The Nama-Karoo of South Africa is a semi-arid biome with infrequent historical fire occurrence and perennial dwarf shrubs and grasses considered to be fire independent and fire sensitive (Mucina et al. 2006, Le Maitre et al. 2014). Recently, fire has become more common in the Nama-Karoo (Du Toit et al. 2016) and other arid ecosystems (Syphard et al. 2017), potentially threatening the persistence of native species. Increases in fire occurrence may result from changing climate and resultant increases in fire-prone weather conditions (Wilson et al. 2010, Kraaij et al. 2013), or changes in the structural composition of vegetation affecting fuel attributes of ecosystems (D’Antonio and Vitousek 1992, Rahlao et al. 2009, Syphard et al. 2017). Increased grassiness in arid shrublands promotes fuel continuity, thereby resulting in increased occurrences of fire (D’Antonio and Vitousek 1992, Brooks et al. 2004, Nicholas et al. 2009). This has been seen in grassy Nama-Karoo, where sufficient grassy fuels may accumulate in high rainfall years to sustain fires (Le Maitre et al. 2014), or with the invasion of arid shrublands by alien grasses in the Karoo (Rahlao et al. 2009) and elsewhere (Brooks et al. 2004, Balch et al. 2013, Syphard et al. 2017). Given that grasses are usually more tolerant of fire than shrubs, the combination of grasses and fire could initiate a persistent negative feedback called the grass-fire cycle (D’Antonio and Vitousek 1992, Brooks et al. 2004, Rahlao et al. 2009, Keeley and Brennan 2012).

Shrub-grass ratios in semi-arid vegetation (including the Nama-Karoo) are influenced by several factors; grassiness mostly increases with the amount and reliability of rainfall, a greater proportion of summer rainfall (i.e., from west to east in the Nama-Karoo), altitude, and long-term reductions in livestock grazing pressure (Kraaij and Milton 2006; Mucina et al. 2006; Nicholas et al. 2009; Brooks and Chambers 2011; Du Toit et al. 2015, 2016). The literature dealing with Karoo vegetation is dominated by phytosociological descriptions of the flora, and studies on vegetation structural and compositional responses to rainfall and utilization by herbivores (Rubin and Palmer 1996, Rubin et al. 2001, Kraaij and Milton 2006, Mucina et al. 2006). However, given the rarity of fire in Karoo systems, plant tolerance of fire is poorly studied (Du Toit et al. 2015). The resilience of arid vegetation to fire is thought to increase along rainfall and nutrient (∼primary productivity) gradients (Brooks and Chambers 2011). Post-fire recruitment abilities of Karoo vegetation would thus be expected to improve with increasing rainfall from west to east and with altitude (cf. Brooks and Chambers 2011). Limited evidence from the eastern Nama-Karoo suggests that fire leads to increases in grassiness (Roux and Vorster 1983, Du Toit et al. 2016) and that most shrubs are able to resprout after fire (Du Toit et al. 2015). However, in Succulent Karoo, a drier vegetation with little grass, very few species of dwarf shrubs were able to resprout post fire (Rahlao et al. 2009, Van der Merwe et al. 2016).

Given the inadequate understanding of plant (shrubs in particular) responses to fire, we aimed to (1) assess changes in plant growth-form composition resulting from fire in westernmost grassy Nama-Karoo vegetation; and (2) characterize fire survival or persistence strategies of perennial shrubs in this vegetation. Improved knowledge of vegetation responses to fire should inform the management of fire to facilitate biodiversity conservation in Nama-Karoo ecosystems.


The study area was the Puttersvlei section of the Karoo National Park on the upper plateau of the Nuweveld Mountains (altitude 1676 m above sea level; 32°13′12″S, 22°31′12″E to 32°16′12″S, 22°28′48″E). The vegetation is a mixture of shrubs and grasses, composed of two vegetation units: Upper Karoo Hardeveld (Nama-Karoo Biome), and the westernmost occurrence of Karoo Escarpment Grassland (Grassland Biome) (Mucina et al. 2006). Mean annual rainfall is 406 mm, mean daily temperatures are 5°C in July (winter) and 18 °C in February (summer), with the absolute minimum-maximum range being −1 °C to 34 °C (Rubin and Palmer 1996). The geology is dolerite rocks with underlying sandstone appearing in terraces (Rubin and Palmer 1996).

Rubin and Palmer (1996) surveyed the vegetation at Puttersvlei in plots (each 10 m × 10 m) as part of a larger study that produced a phytosociological description of the park. We re-surveyed 22 of these plots during April 2016, 2.5 yr after a human-induced fire (8 to 12 October 2013) burned an area of ∼2500 ha. We repeated the methods of Rubin and Palmer (1996) by recording all plant species present in each plot. We assigned each species to a cover value class (Werger 1974) and used the mid-range projected canopy cover values of the respective classes (i.e., 0.01 %, 0.5 %, 2.5 %, 8.5 %, 19.0 %, and 37.5 %). We furthermore classified each species by growth form according to a simplified version of the categorization by Rubin et al. (2001), as follows: graminoid, herb (including herbs, ferns, and geophytes), dwarf shrub, and large shrub (including shrubs and trees). We detected potential discrepancies in species identifications and canopy cover estimates between the 1996 (hereafter pre fire) and 2016 (hereafter post fire) studies. Fifty-eight of 163 species were recorded in both studies, while 55 species were recorded only pre fire and 50 species only post fire. To account for potential discrepancies in projected canopy cover estimates, we assessed changes between pre-fire and post-fire periods in terms of absolute and relative cover abundances of growth forms. Relative canopy cover was calculated as the percentage of the total (summed) vegetation canopy cover in a plot represented by each respective growth form. We compared relative and absolute cover per growth form within the same plots between pre-fire and post-fire periods. The data did not conform to normality; we thus used Wilcoxon matched pairs tests (Statistica v.13, Dell Inc., Tulsa, Oklahoma, USA). To explore species compositional changes, we counted the number of plots in which each species appeared or disappeared from the pre-fire to post-fire periods. To account for potential discrepancies in species identifications between the pre-fire and post-fire studies, we limited this analysis to those species that were recorded in both studies.

We also assessed the fire survival or persistence strategies of seven large shrub species that were abundant in particular localities (outside of plots) prior to the fire. We counted the respective numbers of resprouting, reseeding, and dead individuals post fire in 10 m wide belt transects. The length of transects varied (range 50 m to 300 m) to allow for inclusion of sufficient numbers of individuals (preferably ≥100) per species per locality. Large shrub species were assessed at various localities (one to three localities per species; Table 1), depending on the availability of sufficiently dense populations. For each population sampled, we calculated (1) the ratio of the post-fire population (sum of all reseeding and resprouting individuals) to the pre-fire population (sum of resprouters and mortalities); (2) the percentage resprouting (number of resprouters divided by the pre-fire population × 100); and (3) the percentage reseeding (number of post-fire seedlings divided by the pre-fire population × 100). Nomenclature follows Raimondo et al. (2015) (see Appendix 1).

Table 1 The extent of fire survival or persistence displayed 2.5 years after fire by large shrub species in a Nama-Karoo montane shrubby grassland, South Africa. Measures reported are based on numbers of individuals and calculated as detailed under Methods.


From pre-fire to post-fire periods, absolute cover increased significantly in grasses (Z 1,21 = 2.06, P = 0.039), herbs (Z 1,21 = 3.00, P = 0.002), and dwarf shrubs (Z l,21 = 3.26, P = 0.001), but not in large shrubs (Z 1,21 = 1.42, P = 0.154) (Figure 1). In terms of relative cover, grasses remained unchanged (Z 1,21 = 0.08, P = 0.935) from pre-fire to post-fire, while herbs (Z 1,21 = 2.64, P = 0.008) and dwarf shrubs (Z 1,21 = 2.38, P = 0.017) increased significantly, and large shrubs decreased significantly (Z 1,21 = 2.06, P = 0.039).

Figure 1
figure 1

(a) Absolute and (b) relative projected canopy cover (expressed as a percentage of total vegetation cover) of various growth forms compared between pre-fire (Rubin and Palmer 1996) and post-fire (2016) periods in a Nama-Karoo montane shrubby grassland, South Africa.

The extent to which large shrub species persisted through fire varied widely, with post-fire population sizes ranging from zero to sixfold the pre-fire population sizes (Table 1). All the assessed species, except Elytropappus rhinocerotis, displayed the ability to resprout after fire. Resprouting vigor was high in Helichrysum trilineatum, Diospyros austro-africana and Anthospermum spathulatum, weak in Cliffortia arborea and Euryops annae, and varied among subpopulations of Passerina montana. All assessed species, except Elytropappus rhinocerotis, recruited from seed with large variations among and within species (Table 1). Seven out of 14 large shrub species declined from pre-fire to post-fire in terms of the number of plots that they occupied (Appendix 1).


We showed that fire resulted in increases or maintenance of grass cover in westernmost grassy Nama-Karoo vegetation, in accordance with findings for eastern Nama-Karoo vegetation (De Klerk et al. 2001, Du Toit et al. 2015). Additionally, we found that herbaceous plants and dwarf shrubs increased significantly after fire. Du Toit et al. (2015) did not report on responses of herbs (other than grasses) to fire in the eastern Nama-Karoo, but herbs commonly abound in early successional stages of various vegetation types (Musil and De Witt 1990, Van der Merwe and Van Rooyen 2011, Bachinger et al. 2016). Most dwarf shrubs in Nama-Karoo appeared capable of resprouting after fire (Du Toit et al. 2015), whereas those in Succulent Karoo largely appeared incapable of post-fire resprouting (Rahlao et al. 2009, Van der Merwe et al. 2016). This discrepancy may reflect disparate evolutionary histories and lower fire probabilities under the drier climatic conditions of the Succulent Karoo. We encountered very few succulent dwarf shrubs in our study, but other studies suggest that succulents are incapable of sprouting, although they may establish prolifically from seed after disturbance (Du Toit et al. 2015, Van der Merwe et al. 2016). The finding that dwarf shrubs mostly recovered well after fire in Nama-Karoo is encouraging for the maintenance of these shrublands should fires become more frequent in the future.

Large shrubs were the only growth form that decreased significantly after fire. Despite all large shrub species (except Elytropappus rhinocerotis) assessed in this study exhibiting resprouting abilities (as found by Du Toit et al. 2015), resprouting success varied widely within species. Some of the variation in resprouting appeared to be related to localized differences in fire intensity (T. Kraaij, Nelson Mandela University, George, South Africa, personal observation; cf. Van Wilgen 1980). Likewise, all assessed large shrub species, except E. rhinocerotis, recruited from seed to lesser or greater extents, again showing large variation among subpopulations within species. Collectively, our results suggest that more than half of the large shrub species that were assessed (including the iconic Cliffortia arborea, Red-listed as Vulnerable; Raimondo et al. 2015) may fail to sustain their populations after fire. These large shrubs displayed low post-fire:pre-fire population ratios (Table 1) or post-fire disappearance from the majority of plots (Appendix 1). The lack of post-fire recovery of Elytropappus rhinocerotis was contrary to its prolific reseeding after fires in fire-prone renosterveld (Levyns 1972, Van der Merwe and Van Rooyen 2011). Its lack of recruitment may be due to the timing of fire in relation to seed production, the timing of post-fire rainfall (S.J. Milton, RenuKaroo, Prince Albert, South Africa, personal communication), or high intensity of burning (T. Kraaij, personal observation). We showed that several species may be vulnerable to local extinction through fires despite possessing resprouting and reseeding abilities. This was detected through quantifying, as opposed to merely categorizing, fire survival and persistence responses of large shrubs, and through conducting surveys 2 to 3 years, rather than a few months, post fire.

The above suggests that specific attributes of individual fires and fire regimes (i.e., size, patchiness, fire intensity, fire return interval, and fire season) are likely to affect persistence of species and vegetation structure in the long-term (cf. Vlok and Yeaton 2000, Parr and Andersen 2006, Kraaij and Van Wilgen 2014). Furthermore, abundances of growth forms (and herbaceous species in particular) typically vary widely among years and with post-fire vegetation age (Musil and De Witt 1990, Van der Merwe and Van Rooyen 2011, Van der Merwe et al. 2016). Our study was limited to the effects of a single fire and to single observation events pre fire and post fire. The effects of different types of fires warrant further investigation, as do post-fire Karoo vegetation dynamics in the long term.

We predict that, with increases in the frequency or intensity of fires, the structure and composition of Nama-Karoo montane shrubby grasslands will undergo increases in the smaller growth forms (graminoids, other herbs, and dwarf shrubs) but decreases or potential loss of large shrubs. Such a scenario could mirror the persistent negative feedbacks, or grass-fire cycles, observed in certain other arid and semi-arid areas (D’Antonio and Vitousek 1992, Brooks et al. 2004). Consequently, we recommend that managers largely suppress the spread of man-made fires in protected areas comprising Nama-Karoo ecosystems.