Introduction The Ants of Africa
CHAPTER 3 - Mosaics - Evidence from Ghana - 4
Ghana - 1 Ghana - 2 Ghana - 3 Ghana - 5 Ghana - 6

Jonathan Majer, like Room, was based at the University of Ghana, and did all his research into the ant mosaic at the University Agricultural Research Station, near Kade, which is in the Celtis-Triplochiton moist semi-deciduous forest (see Map 1, point l). His work was carried out shortly after that of Room and differs in emphasis, the prime concern of the first investigation being to map the actual distribution of dominant and co-dominant ants in a single area of cocoa and, so, "test the reality of the mosaic concept" (Majer, 1972). Thus, most of his results were obtained from two adjacent plots of cocoa, both under relatively light shading, and with a total area of 2.25 ha. The greater area, designated as Plot I, had trees planted 11-12 years earlier with a spacing of 5 foot by 5 foot (1.52 m X 1.52 m); and the lesser area, Plot VIII, had trees planted some 8-9 years earlier and a spacing of 10 foot by 10 foot (3.05 m X 3.05 m). Shade trees were sparsely distributed (Majer, 1976c).

Majer, 1972

The initial effort was a marking out of the area into 10 m X 10 m quadrats, and then collecting all the ants found up to head-height on the central tree of each quadrat (Majer, 1972). Numbers of Oecophylla longinoda and Tetramorium aculeatum (as Macromischoides aculeatus) in the canopy of each quadrat were counted, and the position of any Crematogaster carton nests was mapped. The density of the cocoa canopy and the positions and canopy of all shade trees also were surveyed and mapped. Although he noted that 44 species of ant were found, only the ten species which appeared to have dominant or co-dominant status were named or otherwise mentioned. He mapped the quadrats in which the dominants were found and drew attention to co-existence and exclusivity on a quadrat basis. Thus: (the figures in square brackets [total 240] are the number of quadrats in which species were found) -

  1. Oecophylla longinoda [124, given as 126 in Majer, 1976a] occurred in areas lacking Tetramorium aculeatum [68, given as 66 in Majer, 1976a] and vice-versa (when trees were shared, a tree by tree map of a smaller area showed clear separation on such trees).
  2. Six Crematogaster species - Crematogaster buchneri [2], Crematogaster castanea [16], Crematogaster clariventris [6], Crematogaster depressa [40, given as 34 in Majer, 1976a], Crematogaster kneri [4] and Crematogaster striatula [14, given as 18 in Majer, 1976a] - were mutually exclusive and, in most cases, also to Oecophylla longinoda and Tetramorium aculeatum.
  3. Camponotus acvapimensis [38, given as 39 in the text and in Majer, 1976a] and Platythyrea modesta (as Platythyrea frontalis) [10] could also be dominants. The capacity of Camponotus acvapimensis to dominate was described as being limited to areas of broken canopy (applicable to 92% of the quadrats it dominated). Platythyrea modesta was noted as being found in ten quadrats and apparently occupying four of those quadrats as a dominant.
  4. Co-dominance, with no altitudinal separation, was seen between Crematogaster clariventris and Tetramorium aculeatum, and between Oecophylla longinoda and Crematogaster castanea. The association of the latter pair was attributed to an observed mutual toleration mechanism.
  5. As with earlier workers, he found that Crematogaster striatula very aggressively excluded all other dominants and most of the non-dominant species.
  6. Crematogaster kneri was included mainly because of its capacity to dominate the understorey of primary forest (attributed to Leston, pers. comm.).
  7. In twenty quadrats there were none of the ten dominants. Nine quadrats had very broken canopy and were unsuitable for arboreal nesters. The remaining eleven quadrats had a variety of Dolichoderines and small wood- or pod-nesting Myrmicines.

An interesting outcome was that Majer felt that the categorisation of species should be extended to give dominants, co-dominants, non-dominants, and sub-dominants. The non-dominants (following Room's (1971) findings) were felt to be able to co-exist with dominants because of individual properties such as - foraging at different times to the dominant, foraging on different parts of the tree, having mutual toleration mechanisms, having protective cuticle or spines (such as with Cataulacus and Polyrhachis species), or, possibly, simply because of low population levels. The sub-dominants were those, such as Camponotus acvapimensis which matched an environment to their favour. Completing the paper was a concept of how the mosaic could develop, especially bearing in mind Leston's (1970) description of the cocoa forest as being a relatively uniform structure. The key factors were instanced as:

Developing canopy closure with cocoa tree maturity excluded any ants requiring insolated ground.
The presence or absence of the large shade trees on which several of the Crematogaster (Atopogyne) species prefer to nest.
The preference of Oecophylla longinoda for insolated canopy.
The converse preference of Tetramorium aculeatum for shaded canopy.
The requirement for dead wood for nesting by Crematogaster striatula.

Although Majer wrote of a three-dimensional mosaic, this is not clearly demonstrated by his results, as these come from a two-dimensional mapping, with no indication of which ants use the trunk and ground as the third dimension.

Majer, 1976a, b, c

The findings from Majer's subsequent work were published in rather a fragmented fashion and it is necessary to consider the set of three papers (Majer, 1976a, b, c) as a single unit, together with information in his list of the insects on cocoa in the 2.25 ha plot (Majer, 1975). First, he examined the cocoa canopy density using three categories - "dense"; "complete but thin and broken"; and, "absent" (Majer, 1976a). Generally, the conclusions matched those of his earlier (1972) paper, the most notable difference was that the distribution of Oecophylla longinoda was shown not to be correlated with canopy density. Crematogaster depressa, which he described as a forest-dweller, was found mainly in cocoa with a broken canopy or in areas where the canopy was absent. He noted also that this species tended to have its nests high on shade trees, in positions where the nests were exposed to full sunlight. Crematogaster clariventris in contrast was described as shade-loving with its nests on lower trunks.

Next, he provided some quantitative data on the abundance of the dominant ants in the 2.25 ha site at Kade (described in Majer, 1972). The data was obtained by using pkd sprays and counting all the ants, plus other insects, which fell on to cotton sheets, measuring 5 m X 5 m, placed under the sprayed canopy (Majer, 1976b). Within the 2.25 ha, nine plots, each of at least 40 m X 40 m, had been defined in order to study the impact of elimination of dominants (Majer, 1976a). The resultant changes were monitored by using pkd to sample the ant fauna in each of the nine plots once every four weeks over a 16 month period (January 1971 to April 1972). The sixteen samples thus obtained from each plot, form the total of 144 samples (Majer, 1976c). A total of 52 species of ant was collected and these, together with the mean numbers per sample in which they were collected, were listed by Majer (1976b). The frequency of occurrence of the fifteen most commonly found species was given by Majer (1976c) but the frequency of all 52 species was given only by Majer (1975) - see ghana cocoa ants. Incidentally, the full list of ants he gave in Majer (1975) included a further 18 species taken in soil samples, and all are included in the species section of this monograph.

The results of the pkd sampling probably need to be interpreted with constraint because of the bias introduced by the exclusion experiment (Majer, 1976a). That study aimed at exploring how the dominants might occupy territories from which other dominants had been removed. The main outcome was that Oecophylla longinoda and Tetramorium aculeatum were shown to be very capable of moving rapidly into vacated areas. The Crematogaster (Atopogyne) species were not able to recover and did not expand into new areas; indeed the removal of Crematogaster depressa also affected its presence in an adjacent plot in which it had shared territory but from which Tetramorium aculeatum was the removed dominant. Crematogaster striatula proved impossible to remove but showed only limited expansion into Oecophylla longinoda territory, although the plot was adjacent to the secondary forest (as were all quadrats with Crematogaster striatula). Two minor Crematogasterine occasional dominants, Crematogaster gabonensis and Crematogaster julienni, became established in the control plots where no dominant removal had taken place, probably simply indicating how they can exploit small gaps, or lacunae.

To return to the information on abundance, it should be remembered that the 144 pkd samples comprised 16 samples from each of the nine exclusion plots. Moreover, as each plot was formed of at least 16 quadrats and the pkd applications were applied to randomly selected 5 m X 5 m areas of canopy (equivalent to one quarter of a 10 X 10 m quadrat, Majer, 1976c), on average there would have been only one pkd application per quadrat. Firstly, the impact of dominant exclusion (Majer, 1976a) clearly caused the frequency of Crematogaster depressa and Crematogaster clariventris to have been reduced. Both were originally in restricted areas - Crematogaster depressa had a frequency of only 19, but direct loss could have been from 30 (2 quadrats X 15 post-exclusion sampling dates) to 45 (also lost from an adjacent quadrat); Crematogaster clariventris had a frequency of 5, but direct loss would reduce the potential by at least 15 (1 X 15). Oecophylla longinoda and Tetramorium aculeatum would also have suffered some reduction in frequency; adding a compensatory figure (2 X 15) for the former, gives 150 out of 144, and (1 X 15) to the latter gives 116 out of 144. The failure to exclude Crematogaster striatula is emphasised when one sees that the original distribution, in only three quadrats, would have given a potential frequency of 48 (3 X 16) whereas the actual frequency was 53. The impact of the removal of large numbers of workers on the individual ant colonies appears not to have been considered. For example, Majer wrote of a single colony of Crematogaster julienni, with a single pkd containing 1200 workers, and Crematogaster buchneri in only one quadrat with two pkd samples of 2150 and 2192 workers, but nothing was revealed of what may have been a devastating effect on those species. In contrast, the original map of occupied quadrats (Majer, 1972) showed Crematogaster castanea in only 4 quadrats but it was present in 88 pkd samples (purportedly it remained in one of the Oecophylla longinoda exclusion plots but it was not shown as being there). Similarly, Crematogaster kneri was mapped in only one quadrat, i.e. 1/16 of the quadrats in the plot, yet it appeared in 9 samples.

The final major aspect of the work stemming from the pkd sampling was the examination of the results by principal components analysis ordination (Majer, 1976c). Majer's main concerns were to examine the impact of ecological factors and the exclusion attempts on the dominant ants; and to examine the possible associations of 85 species of other insects (the criterion being presence in seven or more of the 144 pkd samples). In terms of the ant mosaic, two comments of caution seem justified - only 15 species of ant were included and three of those were below the "seven occurrences" level; eleven species of ant were excluded, ostensibly because they were non-dominants and "their relationships with the dominants have already been investigated (Room, 1971)".

The initial analysis using the 100 species proved to have only low variance and yielded unconvincing results, the most useful being the separation of Crematogaster striatula (in 53 samples, and a mean of 3420 workers per sample, from Majer, 1976b; data signified as xx/yy for other species) from Oecophylla longinoda (120/4407) and Tetramorium aculeatum (101/1236), with the far less frequent Crematogaster clariventris (5/622), Crematogaster depressa (19/248) and Crematogaster gabonensis (5/710) being within the overall sector of Tetramorium aculeatum.

A second sample ordination, using only the 54 most common species, confirmed the separation of the three common dominants and showed:

  1. The very close association of Crematogaster castanea (88/188) and Oecophylla longinoda. No others of the chosen common ants were grouped with this pair.
  2. Tetramorium aculeatum and Crematogaster depressa appeared to be grouped.
  3. Apparent associates with Crematogaster striatula were Crematogaster gabonensis, Platythyrea modesta (33/4) and Tetraponera anthracina (65/8). Room (1971) had a positive association between Platythyrea modesta and Camponotus chrysurus (not found at Kade), which was linked with Crematogaster striatula.
  4. The other seven ant species all fell without these three groups.

Comparison with the findings of Room (Ghana - 3 - Room)

Comparison with Room (1971) suggests that the territories of the large arboreal Camponotines, Camponotus brutus (27/20) and Camponotus vividus (101/58) are not influenced by the conventional dominants. Room (1971) had a positive association between Camponotus brutus and Crematogaster clariventris and Crematogaster depressa; and a positive association between Camponotus vividus and Crematogaster africana. What may be valid is that Camponotus brutus perhaps shares a preference for forest and shade trees with Crematogaster striatula, Crematogaster depressa and Crematogaster clariventris. Crematogaster africana was not found at Kade and the very high frequency of Camponotus vividus suggests that it was spread throughout the Kade plots. Camponotus acvapimensis (31/207) and Pheidole species E (19/60) both seem simply to fit into appropriate areas. The former requires insolated ground for nesting (Majer, 1972); and the latter needs dead wood on trees (see species list), suggesting perhaps that it would be adversely affected by Crematogaster striatula. Majer (1976b) noted that 750 workers of Pheidole species E were found in one pkd sample but it was less numerous in the remainder (presumably around 21 samples, by extrapolation from Majer 1975, 1976c). The remaining species, Crematogaster clariventris (5), Crematogaster buchneri (2/2150) and Crematogaster kneri (9/29) all can be dominant but their low frequency surely makes their inclusion in the analysis of only marginal merit, indeed arguably dubious for Crematogaster clariventris which was one of the excluded dominants (Majer, 1976a).

Majer's decision to omit eleven ant species seems unfortunate as a test of Room's (1971) results could have had considerable merit. Room found a positive association between Cataulacus mocquerysi (99/25) and Crematogaster clariventris but the very high frequency of the former, in the near absence of the latter, suggests that such an association was not important at Kade.
Perhaps an association with Oecophylla longinoda / Crematogaster castanea is more likely to have been demonstrable.
The assumption of Cataulacus guineensis (117/56) and Polyrhachis decemdentata (99/61) as being positively associated with Oecophylla longinoda/Crematogaster castanea (Room, 1971) may well have been reasonable if simple high frequency is taken as a useful measure, but then Camponotus vividus (101/58) and Polyrhachis laboriosa (68/51) were found to be negatively associated with this dominant pair by Room (1971).
Polyrhachis laboriosa additionally was found by Room (1971) to be positively associated with Tetramorium aculeatum.
The most frequent species at Kade was Cataulacus lujae (as Cataulacus brevisetosus) (137/70), an ant which was not found by Room (1971). Members of other genera, Tapinoma melanocephalum (72/20), Tetramorium species K (44/7), Technomyrmex species 1 (28/14) and Monomorium species G (23/7), also were excluded.

Final aspects of Majer's work

An acknowledgement of the possible weakness of the 10 m X 10 m quadrats, in terms of understanding the more intricate details of ant distribution, led Majer (1976b) to do surveys on an individual tree basis at Kade. In a 0.3 ha area (plot), with cocoa on three sides and a road on the fourth, he demarcated blocks of 16 trees. Then he surveyed each tree to map the canopy, position of Oecophylla longinoda nests (which he noted as matching the foraging range), position of Tetramorium aculeatum nests (he noted the species is nocturnal and, thus, its foraging range was not determined), and position of Crematogaster spp. carton nests. He collected ants from each tree to hand height for later determination. The forest trees and ants on them were also mapped. This was done twice (in June 1971 and January 1972). A nearby 0.4 ha plot of similar cocoa but adjacent to secondary forest was surveyed in September 1971. On the 0.3 ha plot, there were 660 trees, on which the common dominants were -

Crematogaster depressa (on 242 trees),
Tetramorium aculeatum (on 177 trees),
Oecophylla longinoda (on 83 trees) and
Crematogaster clariventris (on 18 trees).

Of those trees, Tetramorium aculeatum was found with Crematogaster depressa on 51, with Oecophylla longinoda on 26, and with Crematogaster clariventris on
Crematogaster clariventris and Oecophylla longinoda were on 17, and Crematogaster depressa and Oecophylla longinoda were on 3.
The remaining 195 trees had no dominant.

The frequency of Tetramorium aculeatum and the Crematogaster species was attributed to the former's nocturnal foraging habits; whereas the co-habitation with Oecophylla longinoda was possible because of territorial separation on the trees. On this plot, Crematogaster depressa was seen to forage over large areas (from the relatively small proportion of the, quite numerous, shade trees on which it had nests). Both the 0.4 ha plot and a second plot at CRIG, surveyed by B. Bolton, were dominated by Crematogaster striatula.
Of 396 trees in the CRIG plot, Crematogaster striatula occupied 287, and 76 had no dominant species. Oecophylla longinoda (13 trees), Crematogaster clariventris (6 trees), Crematogaster gabonensis (6 trees), Pheidole sp. (3 trees), and combined Crematogaster striatula/Pheidole sp. (2 trees) comprised the small remainder.

By using radio-active labelling in a glucose food source injected into nests, Majer (1976b) also examined the spread of workers (or, rather, the foraging range) from the base nest. His paper shows maps of the three species investigated but the maps for "Figure 3" and "Figure 4" are transposed ("Figure 3" showing Crematogaster depressa and "Figure 4" showing Crematogaster clariventris, both areas being shown in the plot map, "Figure 1"). The labelled nest of Crematogaster clariventris was described as being a large carton nest built 4 m above ground on a large silk cotton tree (Ceiba pentrandra). Labelled ants could be found on trees within his original visual survey, visiting a total of 14 trees in under two weeks from the labelling, and within the second nest some 9 m away, on the trunk of a small shade tree. He noted that use of a secondary nest on a separate tree had not previously been reported for the Crematogaster africana group in West Africa. Crematogaster depressa had a nest, 1 m above ground, on the trunk of a small shade tree. Labelled ants could be found on only four linked trees, and a second nest only 5 m away, and adjacent to the foraged trees, was not visited by them. Majer describes finding twin-nested colonies of Crematogaster depressa elsewhere on Kade. For Crematogaster striatula a dead forest tree with a multi-entrance nest was injected. From there, after one month, the labelled ants were found on seven trees in an area measuring 11.5 m X 10 m.

Contents Mosaics - Introduction Mosaics - Ghana - 5
1998, 2003, 2016 - Brian Taylor CBiol FRSB FRES
11, Grazingfield, Wilford, Nottingham, NG11 7FN, U.K.