Chapter 8

Plant Demography and Dynamics: Fire Traps

Besides seasonal changes in biomass dynamics, plants also change in abundance through the establishment of new plants and eventually their death. These demographic processes are difficult to study because of the timescales involved: trees can live for centuries. Among grasses, populations can turn over between one year and the next. Moreover, new grass plants can arise either from seeds or through vegetative growth by established plants. Plants do not progress steadily through life-history stages like mammals and birds do. Seeds may remain dormant for many years until favourable conditions occur, and then all germinate. Most of the seedlings will fade and disappear, unless conditions remain favourable for long enough for roots to be established. Trees may be held in the juvenile stage for many years by recurrent fires. Grass tufts expand and contract through tiller production and death as well as through the establishment of new plants from seeds. Creeping grasses spread by sending out runners, which can root from nodes and break connections so as to appear as new plants. Notions of climax states and stable equilibria have little relevance for savanna vegetation dynamics. Much of plant competition operates like a lottery for access to gaps opened by the death of plants established earlier. Plants typically produce huge numbers of seeds because this increases their chances of finding a gap. Individual trees can vary enormously in size during their growth from newly germinated seedling to mature seed-producing adult, generating a huge shift in biomass without any numerical change.

Nevertheless, impending population changes can be inferred from changes in the stage structure of the plants constituting the population. Populations with lots of small plants are likely to be expanding, while those with mostly big trees and few potential recruits could be in imminent decline. But caution is needed, because size does not necessarily represent age.

Grasses may appear to be ‘Lilliputs’ in comparison with giant trees or ‘Gullivers’,1 but this size contrast is misleading. Much of the mass of a tree is made up of accumulated dead wood, while grasses have a greater portion of their mass underground than appears above. A fairer comparison between the relative abundances of these plant types can be made in terms of leaf canopy cover, because energy captured by leaf surfaces drives the metabolic processes generating growth. Because the grass canopy is continuous while the tree canopy is intermittent, most of the annual production of plant matter in savannas takes place in the form of grasses.

This chapter outlines what is known about the life-history transitions and population dynamics of savanna trees, shrubs and grasses, fragmented in time and space though this knowledge is.

Woody Plant Populations

A healthily expanding tree population is expected to have a size class distribution resembling a reverse-J, i.e. lots of small plants in waiting to replace fewer big plants. Few savanna tree populations show this ideal pattern (Figure 8.1). Humped distributions with many intermediate-sized plants can be produced where seedling establishment takes place episodically, only in years when suitable conditions occur. This generates cohorts of plants that were established in the same year moving through the size classes (Figure 8.2). However, a similar size structure distribution could also arise if some factor blocks plants from growing beyond some size. A true-J distribution with mostly big trees and few recruits can be generated if intervals between successful establishment become sufficiently long. Trees need to produce only 1–2 replacement trees during a lifetime potentially spanning several centuries in order to maintain the population.


Figure 8.1

Size structure distributions of selected tree species comparing protected areas (black) with communal rangelands (grey) in the South African Lowveld region bordering Kruger NP. (A,B) Typical reverse-J distributions shown by silver clusterleaf and sickle bush; (C,D) humped distributions shown by large-leaf bushwillow (Combretum zeyheri) and red bushwillow; (E,F) gap in intermediate size classes shown by marula and russet bushwillow.

(from Kirsten Neke, PhD thesis, University of the Witwatersrand, 2005)


Figure 8.2

Even-aged stand of fever trees 50 years after a flood provided favourable conditions for their establishment in northern Kruger NP, still densely thronged although almost 20 m tall.

(photo by Johna Turner)

Seedlings or juvenile woody plants under 0.5 m in height tend to be under-counted because they are hidden among the grasses, unless the grass cover is removed by grazing, but they then become exposed to being browsed. However, caution is needed in interpreting whether plants are genuinely seedlings, i.e. under a year old. Some little trees could have enormous rootstocks built up over many years, awaiting an opportunity to shoot up beyond the height where they get burnt back. Enormous numbers of seedlings may germinate in favourably wet years, but most fall by the wayside before the end of their first dry season.

Marula trees can show particularly puzzling size structure profiles. A study contrasting the situation between nature reserves with elephants and communal rangelands lacking elephants found a dearth of small trees 2–4 m in height in both localities,2 and this size class was also missing in nearby Kruger NP3 (Figure 8.1). The missing size class was represented within Kruger NP inside fenced areas excluding elephants, but also other herbivores, as well as along road verges outside the park.4 There was no apparent lack of juvenile marula plants <0.5 m in height, either in the nature reserves or in communal lands,2 but inside Kruger NP some places had only big marula trees, whereas elsewhere juvenile-sized plants were common.3,5 How might these patterns be interpreted? (1) Do elephants selectively break plants in the missing size range? Or (2) do plants adaptively grow rapidly through the missing size range to reach the height at which they become less vulnerable to elephant damage? It is also possible that big marula trees are a legacy of a recruitment wave before elephant numbers became substantial, but other tree species similarly vulnerable to elephant damage did not show this missing size class.

In order to reach seed-producing maturity, all savanna trees must escape the fire trap holding them below 2–3 m in height. Variable intervals between hot fires may eventually allow saplings to grow above 3 m.6,7 Juvenile acacias may remain within the fire trap for decades.8 Cooler conditions slow growth rates, and thus contribute to the absence of trees from high-altitude grasslands.9

Savanna trees respond adaptively to the fire trap by growing rapidly in height relative to stem diameter until they get above 3 m and thereafter shift their growth investment towards expanding in stem diameter (Figure 8.3).2 Marula trees exhibited the lowest height relative to diameter during the sapling stage among the tree species measured, despite eventually growing into the biggest trees.


Figure 8.3

Allometry of growth in height versus stem diameter for selected tree species combining protected areas (circles) with communal rangelands (crosses) in the South African Lowveld region bordering Kruger NP. Species names as in Figure 8.1, plus monkey orange (Strychnos madagascariensis). Note how plants grow initially upwards then expand basal diameter (and canopies) once they emerge above the flame zone.

(from Kirsten Neke, PhD thesis, University of the Witwatersrand, 2005)

Baobab trees also show spatially variable size structure profiles. In Kruger NP, plants found on rocky hills showed a range in size, while few small trees were present on the plains below.10 This could be explained either because fewer elephants ascended the hills, or because fires penetrated less frequently, due to lower grass cover among the rocks. Whatever the reason, this pattern suggests a source–sink relationship, with seeds produced on the hills getting dispersed to replace the diminishing population on the plains. Tree species sensitive to being set back by giraffe browsing can likewise become restricted to rocky hillsides.11

Within Serengeti NP, there have evidently been two recruitment episodes of acacia trees. The more recent one began during the 1970s when the extent of the park area burnt annually became greatly reduced by grazing pressure from the resurgent wildebeest population.12,13 An earlier wave of tree establishment apparently occurred during the 1890s, following the rinderpest panzootic and consequent loss of livestock by Maasai herders. Distinct cohorts of splendid acacias were generated during these two periods. Currently, few juvenile acacias occur in the understorey. Instead, shrubs of other species predominate, indicating an impending population turnover.14

Growth rings shown by acacia trees within the Hluhluwe-iMfolozi Park, calibrated for age using radiocarbon dating, revealed spatial variation in age structure profiles along with temporal variation in time of establishment.15 Establishment periods were not synchronised spatially and differed among tree species. They could not be coupled with any particular climatic pattern, fire incidence or change in herbivore populations. In this park, a 2-km strip along the boundary along with areas of dense bush had been cleared of all woody plants during the 1940s to confine tsetse flies (the vector for cattle sleeping sickness) within the park, before aerial spraying with insecticides took place. The cleared areas all reverted to the vegetation formations that had been there previously – acacia trees where there had been acacia savanna, evergreen shrubs in the form of gwarrie bushes (Euclea divinorum) in the lowland pockets where shrub thickets had been present. Some tree species seem to be undergoing a widespread expansion wave presently. They include mopane, silver clusterleaf and tamboti in South Africa’s savanna bushveld, while sweet thorn has been spreading into grassland through the Eastern Cape.16 Ongoing turnover among acacia species has been observed over decades in the Hluhluwe-iMfolozi Park.17

Trees eventually die from competition for resources,18 soil moisture deficits,19 wind throw,20,21 fires burning into their trunks, disease or from cavities in their trunks generated by soil water deficiencies.22 Mortality rates typically decrease with increasing tree size, from 5 percent to 10 percent annually among small saplings towards 1–2 percent among large trees. Wind throw may topple quite large trees, especially those with trunk decay.23 The potential lifespans of savanna trees or shrubs can vary hugely, from over a thousand years for leadwood trees attaining a trunk circumference up to 2.2 m to 30–40 years for sicklebush and sweet thorn.15,24 No acacia tree sampled in Hluhluwe-iMfolozi was older than a century,15 although giraffe thorn trees can live up to 240 years of age elsewhere.25 The oldest reliable age for a baobab tree is ~2500 years.26 During an exceptionally severe drought experienced in Kruger NP, tree mortality was locally high but spatially variable.27 Shrubs and saplings incurred less mortality than taller trees.

Grass Populations

Grass populations turn over much faster than tree populations. During extreme droughts, the ground surface can become locally bared of plants by large grazers plus the activities of harvester termites (Figure 7.3) and yet be covered by green grass within a few weeks after rains resume.28,29 Of course, the grass species may be different – composed mainly of annual grasses, or less nutritious needle grasses, which produce huge numbers of small seeds.30 Various forbs may also contribute to the post-drought cover.28,31 Nevertheless, I have observed three instances where the grass re-establishing was mainly Guinea grass, among the most palatable species for large herbivores (see Chapter 13). Even red grass, with comparatively large seeds, can rapidly fill gaps opened by drought-related mortality among tufts or tillers, provided rodents and seed-eating birds have not consumed all the seeds.32,33 But where do these grass species find refuge during droughts so as to re-establish their seed banks? Low bushes may provide some protection from grazing and desiccation.

Population turnovers among savanna grasses can be surprisingly rapid.34,35 Grass turnover is less in upland or montane grasslands where rainfall varies proportionately less and individual grasses fluctuate in abundance mainly through tuft expansion and contraction. Little is known about the potential lifespan of a grass tuft. Grazing pressure can make a big difference, as will be discussed in Chapter 13, after the herbivores have been introduced.


Savanna tree populations may expand or contract in response to shifting climatic conditions or other influences. Progression through life-history stages by tree saplings is governed largely by variable intervals between fires. Notions of compositionally stable climax states have little relevance. While some savanna tree species can live for several centuries, others have lifespans briefer than a century. While the grass biomass produced annually depends on rainfall, the species contributing can vary from one year to the next.

Studies elsewhere in the world have documented how pine tree woodlands in North America were established episodically at intervals of several hundred years, governed by variable precipitation and fire return intervals.36 This pattern is likely to be widespread, although difficult to document.

In the following chapter, I will expand the time horizon to consider changes in the vegetation cover that have taken place since savanna formations appeared during the late Miocene.


O’Connor, TG; Everson, TM. (1998) Population dynamics of perennial grasses in African savanna and grassland. In Cheplick, P (ed.) Population Biology of Grasses. Cambridge University Press, Cambridge, pp. 333–365.

Staver, AC, et al. (2011) History matters: tree establishment variability and species turnover in an African savanna. Ecosphere 2(4):Art 49.


1.Bond, WJ; van Wilgen, BW. (1996) Fire and Plants. Springer, Berlin.

2.Neke, KS. (2005) The regeneration ecology of savanna woodlands in relation to human utilisation. PhD thesis, University of the Witwatersrand, Johannesburg.

3.Helm, CV; Witkowski, ETF. (2012) Characterising wide spatial variation in population size structure of a keystone African savanna tree. Forest Ecology and Management 263:175–188.

4.Helm, CV; Witkowski, ETF. (2013) Continuing decline of a keystone tree species in the Kruger National Park, South Africa. African Journal of Ecology 51:270–279.

5.Biggs, R; Jacobs, OS. (2002) The status and population structure of the marula in the Kruger National Park. South African Journal of Wildlife Research 32:1–12.

6.Higgins, SI, et al. (2000) Fire, resprouting and variability: a recipe for grass–tree coexistence in savanna. Journal of Ecology 88:213–229.

7.Wakeling, JL, et al. (2011) Simply the best: the transition of savanna saplings to trees. Oikos 120:1448–1451.

8.Bond, WJ. (2008) What limits trees in C-4 grasslands and savannas? Annual Review of Ecology Evolution and Systematics 39:641–659.

9.Wakeling, JL, et al. (2012) The savanna–grassland ‘treeline’: why don’t savanna trees occur in upland grasslands? Journal of Ecology 100:381–391.

10.Edkins, MT, et al. (2008) Baobabs and elephants in Kruger National Park: nowhere to hide. African Journal of Ecology 46:119–125.

11.Bond, WJ; Loffell, D. (2001) Introduction of giraffe changes acacia distribution in a South African savanna. African Journal of Ecology 39:286–294.

12.Sinclair, ARE, et al. (2007) Long‐term ecosystem dynamics in the Serengeti: lessons for conservation. Conservation Biology 21:580–590.

13.Sinclair, ARE, et al. (2008) Historical and future changes to the Serengeti ecosystem. In Sinclair, ARE, et al. (eds) Serengeti III: Human Impacts on Ecosystem Dynamics. University of Chicago Press, Chicago, pp. 7–46.

14.Anderson, TM, et al. (2015) Compositional decoupling of savanna canopy and understory tree communities in Serengeti. Journal of Vegetation Science 26:385–394.

15.Staver, AC, et al. (2011) History matters: tree establishment variability and species turnover in an African savanna. Ecosphere 2:1–12.

16.Bond, WJ; Midgley, GF. (2012) Carbon dioxide and the uneasy interactions of trees and savannah grasses. Philosophical Transactions of the Royal Society B – Biological Sciences 367:601–612.

17.Bond, WJ, et al. (2001) Acacia species turnover in space and time in an African savanna. Journal of Biogeography 28:117–128.

18.Sea, WB; Hanan, NP. (2012) Self‐thinning and tree competition in savannas. Biotropica 44:189–196.

19.Chase, MF, et al. (2019) Severe drought limits trees in a semi-arid savanna. Ecology 100:e02842.

20.Spinage, CA; Guinness, FE. (1971) Tree survival in the absence of elephants in the Akagera National Park, Rwanda. Journal of Applied Ecology 8:723–728.

21.Williams, RJ; Douglas, M. (1995) Windthrow in a tropical savanna in Kakadu National Park, northern Australia. Journal of Tropical Ecology 11:547–558.

22.Shackleton, CM. (1997) The prediction of woody productivity in thie savanna biome, South Africa. PhD thesis, University of the Witwatersrand, Johannesburg.

23.Teren, G, et al. (2018) Elephant‐mediated compositional changes in riparian canopy trees over more than two decades in northern Botswana. Journal of Vegetation Science 29:585–595.

24.Fuls, A; Vogel, JC. (2005) The life-span of leadwood trees. South African Journal of Science 101:98–100.

25.Seymour, CL. (2008) Grass, rainfall and herbivores as determinants of Acacia erioloba (Meyer) recruitment in an African savanna. Plant Ecology 197:131–138.

26.Patrut, A, et al. (2018) The demise of the largest and oldest African baobabs. Nature Plants 4:423–426.

27.Swemmer, AM. (2020) Locally high, but regionally low: the impact of the 2014–2016 drought on the trees of semi-arid savannas, South Africa. African Journal of Range & Forage Science 37:31–42.

28.Wilcox, KR. (2020) Rapid recovery of ecosystem function following extreme drought in a South African savanna grassland. Ecology 101:e02983.

29.Wigley-Coetsee, C; Staver, AC. (2020) Grass community responses to drought in an African savanna. African Journal of Range & Forage Science 37:43–52.

30.Abbas, HA, et al. (2019) The worst drought in 50 years in a South African savannah: limited impact on vegetation. African Journal of Ecology 57:490–499.

31.Novellie, PA; Bezuidenhout, H. (1994) The influence of rainfall and grazing on vegetation changes in the Mountain Zebra National Park. South African Journal of Wildlife Research 24:60–71.

32.O’Connor, TG. (1991) Local extinction in perennial grasslands: a life-history approach. The American Naturalist 137:753–773.

33.O’Connor, TG; Pickett, GA. (1992) The influence of grazing on seed production and seed banks of some African savanna grasslands. Journal of Applied Ecology 29:247–260.

34.O’Connor, TG. (1993) The influence of rainfall and grazing on the demography of some African savanna grasses: a matrix modelling approach. Journal of Applied Ecology 30:119–132.

35.Anderson, TM. (2008) Plant compositional change over time increases with rainfall in Serengeti grasslands. Oikos 117:675–682.

36.Brown, PM; Wu, R. (2005) Climate and disturbance forcing of episodic tree recruitment in a southwestern ponderosa pine landscape. Ecology 86:3030–3038.

If you find an error or have any questions, please email us at Thank you!