Reconstruction of periodicity of repetitive linear enamel hypoplasia from perikymata counts on imbricational enamel among dry-adapted chimpanzees (Pan troglodytes verus) from Fongoli, Senegal

Repetitive episodes of linear enamel hypoplasia (rLEH) are commonly observed on teeth of recent and fossil apes (Skinner,1986; Moggi-Cecchi and Crovella,1991, 1992; Skinner et al.,1995; Guatelli-Steinberg,1998, 2000; Guatelli-Steinberg and Skinner,2000; Skinner and Hopwood,2004). They reflect developmental disturbance of enamel formation in the immature animal (Goodman and Rose,1990; Skinner and Goodman,1992). Despite their striking appearance and ubiquity, their cause(s) and, hence, significance remain uncertain. Some consider linear hypoplastic events to have an ‘irregular periodicity’ (Smith,2008) due to an exogenous etiology, whereas others are so struck by the apparent regularity of LEH in apes they link their timing to seasonality (Skinner,1986; Brunet et al.,2002; Skinner and Hopwood,2004). The impression that uniformity of spacing between episodes of LEH means the passage of equal amounts of time between stress events can be tested by counting perikymata between events, since the periodicity of perikymata is believed to be invariant within the tooth (Smith et al.,2007a). Perikymata are the surface expression of striae of Retzius whose periodicity can be determined by counting circadian cross-striations between striae (AKA “long count”) in thin-sections of dental crowns. Since hominoid long counts are known to range among genera from 4 to 12 days, it is recommended that taxon-specific ranges of variation in the timing of striae formation be utilized (Smith et al.,2007a; Smith,2008) for the prediction of time-related events in dental formation. Chimpanzee long counts are reported to range from five to nine (Reid et al.,1998; Dean and Reid,2001; Schwartz et al.,2001a; Smith et al.,2007b); however the large majority (ca. 90%) have modal periodicities of six or seven (Ramirez-Rozzi and Lacruz,2007; Smith et al.,2007a; Smith et al.,2010). Using these values, we can predict the number of perikymata that should occur between allegedly annual or semi-annual repetitive episodes of rLEH.

Skinner and Hopwood (2004) hypothesized that, fundamentally, rLEH episodes in low latitude great apes from Africa and Asia are linked to the animal's experience with yearly moisture cycles. They report that chimpanzees (Pan troglodytes) and gorillas (Gorilla gorilla) from West Africa and orangutans (Pongo pygmaeus) from Borneo typically experienced twice-yearly episodes of developmental stress, lasting about 6 weeks, correlated with double rainy minima and maxima per year due to latitudinal migration of the inter-tropical convergence zone. They considered both seasonal malnutrition and disease as possible explanations for rLEH. While not mutually exclusive etiological factors, it is likely that phenologically linked stress would be more long lasting than are febrile diseases. Thus, if one could locate apes, which are ripe fruit specialists (Wrangham et al.,1998), living in habitats with a single, marked dry season, the occurrence of rLEH should be annual and of long duration. We define “circannual” as meaning “occurring approximately or on average every 12 months” (Heideman and Bronson,1994; Paul et al.,2008). We define “stress” in this article as a physiological disruption of the secretory phase of enamel formation that leaves macroscopically visible areas of thinned enamel, commonly termed enamel hypoplasia.

Our study is meant to test whether rLEH observed in chimpanzees from Senegal, who experience a marked annual dry season, is circannual. In that imbricational enamel on canine teeth among chimpanzees is estimated to form in the range of 5.5–7.1 years in males and 4.7–5.9 years in females (Schwartz and Dean,2001) it is predicted that, in those instances where LEH events span the imbricational enamel, there should be six or seven marked LEH events per canine crown. Given documented long counts of typically 6 and 7 days/stria among chimpanzees (noted above), we predict that the average number of perikymata between what are suspected to be annual events should range between about 52 and 61.

Finally, in addition to finding the interval between stressful episodes, we hope to be able to reconstruct the duration of the stressful episodes themselves. The greatest percentage of fruit at Fongoli is eaten in the late dry season, whereas there is a relative lack of fruit at the height of the wet season (Pruetz,2006). Theoretically, then, the more stressful times physiologically for an immature ape in this habitat could be the dry season (lack of moisture, high heat) and/or the wet season (under-nutrition). Given the remarkable length of the dry season at Fongoli, we predict that the reconstructed duration of enamel hypoplasia can provide a means of choosing between these two scenarios.

MATERIALS

High-resolution casts of teeth were obtained from three recently deceased chimpanzees (Pan troglodytes verus) (Table 1) at the Fongoli study site in southeastern Senegal (12°39' N, 12°13'W). Fongoli is a mosaic savanna environment, including open (woodland, bamboo, and short and tall grassland) and closed (gallery forest and eco-tone forest) canopy habitat types (Pruetz et al.,2002; Pruetz,2006) at the extreme geographic range of chimpanzee distribution. This habitat is unusual compared to that of other chimpanzees (Pruetz,2006, 2007; Pruetz and Bertolani,2009). Fongoli's climate encompasses an extraordinary seven-month dry season from November to May, during which time streams dry up, and chimpanzees rely on a few sparsely dispersed waterholes within their 86 km² home range (Pruetz, unpublished data). The Fongoli community of chimpanzees has been studied continuously since April 2001 but was not habituated for systematic behavioral data collection until 2005. Based on estimates from nest surveys, chimpanzees occur here at a density of 0.09 individuals per km² (Pruetz et al.,2002) with an estimated community size that has averaged 31 over the last 6 years. The Fongoli diet is less diverse than that of chimpanzees living in more forested sites but is similarly frugivorous, with some fruit, usually figs, ingested in every month (Pruetz,2006).

Our study is based on six canine teeth from three animals (Table 1). Subject A is a sub-adult male whose partial remains were recovered at one of the few water sources available at the peak of the dry season (ca. April 2003). He was found prior to habituation so his identity and history are unknown. Subject B is an older adult female represented only by some loose teeth and a skullcap found in a cave in June or July 2007. The bone was dry so year of death is uncertain. Her skull sutures were completely fused. Frito is an adolescent male estimated to be 10–11 years old at death in April 2010; his remains were found between two distant water sources.

At the time of our analysis, it is not known precisely when these animals were forming their canine teeth; consequently, there is no implication here that they necessarily were responding to the same environmental conditions. In other words, there is no assumed correspondence among the three events labeled as LEH 4 (for example) on each of the three animal's canine teeth. However, in the case of Frito, for whom all four canine teeth are preserved, there is clear correspondence among his teeth in numbered LEH; as most easily seen in LEH 5, which is imaged (Fig. 1) and discussed below.

Abbreviation

METHODS

Dental imaging procedures

For this study, canine teeth only were selected for imaging, as this tooth type maximizes recovery of developmental information (Skinner,1986). Perikymata are clearly visible on unworn chimpanzee canine teeth; contra Schwartz et al. (2006). Perikymata are the surface expression of internal striae of Retzius whose periodicity in chimpanzees is reported to range from 5–9 days per perikyma (details in Table 3). Thus, in theory, the timing and duration of episodes of LEH are reconstructible by multiplying the number of perikymata (Pk) within and between hypoplastic grooves by their known or inferred number of days required to form a single perikyma. Since we are using casts and are not in a position to section these precious teeth as yet, we relied on the literature for these figures following the practice advocated by Guatelli-Steinberg (2008; Guatelli-Steinberg and Reid,2008) where ranges and modal values for long counts are considered.

Table 3. Reconstructed periodicity of rLEH events (months) in imbricational enamel based on published long counts (number of ‘circadian’ cross-striations between striae of Retzius) in Pan

LEH					Possible long counts between striae
Interval	Animal		PK counta		If 5b	if 6c	if 7d	if 8e	if 9f	if 10g
1–2	Adult female (B)		29.5		4.8	5.8	6.8	7.8	8.7	9.7
	Frito		30.8h		5.1	6.1	7.1	8.1	9.1	10.1
2–3	Adult female (B)		36		5.9	7.1	8.3	9.5	10.7	11.8
	Sub-adult male (A)		38		6.2	7.5	8.7	10.0	11.2	12.5
	Frito		38.8h		6.4	7.6	8.9	10.2	11.5	12.7
3–4	Adult female (B)		41.5		6.8	8.2	9.6	10.9	12.3	13.6
	Sub-adult male (A)		36		5.9	7.1	8.3	9.5	10.7	11.8
	Frito		38.3h		6.3	7.5	8.8	10.1	11.3	12.6
4–5	Sub-adult male (A)		44		7.2	8.7	10.1	11.6	13.0	14.5
5–6	Sub-adult male (A)		41		6.7	8.1	9.4	10.8	12.1	13.5
6–7	Sub-adult male (A)		32.7		5.4	6.5	7.5	8.6	9.7	10.8
	Mean		37.0		6.1	7.3	8.5	9.7	10.9	12.1

• a Average of two or more counts
• b Long count of five: one of three Tai animals with this periodicity (Smith et al., 2010); also could not be ruled out in two of 75 animals (Smith et al., 2007b).
• c Long count of six: 41 of 75 animals (Smith et al., 2007b); (Schwartz et al., 2001) ; mode in 61 (of 75 tested) animals (Smith et al., 2007a).
• d Long count of seven: 20 of 75 animals (Smith et al., 2007b); modal value among 20 chimpanzees (Schwartz et al., 2001); two of four animals (Reid et al., 1998).
• e Long count of eight: could not be ruled out in 1 of 75 animals (Smith et al., 2007b); (Schwartz et al., 2001); two of four animals (Dean and Reid, 2001); (Reid et al., 1998); this was reduced to a count of seven in one animal by T. Smith (2004) cited in (Ramirez-Rozzi and Lacruz, 2007).
• f Long count of nine: (Schwartz et al., 2001); (Dean and Reid, 2001).
• g Long count of 10: not reported in chimpanzees.
• h Average of counts on homologous LEH for both lower canines.

Molds of affected enamel surfaces were taken in the field using Coltene President Plus Jet impression material. Casts were later made in Araldite MY 753 epoxy resin with XD 716 hardener (Ciba-Geigy) for examination by digital microscopy. Photographic images, covering the whole crown surface were recorded at high magnifications ranging typically from 50 to 100× with a VHX-100 Keyence digital microscope. Separate images were combined with Adobe Photoshop Elements 6 to form a photomontage of the enamel surface; these could be examined at very high enlargements allowing each perikyma to be labeled and counted. Often, rLEH events that were obvious at low or no magnification could not be visualized under the lighting conditions required to show perikymata. Consequently, low power (ca. 10–15×) pictures of a whole crown were recorded and the rLEH numbered consecutively starting at the apex: 1, 2, etc. Registration of an LEH between the low and high magnification pictures was accomplished by noting small irregularities or imperfections in the cast (e.g., bubbles/scratches) that could be located on both. Perikymata were labeled and counted at least twice on occasions separated by 6 weeks or more. Further checks on replicability of perikymata counts were obtained where possible by imaging more mesial and distal locations on each crown. Boundaries of a hypoplastic groove are often not sharp but gradual which requires repeated counts within and between hypoplastic grooves (e.g., from deepest perikyma on one groove to the deepest on the next groove, or from the highest perikyma on the perceived shoulder of one groove to the highest perikyma on the next.

An alternative approach, developed here, to avoid the difficulty of defining the boundaries of a hypoplastic groove is termed, by us, a “waveform” model. This is a new approach to the analysis of hypoplastic enamel predicated on the assumption that growing mammals usually experience sub-optimal nutrition (White,1978) while growing their teeth. It is a working model created for this study to deal with the observed enamel surfaces of the Fongoli chimpanzee teeth and is not necessarily applicable to other chimpanzees. The waveform model treats the undulating enamel surface as an (a)symmetric wave, a portion of which (deemed normal) is above a hypothetical flat surface and a portion below (deemed hypoplastic). The hypothetical flat surface is defined as halfway between the highest and lowest points of the undulating surface (Fig. 2). Profiles of undulating enamel surfaces were created by taking precisely focused pictures with the digital microscope of side-views of crown surfaces and measuring distances between high and low (LEH groove depth) points. Calibration was performed for every magnification using objects of known dimension.

In addition, depth of a hypoplastic groove was measured at high magnification (X300) using the “depth to defocus” capability of the digital microscope. These latter depth measures are typically less than those measured by the previous method since they measure only the highest points captured within the field of view of the lens (discussed further below). Apparent severity of a hypoplastic groove was assessed by comparison of measured depth with an ordinal score of visibility of a hypoplastic groove (ranging from none, through trace, mild, moderate, and strong [modified from Skinner and Hopwood (2004)]. Perikymata counts were obtained where possible for the occlusal wall of defects as recommended by (Hillson and Bond,1997); this area representing the period of disrupted growth. We also counted perikymata in the cervical wall of defects whenever possible as these reflect the recovery phase of disrupted enamel formation (Hillson and Bond,1997) and are part of the continuum of departures from homeostasis or theoretical optimum enamel formation. Based on our experience with the Fongoli chimpanzees as detailed below, we feel that enamel formation cannot be simply dichotomized as normal versus hypoplastic but rather as an undulating surface of reduced enamel formation with more marked areas, visible as rLEH, superimposed on the former (Fig. 2). Undulations of the outer enamel surface can be due to enamel thinning and/or undulations in the enamel-dentin junction (Skinner et al., in press).

RESULTS

Our results are organized as follows: periodicity of recurrent rLEH based on perikymata counts between events; duration of LEH events themselves also based on perikymata counts; comparative severity of rLEH within and between teeth of individuals; temporal patterns of disrupted enamel formation.

Periodicity

As shown in Figure 1, the Fongoli canine teeth show repetitive episodes of quite evenly spaced hypoplastic enamel. Where possible, counts were made of the number of perikymata between successive grooves. As described above, recourse was made to low and high power photographs of each tooth to ensure that the location of each groove was consistently identified when counts were being performed and replicated. The results are shown in Table 3. In that, as shown in Figure 3, canine tooth 348 with a virtually complete crown (ca. 94%) shows seven grooves delineating six intervals and that, according to Schwartz and Dean (2001), male chimpanzees form imbricational enamel in about 6.25 (5.4–7.1) years, we can conclude that stressful events experienced by the Fongoli chimpanzees are not recurring semi-annually, as shown for chimpanzees from Cameroon (Skinner and Hopwood,2004), but are likely annual events. Earlier we had predicted that, based on current knowledge and the assumption that rLEH at Fongoli is circannual, there should be about 52–61 perikymata between events. Table 3 show that all counts are far below these predictions, ranging from 29.5 to 44, with an average count of 37.0. Table 3 includes reconstructions of the possible periodicities of rLEH at Fongoli based on the known range of long counts. On current evidence, contrary to the number of LEH/canine crown imbricational enamel, our perikymata counts do NOT support a finding of annual stress based on enamel hypoplasia among the Fongoli chimpanzees.

Duration

Duration of stress, as revealed by rLEH, can be estimated several ways (Skinner and Hopwood,2004; Guatelli-Steinberg,2008). In that crown formation typically slows towards the cervix, especially in hominins (Skinner and Hopwood,2004; Guatelli-Steinberg,2008), it is generally felt that perikymata counts within events (hypoplastic grooves) yield a more reliable and consistent estimate of duration than do spatial measurements. Here, we employ and compare both approaches.

Tooth 347 from the adult female (Subject B) is only mildly affected by rLEH but shows an interesting pattern of recurring pairs of LEH close together, separated by a larger distance from the next main event (Fig. 1). We label these pairs as 1bis, 1 [space], 2bis, 2 [space], and so on. The observed number of perikymata between the “bis” and main events averages 8.3 perikymata [(7.0, 8.5, 9.5)/3] which, depending on a long count range from five to nine, translates into an interval of about 1.5 up to 2.5 months. Noteworthy is the apparent partial recovery from stress during the reconstructed stressful period. This is shown in Figure 4, which provides side-views of canine crown surfaces rendered as silhouettes. The relatively unaffected enamel (that portion without marked LEH) averages 27.3 perikymata in duration (4.5–8.1 months, depending on long count).

All canine teeth from Frito (adolescent male) show an extremely severe hypoplastic groove near the cervix, labeled here as LEH 5 (Fig. 1). It is a plane-form defect (Hillson and Bond,1997) followed by a long period of rugose, thinned enamel with poorly countable perikymata (see Fig. 5). This LEH event varies in its expression within and between teeth (Fig. 6). Teeth 373 and 380 (upper and lower right canines from Frito) have countable perikymata from LEH 4 to the onset of LEH 5. This period of relatively unaffected enamel averages 27.25 PK [(26 + 28.5)/2) (4.5–8.1 months, depending on long count]; subtracting this count from the average interval of 37, the estimated period of severe stress experienced by Frito in this particular cycle (LEH 5) lasts 1.4 up to 2.6 months (depending on long count).

A third method applied to tooth 348 (Subject A-subadult male) for determining the duration of stressful versus non-stressful times during the cycle, as described earlier, is to calculate the width of enamel below a hypothetical plane (midpoint between highest and lowest points on the enamel contour forming the groove) as a fraction of the width of enamel above the plane. This method was developed here to address the subjectivity of defining the boundaries of hypoplastic grooves whose margins are gradual. The measurements were taken on a Keyence VHX-100 digital microscope precisely focused on the very edge of the canine crown looked at from the mesial or distal side. A straight line was drawn between the two highest points on either side of a hypoplastic groove and a second straight line drawn tangentially from the first line to the deepest point of the groove. The mid-point of this line was located and a line drawn parallel to the first line. Where this ‘hypothetical plane’ intersects the sides of a hypoplastic groove defines the width/duration of the stress event; while the length of the line between two hypoplastic grooves defines the whole cycle. The proportion of the stress event, so defined as a percentage of the whole cycle, yielded the estimate of stressful months within a cycle (Fig. 2). Results of this analysis are shown in the second last column of Table 4 where the proportion of hypoplastic enamel averages 20.8% of the cycle which, were the cycle circannual, equates with 2.5 months. Also shown in Table 4 are the measured depths of the hypoplastic grooves for Tooth 348 as well as the average width of a single perikyma. These can be seen to be not more closely spaced cervically, unlike the situation reported for hominins (Dean and Reid,2001). Calculated this way, the average hypoplastic groove is 0.45-mm wide; the average number of perikymata forming a hypoplastic groove on this average tooth would be 7.9 perikymata (1.3 up to 2.3 months, depending on long count).

Table 4. Evaluation of stressful portion (%) of the cycle for tooth #348

	Depth @300× (μm)	Interval	Pk counta	Space (mm) between grooves	Groove width (mm) at half-height	Col6/Col5 %	Mean Pk width (μm)
LEH1	−23				0.33
LEH2	−75	LEH 1–2	–	2.25	0.44	19.5	–
LEH3	−59	LEH 2–3	38	2.14	0.39	18.2	56.3
LEH4	−66	LEH 3–4	36	2.04	0.43	21.1	56.7
LEH5	−51	LEH 4–5	44	2.48	0.59	23.8	56.4
LEH6	−35	LEH 5–6	41	2.37	0.51	21.5	57.8
LEH7	−20	–	–	–	–	–	–
Mean	−45.5		38.3	2.26	0.45	20.8

• a Average of three ‘shoulder onset to shoulder onset’ counts, taken two months apart.

In sum, using three different methods, all three animals experienced an episode of stress while forming their canine teeth that fell within a span of 7.9–9.4 perikymata; depending on the actual long count, this translates into 1.3–2.8 months within a complete cycle. Notably, this interval is longer than that previously reported for low latitude apes (Skinner and Hopwood,2004).

Pattern

The realization that apparent groove depth varied as a function of width of field of view alerted us to the fact that enamel contour is not uniform but tends to peak at some point within the relatively smooth enamel; that period characterized as non-stressed. We have tried to capture this subtle phenomenon in Figures 4 and 5, which show silhouetted side-views of “between LEH groove” enamel and “groove” enamel, respectively. In several cases, it can be seen that the enamel contour dips slightly about midway through the interval separating LEH events (e.g., between LEH 2 and 3, 3 and 4 on Tooth 348, between LEH 3 and 4 on Tooth 380 and 382. These can be seen also in Figure 1. There must be some slight stressor occurring at such, otherwise, good times.

As seen in Figure 5, most LEH grooves have a steeper occlusal border followed by a more gradually climbing cervical border (e.g., LEH 4 in Tooth 347, LEH 4 on Tooth 380 and 382 and, not surprisingly plane-form LEH 5 on Frito's canines). In some instances, the groove is quite symmetrical (e.g., LEH 3 on Tooth 348) while in other instances the recovery slope rises and drops slightly over an extended period (e.g., LEH 2 on Tooth 348). In such instances, defining groove boundaries is quite problematic. This is shown schematically in Figure 2.

The observed undulations of enamel contour adjacent to more salient grooves suggested to us a more variable physiological experience of the immature chimpanzee striving to form its canine crowns than could be captured by a simple model of flat, or gently curving, enamel with grooves in it. For this reason, we created Figure 7, which shows enamel contour for Tooth 348 (sub-adult male) and Teeth 377 and 380 (Frito-adolescent male). We reconstructed approximate times of crown formation from published data on captive chimpanzees (Kuykendall,1996; Reid et al.,1998; Smith et al.,2010). The sub-adult male (Subject A), particularly, shows an apparent decrement of enamel commencing at a reconstructed age of about 3.5 years continuing for 3 years, and also marked by recurrent, evenly spaced grooves in the already depressed enamel. Frito shows a severe enamel defect at about age 5 years with very slow recovery. The adult female chimpanzee (Tooth 347) shows a smooth enamel contour marked only by mild LEH with no indication of relative enamel decrement (not shown in Fig. 7).

DISCUSSION

When commencing this research, our focus was on detecting the duration and severity of the dry season as expressed in the dental enamel of chimpanzees from Fongoli, Senegal. We had anticipated that perikymata counts would fit conformably into a model of circannual stress. This is not the case. The individually weighted average perikymata count between LEH events for all three animals [(35.7 + 38.3 + 36.0)/3] is 36.7. This average figure, while consistent among teeth and individuals from Fongoli, is unexpectedly low in light of reported daily cross-striation counts between striae of Retzius, noted earlier, ranging from five to nine in chimpanzees from other locations (Reid et al.,1998; Dean and Reid,2001; Schwartz et al.,2001b; Smith et al.,2007b). Perhaps, the most relevant long counts for this study are the modal values for Liberian chimpanzees of six and seven (Smith et al.,2007b). In other words, all else being equal, the number of perikymata between hypothetically circannual events in chimpanzees should be about 52 or 61 not ca. 37 as observed here. Having ruled out instrument and observer error, we can think of four possible explanations for such an anomalously low count.

1. Stressor is semi-annual not annual. As may be seen in Table 3, a long count of five results in a reconstruction of semi-annual stress. This is consistent with a finding of semi-annual rLEH periodicity among apes from Cameroon (Skinner and Hopwood,2004). It is inconsistent with the seasonal cycle at Fongoli and with the rarity of a long count of five; although one in three studied Taï chimpanzees showed this count (Smith et al.,2010). She also could not rule out a long cycle of five for two of 75 animals, mostly from Liberia (Smith et al.,2007b). We can see from Figure 1 that for those two teeth where the cast of the canine crown included the cervix (348, 373) (Table 1), there are seven and six LEH visible. Were these only 6 months apart, canine crown formation in Fongoli chimpanzees would be on the order of 2.5–3.5 years which is inconsistent with known canine crown and imbricational enamel formation times (e.g., (Kuykendall,1996; Reid et al.,1998). On balance we think this explanation is untenable.

2. These chimpanzees have teeth with high periodicity (i.e., a long count of ten or so). Such an explanation is consistent with the marked annual stress at Fongoli but inconsistent with every known study of stria periodicity in chimpanzees (Table 3). Smith et al. state, in their careful and comprehensive study of cross-striations between striae, that they counted only those that “clearly met the tooth surface”; consequently we can be confident that the long count among the majority of chimpanzees centers on six or seven and is never as high as 10 (2007b). Therefore, we consider very high periodicity (long counts) among the Fongoli chimpanzees as statistically improbable.

3. Stress at Fongoli recurs sub-annually. It is likely that the long count at Fongoli is equal to those reported from the same subspecies in nearby Liberia and from the great majority of other chimpanzees where this has been determined; i.e., modal values of six and seven (Smith et al.,2007b). Taking an average long count of 6.4 for Pan troglodytes (Smith et al.,2007b), the observed interval between rLEH events at Fongoli of 36.6 equates with stress every 7.7 months. Such a periodicity has little support in tropical ecology or primatological literature. Heideman and Bronson (1994) report an endogenous 7.3 month reproductive cycle in a tropical bat (Anoura geoffroyi) but only under laboratory conditions; normally the cycle is entrained by cues in the natural environment, possibly plant compounds (Paul Heideman, personal communication), to 1 year (Heideman et al.,1992). At Kanyawara, Kibale, a chimpanzee habitat in Uganda, Chapman et al. (1999) report a 9-month flowering pattern among mid-story and emergent tree species; such has not been reported at Fongoli. A sub-annual periodicity of about 8 months in phenological cycles or chimpanzee physiology and behavior at Fongoli seems unlikely.

4. Non-emergent imbricational striae. Having deemed stress recurrence at Fongoli every 7 or 8 months on average as unlikely, the discrepancy between expected and observed perikymata counts is so large that we have termed it the problem of “missing perikymata.” We now strongly suspect that not all striae of Retzius in the imbricational enamel of the canine teeth from chimpanzees at Fongoli reach the outer enamel surface. Apart from the well-known occurrence of buried striae in cuspal enamel; e.g., (Reid et al.,1998), we can find no explicit recognition in the literature of buried imbricational striae; although convergence of striae at the surface is a well-recognized phenomenon (Smith,2008). In a review of the link between perikymata and crown formation, Mann et al. (1990) remarked that perikymata counts on Spitalfield incisors were too low yielding ages that were 75% of actual ages and opined that “It is possible that incomplete perikymata counts result from a greater number of striae of Retzius that do not reach the surface than has been previously assumed …” 1990:127 (our emphasis). From our reading of the literature on perikymata, it is our impression that it is widely assumed that ALL striae of Retzius reach the lateral surface. Nevertheless, Witzel et al. (2008) illustrate (e.g., their Fig. 6) striae which bend so much as they come to the surface in regions of enamel hypoplasia that they are uncountable (see similar figures in Gustafson,1959). However, the enamel hypoplasia has to be fairly marked for this phenomenon to pertain, while the LEH on our chimpanzees is usually mild to moderate at most. How likely is the occurrence of non-emergent striae in imbricational enamel? Our earlier study of LEH periodicity among chimpanzees from low latitude (Skinner and Hopwood,2004) relied on spatial measurements not perikymata counts; and so is not informative about this issue. However, a separate study of perikymata counts between rLEH events among Taï chimpanzees from the Côte D'Ivoire (Skinner et al., in preparation) finds counts that range from 30 to 49 (mean = 37.8, n = 35 intervals from six animals). This result is virtually identical to the situation observed among the Fongoli chimpanzees (mean = 37.0, range 29.5–44, 11 intervals) (Table 3). Second, we have observed a phenomenon on fortuitously flaked surface enamel on a canine tooth from Côte d'Ivoire of what may be buried perikymata that peter out as they approach the surface (Fig. 8). In Figure 8, it can be seen that while the outer enamel is rather too worn to count perikymata with confidence, in the area of missing outer enamel the striae of Retzius are clearly visible. Most are evenly spaced at about 55 μm apart (upper right of figure) but between some of these are minor striae, which are only just visible, much less salient and squeezed between major striae. While these may be accentuated striae we think, rather, that they are normal striae of Retzius that fail to reach the surface; i.e., are the “missing perikymata.”

Both “semi-annual” stress and “high periodicity teeth” are unlikely explanations based on current knowledge. Neither “sub-annual stress” nor “non-emergent striae” has support in the literature. Nevertheless, of the four possible explanations outlined above for the lower than expected perikymata count between LEH among the Fongoli chimpanzee canine teeth, we think non-emergent striae is best for the reasons given. However, this is an unprecedented conclusion. Complete resolution of this problem will require thin sectioning of several canine teeth from different chimpanzee populations; a step that is not appropriate for the Fongoli teeth at the present time.

Our results have been presented so far to acknowledge the documented range of long counts (five to nine) in chimpanzees; that is reconstructed intervals between stress events ranged between 6.1 and 10.9 months, while duration ranged between 1.3 and 2.8 months. These reconstructions are not supported by other evidence drawn from tropical ecology, nor the seasonal habitat at Fongoli, nor from what we know about chimpanzee physiology and behavioral ecology in general. Given that known canine crown imbricational enamel formation spans in chimpanzees (see above) are entirely consistent with the number of LEH events (i.e., are apparently circannual), it would seem fair at this point to entertain the possibility that the observed perikymata counts between LEH are formed in about one year. This notion accepts that not all striae reach the surface and that, for the average individual chimpanzee, about 36.7 perikymata are deposited in 365 days; in other words, in this sample at least, one perikyma represents about 10 days. To be very clear, we are NOT claiming that one perikyma in this series forms in 10 days, only that each represents about 10 days. The remaining discussion uses the scenario that one perikyma represents 10 days.

As shown in Table 3, the number of perikymata varies between grooves from 29.5 to 44. Some of this variation is attributable to observer error that, as noted above, averages 2.5 perikymata. The rest, amounting to about 12 perikymata (i.e., reconstructed to be equal to 120 days) must represent the maximum variation in the onset of recurring stressful events; hence our use of the term “circannual” to describe the periodicity. The standard deviation for 11 events is 4.6 perikymata; in other words, the stressor is reconstructed to recur typically within 6 weeks on either side of the average of one year. If so, duration of stress lasts about 2.6–3.1 months.

At the beginning of this article, we predicted dichotomized stressful episodes during the annual cycle at Fongoli as “short duration, wet-season fruit shortage” and “long duration, dry-season heat and lack of water.” This scenario is supported by the occasional episode of mild stress during the longer interval of raised enamel noted earlier (Figs. 1 and 4). We think it germane that a period of relative fruit shortage at Fongoli tends to occur in the middle of the rainy season, when chimpanzees eat little fruit and spend a significant amount of time focusing on fallback foods such as cambium. Our reconstruction of a protracted interval of stress lasting almost 3 months is consistent with a long dry season. Similarly, looking at a single cycle, there is a clear tendency for the enamel contour to rise gradually from a time of stress (LEH) and then just as slowly decrease to the next LEH. We attribute this pattern to phenological factors at Fongoli where there is a general increase in available nutrition as the rainy season progresses (with the exception noted above of fruit shortage mid-rainy cycle).

In only one animal do we have specific life history information to compare with his pattern of LEH. Frito, the adolescent male who died in April 2010, experienced an episode of stress at a reconstructed age of about age 5.5 years which appears to have been very severe and lasted for several months. It occurred before crown completion, which is achieved in captive male chimpanzees at 7.28 (6.37–8.20) years (Kuykendall,1996). We are aware of three stressful events in Frito's short life: weaning, disease, and violence. The last event, aggression on him by an adult male, occurred just months before his death and so cannot be linked to the stress recorded on his canine crown which formed earlier in life. In 2007, Frito was observed with a disease, possibly fungal in nature, which severely affected his ability to keep up with his mother and the rest of the party. There were times when he spent half an hour whimpering and/or screaming for either his mother or others to wait for him or simply because it hurt to walk. His inability to keep up basically slowed the whole party. He had lesions on his face and on his hands and feet that made moving painful; a condition that lasted at least several months before an apparent full recovery. If the severe LEH that we think occurred at age ca. 5.5–6 (±1.3) years (Kuykendall,1996) reflects this infection, then he would have been about 7.2–10.3 years old at death. However, field observations of behavior, size, and testes development in Frito indicated an age at death in April, 2010 of 11 years. Smith et al. (2010) have re-evaluated dental formation among known age or histologically aged Taï chimpanzees. They conclude that Kuykendall's (1996) radiographic standards of tooth formation among captive chimpanzees remain appropriate for estimating the timing of dental formation in wild chimpanzees. In terms of dental eruption, H. Smith has concluded that there is indeed a mild wild-effect amounting to about 1 year (Smith and Boesch,2011). After Frito's death, a video recording was made of his exhumed skeleton. Using maturity indicators provided for captive or wild chimpanzees of known age (Nissen and Riesen,1964; Conroy and Mahoney,1991; Hamada et al.,1998; Hamada et al.,2003; Zihlman et al.,2007), Frito at the time of his death in 2010 must have been older than 8 years and younger than 13.5. His lower third molar was still deep in its crypt and partially overhung by bone indicating an age of <9.0–11.1 years (Nissen and Riesen,1964 captives) while his canines were partially erupted suggesting an age >10.5–13.5 (Zihlman et al.,2007 wild animals). The lack of fusion of secondary ossification centers in the limbs and hands is consistent with an age <10.5–12.5 years (Zihlman et al.,2007) (assuming somewhat delayed maturation in wild chimpanzees, the so-called “wild-effect”) (Zihlman et al.,2004; Zihlman et al.,2007). On balance, we think Frito's age at death in 2010 was about 10–11 years. Therefore, the severe stress experienced by Frito at about age 5.5 years affecting the cervical enamel of his canine teeth was something else entirely than the observed episode of disease in 2007 at about age seven to eight but rather occurred, we estimate, in 2004 or 2005.

Frito's younger sibling was born by spring 2005, but it is possible that she was born in late 2004 as individual animals were still being identified at that stage of our field project. Frito ranged with his mother relatively longer than did males about his own age. One of the first times, of which we are aware, when he was observed to set off with the other males, he literally cried for his older, sub-adult brother to wait for him. He returned, to continue ranging with his mother, and he was not observed to set off again without her for months. While mother chimpanzees may start to discourage suckling starting at age three, actual weaning occurs fairly abruptly at about age 5 years when the mother resumes cycling and becomes pregnant (Clark,1977). Our provisional explanation is that the stress that caused LEH 5 to occur in Frito's teeth at about age 5.5 years is related to birth of a sibling and disruption of the formerly close relationship with his mother. That emotional stress can impact forming enamel in a captive gorilla was shown by Schwartz et al. (2006) who linked a marked Wilson band with enclosure transfer.

Without the ability to link specific stressful episodes in the enamel to the Fongoli environment (with the exception of Frito's singular episode of stress which we have tentatively linked temporally to weaning), we can only seek correspondences between the pattern of stress and seasonal changes in the habitat. In that the dry season lasts 7 months, while we reconstructed only 2–3 months of stress at most, we conclude that another factor may be influencing variation in the periodicity of LEH and its duration. We suggest that only when drought has lasted several months or reached a threshold does access to water become a problem. Figure 9 shows the annual cycle of rainfall and evapotranspiration at Kedougou, Senegal (12°34'N, 12°13'W) less than 10 km from Fongoli. Evapotranspiration is defined as the amount of water transpired from an actively growing, short green plant cover (usually grass) with a full crop cover and a continuously adequate moisture supply (Virmani et al.,1980). There is a marked peak in evapotranspiration from March through May that we think corresponds to episodes of LEH. In other words, we would expect to see, in future studies at Fongoli, an interval of about 3 months in which chimpanzees will show marked behavioral responses to the heat and lack of water. At the peak of the dry season, Fongoli chimpanzees may have access to only one to three water sources. Although Fongoli chimpanzees use caves at the hottest time of year, no available water has been found in these caves (Pruetz,2007).

The effects of drought at Fongoli are not felt equally by all chimpanzees. The community is very cohesive, even in the dry season, with parties averaging about 12 individuals (Pruetz and Bertolani,2009). Adult males can enjoy the choicest feeding spots, but have to move further and further from water sources as the dry season endures. Adult males and adolescents can go a day without drinking, but females drink once a day (Pruetz, unpublished data). During the dry season, aged adult females lose body mass and usually remain in the vicinity of a permanent water source whether or not most other individuals travel away. We suspect that sub-adult males would be stressed since their lack of status affects access to food; in addition, they are not usually allowed to feed around the adult males. There are other indicators that the dry season is especially stressful. Some deaths, as well as disappearances, have occurred at this time. One sub-adult male went missing (presumably dead) near the beginning of the rainy season/end of the dry season in 2007. A second sub-adult male was found dead during the late dry season (Subject A in this study). Frito had been in fine health when last seen, having recovered from an attack by an adult male. When found several months later his body showed evidence of another attack, with canine tooth wounds evident in bone, although his body was decomposed to the point that soft tissue assessment was not possible. His body was located between two distant water sources, more than 1 km from the nearest one, and we suspect that dehydration exacerbated his condition.

We do not wish to be simplistic in interpreting the evidence for dry season stress. Sheer water shortage may indeed be a problem, but it should be recognized that during the dry season, while there may be widespread fruit abundance generally, in the vicinity of dwindling water holes there may be localized fruit shortage due to intensive exploitation by chimpanzees that dare not go too far afield. Water availability therefore exacerbates even minimal food stress, especially as the dry season advances.

Enamel hypoplasia studies have usually treated enamel formation in a simple dichotomous fashion; grooved or not. We are beginning to question this model, suspecting that full enamel thickness may only rarely be achieved (at least in the chimpanzee habitat reported here). Altmann (1991) found that diets of yearling baboon females fell consistently short of their optima and that protein intake was a major determinant of lifespan and fitness. Apes have a slow life history and, as a consequence, are limited by foods that ensure survivorship (Marshall et al.,2009). In an essay on the importance of relative food shortage in animal ecology, White (1978) stated that:

“…the component of the environment which seems to exert the major influence on the abundance of animals is a relative shortage of food for the very young; that for most of the time there is an inadequate supply of too-thinly or too-patchily dispersed nitrogenous food in the environment, and most young mammals cannot get enough of this food to maintain their very rapid growth. [ ] Variation in the weather (especially of the amount of rainfall) seems to be the major factor influencing the amount and nutritional quality of the food available for herbivores (p. 72).”

We think this scenario may very well apply to infant and juvenile Fongoli chimpanzees who, to use White's evocative phrasing, “…must struggle merely to persist in a passively hostile inadequate environment” (White,1978, p. 73). With that said, we must remember that the chimpanzees reported here form a mortality cohort whose developmental experiences may not be shared with survivors (Smith and Boesch,2011).

CONCLUSIONS

In this study of three chimpanzees from southeastern Senegal, growing their canine teeth in an extreme environment, we have shown that they experienced marked, recurrent physiological stress as evidenced by rLEH. Groove depth is more pronounced in the cervical half of the crown. The number of LEH events per canine crown is consistent with the number of years required to form imbricational enamel; i.e., are circannual. Our prediction of perikymata counts between LEH events, were they circannual in periodicity (ca. 52–61), is not supported. On average, only 37 perikymata are expressed on surface enamel between what we believe are circannual events. We have considered four possible explanations for this unanticipated observation: a cycle of stress recurring every 6 months, rather than twelve; truly sub-annual stressors in the Fongoli environment; high periodicity teeth; and non-emergent imbricational striae. We find only the last hypothesis is supportable. We reason that we can reconstruct time-related events, such as duration of stress resulting in hypoplasia, using the inference that one perikyma represents about 10 days. Stress lasts just under 3 months, shorter than the duration of the dry season but longer than reported for apes at lower latitudes (Skinner and Hopwood,2004).

In that we have previously shown rLEH occurs semi-annually at low latitude with double seasonal maxima in rainfall (Skinner and Hopwood,2004) but, we think, annually at Fongoli where there is only a single cycle of dry/wet seasons, it seems reasonable to conclude that rLEH is indeed informative of basic meteorological phenomena which could be applied in paleontological contexts from other latitudes and landmasses. While it is clear that the detailed observation afforded by perikymata expressed on the enamel surface, over the span of five or more years, is an indirect form of longitudinal study of growing chimpanzees of potential precision, the apparent problem of “missing perikymata” needs to be resolved.