Secondary indicators of sex — such as the diameter of the femoral head (Thieme 1957), mastoid size, the prominence of the brow ridges, and the prominence of the submental arch on the mandible — were used to corroborate initial sex identification, and to establish the sex of those individuals lacking a pelvis.
Ages of immature individuals were based on the degree of epiphyseal bone union, following McKern (1972). Adult ages were estimated from tooth wear by relating known epiphyseal fusion events of bones in immature individuals to tooth wear. An assessment of the rate of tooth wear in relation to age was then made and extrapolated to the adult population. This method has greater accuracy in small homogeneous populations than other adult ageing methods (Miles 1962; Nowell 1978; Richards and Brown 1981). Secondary assessment of adult age was based on comparative advancement of age-related, degenerative diseases, for example, the degree of synovial joint degeneration in the vertebræ and cortical bone loss in the tibia.- 304
Most of the Sigatoka skulls are in a poorly preserved state, and were therefore excluded from the analyses. Three mandibular measurements — ramus breadth, ramus height, and mandible angle — were used to identify whether differences in the size of the masseter muscle occurred between males and females. The measurements taken follow Martin (1928).
A subjective scoring scale was developed for progressive remodelling on the temporal eminence. Points on this scale are:—
The different stages of progressive remodelling are illustrated in Fig. 4. To locate the area of remodelling, the surface was divided into lateral, central and medial areas.
Figure 4. Progressive remodelling in the temporomandibular joint. Sagittal section of the mandibular foss and temporal eminence, illustrating the four stages of progressive remodelling on the temporal eminence.
Degrees of TMJ regressive remodelling were scaled according to Richards and Brown's ratings (1981:297). These scalings are:—
Figure 5. The stages of regressive remodelling on the mandible condyles and temporal articular surface of the temporomandibular joint.
Fig. 5 illustrates each stage of regressive remodelling of the condylar and temporal surfaces. Each articular surface was divided into nine regions. On the temporal articular surface three anteroposterior divisions are represented: the temporal eminence, mandibular fossa, and postglenoid region. These are further subdivided into lateral, central and medial areas. On the mandibular condyles the divisions are taken as three 60° sectors from a hypothetical axis (Richards and Brown 1981).
The mediolateral orientation of the condyles was assessed to determine a possible correlation between the angle of condyle skewing and TMJ changes. The angle of orientation was calculated by measuring the midpoint of the maximum mediolateral breadth of the condyle (Fig. 6). The midpoint of the condyle breadth was aligned with the first premolar to create a 90° angle. When the angle deviated more than 10° from the 90° angle, the deviation was taken as indicating condyle skewing. Condyle skewing was in intervals to the nearest five degrees.
Tooth wear is often cited as a cause of TMJ regressive remodelling (Richards and Brown 1981; Whittaker et al. 1990). To assess this association in the Sigatoka sample, the rates of tooth wear were rated according to Molnar
Figure 6. Orientation of the mandibular condyles, The angle of condyle orientation is calculated from the mid point of the maximum condyle breadth to the second premolar. The angle at which the condyle skews more than 10° from the 90° angle is taken to indicate skewing.
(1971). Molnar's scale ranges from one, for no tooth wear, to eight, for wear on the root surface. Rank correlations were used to calculate whether an association existed between mean tooth wear and TMJ regressive remodelling. A separate correlation was made between molar tooth wear and TMJ regressive remodelling. Molar tooth wear was assessed separately, because of the important role that molars play in crushing food, and the extent to which they absorb the force exerted during chewing, the intention being to test the theory that severe wear in the molar teeth is closely correlated to TMJ regressive remodelling.
Of a sample of 13 females only two individuals showed slight TMJ regressive remodelling. Although regressive remodelling was found on the anterior eminence and central posterior condyle, 58% of the females from this same sample exhibited progressive remodelling on the central temporal eminence (Table 1). No changes were observed on the mandibular fossæ or postglenoid tubercle.
Table 1. Incidence of progressive remodelling on the temporal eminences of women.
In contrast with females, males experienced only regressive remodelling. Of the 15 males, 12 showed differing stages of TMJ regressive remodelling, which occurred equally on both sides of the joint. A high frequency of regressive remodelling (70%) occurred on the temporal articular eminence (Fig. 7). A lower frequency (20%) was noted in the mandibular fossæ, where it was mostly restricted to the central areas. Little regressive remodelling was observed in the postglenoid fossæ areas.
In severe cases, regressive remodelling of the temporal eminence was such that, either the eminences were eroded to the level of the mandibular fossæ, or the reactive bone on the posterior aspect of the degenerated temporal eminences extended over the mandibular fossæ (Fig. 5). This made it impossible for the condyles to articulate within the mandibular fossæ, thereby forcing them to articulate on the eroded eminences.- 308
Figure 7. Frequency distribution of regressive remodelling on the nine regions of the male temporal articular surface. Sample size of 26 cases in each category.
Figure 8. Frequency distribution of regressive remodelling on nine regions of male mandibular condyles. Sample size of 24 cases in each category.
On male mandibular condyles regressive remodelling was greatest on the central parts of the condyles, where 44% of the observed changes occurred. Elsewhere on the condyle, regressive remodelling was evenly distributed (Fig. 8).- 309
A characteristic pattern of tooth wear among the Sigatoka population was that, as wear advanced, there was a corresponding increase in tooth loss, particularly of the molars. The initial effects of this wear were found on individuals at a young age. To establish whether advancing dental attrition and loss of molar tooth support among males was concomitant with advancing TMJ regressive remodelling, rank correlations were calculated (Table 2). The results show that, among males, there is a very significant correlation between total tooth wear and TMJ degeneration (P≤0.002). This correlation is equally strong when only the molar teeth are included in the calculation (P≤0.002). When we consider this association among females, TMJ regressive remodelling was negligible even though they experienced similar wear on all teeth. Further, no statistical correlation was found between total tooth wear and TMJ degeneration in females, nor was there any association between molar wear and TMJ regressive remodelling (Table 2). Certainly, the causal relationship between tooth loss, tooth wear and TMJ regressive remodelling is not secure.
Table 2. Male TMJ regressive remodelling correlations.
During analysis, it was noted that males who experienced more severe TMJ regressive remodelling also displayed severe unilateral skewing of the condyle. Characteristically, the medial end of the condyle had rotated posteriorly. Skewing was observed in male cases with only stage 1 degeneration (Table 3). Degeneration and condyle skewing showed a very significant correlation (P≤0.002). No condyle skewing was observed among females.
Table 3. Female TMJ regressive remodelling correlations
Although female mandible dimensions were more gracile than males (Table 4), only the minimum ramus breadth was found to be significantly dimorphic (P≤0.03). Ramus height approached significance (P≤0.08). The larger male ramal dimensions demonstrate that they had greater masseteric musculature which enabled them to achieve a more forceful bite than females.
Table 4. Student t test of male and female mandibular variables.
The different correlation results, from male and female tooth wear and tooth loss patterns compared with TMJ regressive remodelling, show that there is no primary relationship between teeth and joint degeneration. The patterns of remodelling between the sexes are too different. In light of this, there must be another causal mechanism contributing to TMJ regressive remodelling.
Among males, the temporal articular eminences and the lateral portion of the condyles absorbed most of the force generated during chewing. Early indications of regressive remodelling were evident in young adult males, and rapidly became more severe in older male age groups, suggesting repetitive and forceful chewing.
Among the Sigatoka males, there is a characteristic pattern of temporal eminence and condyle destruction, indicating that mandibular movement occurred in a scything movement. One condyle acted as a pivot in such a way that, as it was pulled out of the mandibular fossa, it was rotated slightly. Simultaneously, the other condyle was pulled out of the mandibular fossa and drawn forcefully on an angle across the eminence. The constant repetitive skewing force on the pivot condyle led to an adaptive adjustment where the condyle head was remodelled, twisting it from the line of usual orientation.
One study has demonstrated that, even after relatively short periods of forceful chewing, short-term changes can occur in the mandible's position (Tzarkis et al. 1989). Masticatory muscles relax within 30 minutes of a forceful chewing episode, enabling the jaw to becomes more mobile, and move over the temporal eminences more freely. Although the clinical trial by Tzarkis et al. (1989) did not identify any long-term changes, it is interesting to consider the effects of long-term, forceful, asymetrical chewing and elastic - 311 mandibular movements on the TMJ. It is well known that constant pivoting on one condyle, while the other moves forward and medially over the temporal eminence in a scything action, can initiate degenerative changes within the joint. Such changes occur because soft tissue attempts to resist shear forces placed upon it, which act to twist the cartilage. In such cases, especially among atypical one-sided chewers, the joint tissue is unable to cope with the force loading and shear stress, leading to cartilage dysfunction. Continuous forceful loading can initiate changes in the TMJ bony surface, which attempts to adapt to the changing conditions (Westesson 1985).
In most males, the speed of destruction of the TMJ proceeded at such a rate that bony deposits were not laid down fast enough to prevent erosion. Consequently, most males experienced varying degrees of cortical erosion and osteoarthritis. In some cases there is clear evidence of severe eburnation (Fig. 5). In most females, the force placed on the TMJ was such that bony remodelling proceeded at a rate fast enough to adapt and compensate for the changing bony status.
There does not seem to be any histological reason why female TMJ tissue should have been more resistant to these forces. On the contrary, clinical and skeletal evidence points to females being more susceptible to TMJ degeneration (Grosfeld et al. 1985; Hodges 1991; Sheridan et al. 1991). Moreover, there are indications that condyle skewing is also more common among females (Oberg et al. 1971).
Among the Sigatoka population, TMJ regressive remodelling occurred 12 times more frequently in males than females. It would seem that although both sexes in the Sigatoka population had an equally abrasive diet, the chewing action of females did not involve the repetitive, forceful, and shearing movements that males experienced. Therefore, it is clear that sex-specific behavioural activities were responsible for this difference. This raises two questions: what activities were the males doing, and why did they continue to chew forcefully when, in many cases, each mandibular movement might have been accompanied by considerable pain?
A resolution to these questions may lie in cultural activities. Tooth wear patterns among males excludes the possibility that TMJ regressive remodelling was the consequence of processing fibrous plant material for fishing nets, lines, or any similar activity.Industrial wear patterns were evident only among some of the women. An answer to the greater frequency of TMJ regressiveremodelling in males compared with females can be found in kava processing and use.
Kava is a narcotic beverage made by pulverising the roots of the plant Piper methysticum Forst. f. either by pounding or, more commonly in the past, chewing the root. The processed product is then mixed with water and drunk. - 312 The kava drink acts as a tranquilliser. Processing the roots releases the relaxants, pyrones, methysticin, dihydromethysticin and dihydrokawain. Their general effect inhibits sensory stimuli and motor responses (Pfeiffer et al. 1967).
Kava, Piper methysticum Forst. f., is a binominal of a wild form of Piper species, P. wichmannii c. DC. (Lebot and Cabalion 1988). P. wichmannii c. DC. is a member of the genus Piper indigenous to Island South-east Asia and Near Oceania; however, P. wichmannii, and its binominal cultivated form P. methysticum, is the only species from which a particular chemotype, kavalactones, has been found (Lebot 1991). The results of the chemotype, and genetic studies led Lebot and Levesque (1989) to suggest that kava was selectively cloned and developed from an ancestral plant in Northern and Central Vanuatu. Vanuatu continues to have the greatest range of the wild forms of Piper species and the cultivars of P. methysticum (Lebot and Cabalion 1988). P. methysticum plants found on islands to the east of Vanuatu, including Fiji, and in localities in New Guinea, could have spread there only through human agency from North and Central Vanuatu.
Kava seems likely to have been domesticated in Northern or Central Vanuatu after first settlement of the area by the Lapita people. Glottochronology suggests that domestication occurred sometime after 1200 B.C. (Crowley 1990; Lebot 1991). Domestication of kava would have occurred at about the same time that Fiji was settled, around 1100 B.C. probably by the Lapita people (Palmer 1968). However, kava was very unlikely to have been taken to Fiji then. Linguistic reconstructions indicate that kava was introduced into Fiji after first settlement, but still within the Lapita cultural complex period, which may have lasted to sometime between 510 and 110 B.C. in Fiji (Birks 1973; Frost 1979; Palmer 1968). The evidence for a date of kava introduction into Fiji is based on the fact that the Fijian name for kava is yaqona. The word yaqona is not related to the languages of Northern or Central Vanuatu (Crowley 1990).
When Proto-Polynesian became a distinct language between 800 and 500 B.C., the word of which kava is a modern reflex appeared. It probably originated from the Proto-Polynesian word *kawa ‘sour’ Words associated with kava drinking also appeared at this time (Crowley 1990; Geraghty 1983; Grace 1969). At about the same time cultural paraphernalia associated with kava-drinking appeared: the pottery drinking cup, and mixing and serving bowls have been found in the archæological record (Green 1974).
Ceremonial kava bowls used in Fiji are wooden, flat-rimmed, open bowls with a shallow slightly, rounded base (Gordon-Cumming 1882:208a). Similar shaped pottery bowls have been recovered from early Lapita sites in Tonga and later sites in Samoa; these may have been kava bowls (Green 1974; Green - 313 personal communication 1993). Similar flat-rimmed, shallow bowls have been recovered from a Lapita site at Yanuca, near Sigatoka, which has been dated at 710 B.C. (Palmer 1968). A problem with the idea that the pots from Yanuca may have been associated with kava use is that the linguistic evidence does not support a very early introduction of kava in Fiji (Crowley 1990; Geraghty 1983; Green 1974). The pots' presence would not have allowed sufficient time for the Proto-Fijian language to develop, or allow sufficient time for kava to be domesticated and gain recognition in North and Central Vanuatu.
No flat-rimmed, shallow bowls have been recovered from Sigatoka. However, there are a number of bowls recovered from the early period (510 B.C.±90) which have the correct dimensions for kava bowls, but lack the flat-rimmed sides (Birks 1973: Figs 46, 55). It is possible that the flat-rimmed kava bowls used today are a more recent innovation, and perhaps owe their origin to Tonga. Certainly, the Fijian word for the serving bowl, taanoa, is representative of a Tongan word borrowing (Crowley 1990), and there is also a suggestion that different kava strains were also introduced from Tonga (Lebot and Levesque 1989). One conclusion that can be drawn from this is that interaction between Fiji and Tonga occurred after kava was first introduced in Fiji, and that a different kava bowl style was introduced after Sigatoka was abandoned.
The social importance of kava in the past was noted by many 19th century visitors to Fiji (Gordon-Cumming 1882; Rowe 1860; Seemann 1862; Williams 1858). According to their accounts, kava was used overwhelmingly by males and the suggestion is that this was also the case in the prehistoric era (Rowe 1860). Kava drinking was, and still is, an important feature of Fijian social occasions. The elaborateness of a kava drinking ceremony depends on the importance of the occasion. It seems likely that kava drinking was just as important on social occasions historically, and there is no reason to assume that this situation was different in the prehistoric era.
At the time of early European contact, kava root was processed by chewing or grating (Williams 1858:141). The chewing method involved cutting the kava root up into small pieces and then distributing it to a group of young men to chew (Rowe 1860:141; Williams 1858:141). They chewed the kava into a finely masticated fibre which was seen to “take some time” (Gordon-Cumming 1882:51). On the basis of these observations, processing kava by chewing would require repetitive force and a reasonable amount of time. This would certainly involve the necessary conditions to initiate TMJ dysfunction.
One of the effects of kava chewing is that it inhibits sensory stimuli to the brain from the proprioceptor nerves, located in the periodontal ligaments and TMJ. As a result, the shear and compressive forces produced during chewing - 314 would be somewhat uncontrolled. Any pain experienced would be at least partially inhibited by the released relaxants. Even small amounts of kava inhibit the recognition of external stimuli and produce a lack of muscle co-ordination (Frater 1952). A reduction in sensory stimuli and a lack of muscle control while chewing forcefully, particularly in a scythe-like action, could quickly lead to cartilage deterioration and TMJ dysfunction.
This study has found that the age of onset, severity and occurrence of TMJ regressive remodelling in the Sigatoka population were primarily dependent upon forceful and dysfunctional chewing action. Mediolateral condyle movement placed shear force on the TMJ which quickly caused cartilage degeneration. In cases where this movement was prolonged and excessive, cortical bony erosion occurred. Frequently, remodelling was unable to respond quickly enough and in these cases osteoarthritic degeneration occurred. Attempts to adapt to the occluding movement continued through changes in mandibular condylar orientation. These changes were most obvious in individuals with advanced degeneration. The loss of molar support did not seem to have a primary causal effect.
Contrary to other skeletal population studies, degenerative osteoarthritis was found almost exclusively in males. The reason for this is that chewing kava was a male specific activity.
I should like to thank several people who read drafts of this paper, particularly John Dennison, Alistair Hay, Philip Houghton, Nancy Tayles and Richard Walter. I should also like to thank Roger Green for his help and, Robbie McPhee for his assistance with the illustrations. The Sigatoka skeletons were in an extremely degraded condition and, without the skills and tireless work of Dilys Johns, the bones would not have survived transport or been available for analysis. Dr Simon Best directed the excavation. I also acknowledge with gratitude the financial assistance of the Lounsbery Foundation.