Possible mineral contributions to the diet and health of wild chimpanzees in three East African forests
Abstract
We present new data on the ingestion of minerals from termite mound soil by East African chimpanzees (Pan troglodytes schweinfurthii) living in the Budongo Forest Reserve, Uganda, the Gombe National Park and the Mahale Mountains National Park, Tanzania. Termite mound soil is here shown to be a rich source of minerals, containing high concentrations of iron and aluminum. Termite mound soil is not, however, a source of sodium. The concentrations of iron and aluminum are the highest yet found in any of the mineral sources consumed. Levels of manganese and copper, though not so high as for iron and aluminum, are also higher than in other dietary sources. We focus on the contribution of termite mound soil to other known sources of mineral elements consumed by these apes, and compare the mineral content of termite soil with that of control forest soil, decaying wood, clay, and the normal plant-based chimpanzee diet at Budongo. Samples obtained from Mahale Mountains National Park and Gombe National Park, both in Tanzania, show similar mineral distribution across sources. We suggest three distinct but related mechanisms by which minerals may come to be concentrated in the above-mentioned sources, serving as potentially important sources of essential minerals in the chimpanzee diet.
HIGHLIGHTS
-
Termite mound soil provides the highest concentrations of aluminum and iron found in any of the dietary items at the sites studied hitherto.
-
We describe concentrating mechanisms in termite mound soil for some minerals.
-
Chimpanzees have discovered these sources of minerals and how to exploit them.
1 INTRODUCTION
Some bird and mammalian species, including elephants, macaques, tamarins, gorillas, chimpanzees, and humans (Wilson 2003), consume soil of a variety of kinds, often in the form of clay. Geophagy is widespread and has been observed on all continents inhabited by humans and nonhuman primates (Pebsworth, Huffman, Lambert, & Young, 2018), with archeological evidence suggesting its practice to be as old as 2 million years (Clark, 2001). Though the most prominent causes of geophagy remain unclear (Pebsworth et al., 2018), the practice of geophagy increases micronutrient intake, which may have nutritional value, and other benefits such as the detoxification of harmful compounds such as alkaloids in the diet (Klaus, Klaus-Hugi, & Schmid, 1998), protection against infection by parasites and pathogens (Knezevich 1998), and alleviation of gastro-intestinal upsets (Mahaney, Hancock, Aufreiter, & Huffman, 1996; Young 2010). As pointed out by Pebsworth et al. (2018), in a review of the literature in this field, the total elemental composition of soil may not reflect the amount of minerals available for the consumer, and in vitro studies are needed to determine bioavailability of mineral elements eaten in the course of geophagy (Pebsworth et al., 2013; Seim et al., 2013; Wilson 2003). Probably no single characteristic of soils eaten by animals, including humans, can account for their consumption (Abrahams, 1999; Wilson, 2003; Young, Sherman, Lucks, & Pelto, 2011), with mineral supplementation, medical, and detoxification functions all playing a part (Aufreiter et al., 2001; Aufreiter, Hancock, Mahaney, Strambolic-Robb, & Sanmagudas, 1997; Ketch, Malloch, Mahaney, & Huffman, 2001; Mahaney, 1993; Mahaney et al., 1999; Pebsworth et al., 2018; Vermeer & Ferrell, 1985; Wilson, 2003; Young 2010). Furthermore, geophagy may not always be beneficial as soil may contain soil-transmitted helminths, heavy metals, and increase the risk of predation (Link, de Luna, Arango, & Diaz, 2011; Matsubayashi et al., 2007., Pebsworth, Archer, Appleton, & Huffman, 2012).
The diet of wild chimpanzees in the Budongo Forest, Uganda, is typical of East African chimpanzee groups, and consists primarily of fruits and leaves, with additional flowers, bark, and pith (Reynolds 2005). Besides these plant-based items, meat and insects are eaten sporadically when they become available. Both meat, obtained primarily by killing monkeys (Goodall, 1986; Mitani & Watts, 2001; Newton-Fisher, Notman, & Reynolds, 2002; Nishida, Uehara, & Nyundo, 1979) and insects, for example termites (O'Malley & Power 2014), are highly nutritious sources of minerals as well as proteins, fats and other dietary requirements. However, the bulk of the food eaten by wild chimpanzees is plant-based and this constitutes 80% or more of the daily diet of most individuals. While high in some minerals for example, potassium and calcium, the Budongo chimpanzees' diet lacks (or has low quantities) of others for example, copper, manganese, and sodium, and, as a result, they need to locate these minerals from other sources (Reynolds, Lloyd, Babweteera, & English, 2009). Earlier work (Reynolds et al., 2009; Reynolds et al., 2015; Reynolds, Lloyd, & English, 2012) explored a number of dietary supplements for mineral acquisition, namely decaying pith of Raphia farinifera and the decaying wood of Cleistopholis patens, which provide appreciable amounts of sodium (Reynolds et al., 2009, 2012), and clay, which provides substantial amounts of iron (Reynolds et al., 2015). In this paper, we show that termite mound soil is a further valuable source of minerals eaten by chimpanzees in the Budongo Forest Reserve, Uganda, by the Kasekela group at Gombe National Park and by the M group at the Mahale Mountains National Park (Aufreiter et al., 2001).
Some discussion revolves around the extent of bioavailability of the iron ingested in soils, including termite mound soil (Aufreiter et al., 2001; Seim et al., 2013). In part this resolves itself into the question of whether the iron is in ferric (Fe3+) or ferrous (Fe2+) form. If the former, it is not bioavailable; if the latter it is. Experimental work (Aufreiter et al., 2001) using a medium with low pH to simulate digestive conditions suggests that most of the iron in soil is in ferric form and only a small part is ferrous. This finding suggests that the nutritional value of ingested termite mound soil may be limited. However, we should note that in humans a ferric reductase enzyme, duodenal cytochrome B, reduces ferric Fe3+ to Fe2+ (McKie et al., 2001). This enzyme, if present in chimpanzees, as seems likely, serves to increase the bioavailability of iron ingested in termite mound soil. If present, ferrihydrite, a hydrous ferric oxide mineral, is likely to be solubilized (Wilson, 2003). Mahaney et al (1997) concluded that in geophagy soils eaten by chimpanzees in the Kibale Forest, Uganda, 20% of ingested iron was bioavailable, sufficient for nutritional significance. In a study of soils eaten by humans and sold in local markets in Uganda, it was concluded that consumption of 5 g of soil contributed 19–25% of daily needs for iron (Abrahams 1997; Abrahams & Parsons, 1997); however, more recent work suggests that some iron in soil may not be bioavailable and that some soil types may inhibit iron absorption from food (Seim et al., 2013). Geissler, Mwaniki, Thiong'o, Michaelsen, and Friis (1998), by contrast, found that despite consuming 30 g daily of iron-rich termite mound soil, anemia remained prevalent in a human population in Kenya. Pregnant women were particularly prone to eating clays in Uganda and other tropical countries, although consumption occurs in nonpregnant women and men (Huebl, Leick, Guetti, Akello, & Kutalek, 2016). In western Kenya, approximately half of pregnant women preferred termite soil (van Huis 2017). In northern Uganda a greater diversity of soil types were eaten during gestation, and only pregnant women regularly ate termite soil (Huebl et al., 2016). Pregnant Chacma baboons (Papio ursinus) spent more time consuming iron-rich clay at monitored geophagy sites in Western Cape, South Africa than baboons of other age-sex classes (Pebsworth, Bardi, & Huffman, 2011).
Whereas the majority of minerals discussed in this paper can be regarded as either major minerals essential for life or minor minerals required only as trace elements, aluminum is neither of these and is not essential for life. Its ingestion in termite mound soil, probably in the form of kaolinite (Johns & Duquette, 1991; Mahaney et al., 1995) and in some cases gibbsite (Bolton, Campbell, & Burton, 1998), probably serves medicinal functions, by reducing acidity in the gut and neutralizing plant toxins such as condensed tannins (Goodall, 1986; Hladik & Clutton-Brock, 1977). Condensed tannins are ingested by chimpanzees on a daily basis at Budongo, being found at high concentrations in several species of figs (Ficus sp), particularly in the seed component. One fig species with a high concentration of condensed tannins, Ficus sur, is the second most frequently eaten food of the Budongo chimpanzees. Condensed tannins thus appear to be well tolerated by chimpanzees (Reynolds, Plumptre, Greenham, & Harborne, 1998; Wrangham, 1993; Aufreiter et al., 2001).
Termite mound soil eating is directed to specific species of termites (Uehara, 1982) and appears to be an opportunist, brief, and largely individual activity, occurring when the animals pass by a termite mound in the forest, often moving from one vegetative feeding site to another (Goodall, 1986; Nishida & Uehara 1983). Observations by researchers and field assistants indicate that “Gombe chimpanzees eat termite mound soil, on average, once a day” (Wrangham & Clutton-Brock, 1977) and the same may be true at Mahale and Budongo. Anecdotal reports suggest that at all three sites termite mound soil eating is more frequent among females than males, but quantitative data are lacking. Termite mounds present a hard surface (Figure 1) and chimpanzees either bite off a piece with their teeth or break off a piece with their fingers (Figure 2). At Mahale, chimpanzees eat the soil of termite mounds frequently through the year. While consumption can be sometimes linked to times of gastrointestinal distress (Mahaney et al. 1996), it may also allow chimpanzees to assess additional feeding opportunities. The chimpanzees of the K-group at Mahale were reported, before their disappearance, to vary the technique they used to feed on termites with the colony's reproductive cycle. In addition to direct nutritional benefits, feeding on termite soil may provide additional cues that allow selection of the most effective technique for subsequent consumption of the termites themselves (Uehara, 1982). At Gombe, about once a day, as they pass termite mounds, chimpanzees pick off and eat a “walnut” sized piece of termite mound soil (Goodall, 1986; Huffman, 1997; Mahaney et al. 1996). Time spent feeding on termite mound soil is short: at Mahale, 32 bouts of geophagy were measured and the mean duration was 1.7 min, range 1–8 min (Uehara, 1982). Co-feeding in large groups on termite mound soil, seen for example when feeding on other soils such as clay, has not been observed. And, unlike clay, termite mound soil is not eaten with leaves. At Budongo, if termites are present in termite mound soil, they are also eaten (Newton-Fisher, 1999), but use of tools for termite fishing has not been observed at Budongo, possibly because termite mounds of Pseudacanthotermes are less fishable, having few or no external holes (Collins & McGrew, 1985), unlike those of Macrotermes species. At Mahale, use of tools for termite fishing by the M group has only been seen occasionally (Takahata, 1982); whereas at Gombe, chimpanzees termite fish year around, though concentrate this activity around the wet months (Goodall, 1986; Uehara, 1982). Goodall (1986):256) also refers to Wrangham and Clutton-Brock (1977) study at Gombe: “Analysis of samples of termite clay … revealed substantial quantities of potassium, magnesium and calcium and traces of copper, manganese, zinc, and sodium … feeding on termite clay may be to neutralize tannins and other poisons present in plant foods (Hladik & Clutton-Brock, 1977)”. Soil recovered from a termite mound eaten by chimpanzees at Mahale contained a relatively high concentration of aluminum (10%), iron (3%), and sodium (0.5%). Metahalloysite was the dominant mineral found, which authors attribute a possible role as a pharmaceutical agent to alleviate intestinal upset (Mahaney et al. 1996).
Termite mound (Pseudacanthotermes spiniger) in the Budongo Forest, Uganda
Site where chimpanzee has removed a piece of termite mound soil, Budongo Forest, Uganda
In this paper we explore the concentrations of mineral elements in termite mound soil across three sites where chimpanzee have been well studied for decades: Gombe and Mahale, Tanzania (Goodall, 1968; Nishida, 1968) and Budongo, Uganda (Reynolds, 2005), as compared with control soil samples and other dietary sources. We go on to provide possible explanations for the mechanisms by which mineral elements are concentrated in different soil and plant-based sources.
2 METHODS
2.1 Subjects and sites
Data were collected in the Budongo Forest Reserve, in north-western Uganda, and at the Gombe National Park and the Mahale Mountains National Park, both in western Tanzania. Subjects at each of the three sites sampled were all well identified wild East African chimpanzees (Pan troglodytes schweinfurthii), whose communities have been habituated to observation for several decades, (Budongo, 28-years; Hobaiter, Samuni, Mullins, Akankwasa, & Zuberbühler, 2017; Newton-Fisher, 1999; Reynolds, 2005; Reynolds et al., 2015. Gombe, 58-years, Goodall, 1968, 1986; Wrangham & Clutton-Brock, 1977. Mahale M-group, 51-years, Mahaney et al., 1999; Nakamura, Nishida, Kappeler, & Watts, 2012; Nishida, 1968; Nishida et al., 1979, 1983; Uehara, 1982). Males and females of all age groups except infants (aged 0–5 years old) were seen eating at the termite mounds from which samples were collected. Unfortunately, consumption of soil was not reliably recorded with the long-term behavioral observations, so we are unable to provide frequency or rates of soil consumption behavior. Samples described here were collected between July 2015 and October 2017. Termite species are shown in Table 1.
| Site | Date(s) collected | Samples (N) | Termite species | Collectors |
|---|---|---|---|---|
| Budongo | July 2015 – Oct 2017 | 39 TMS, 27 CTRL | Pseudacanthotermes spiniger and Cubitermes ugandensis | VR |
| Gombe | Dec 2015 | 12 TMS, 7 CTRL | Macrotermes bellicosus, Macrotermes michaelseni, and Macrotermes subhyalinus | APG |
| Mahale | Aug – Sept 2015 | 11 TMS, 0 CTRL | Likely Pseudacanthotermes spp. | KH |
| MS |
- Note. APG: A Pascual-Garrido; CTRL: control soil; KH: K Hosaka; MS: M Shimada; TMS: termite mound soil; VR: V Reynolds.
2.2 Soil sample collection
Across sites, termite mound soil samples were collected by removing a 10–15 g piece of mound soil from a termite mound, using a sterile knife. None of the collected samples contained termites. Clean gloves were worn to prevent contamination from human sweat. In addition, control samples were collected of forest soil. At Budongo, control samples were taken from forest soil 1–3 m laterally from the termite mound and 15–20 cm deep. At Gombe control samples were taken from forest soil 1 m laterally from the termite mound and 15–20 cm deep. Control samples were not collected at Mahale. All samples were put into individual new plastic bags, marked with the date, collector, block number (an indication of a location within the chimpanzee territory), and sample number, and taken back to base camp where they were dried at a temperature of 40°C until fully dry. Five grams of each dried sample was then transferred to new sterile plastic container tubes for onward shipment to the UK under license.
2.3 Laboratory analysis of soil samples
The soil samples were dried to constant weight in an oven at 105°C for 6 hr. The total mass of the dried material was determined. Duplicate samples were prepared by taking 0.1 g of the material and 3 ml of Aqua Regia in a 10 ml centrifuge tube. The samples were digested in a water bath at 85°C for 3 hr. A total of 7 ml of ultrapure Type 1 water was then added to each sample and the samples mixed using a vortex mixer. A 1 ml aliquot of each sample was diluted 10-fold with Type 1 water for analysis. The elemental content of each sample was then determined using a Perkin Elmer Optima 2100 DV Inductively Coupled Plasma Optical Emission Spectrometer (ICP-OES). Standards and a blank were made up at 2, 4, 6, 8, and 10 ppm concentrations with 3% HNO3 and three replicates of each element were measured. Each sample was analyzed in triplicate and the average of the triplicate analysis taken for each duplicate. The mean of the duplicate analyses of the individual soil samples was then taken to be representative of that soil sample. The elemental content per kg of dried material was calculated from the raw data. In addition, we undertook preliminary X-ray Photoelectron Spectroscopy (XPS) analysis of one paired control and termite soil sample using a ThermoFisher ESCALAB 250Xi X-ray Photoelectron Spectrometer to investigate any differences in iron speciation. We include comparison data from two published studies that explored the mineral content of decaying wood fed on by Sonso chimpanzees (Reynolds et al., 2015) and the typical diet of Sonso chimpanzees (including fruits, leaves, and other plant parts; Reynolds et al., 2012). However, we do not have accurate data available on the relative quantity of these items consumed by the Sonso community; thus we are unable to calculate the relative contribution specific food types, such as termite soil, make to total mineral consumption.
2.4 Statistical analyses
The data for each variable were tested for normality of distributions and equality of error variances. Where these assumptions were not upheld nonparametric tests were used. Results were considered significant at α = 0.05. All data were analyzed using SPSS v24 (IBM Corp. Armonk, NY).
3 RESULTS
Values are mg/kg except where otherwise stated. We found a wide variation in the concentration of the mineral elements measured in termite mound and control soil samples (Table 2). Iron, aluminum, and potassium were the highest in both termite mound soil and control samples across sites. Zinc, sodium, and copper had the lowest concentrations in both soil types (with the exception of Mahale where zinc was more abundant in termite mound soil, see Table 2).
| Mineral element | Budongo | Gombe | Mahale | NRC | ||||
|---|---|---|---|---|---|---|---|---|
| TMS (n = 39) | CTRL (n = 27) | Kruskal–Wallis | TMS (n = 12) | CTRL (n = 7) | Kruskal–Wallis | TMS (n = 11) | ||
| Na | 5 ± 15 | 14 ± 27 | χ2 = 1.43; p = 0.232 | 0 | 47.1 ± 8 | χ2 = 16.84; p < 0.0001 | 41.9 ± 43 | 0.2% |
| K | 1080 ± 395 | 685 ± 90 | χ2 = 25.5; p < 0.001 | 1980 ± 724 | 1197 ± 291 | χ2 = 7.78; p = 0.005 | 5140 ± 2659 | 0.4% |
| S | 237 ± 171 | 169 ± 188 | χ2 = 2.94;p = 0.86 | 119 ± 50 | 339 ± 27 | χ2 = 12.60; p < 0.0001 | 279 ± 133 | – |
| P | 694 ± 219 | 524 ± 109 | χ2 = 9.92; p = 0.002 | 422 ± 115 | 329 ± 35 | χ2 = 2.86; p = 0.091 | 264 ± 123 | 0.6% |
| Ca | 3270 ± 3179 | 2310 ± 1463 | χ2 = 0.83; p = 0.361 | 1030 ± 939 | 466 ± 257 | χ2 = 3.46; p = 0.063 | 1720 ± 648 | 0.8% |
| Fe | 49100 ± 19576 | 43657 ± 15489 | χ2 = 0.80; p = 0.372 | 44500 ± 6380 | 28200 ± 4728 | χ2 = 12.00; p = 0.001 | 32100 ± 3235 | 100 |
| Zn | 4.06 ± 15 | 0 | χ2 = 3.34; p = 0.068 | 0 | 0 | N/A | 455 ± 293 | 20 |
| Mn | 1050 ± 421 | 1130 ± 418 | χ2 = 0.46; p = 0.498 | 383 ± 244 | 357 ± 119 | χ2 = 0.00; p = 1.00 | 585 ± 242 | 20 |
| Al | 18100 ± 4690 | 15300 ± 4182 | χ2 = 5.36; p = 0.021 | 19400 ± 5428 | 11700 ± 2327 | χ2 = 7.76; p = 0.005 | 32600 ± 8016 | – |
| Cu | 20.86 ± 27 | 1.41 ± 4.5 | χ2 = 12.62; p < 0.0001 | 92.3 ± 62 | 18.8 ± 29 | χ2 = 7.39; p = 0.007 | 10.2 ± 12 | 20 |
| Mg | 670 ± 294 | 604 ± 125 | χ2 = 0.12; p = 0.912 | 3520 ± 2996 | 1600 ± 775 | χ2 = 2.06; p = 0.151 | 5210 ± 2751 | 0.08% |
- Note. All mineral concentrations reported in mean mg/kg ± standard deviations; Significant differences between termite mound and control soil are indicated in bold. We provide the NRC nutritional recommendations for comparison as % (where indicated) or mg.kg−1 (National Research Council, 2003). Element key: Al: aluminum; Ca: calcium, Cu: copper; Fe: iron; K: potassium; Mg: magnesium; Mn: manganese; Na: sodium; P: phosphorus; S: sulfur; Zn: zinc.
- TMS: termite mound soil.
3.1 Budongo
Potassium, phosphorus, aluminum, and copper were all more concentrated in termite mound soil than in control soil; no other minerals varied in their abundance between soil types (Table 2). When compared with mineral concentration in the normal diet (data taken from Reynolds et al., 2012, Table 3), potassium (Kruskal–Wallis: χ2 = 0.95; p = 0.329) and phosphorus (Kruskal Wallis: χ2 = 0.80; p = 0.373) are found at similar concentrations in termite mound soil. Concentrations of all other minerals measured differed. Termite mound soil had concentrations of iron over 75 times higher (49.1 ± 19.6 g/kg; n = 39) than found in the normal diet (649 ± 1309 mg/kg; n = 24; Kruskal Wallis: χ2 = 44.1; p < 0.001); and a very large concentration of aluminum (termite mound soil 15,300 ± 4690 mg/kg; n = 39), which is completely absent from the normal diet (n = 24; Kruskal–Wallis: χ2 = 46.4; p < 0.001). Of other minerals, calcium (χ2 = 9.09; p = 0.003), magnesium (χ2 = 5.13; p = 0.024) and sodium (χ2 = 44.1; p < 0.001) were higher in the normal diet, while manganese (χ2 = 43.9; p < 0.001) and copper (χ2 = 18.6; p < 0.001) were higher in termite mound soil.
| Mineral element | Termite mound soil (n = 39) | Clay soil a(n = 10) | Decaying wooda, b (n = 31) | Normal dietb (n = 24) | NRC | Kruskal-Wallis |
|---|---|---|---|---|---|---|
| Na | 5 ± 15 | 234 ± 228 | 3032 ± 3826 | 293 ± 507 | 0.2% | χ2 = 84.33; p < 0.0001 |
| K | 1080 ± 395 | 2528 ± 3613 | 9478 ± 14282 | 4074 ± 6485 | 0.4% | χ2 = 37.13; p < 0.0001 |
| P | 694 ± 219 | 414 ± 534 | 1049 ± 2107 | 851 ± 964 | 0.6% | χ2 = 9.36; p < 0.025 |
| Ca | 3270 ± 3179 | 2381 ± 3003 | 4221 ± 5675 | 13315 ± 30648 | 0.8% | χ2 = 17.75; p < 0.0001 |
| Fe | 49100 ± 19576 | 8720 ± 3080 | 141 ± 152 | 649 ± 1310 | 100 | χ2 = 82.04; p < 0.0001 |
| Mn | 1050 ± 421 | 306 ± 252 | 183 ± 369 | 66 ± 69 | 20 | χ2 = 67.67; p < 0.0001 |
| Al | 18100 ± 4690 | 7885 ± 5245 | 0 | 0 | – | χ2 = 94.83; p < 0.0001 |
| Cu | 20.9 ± 27 | 17 ± 13 | 0 | 0 | 20 | χ2 = 40.36; p < 0.0001 |
| Mg | 670 ± 294 | 1012 ± 1165 | 2240 ± 2071 | 1557 ± 1272 | 0.08% | χ2 = 18.71; p < 0.0001 |
- Note. All mineral concentrations reported in mean mg/kg ± standard deviations. Significant differences between termite mound and other sources are indicated in bold. We provide the NRC nutritional recommendation for comparison as % (where indicated) or mg.kg−1 (National Research Council, 2003). Element key: Al: aluminum; Ca: calcium; Cu: copper; Fe: iron; K: potassium; Mg: magnesium; Mn: manganese; Na: sodium; P: phosphorus.
- a Data taken from Reynolds et al. (2015).
- b Data on the normal diet of Sonso chimpanzees includes fruits, leaves, and other plant parts; taken from Reynolds et al. (2012).
3.2 Gombe
As at Budongo, iron had the highest concentrations in both termite mound soil and control samples from Gombe, followed by aluminum (see Table 2). Preliminary XPS analysis of the speciation of iron showed no differences in the ratio of Fe3+ to Fe2+ between the termite mound soil and the control samples but provided a strong indication of the removal of organic matter in the termite mound soil. Levels of magnesium were higher across Gombe soil samples (n = 19) than in Budongo soil samples (n = 66; Mann–Whitney: U = 71; p < 0.001); with concentrations in termite mound soil over five times higher in Gombe (Table 2; Mann–Whitney: U = 22; p < 0.001). As at Budongo, zinc, sodium, and copper had the lowest concentrations. Sodium was completely absent from termite mound soil at Gombe but was present in small amounts in control samples. So, as at Budongo, Gombe termite mound soil provided high concentrations of iron and aluminum, together with some magnesium and other minerals, with the notable exception of sodium. Concentrations of potassium, iron, aluminum, and copper were all higher in termite mound than in control soil samples at Gombe; concentrations of sodium and sulfur were lower (Table 2).
3.3 Mahale
As at Budongo and Gombe, iron and aluminum were present in the highest concentrations, although at Mahale aluminum, rather than iron, was highest; at almost double the concentrations present in Budongo or Gombe (Table 2; Kruskal–Wallis: χ2 = 25.13; p < 0.001). Also, as at Budongo and Gombe, sodium and copper had the lowest concentrations at Mahale. None of the three sites compared had a consistently higher or lower overall concentration of minerals in any particular soil type.
3.4 Comparisons between termite mound soil, clay, decaying wood, and the normal diet of fruit and leaves at budongo
We compare the mineral content in termite mound soil with that present in clay (data from Reynolds et al 2015; Table 3), decaying wood (Raphia farinifera and Cleistopholis patens; data from Reynolds et al., 2012; Table 1 and 2 combined), and the normal diet of fruit and leaves at Budongo (data from Reynolds et al., 2012 ; Table 3). The differences between means shown in Table 3 are significant for all minerals shown.
4 DISCUSSION
Given the distance between the three sites (Budongo to Gombe 740 km, Gombe to Mahale 180 km) there is a high degree of similarity in the concentration of soil minerals between them. Termite mound soil represents a rich potential source of iron (Figure 3a) and aluminum (Figure 3b), which are present in high concentrations at all three sites. Iron, if bioavailable, is an essential dietary mineral, and aluminum may serve an important role in detoxification or regulation of the gastrointestinal system (Abrahams, 1997; Johns & Duquette, 1991; Vermeer & Ferrell, 1985). Other minerals are present and potentially available at lower concentrations, and there is absence or near absence of sodium in soils across all three sites. Thus, a clear picture emerges of the potential contribution of termite mound soil to the mineral intake of chimpanzees in East Africa and possibly elsewhere. While it has been suggested that consumption of the soil may provide additional cues for subsequent consumption of termites (Uehara, 1982), we did not observe feeding on termites during this study and termites were not present in the soil samples collected, and so we were unable to assess this as a possible motivation for soil consumption.
Mineral content across sources in chimpanzee diet. Panel (a) Iron content of termite mound soil at Budongo, Gombe, Mahale and in the normal diet of chimpanzees at Budongo. Panel (b) Aluminum content of termite mound soil at Budongo, Gombe, Mahale, and in the normal diet. Panel (c) Presence of sodium in four sources of chimpanzee diet at Budongo
The differences between termite mound soil and control samples observed in our data are consistent with those found by Adams, Rehg, and Watsa (2017), Mahaney et al. (1996), (1999), Aufreiter et al. (2001), and Sarcinelli et al. (2009). This widespread difference indicates a process whereby some mineral elements become concentrated in soil of fungus-culturing termite mounds (Mills, Milewski, Fey, Groengroeft, & Pertersen, 2009; Seymour et al. 2014). What is the process? It could take place at the stage of acquisition of soil by termites, which involves a prolonged process of embedding grains of soil in ingested water and salivary secretions (Turner, 2005) after which they are carried up into the mound to the building point. However, minerals that are relatively scarce in control forest soil are also relatively scarce in termite mound soil. Sodium in particular, scarce in forest soil, is very low or absent (i.e., below measurement detection limits) in termite mound soil (see also Tweheyo et al., 2006). The main process whereby minerals become concentrated in termite mound soil is therefore unlikely to be selected by termites and more likely, based on preliminary XPS data, to be due to the removal of organic matter.
Low values (or absence) of sodium in termite mound soil were found in the initial samples of termite mound soil collected as part of a study of minerals in clay (n = 5; Reynolds et al., 2015). This finding is now validated by a larger sample size across three different sites. The complete absence of sodium from termite mound soil at Gombe, whereas present in control samples, could indicate avoidance or rejection of sodium by termites or that they consume sodium for their own requirements. The latter may be the correct explanation. Kaspari, Yanoviak, Dudley, Yuan, and Clay (2009); Kaspari, Clay, Donoso, and Yanoviak (2014) showed experimentally that numbers of termites in the soil and litter decomposition rates were higher in Amazonian forest plots to which sodium had been applied than in control plots. Whether sodium consumption is a common attribute of termites or can explain the relative lack of sodium in Gombe termite mound soil is not known (Scheffrahn pers. comm.).
High values of aluminum and iron and low values of sodium were also found by Mahaney et al. (1996), (1997), (1999) and Tweheyo et al. (2006) who emphasized the possible medicinal use of aluminum in clay in the form of metahalloysite. Metahalloysite has the same formula as kaolinite, Al2Si2O5(OH)4 (Brindley, Robinson, & MacEwan, 1946) and is used by humans (commercially in the form of Kaopectate) to treat gastrointestinal complaints (Fairhead, 2016; Hunter, 1973; Johns & Duquette, 1991; Mahaney et al., 1997, 1999; Wilson, 2003). Smectite and gibbsite are further possible contributors to the efficacy of termite mound soil (Wilson, 2003). Higher concentrations of mineral elements in termite mound soil than in surrounding control soil were found by Aufreiter et al. (2001) and Adams et al. (2017) in a study of arboreal termitaria in Peru.
4.1 Mineral accretion
It is of great interest that chimpanzees appear to have discovered these three “hidden” sources of minerals: plant-based, soil-based, and animal-generated. In two of the three (plant-based and animal-generated) mineral concentration comes about as a result of water evaporation. In each case, water containing minerals is drawn up in decaying wood by capillary action, in the case of termite mounds transported by termites. In the third case, clay, low levels of minerals occur in the forest substrate and these are leached out of the soil by rain-water that collects in holes under trees.
At Raphia farinifera and Cleistopholis patens sites, chimpanzees chew the fibrous, decaying wood containing minerals left behind after evaporation, following which they spit out “wadges” of fibrous matter. At clay sites, it appears that the minerals are ingested by chimpanzees by chewing the clay when it is in semi-solid form or extracting it from clay-water with the use of leaf or moss sponges (Reynolds et al., 2015). At termite mound soil sites, chimpanzees chew pieces of mound soil in a similar way to the way they chew clay.
- (a)
In the case of decaying Raphia farinifera palms, and Cleistopholis patens trees, these are located in swamp forest which periodically floods, bringing in river water which contains low levels of mineral elements leached from the soil and rocks along its course. These elements are in low concentration (Reynolds et al., 2009, 2012, 2015). We suggest that the decaying roots and pith of Raphia use capillary action to draw swamp water upwards inside the tree's vertical, fibrous, pith-filled trunk. Because the head of the Raphia palm has previously fallen off after the tree fruited, the top of the trunk is now open and the whole trunk forms a cylinder filled with fibrous pith. Water containing low levels of minerals can enter this cylinder from below and rises up the fibers. As water evaporates from the top of the cylinder, it will leave its mineral content behind. As a result, we speculate that this becomes concentrated, and it is this source that the chimpanzees have learned to access by making a hole in the bark of the lower trunk (see Reynolds et al., 2009). In the case of Cleistopholis patens, we believe minerals become concentrated in a similar way but without the cylindrical process, merely by the adsorption by the decaying tree of mineral-containing water, which evaporates upwards from the tree, leaving behind concentrated minerals, which are then accessed by chimpanzees chewing the decaying wood.
- (b)
In the case, of clay, we don't believe evaporation plays a part. The action of rainwater and/or river water on forest soil, especially in hollows under trees, leads to dissolution and/or dispersion of minerals from the clay material which contains a high level of aluminum and surrounding soil which has a high iron content (Aufreiter et al, 1997; Eggeling 1947).
- (c)
In the case of termite mound soil, the actions of the termites themselves serve to concentrate the mineral elements in surrounding soil. The mechanisms by which this happens are not clear and require further study. Studies by Sieber and Kokwaro (1982) and Hesse (1955) focus on the use of water by termites in processing surrounding soil before carrying it to the surface of the mound. Turner (2005), (2011) describes, with associated videos, the process of drinking and carrying soil by termites. In the case of forest termites, a further process may be important: the ingestion of organic matter in forest soil, thus having the incidental effect of increasing the proportion of the mineral component and potentially making the termite mound soil more palatable following the removal of unpalatable organic components. Further work is needed to elucidate the causes of the differences between forest soil and termite mound soil.
5 SUMMARY AND CONCLUSIONS
Termite mound soil provides the highest concentrations of aluminum and iron found in any of the dietary items at the sites studied here. The normal diet of chimpanzees, whereas high in calcium and moderately high in potassium and magnesium, lacks aluminum and copper and is low in other minerals. Sodium, low in the normal diet, is absent or in low concentration in termite mound soil, which is thus not a dietary source of sodium for chimpanzees. This absence is in stark contrast to the high concentration of sodium found in decaying wood, which is eaten (Figure.3c; see also Reynolds et al. 2009). Thus, geophagy, meat-eating, and insectivory (O'Malley & Power, 2014) all add potential sources of important minerals for chimpanzees. In both Budongo and Gombe, control forest soil taken from just a few meters away from the termite mounds contains substantially lower concentrations of potassium, aluminum, and copper. Thus we can see a concentrating effect in termite mound soil for some minerals, with the notable exception of sodium. Termite mound soil at Mahale shows a similar pattern of minerals to those at Budongo and Gombe, with high levels of iron and aluminum, and moderate levels of potassium and magnesium. We suggest three possible mechanisms by which minerals become concentrated: evaporation of water in decaying wood, concentration after transport by termites, and dissolution or dispersion of mineral elements in clay after leaching of soil by water. Chimpanzees have discovered these potentially rich sources of minerals. If bioavailable, they would represent important additional opportunities to supplement the intake of nutritive minerals available in their normal diet of fruits, leaves and other plant parts, or (in the case of aluminum) otherwise regulate the functioning of the gastrointestinal system.
ACKNOWLEDGMENTS
The authors thank Jessica Rothman and an anonymous reviewer for their helpful comments that substantially improved our manuscript. At the Budongo Conservation Field Station we are indebted to Aliyo Jacob and Chandiya Bosco, Geresomu Muhumuza, Monday Gideon, Geoffrey Muhanguzi, and Caroline Asiimwe. We thank Philip Nyeko for termite identification. For research permission and export permits in Uganda, we are grateful to the Uganda National Council for Science and Technology, the National Forest Authority, and the Uganda Wildlife Authority. For securing the necessary permission to import samples to the UK for analysis we are grateful to Raymond Ward of the University of Brighton. For undertaking the preliminary XPS analysis at short notice we are indebted to Dr Santanu Ray of the Surface Analysis Laboratory at the University of Brighton. For help at Gombe National Park, we are indebted to Anthony Collins, Deus Mjungu, Nuhu Japhary Buke, Matendo Msafiri Katoto, and to Rudolf Scheffrahn, University of Florida, USA, for termite identification. At Mahale Mountains National Park, the authors are grateful to Michio Nakamura, Kazuhiko Hosaka, and Masaki Shimada. For research permission in Gombe National Park and Mahale Mountains National Park, we are indebted to the Tanzania National Parks Service, the Tanzania Wildlife Research Institute, and the Tanzania Commission for Science and Technology. For logistical assistance, we acknowledge the help of Anthony Collins and The Jane Goodall Institute at Gombe and the Mahale Mountains Chimpanzee Research Project at Mahale. The authors thank Anne-Marijke Schel for permission to use the video material in Supporting Information, and J.P.E.C. Darlington, Derek Pomeroy, S. Sutherland, and Andrew Smith for their comments and advice. For financial support, the authors acknowledge the Mohamed bin Zayed Species Conservation Fund grant numbers 0925272, 10251055, 11252562, 12254904, the Royal Zoological Society of Scotland, the Leverhulme Trust grant number ECF-2013-507, and the Boise Fund.
