University of Groningen
Spatial and temporal hotspots of termite-driven decomposition in the Serengeti Freymann, Bernd P.; de Visser, Sara N.; Olff, Han; Spence, John R.
Published in: Ecography
DOI: 10.1111/j.1600-0587.2009.05960.x
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record
Publication date: 2010
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA): Freymann, B. P., de Visser, S. N., Olff, H., & Spence, J. R. (Ed.) (2010). Spatial and temporal hotspots of termite-driven decomposition in the Serengeti. Ecography, 33(3), 443-450. DOI: 10.1111/j.1600- 0587.2009.05960.x
Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Download date: 10-02-2018 Ecography 33: 443 450, 2010 doi: 10.1111/j.1600-0587.2009.05960.x # 2010 The Authors. Journal compilation # 2010 Ecography Subject Editor: John R. Spence. Accepted 17 July 2009
Spatial and temporal hotspots of termite-driven decomposition in the Serengeti
Bernd P. Freymann, Sara N. de Visser and Han Olff
B. P. Freymann ([email protected]), S. N. de Visser and H. Olff, Centre for Ecological and Evolutionary Studies, Community and Conservation Ecology Group, Univ. of Groningen, PO Box 14, NL-9750 AA Haren, The Netherlands.
Ecosystem engineers are organisms that directly or indirectly control the availability of resources to other organisms by causing physical state changes in biotic or abiotic materials. Termites (Insecta, Isoptera) are among the most important ecosystem engineers in tropical ecosystems. We used a field experiment in the tall grasslands of Serengeti National Park, Tanzania, to investigate 1) the consumption by termites of grass litter and dung baits along the landscape gradient of catena position, and 2) seasonal variation in litter and dung removal. Our maps of termitaria and patterns of bait removal revealed clear spatial and temporal hotspots of termite activity. In the dry season termites removed more baits at the top- catena positions than at the bottom positions, but there was no effect of catena position in the wet season. Spatial hotspots of termite activity overlapped with those of both mammalian herbivores and predators. Within the framework of ecosystem engineering, this study suggests that intraspecific aspects of spatial heterogeneity and temporal variability deserve much greater consideration.
Every living organism interacts with its environment, 2006, Jime´nez and Decae¨ns 2006). Some termites act as including other individuals of its own species and other herbivores as well as decomposers, feeding on a wide range biotic elements through competition, mutualism, parasitism of living, dead or decaying plant materials and soil and predation (Krebs 2001). Recently, the concept of (Adamson 1943, Lee and Wood 1971, Wood 1976, interaction types was extended to include the conceptual 1978, Bignell and Eggleton 2000, Traniello and Leuthold framework of ecosystem engineering (Jones et al. 1994, 2000). Also, termites may play an important role in 1997, Wright and Jones 2006). Ecosystem engineers are recycling herbivore dung in tropical ecosystems (Freymann organisms that directly or indirectly control the availability et al. 2008). Thus, termites play a crucial role in the of resources to other organisms by causing physical state decomposition processes of tropical ecosystems, providing changes in biotic or abiotic materials (Jones et al. 1997). ecosystem services via nutrient cycling (Lavelle et al. 2006). Termites (Insecta, Isoptera) and earthworms are among Termite mounds are common but heterogeneously the most important ecosystem engineers in terrestrial distributed landscape elements in several regions of Africa ecosystems (Lavelle et al. 1997). As eusocial insects, termites (Pomeroy 1977, Kaib et al. 1997, Meyer et al. 1999), but build nests of great architectural diversity ranging from the implications of this spatial and temporal heterogeneity inconspicuous below-ground nests to the massive above- for decomposition activity has only rarely been studied ground mounds of Macrotermes spp. (Noirot and Darling- (Schuurman 2005). At the landscape scale, the concept of ton 2000). Adamson (1943) suggested that termites are soil catenae is particularly important in semi-arid ecosys- crucial for maintaining the fertility of tropical soils and the tems such as the African plains (Jarman and Sinclair 1979). productivity of the tropical ecosystems they inhabit because Milne (1935) introduced the pedological term catena they promote aeration and mixing of the soil, drainage and (Latin: chain) in senso stricto to describe a sequence of penetration of plant roots. In addition, termites accelerate repetitive soil patterns along the slope of a hill. Soils at the formation of humus and nutrient cycling by consuming a top position of a catena are characterised by a higher sand variety of cellulose containing material. content and a lower clay and water content in comparison Many studies have demonstrated the beneficial ecosys- to the soils at the bottom position; this pattern is common tem impacts of termites (Wood and Sands 1978, Anderson in the central region of Serengeti National Park (SNP) and Wood 1984, Abbadie et al. 1992, Holt and Lepage (Anderson et al. 2006). Despite the ecological importance 2000), some of which deal explicitly with their ecosystem of these landscape features, we know of no studies of termite engineering activities (Black and Okwakol 1997, Lavelle activity in relation to variation in soil texture and depth et al. 1997, Dangerfield et al. 1998, Jouquet et al. 2005, along catenae.
443 In this study we explored the spatial and temporal Field experiment heterogeneity of termite activity in the long-grass plains of Serengeti National Park, Tanzania. We hypothesised that Experimental sites were located randomly within the long spatial and temporal hotspots of termite activity and grass region of SNP, stratified according to top or bottom therefore decomposition would follow from the reported position of a catena. In total, eight experimental plots were spatio-temporal heterogeneity of termite mounds (Sinclair established on each ridge position. The minimum distance 1979) in this savanna ecosystem. We used a field experi- between plots located on the same hill was 500 m. The ment to investigate whether termite consumption of grass distance between top and bottom positions was about 1 km. litter and dung differed between the top and bottom We randomly chose a point in the field as the centre of a positions of catena ridges. We also compared termite 3 3 m plot. We marked the plots and subdivided them activity along the catena between the dry and wet seasons. into nine 1 1 m subplots in which we placed the baits. In addition, we quantified the spatial distribution of Each 3 3 m experimental plot formed the centre of a termitaria in this ecosystem to establish relationships wider 100 100 m plot used for the later mapping of the between nest location and ecologically significant activities present termite mounds. related to decomposition. In a pilot experiment we identified a mesh-size of 1.5 mm to be suitable to allow the dominant termite taxa of SNP to access baits placed in metal mesh-bags (10 Material and methods 10 cm), whereas the majority of the remaining arthropod fauna, especially dung beetles, could be excluded. Study area Baits were collected and processed as follows: from the vicinity of each experimental plot we randomly harvested This study was conducted in the long grass region in the nine samples of the characteristic grass species of the southern part of Serengeti National Park (SNP), Tanzania particular site and air-dried them for three weeks. This (2828?S, 34853??E; 1570 m a.s.l.; Fig. 1A). The prevailing treatment was necessary since the dominant termite taxa in habitat type of the study area is a semi-arid grassland, this area (Odontotermes sp., Macrotermes sp.) consume dry dominated by long-grass species such as Themeda triandra grass litter. Of each grass sample a portion of 15 20 g was and Pennisetum mezianum. For further details see Sinclair randomly chosen, filled in a mesh-bag and labelled. The (1979). The study site is characterised by an undulating exact filling weight of each of the 144 litter baits was topography with small ridges, catenae, showing an elevational determined using a digital scale accurate to 0.001g. We difference of 15 25 m between hill tops and bases. In contrast collected 144 fresh wildebeest (Connochaetes taurinus) to these catenae located throughout the study area, the five droppings. Of each dropping a random sample of about hills of the so-called five-hills-track, also situated in the long 50 g was put into a mesh-bag and processed in the same grass region of SNP, show considerably larger elevational way as the litter baits. Placement of the baits took place differences (75 100 m) between top and bottom. The dry within one day (day 0 of the experiment). In each 1 1m season component of the experiment was conducted in late subplot we placed one litter and one dung mesh-bag in July 2005, and the wet season component in early May 2006. close contact with the soil. Prior to the re-collection of the
(A) (B)
KENYA
SNP TANZANIA
LGP
05025 Kilometers
Figure 1. (A) Location of the study area (LGP long-grass plains) within Serengeti National Park (SNP), Tanzania. (B) Location of potential hotspots of termite-driven decomposition (dark shaded areas) within the study region.
444 mesh-bags from the field we determined a random order, of termites in our study plots we calculated the mean daily which was used consistently for all plots, in which the bags mass loss (%) for each 3 3 m experimental plot and for were retrieved. We collected baits at day 1, 2, 3, 4, 5, 6, 10, each location (top and bottom plots combined). We 15, 20 of the experiment. Before we approached the plots assumed a constant rate of relative mass loss over time we searched in a 250 m diameter around each plot with caused by termites, a valid underlying assumption (Collins binoculars for the presence of mammals for five minutes. At 1981, Schuurman 2005). We tested the validity of this one top and one bottom ridge position we excavated a soil assumption with our own data by plotting the mass loss (%) pit of about 1.5 1.5 m to the depth of the underlying of those baits placed in all locations and in both seasons in CaCO3 hardpan that is situated beneath the soil surface in the field that showed signs of termite consumption against this part of SNP. time through the course of the experiment, and tested for a significant association using linear regression. Termite mound densities obtained with the two differ- Mapping ent methodologies described above were calculated as the number of mounds ha 1. Since the five hills of the five- We recorded the number of termite mounds per ha in the hill-track differ in absolute size, for standardisation we long grass region of SNP using two different methodolo- calculated the number of termite mounds ha 1 for five gies. At the sites of the baiting experiment we mapped the termite mounds in the entire 100 100 m area around the distance classes, each covering 20% of the distance, starting 3 3 m experimental plots as a centre (n 16). We also with class one at the bottom of each hill. It was not possible mapped the termite mounds in the much higher hills of the to statistically analyse the cumulative numbers of individual so-called five-hills-track. Recording started 200 m away, mammals encountered due to the possibility of pseudo- covered the entire hill, and stopped 200 m away from the replication caused by philopatric individuals being counted bases of each of the five hills (n 5). multiple times. To determine if data were normally distributed we used the criterion of p 0.05 in the Shapiro-Wilk W test using Statistica 6.1 (StatSoft 2003). Laboratory analysis In the case of comparisons of daily mass loss non- parametric Mann-Whitney U tests were used, since the After retrieval of the mesh bags we air-dried them for four data were not normally distributed, even after transforma- weeks in the laboratory of the Seronera Research Centre, tion. We compared the termitaria densities using general Tanzania. Overall, the daily relative mass loss of the grass linear models (GLMs). Mean and standard errors were litter/dung baits served as the parameter indicating the calculated from untransformed data. termites’ decomposition activity. It was calculated as the difference of the baits’ weight at the beginning of the experiment and after completion of the drying. The Results computed values needed to be corrected for two factors: First for the initial water content of the baits. For this, five Field experiment sub-samples, each of 5 10 g, were taken at the beginning of the experiment from each litter/dung sample before placing In the dry season, mean daily litter mass loss (%) per plot at them into the mesh bags. We weighed these sub-samples at the top of ridges was significantly higher than at the bottom the start of the experiment and after four weeks of air- (Fig. 2A). We also found a significantly higher mean daily drying. Second, the weight of the so-called termite ‘‘sheet- relative mass loss per plot of the dung baits located at the ing’’ that was attached to those mesh bags utilised by top positions than at the bottom in the dry season, which termites needed to be corrected for. Sheeting consists of a showed no detectable relative mass loss caused by termites mixture of soil particles, termite saliva, and termite faeces. It (Fig. 2A). However, the wet season results were different. is used by termites to construct foraging tunnels to protect For both litter and dung baits, there was no difference in themselves from harmful direct exposure to sunlight and relative mass loss between the top and bottom hill positions predators (Bagine 1984, Ndiaye et al. 2004). The covering (Fig. 2B). Overall, the mean daily mass loss per plot in the of a given material with sheeting indicates that termites have top positions was less than in the dry season (Fig. 2B). For a used the material as a food source. To determine the weight given catena position, however, there was no significant of the applied sheeting, we transferred the baits to the difference in the daily relative mass loss of either bait type laboratory of the Biological Centre, Univ. of Groningen, across seasons. Despite considerable variance in the relative Netherlands and combusted all organic material by burning mass loss data (both seasons, both ridge positions, both bait the baits for 4 h at 5508C using Nabertherm Controller B 2 170 ovens. Additionally 20 further samples per bait type types pooled; r 0.19; n 42), the underlying assumption without any field exposure and therefore without any of linear increase of relative mass loss of the exposed baits sheeting were treated in an equivalent way to correct for over time is also met by our data set (y 0.89x 5.6; p the weight of the remaining ash residues after combustion. 0.004). Abiotic factors, i.e. water loss over time, were accounted for in our analyses, and therefore cannot explain the observed linear increase. Comparison of the relative Statistical analysis mass loss of one particular bait type per location (combin- ing plots located at top and bottom) across seasons showed To detect any spatial (top vs bottom of hills) and temporal no statistically significant differences (Table 1). Absolute (dry vs wet season) differences in the decomposition activity bait mass losses averaged per location are given in Table 1.
445 (A) 0.6 22 b 0.5 a 20 a 18 0.4 16 0.3 14 a 0.2 12 Mounds per ha per plot 0.1 10
Mean daily mass loss (%) b b 8 0.0 0 Litter - Top Dung - Top Bottom Top Litter - Bottom Dung - Bottom Catena position Bait - Position Figure 3. Number of termite mounds (Odontotermes sp.) found at (B) 0.6 the beginning of the study in the 100 100 m experimental plots located in the long grass region of Serengeti National Park, 0.5 Tanzania. Error bars illustrate means9SE. Different letters indicate significant (pB0.05) differences. n 8 for bottom, a 0.4 respectively top, of catena position.
0.3 hill (not compared to the top of the hill; Fig. 4). Accord- a ingly, the highest density of termitaria was found in this 0.2 a a region, which formed a ring directly below the top of the
per plot hills. Identification of voucher specimens for a related 0.1 experiment (De Visser et al. 2008) carried out in the same Mean daily mass loss (%) study area, revealed that ca 64% of the studied termite 0.0 mounds were inhabited by Odontotermes sp., ca 29% by Litter - Top Dung - Top Trinervitermes sp. and ca 7% by Macrotermes subhyalinus. Litter - Bottom Dung - Bottom The study area is characterised by a pedological feature Bait - Position found in many semi-arid areas. Rainfall has leached salts out of the sandy, highly porous top layers and redeposited them Figure 2. Dry season (A) and wet season (B) mean daily as an impermeable calcium carbonate hardpan about one percentage mass loss (%) of litter and dung baits per plot (n 8 for each possible bait catena position combination). Error bars meter below the surface (petrocalcic horizon) (Sinclair illustrate means9SE. Different letters indicate significant (pB 1979). In our soil pit located at the top of a catena we 0.05) differences within one season. observed the hardpan at a depth of 90 cm compared to
Mapping and soil pits 22 20 b At the 100 100 m plot scale, we found a higher density of 18 termite mounds at the top than at the bottom positions of 16 the ridges (Fig. 3). On the spatial scale of the ‘‘five-hills- 14 ab track’’ we found significantly fewer termite mounds at the 12 ab bottom of the hills compared to the upper 60 80% of the 10 ab 8
Mounds per ha a 6 Table 1. Relative and absolute mass loss of litter and dung baits per location (plots of top and bottom catena position combined). Given 4 are mean values9SE, with sample size in parentheses. Different 2 letters indicate significant (pB0.05, Mann-Whitney U test) seasonal 0 differences within one bait type. 0-20% 21-40% 41-60% 61-80% 81-100% Bait/season Mean daily mass loss (%) Mean daily mass loss (g) Relative distance class to hilltop per location per location Bottom of hill Top of hill
Litter/dry 0.16a90.05 (8) 0.2290.07 (8) Figure 4. Number of termite mounds per ha on major hills (n season 5) situated in the long grass region of Serengeti National Park, Litter/wet 0.12a90.05 (8) 0.1690.08 (8) Tanzania (‘‘5-hills-track’’). Since the hills mapped are of different season a absolute height, termite mound densities are categorised in relative Dung/dry 0.17 90.07 (8) 0.4290.18 (8) distance classes beginning with the lowest 0 20% of distance along season a transect (starting from bottom of hill heading towards the top). Dung/wet 0.15a90.09 (8) 0.3890.24 (8) season Error bars illustrate means9SE. Different letters indicate sig- nificant (pB0.05) differences.
446 Table 2. Abundance and diversity of mammals sighted during the nine (day 1, 2, 3, 4, 5, 6, 10, 15, 20) controls of the termite experimental plots (n 8 for each catena position). Mammals were searched for using binoculars for 5 min per site in a radius of 250 m around the experimental plots prior to approaching the plots. Data for dry and wet season were recorded in an identical way.
Location/season Cumulative individual Number of herbivore Cumulative individual Number of predator sightings of herbivores species sighted sightings of predators species sighted
Catena top/dry 1416 7 9 4 season Catena bottom/ 397 5 0 0 dry season Catena top/wet 73 6 13 1 season Catena bottom/ 24 3 0 0 wet season
140 cm at the bottom of the same catena. It follows that the an indicator of the actual ecosystem engineering activities of actual soil layer is shallower at the top than at the bottom termites, a quantity that is difficult to monitor. position. Although our rates of daily mass loss of baits (Fig. 2) may seem small, they are in the same order of a published study (Schuurman 2005) using a similar experimental and Mammal observations analytical approach. In his study, conducted in the Oka- vango Delta in Botswana, Schuurman (2005) found daily Abundances and diversity of both mammalian herbivores relative mass losses of the used wood litter baits ranging from and predators were higher in both seasons at the top 0.15 to 0.88%, exceptionally up to 1.01%, with higher daily positions of ridges than at the bottom positions (Table 2). mass loss values in the wet season. If we compute the mean The sum of the cumulative sightings of herbivores in both of both catena positions per bait type and season (Table 1), seasons was 3 to 3.5 times greater at the tops than at the we get values near the lower range of those found in bottoms of the hills studied (Table 2). Inter-seasonal Schuurman’s study. A possible explanation for this finding comparisons revealed that the cumulative number of might be that the dominant termite taxon in our study herbivore sightings was greater in the dry than in the wet (Odontotermes sp.) is predominantly xylophagous, although season by a factor of 19 in the case of the top positions and grass-litter and even mammalian hooves are also consumed by a factor of 17 in the case of the bottom positions. These to a lesser extent (Freymann et al. 2007). Macrotermes, in large differences were predominantly caused by the presence turn, is considered to be a mixed wood/grass-litter feeding of large herds of Thomson’s gazelles Gazella thomsoni taxon, while Trinervitermes is a grass-litter feeder. Given that throughout the dry season, which were absent in the wet our study area is almost treeless, it represents a suboptimal season. Other species of mammalian herbivores observed habitat for Odontotermes. Nevertheless, we found consider- were: African buffalo Syncerus caffer, Coke’s hartebeest ably high densities of mounds of these species. This paradox Alcelaphus buselaphus, giraffe Giraffa camelopardalis, Grant’s could indicate an adaptive shift in the food preferences of gazelle Gazella granti, hare Lepus sp., bohor reedbuck Odontotermes, an hypothesis that is supported by the results Redunca redunca, topi Damaliscus korrigum, and warthog Phacochoerus aethiopicus. The encountered species of mam- of a separate stable isotope study (De Visser et al. 2008) malian predators were: bat-eared fox Otocyon megalotis, showing that all dominant mound-building taxa (Odonto- cheetah Acinonyx jubatus, spotted hyena Crocuta crocuta, termes, Trinervitermes, Macrotermes) in our study area feed and black-backed jackal Canis mesomelas. exclusively on grasses. Interestingly, we found similar relative biomass removal values for grass-litter and dung baits (Fig. 2). This supports the conclusion of Freymann and co-workers Discussion (2008) that the functional role of termites in the removal of mammalian dung in tropical ecosystems had previously Termite-driven decomposition been widely underestimated. It should be noted that other fungus-growing, but not mound-building termite taxa (e.g. We found clear experimental evidence for the existence of Ancistrotermes, Pseudacanthotermes, Microtermes) as well as spatial and temporal hotspots of termite activity in the long harvester termites (Hodotermes) also play crucial roles in the grass region of Serengeti National Park (Fig. 1B). At the decomposition of plant materials in addition to those landscape scale of a soil catena, we found significant constructing epigeal mounds. differences in the grass litter and dung biomass removal How can the observed spatio-temporal patterns of by termites; a proxy for the decomposition activities of these termite-driven decomposition be explained? The obvious insects. In the dry season termites removed more bait proximate mechanism underlying the observed differences in biomass at the hilltop positions than at the bottom litter and dung biomass removed in the dry season by positions, but this difference disappeared during the wet termites at the top versus the bottom positions of catenae is season. This corresponded with a higher density of termite the different densities of termitaria in these positions (Fig. 3). mounds at the top of ridges (GLM: F1,14 4.98, pB0.05). A higher density of termitaria in the top positions appears to As outlined earlier, the spatial and temporal heterogeneity cause a higher overall decomposition activity per given area in of the ecosystem service provided by these invertebrates these regions in the dry season (Fig. 2A). While this study is the decomposition of organic material serves moreover as the first to report statistically significant differences in this
447 parameter in relation to topographic variation at this spatial frequently breaks through a layer of soil that is only a few scale, the overall range of termite mound densities we centimetres thick. The soil layer in the ring below the top is documented fall within the range reported in previous likely deeper, given that the hardpan does not penetrate the studies conducted in comparable African savanna habitats. surface there. We may therefore hypothesize that soil depth is Josens (1972 in Baroni-Urbani et al. 1978) found 17 the ultimate explanation for the documented heterogeneous Odontotermes sp. nests per hectare in Lamto savanna (Ivory distribution of termite mounds at this spatial scale. The focal Coast). Numerous publications report mound-densities of termite taxa may prefer to nest in soil of a medium depth of Macrotermes sp.: Bouillon (1970) 3 10 ha 1, Bouillon up to one meter. This optimal soil depth for termites to build and Kidieri (1964) 2 3ha 1, Hesse (1955) 3 4ha 1, nests in may be interlinked with humidity; an interaction Kaib et al. (1997) up to 2 ha 1, Korb and Linsenmair causing the differences in seasonal termite activity. (1998a) up to 6.5 ha 1 in gallery forest, up to 22.7 ha 1 In the rainy season rainfall accumulates at the bottom of in shrub savanna, Lepage (1984) 4 22.4 ha 1. Sands the catena due to run-off effects. This may result in (1965) reports densities of 12.4 (Trinervitermes carbonarius), comparatively more favourable conditions for termites 9.9 (T. suspensus) and 7.4 (T. oeconomus) mounds or nests per (Sheppe 1970) at the bottom positions, a hypothesis hectare for a grassland area. Given that the mound density supported by our biomass removal data (Fig. 2). Although values we present here (mean of 17.8 mounds ha 1 at top the differences are not statistically significant, during the positions and 11.4 mounds ha 1 at bottom positions) wet season we documented a trend toward down-regulation include all three termite genera, our study area appears to of decomposition activity at the top position and an up- be typical when compared to data from other sites. regulation at the bottom positions. This could be the result Korb and Linsenmair (1998a) explain the differences of a seasonal change in foraging behaviour, with termites they found between the studied shrub savanna and gallery foraging over a wider area in the wet season resulting in forest of Comoe´ National Park (Ivory Coast) by differences lower bait mass losses. Nevertheless, no significant differ- in ambient temperature between these habitats. According ences between the seasons were detected in the cross to these authors, lower Macrotermes mound densities are seasonal comparison per location (Table 1) for the bait found in the forest compared to the savanna due to cooler types used. Our failure to detect a significant difference microclimate conditions in the forest, resulting in sub- might indicate that the termite taxa we studied do not optimal nest temperatures. Given the close spatial proximity change their foraging behaviour seasonally, or it might be of the top and bottom positions of the catenae we studied, it due to a low sample size. In the case of the litter baits, the is unlikely that there is a significant difference in ambient seasonal difference was nearly significant (p 0.09). We did temperature between these positions. Differences in micro- not observe more termitaria at the bottom positions, climate can therefore not explain our observed differences in presumably due to the risk of water-logging caused by mound densities, but other factors, such as pedology and extreme rainfall events during the wet seasons and therefore hydrology, must be considered. the potential danger for entire colonies to drown. Termite mound density alone cannot explain the ob- In the dry season, termites may face another challenge as served seasonal variability in termite-driven decomposition they require access to ground water to regulate the internal given that the number of termite mounds in the experimental nest humidity. Water running off the top of the catena will plots was constant across seasons. To understand this probably provide an optimal zone near the top, providing a disparity, we explored potential ultimate explanations for compromise between soil depth and sufficient availability of the observed spatio-temporal pattern of termite-driven ground water. In sum, we suggest an operational link decomposition. Odontotermes sp. and Macrotermes sp., which between soil depth and humidity requirements as the are the dominant mound-building termites in our experi- ultimate explanations for the observed spatial and seasonal mental plots, are representatives of the fungus-cultivating differences in biomass removal in this undulating, semi-arid Macrotermitinae. Based on data from the most intensively environment. studied genus, Macrotermes, we know that the apparently actively regulated humidity inside the nest is maintained at levels that are higher than the atmospheric humidity (Turner Ecosystem-wide consequences 2006). Despite uncertainties regarding how this regulation is achieved, the result of the termites’ mound-building, soil- The spatial concentration of litter- and dung-decomposition particle-redistributing activity is a homeostatic internal activity by termites in activity-hotspots in SNP may have environment within the mound, providing optimal condi- direct and indirect effects on the surrounding ecosystem. tions for the termites’ obligatory exo-symbionts (Turner Various authors (reviewed by Wood and Sands 1978) have 2006). It therefore seems that in addition to a relative acknowledged that termites facilitate the re-entry of macro- constant air temperature inside the mounds (Korb and and micro-nutrients contained in their food items back into Linsenmair 1998a, b, 2000a, b), Macrotermitinae appear to the general nutrient cycle via the decomposition of grass require comparatively high levels of humidity in their nests to litter and dung. Whether these processes of nutrient release grow their symbiotic fungi. This notion alone can not take place immediately via defecation or in a time-delayed explain our interseasonal differences in biomass removal by lag via erosion of termitaria over years or decades (Coventry termites. Our soil pits showed that the soil layer at the top of et al. 1988) varies by taxon. The point centered pedoturba- the catena has a depth of 90 cm and 140 cm at the bottom. tion of termites contributes also to the short range variability In case of the five-hills-track, we found the highest termitaria of the soil texture (Obi and Ogunkunle 2009). By enhancing density along a ring below the top of the hills. At the top of soil nutrient levels and improving soil water availability these larger hills the termite-impermeable calcium-hardpan (Konate´ et al. 1999), termites have positive overall effects on
448 plant growth (review Holt and Lepage 2000). Odontotermes Acknowledgements We thank the Tanzanian Wildlife Research n. pauperans was shown to also influence the spatial pattern Inst. (TAWIRI), Tanzania National Parks (TANAPA) and the of savanna grass species (Jouquet et al. 2004). Tanzanian Commission of Science and Technology (COSTECH) The spatial clustering of termite activity in hotspots for their permission to work in Serengeti National Park. We are grateful to E. Mayemba and Y. Byarugaba for help in the field, to (Table 2) may also have implications for spatial clustering P. Eggleton for an introduction to termite identification and to of mammalian herbivores. While we are beginning to N. D. Eck for help with the laboratory analyses. We thank J. R. understand the abiotic determinants of large herbivore Spence, J. S. Turner and an anonymous referee for very helpful diversity on the global scale (Olff et al. 2002), our comments on an earlier version of the manuscript. We also thank knowledge of the role of termites in causing herbivore S. E. Solomon who greatly improved the English of the manu- clustering on the landscape spatial scale is limited. Coe and script. B. P. F. and H. O. were financially supported by the Robert Carr (1978) report a positive spatial correlation between the Bosch Foundation (Germany) and the Univ. of Groningen as well location of Trinervitermes trinervoides mounds and blesbok as the Netherlands Organization for Scientific Research (NWO), Damaliscus dorcas phillipsi dung middens used as indicators S. N. d. V. by the Univ. of Groningen (Marco Polo Fonds and of the small spatial geographic distribution of this mam- Stichting Groninger Universiteitsfonds). malian herbivore. Fleming and Loveridge (2003) found an increased abundance and diversity of small mammals on Macrotermes mounds compared to the adjacent woodland. References Mobaek and co-workers (2005) report that six out of seven ungulate species studied grazed preferentially on Macro- Abbadie, L. et al. 1992. Soil fauna of the forest-savanna boundary: termes mound areas compared to the adjacent savanna. role of termite mounds in nutrient cycling. In: Furley, P. A. et al. (eds), Nature and dynamics of forest-savanna boundaries. Similarly, throughout both seasons we found considerably Chapman and Hall, pp. 473 484. greater abundances of mammalian herbivores as well as Adamson, A. M. 1943. Termites and the fertility of soils. Trop. mammalian predators in the hotspots areas of termite Agric. 20: 107 112. decomposition activity. This aggregation may be caused by Anderson, J. M. and Wood, T. G. 1984. Mound composition and the enhanced primary productivity within these areas. Also, soil modification by two soil-feeding termites (Termitinae, the removal of dead grass litter by termites and the Termitidae) in a riparian Nigerian forest. Pedobiologia 26: promotion of nutrient cycling could result in an increased 77 82. quality per mammalian herbivore bite, making the tops of Anderson, T. M. et al. 2006. Nutrient acquisition and physiolo- ridges attractive to mammalian herbivores and, by exten- gical responses of dominant Serengeti grasses to variation in sion, to predators. The question of why this spatial pattern soil texture and grazing. J. Ecol. 94: 1164 1175. Bagine, R. K. N. 1984. Soil translocation by termites of the genus is not reversed in the wet season remains open. Potentially, Odontotermes (Holmgren) (Isoptera: Macrotermitinae) in an the tops of hills could be more attractive to mammalian arid area of northern Kenya. Oecologia 64: 263 266. herbivores during the wet season because the termites Baroni-Urbani, C. et al. 1978. Empirical data and demographic removed large amounts of standing dead grass during the parameters. In: Brian, M. V. (ed.), Production ecology of dry season, increasing visibility. ants and termites. Cambridge Univ. Press, pp. 5 44. Our study suggests that termites may alter the spatial Bignell, D. E. and Eggleton, P. 2000. Termites in ecosystems. distribution of mammalian herbivores. Interestingly, this In: Abe, T. et al. (eds), Termites: evolution, sociality, interaction forms a positive feedback loop since the spatial symbioses, ecology. Kluwer, pp. 363 387. aggregation of mammalian herbivores results in a spatially Black, H. I. J. and Okwakol, M. J. N. 1997. Agricultural clustered accumulation of deposited dung, which in turn intensification, soil biodiversity and agroecosystem function in the tropics: the role of termites. Appl. Soil Ecol. 6: 37 53. would promote termite activity (see also Freymann et al. Bouillon, A. 1970. Termites of the Ethiopian region. In: 2008). Clearly, this interplay between termite activity and Krishna, K. and Weesner, F. M. (eds), Biology of termites. mammal distribution deserves further investigation. Vol. II. Academic Press, pp. 153 280. Bouillon, A. and Kidieri, S. 1964. Re´partition des termitie`res de Bellicositermes bellicosus rex Grasse´ et Noirot dans l’Ubangi, Concluding remarks d’apre`s les photos ae´riennes. Corre´lations e´cologiques qu’elle re´ve`le. In: Bouillon, A. (ed.), Etudes sur les Termites Africains. Leopoldville Univ., pp. 373 376. Although recent studies (reviewed by Hastings et al. 2007) Coe, M. and Carr, R. D. 1978. The association between dung have highlighted the importance of interspecific differences middens of the blesbok (Damaliscu dorcas philipsi Harper) and in spatial and temporal scale in the context of ecosystem mounds of the harvester termite (Trinervitermes trinervoides engineering, intraspecific variability of ecosystem engineer- Sjoestedt). S. Afr. J. Wildl. Res. 8: 65 69. ing activities is currently underappreciated. The results of Collins, N. M. 1981. The role of termites in the decomposition of this study demonstrate the fine-scale spatio-temporal varia- wood and leaf litter in the southern Guinea savanna of Nigeria. bility in ecosystem engineering activity by termites. The Oecologia 51: 389 399. differences in litter and dung removal observed among hill Coventry, R. J. et al. 1988. Nutrient cycling by mound-building positions and between seasons resulted in differential impacts termites in low-fertility soils of semi-arid tropical Australia. on the environmental matrix surrounding the termitaria. Aust. J. Soil Res. 26: 375 390. Dangerfield, J. M. et al. 1998. The mound-building termite Although patterns of intraspecific spatial and temporal Macrotermes michaelseni as an ecosystem engineer. J. Trop. heterogeneity are well documented in the long history of Ecol. 14: 507 520. ecological research, this study suggests that the ecosystem- De Visser, S. N. et al. 2008. Trophic interactions among wide importance of this variability within the framework of invertebrates in termitaria in the African savanna: a stable ecosystem engineering deserves much greater consideration. isotope approach. Ecol. Entomol. 33: 758 764.
449 Fleming, P. A. and Loveridge, J. P. 2003. Miombo woodland Lee, K. E. and Wood, T. G. 1971. Termites and soils. Academic termite mounds: resource islands for small vertebrates? J. Press. Zool. Lond. 259: 161 168. Lepage, M. 1984. Distribution, density and evolution of Macro- Freymann, B. P. et al. 2007. Termites of the genus Odontotermes termes bellicosus nests (Isoptera: Macrotermitinae) in the north- are optionally keratophagous. Ecotropica 13: 143 147. east of Ivory Coast. J. Anim. Ecol. 53: 107 117. Freymann, B. P. et al. 2008. The importance of termites (Isoptera) Meyer, V. W. et al. 1999. Distribution and density of termite for the recycling of herbivore dung in tropical ecosystems: a mounds in the northern Kruger National Park, with specific review. Eur. J. Entomol. 105: 165 173. reference to those constructed by Macrotermes Holmgren Hastings, A. et al. 2007. Ecosystem engineering in space and time. (Isoptera: Termitidae). Afr. Entomol. 7: 123 130. Ecol. Lett. 10: 153 164. Milne, G. 1935. Some suggested units of classification and Hesse, P. R. 1955. A chemical and physical study of the soils of mapping, particularly for East African soils. Soil Res. 4: termite mounds in East Africa. J. Ecol. 43: 449 461. 183 198. Holt, J. A. and Lepage, M. 2000. Termites and soil properties. Mobaek, R. et al. 2005. Termitaria are focal feeding sites for large In: Abe, T. et al. (eds), Termites: evolution, sociality, ungulates in Lake Mburo National Park. J. Zool. 267: 97 symbioses, ecology. Kluwer, pp. 389 407. 102. Jarman, P. J. and Sinclair, A. R. E. 1979. Feeding strategy and the Ndiaye, D. et al. 2004. Nitrogen transformations associated with pattern of resource partitioning in ungulates. In: Sinclair, termite biogenic structures in a dry savanna ecosystem. Plant A. R. E. and Norton-Griffiths, M. (eds), Serengeti, dynamics of an ecosystem. Univ. of Chicago Press, pp. 130 163. Soil 265: 189 196. Noirot, C. and Darlington, J. P. E. C. 2000. Termite nests: Jime´nez, J. J. and Decae¨ns, T. 2006. Chemical variations in the biostructures produced by soil ecosystem engineers. Examples architecture, regulation and defence. In: Abe, T. et al. (eds), for the neotropical savannas. Eur. J. Soil Biol. 42: S92 S102. Termites: evolution, sociality, symbioses, ecology. Kluwer, pp. Jones, C. G. et al. 1994. Organisms as ecosystem engineers. 121 139. Oikos 69: 373 386. Obi, J. C. and Ogunkunle, A. O. 2009. Influence of termite Jones, C. G. et al. 1997. Positive and negative effects of organisms infestation on the spatial variability of soil properties in the as physical ecosystem engineers. Ecology 78: 1946 1957. Guinea savanna region of Nigeria. Geoderma 148: 357 363. Josens, G. 1972. Etudes biologiques et e´cologiques des termites Olff, H. et al. 2002. Global environmental controls of diversity in (Isoptera) de la savanna de Lamto-Pakobo (Coˆte d’Ivoire). large herbivores. Nature 415: 901 904. Ph.D. thesis, Univ. of Brussels, Brussels, Belgium. Pomeroy, D. E. 1977. The distribution and abundance of large Jouquet, P. et al. 2004. Association between subterranean termites termite mounds in Uganda. J. Appl. Ecol. 14: 465 475. and grasses in a West African savanna: spatial pattern analysis Sands, W. A. 1965. Termite distribution in man-modified habitats shows a significant role for Odontotermes n. pauperans. Appl. in West Africa, with special reference to species segregation in Soil Ecol. 27: 99 107. the genus Trinervitermes (Isoptera, Termitidae, Nasutitermiti- Jouquet, P. et al. 2005. Impact of subterranean fungus-growing nae). J. Anim. Ecol. 34: 557 571. termites (Isoptera, Macrotermitinae) on chosen soil properties Schuurman, G. 2005. Decomposition rates and termite assem- in a West African savanna. Biol. Fertil. Soils 41: 365 370. blage composition in semiarid Africa. Ecology 86: 1236 Jouquet, P. et al. 2006. Soil invertebrates as ecosystem engineers: 1249. intended and accidental effects on soil and feedback loops. Sheppe, W. 1970. Daily and seasonal patterns of construction Appl. Soil Ecol. 32: 153 164. activity by Odontotermes latericius (Isoptera: Termitidae). Kaib, M. et al. 1997. Distribution of Macrotermes michaelseni Insectes Soc. 17: 225 232. colonies on different spatial scales. Mitt. Dtsch. Ges. Allg. Sinclair, A. R. E. 1979. The Serengeti environment. In: Sinclair, Angew. Ent. 11: 189 192. A. R. E. and Norton-Griffiths, M. (eds), Serengeti, dynamics Konate´, S. et al. 1999. Influence of large termitaria on soil of an ecosystem. Univ. of Chicago Press, pp. 31 45. characteristics, soil water regime, and tree leaf shedding pattern StatSoft 2003. STATISTICA. Version 6.1. StatSoft, Tulsa, OK, in a West African savanna. Plant Soil 206: 47 60. USA. Korb, J. and Linsenmair, K. E. 1998a. The effects of temperature Traniello, J. F. A. and Leuthold, R. H. 2000. Behavior and on the architecture and distribution of Macrotermes bellicosus ecology of foraging in termites. In: Abe, T. et al. (eds), (Isoptera, Macrotermitinae) mounds in different habitats of a Termites: evolution, sociality, symbioses, ecology. Kluwer, pp. West African Guinea savanna. Insectes Soc. 45: 51 65. 141 168. Korb, J. and Linsenmair, K. E. 1998b. Experimental heating of Turner, J. S. 2006. Termites as mediators of the water economy of Macrotermes bellicosus (Isoptera, Macrotermitinae) mounds: arid savanna ecosystems. In: D’Odorico, P. and Porporato, A. what role does microclimate play in influencing mound architecture? Insectes Soc. 45: 335 342. (eds), Dryland ecohydrology. Springer, pp. 303 313. Korb, J. and Linsenmair, K. E. 2000a. Ventilation of termite Wood, T. G. 1976. The role of termites (Isoptera) in decomposi- mounds: new results require a new model. Behav. Ecol. 11: tion processes. In: Anderson, J. M. and Macfadyen, A. (eds), 486 494. The role of terrestrial and aquatic organisms in decomposition Korb, J. and Linsenmair, K. E. 2000b. Thermoregulation of processes. Blackwell, pp. 145 168. termite mounds: what role does ambient temperature and Wood, T. G. 1978. Food and feeding habits of termites. In: metabolism of the colony play? Insectes Soc. 47: 357 363. Brian, M. V. (ed.), Production ecology of ants and termites. Krebs, C. J. 2001. Ecology: the experimental analysis of distribu- Cambridge Univ. Press, pp. 55 80. tion and abundance. Benjamin Cummings. Wood, T. G. and Sands, W. A. 1978. The role of termites in Lavelle, P. et al. 1997. Soil function in a changing world: the role ecosystems. In: Brian, M. V. (ed.), Production ecology of of invertebrate ecosystem engineers. Eur. J. Soil Biol. 33: ants and termites. Cambridge Univ. Press, pp. 245 292. 159 193. Wright, J. P. and Jones, C. G. 2006. The concept of organisms as Lavelle, P. et al. 2006. Soil invertebrates and ecosystem services. ecosystem engineers ten years on: progress, limitations, and Eur. J. Soil Biol. 42: S3 S15. challenges. BioScience 56: 203 209.
450