Forest Tree and Fern Species As Indicators of an Unnatural Fire Event in a Southern Cape Mountain Forest

CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

Forest tree and fern species as indicators of an unnatural fire event in a southern Cape mountain forest

LH Watson* and MJ Cameron

Department of Nature Conservation, Port Elizabeth Technikon, Saasveld Campus, Private Bag X6531, George 6530, South Africa * Corresponding author, e-mail: [email protected]

Received 14 March 2001, accepted in revised form 21 January 2002

Afromontane forests in the southern Cape are typically bullata in the margin from single stemmed to multi- surrounded by fire prone fynbos. Although the impact of stemmed trees and reduced the number of Cyathea fire on fynbos has been extensively studied, little is capensis in the margin by 68%. Using these species as known about the impact of fire on adjacent forest. Fire indicators, we argue that similar fires will reduce the swept through a mountain forest in the southern Cape structure and diversity of mountain forests as a whole. in 1996. Our study indicated that this fire was one of at The 1996 fire was fuelled by abnormally high fuel loads least two fires that influenced the recent history of this associated with surrounding pine plantations and forest; the other fire was probably the great fire of 1869. mountain fynbos invaded by woody aliens. Given that We consider the latter fire typical of fires that burn large areas of mountain fynbos are invaded by woody mountain forest and suggest a return period for these aliens, that large areas of the southern Cape are fires of between a 100 and 200 years. In contrast to the afforested and that global climate change is likely to 1869 fire, evidence from our study suggests that the lead to an increase in the conditions favouring intense 1996 fire was unprecedented in the recent history of the fires, we consider effective management of forest. This fire changed a large proportion of Ocotea Afromontane mountain forests essential.

Introduction

Afromontane forests in the southern Cape are commonly fire (Watson and Cameron 2001). Inspection of the forest, surrounded by fire-prone fynbos which typically burns once however, suggested that the fire was exceptionally intense. every 15–20 years (Kruger and Bigalke 1984, Van Wilgen et Our study had the following objectives: (a) To assess the al. 1992). Although work in western Cape forests indicated intensity of the fire. This we did by studying the influence of that they are unlikely to burn under conditions favouring fire the fire on the tree Ocotea bullata1 and the tree fern Cyathea in fynbos (Van Wilgen et al. 1990), soils in southern Cape capensis. (b) To determine the interval between the 1996 fire forests contain charcoal (Scholtz 1983, Geldenhuys 1993), and the previous major fire. Podocarpus latifolius is fire-sen- indicating that they do burn. Geldenhuys (1996) studied sitive (Watson and Cameron 2001) and we use the diameter southern Cape forests at a regional scale and indicated that distributions of this species to determine the interval in contrast to plateau forests, mountain forests frequently between these fires. burn. Working in mountain forest after a burn, Watson and Cameron (2001) recorded differences in the composition, Materials and Methods diversity and mortality of tree species in the margin and core of the forest. They related these differences to fire and sug- Study area gested that the dynamics of mountain forests could not be understood without considering fire. Although fire is consid- Geldenhuys (1993) subdivided southern Cape forests into ered important in these forests, no work is available on their three major types according to species composition: wet fire regimes (see Midgley et al. 1997). mountain forests, moist plateau forests and dry valley slope In June 1996, fire burned through the mountain forest or coastal forests. PWB is a patch of wet mountain forest Paddawaterbos (PWB) and left the trees charred and with- (approximately 15ha) situated on Groenkop. Groenkop is out leaves. Three years after the fire, most trees in PWB had part of the Outeniqua mountain range and lies near George leaves and the forest seemed to have recovered from the in the southern Cape.

1 Nomenclature follows that of Arnold and De Wet (1993) 358 Watson and Cameron

PWB has sandy, loamy soils (Von Breitenbach 1968) tor of stress due to fire. The proportion of bark peel was esti- underlain by sandstone of the Table Mountain Group (Toerien mated using the eight point scale of Walker (1976). When 1979). No climatic data are available for the site, but rainfall is calculating the mean proportion of bark peel per O. bullata orographic and probably ranges between 1 000–1 200mm per site, the midpoint of each rank was used (Walker 1976). per year (Von Breitenbach 1968). Temperatures are moderate but hot ‘berg’ winds are common during the months May to Cyathea capensis August (Von Breitenbach 1968). The margin of PWB is dom- inated by the trees O. bullata and Cunonia capensis but these For C. capensis, we sampled one plot (20 x 20m) in the species are less common in the forest core (Watson and margin and one in the core, and recorded the height of each Cameron 2001). Prior to the fire, PWB was bounded on the individual (the distance between the apex of the stem and upper slopes by fynbos, invaded by Pinus pinaster and Hakea the ground) and whether the individual was dead or alive. sericea and on the lower slopes by P. pinaster plantations which had a dense mat of the fern Gleichenia polypodioides. Podocarpus latifolius Our study was carried out in March 1999 and June 2000. For P. latifolius we walked random transects in the forest Ocotea bullata margin and recorded the DBH of stems ≥5cm and whether the individual was dead or living. Bark characteristics were We selected three sites with respect to distance from the for- used to identify dead P. latifolius. est margin (margin, intermediate and core) and sampled two plots (20 x 20m) per site. In the margin and intermediate Statistical analyses sites, the species composition and the proportion of trees recorded as dead (all species) was similar (Watson and Continency tables with the Yates correction for continuity Cameron 2001). Data collected in the margin and intermedi- (Zar 1984) were used to test for difference in the frequency ate sites were therefore grouped for analysis and referred to of dead and living O. bullata and C. capensis recorded in as the margin. Defined as such, the margin extended >40m each site. Proportions of bark peel on O. bullata were arc- into the forest. Data collected in the core were referred to as sine transformed for analysis and the t test (Zar 1984) was the core. The following variables were recorded for O. bulla- used to test for difference in the mean proportion of bark ta in each plot: (a) The diameter at breast height (DBH) of peel on trees between sites. stems ≥5cm. Fire and other damage may cause O. bullata to coppice and form multi-stemmed trees (Lübbe 1990). For Results multi-stemmed trees, each stem was recorded separately but indicated as belonging to an individual tree. (b) Whether Ocotea bullata a multi-stemmed tree was multi-stemmed prior to the 1996 fire or whether this was due to the 1996 fire. In the latter In the forest core, living O. bullata on average had signifi- case, a multi-stemmed tree was defined as a tree with cop- cantly lower proportions (8.1%) of bark peel than living O. pice from the base and a primary stem that was dead (i.e. no bullata in the margin (42.7%; t = 6.249, df = 57, P<0.0001). leaves and no coppice from the aerial portion of the primary We interpret this as indicating a fire of considerably lower stem). In the margin, two trees were multi-stemmed owing to intensity in the core than in the margin. Given this difference harvesting and were excluded from the analysis. (c) in intensity, the mortality of O. bullata in each site was low Whether the tree was dead or alive. Dead trees were those and the frequency of O. bullata recorded as dead in each with dead stems and no coppice from the base. Bark char- site was similar (Table 1; χ2 = 0.413, df = 1, P>0.50). acteristics were used to identify dead O. bullata. (d) For each Therefore in terms of mortality, the fire had little influence on living O. bullata, we recorded the proportion of stem with bark O. bullata in both the core and margin of the forest. peel. Lübbe (1990) indicated that for O. bullata recovering The fire did however influence the structure of O. bullata after a fire, crown health status was significantly negatively in the margin. Of the trees in the margin, only one (5%) was correlated with bark peel and we use bark peel as an indica- multi-stemmed due to damage before the 1996 fire, com-

Table 1: The number of Ocotea bullata trees recorded in each DBH class in the core and margin of Paddawatersbos. The number of trees recorded as multi-stemmed are shown in brackets. For the category ‘living individuals’, multi-stemmed individuals are those that resprouted after the 1996 fire. n = number of plots sampled

Position DBH class (cm) Total 5–14 15–24 25–34 35–44 45–54 ≥55 Core (n = 2) Total individuals 6 (0) 5 (0) 2 (0) 3 (0) 2 (0) 1 (0) 19 (0) Living individuals 5 (0) 5 (2) 2 (0) 3 (0) 2 (0) 1 (0) 18 (2) Margin (n = 4) Total individuals 10 (0) 15 9 (1) 5 (0) 4 (0) 5 (0) 48 (1) Living individuals 5 (5) 13 (7) 9 (3) 5 (1) 4 (0) 5 (0) 41(16) South African Journal of Botany 2002, 68: 357–361 359 pared to 16 (39%) that were multi-stemmed owing to the fires. Taking 14cm as the largest of the small P. latifolius in 1996 fire (Table 1; χ2 = 11.40, df = 1, P < 0.001). In the core, the margin and using an average diameter increment of no individuals were recorded as multi-stemmed before the 0.1014cm year -1 (DBH class 10–20cm, Seydack 1995), we fire and only two after the fire (Table 1). calculate an interval of 138 years between these two fires. Although this estimate suffers from the use of data col- Cyathea capensis lected in plateau forest, in the DBH class 10–20cm, it is interesting to note that this estimate approximates the 127 In contrast to O. bullata, the mortality of C. capensis, espe- year interval between the 1996 fire and the great fire of cially in the forest margin, was greatly influenced by the fire. 1869. The latter fire occurred under hot dry conditions, was In the core and margin of the forest, the density of C. capen- driven by strong winds and was fuelled by high fuel loads sis before the fire was similar, but more C. capensis were that resulted from high rainfall and policies protecting vege- killed by the fire in the margin (68%) than in the core (8%; χ2 tation against burning (Edwards 1984). These factors pro- = 25.246, df = 1, P<0.001; Table 2). In the margin, the mor- duced high intensity fires which swept through much of the tality of C. capensis mainly occurred in the height class southern Cape (approximately 640km from Swellendam to 1–150cm (76%; Table 2). Humansdorp), including the George region and caused extensive loss of human life and property (Edwards 1984). Podocarpus latifolius We suggest that fires similar to the 1869 fire are relatively common in mountain forest, and from our study and that of In the margin, 76% of P. latifolius were recorded as dead Geldenhuys (1994), probably have return periods of (Table 3), but mortality was significantly higher among small between a 100 and 200 years. trees (5–14cm DBH; 87%) than among large trees (≥15cm DBH; 14%; χ2 = 13.56, df = 1, P<0.001; Table 3). The small Fire intensity trees also formed most (85%) of the P. latifolius recorded in the margin (Table 3). We know of no work measuring the growth rates of C. capensis. However, given the relatively smooth bell-shaped Discussion height distribution of C. capensis in the margin before the 1996 fire, the low number of living C. capensis in the height Fire frequency class 1–150cm after the fire, is anomalous. This suggests that the 1996 fire had an intensity unprecedented in the Although Geldenhuys (1996), and Watson and Cameron recent history of PWB. (2001) suggest that mountain forests frequently burn, there Using data from single stemmed O. bullata in plateau are no data on the fire regimes of these forests. Given our forests (average diameter increment for DBH classes finding that small P. latifolius trees are more susceptible to 10–60cm = 0.1145cm year -1, Seydack 1995), we estimate fire than large trees, we interpret the low number of large P. the oldest single stemmed O. bullata in the margin (DBH = latifolius trees in the margin as resulting from a major fire 54cm, Table 1) at approximately 470 years. One multi- prior to the 1996 fire, and suggest that most of the P. latifolius stemmed O. bullata therefore accumulated within the margin trees in the margin (i.e. the small trees) represent recruitment plots over the last 450 years, compared to the 16 after the after this fire. Given these assumptions, we use average 1996 fire. Although our sample size is small and yearly diam- diameter increment data for P. latifolius in plateau forests eter increments of O. bullata in plateau and mountain forest (Seydack 1995) to estimate the period between these two are unlikely to be similar, we suggest that the 1996 fire was

Table 2: The number of Cyathea capensis recorded in each height class in the core and margin of Paddawatersbos. n = number of plots sampled

Position Height class (cm) Total 1–50 51–100 101–150 151–200 201–250 >250 Core (n = 1) Total individuals 9 11 12 7 3 6 48 Living individuals 8 7 11 6 3 6 41 Margin (n = 1) Total individuals 6 18 13 7 – – 44 Living individuals 1 0 7 6 – – 14

Table 3: The number of Podocarpus latifolius recorded in each DBH class, in random transects in the margin of Paddawatersbos

Individuals DBH class (cm) Total 5–14 15–24 25–34 35–44 45–54 ≥55 Total 39321–146 Living 5 2 2 1 – 1 11 360 Watson and Cameron exceptionally intense and that a fire of similar intensity did Acknowledgements — We thank RM Cowling, CJ Geldenhuys, G not occur in PWB in the order of 400 years. von dem Bussche for comments on earlier drafts of the manuscript. The assumed intensity of the 1869 fire and the likelihood The South African Forestry Company Limited allowed us access to that it burned PWB, appears to contradict the above. Paddawatersbos. However, we relate the exceptional intensity of the 1996 fire to the abnormal fuel loads associated with the fire and to the References observation that most of this fuel was consumed by the fire. Prior to the fire, PWB was surrounded by fynbos and plan- Arnold TH, De Wet BC (1993) Plants of southern Africa: names and distribution. Memoirs of the Botanical Survey of South Africa tations of P. pinaster. The fynbos was >50 years old and 62: 71–77 invaded by dense stands of H. sericea and P. pinaster (pers. Cowling RM (ed) (1992) The Ecology of Fynbos: Nutrients, Fire and comm. G von dem Bussche; pers. comm. C Geldenhuys). Diversity. Oxford University Press, Cape Town, South Africa The P. pinaster plantations contained a dense, thick mat (up Edwards D (1984) Fire regimes in the biomes of South Africa. In: to 2m thick) of the fern G. polypodioides. Fynbos invaded by Booysen P de V, Tainton NM (eds) Ecological Effects of Fire in woody aliens may contain fuel loads three times that of fyn- South African Ecosystems. Ecological Studies 48, Springer- bos, while pine plantations may support fuel loads 3–10 Verlag, Berlin, Germany times that of fynbos (Van Wilgen and Richardson 1985, Geldenhuys CJ (1993) Composition and Dynamics of Plant Versfeld and Van Wilgen 1986). Dense, thick expanses of G. Communities in the Southern Cape Forests. FOR-DEA 612, polypodioides are uncommon under the natural fynbos fire Division of Forest Science and Technology, CSIR, Pretoria Geldenhuys CJ (1994) Bergwind fires and the location pattern of regime. This abnormal fuel load was ignited by an acciden- forest patches in the southern Cape landscape, South Africa. tal fire and the fire driven by a strong, hot berg wind. The fire Journal of Biogeography 21: 49–62 produced a clean burn in the alien infested fynbos, con- Geldenhuys CJ (1996) Forest management systems to sustain sumed the crowns of most P. pinaster trees in the plantations resource use and biodiversity: examples from the southern Cape, and reduced the thick layers of G. polypodioides to ash. South Africa. In: Van der Maesen LJG, Van der Burgt XM, Van Medenbach de Rooy JM (eds) The Biodiversity of African Plants. Management considerations Kluwer Academic Publishers, Netherlands Kruger FJ, Bigalke RC (1984) Fire in fynbos. In: Booysen P de V, The impact of fire on fynbos, invaded or uninvaded by Tainton NM (eds) Ecological Effects of Fire in South African woody aliens, has been extensively studied (Cowling 1992, Ecosystems. Ecological Studies 48, Springer-Verlag, Berlin, Germany Van Wilgen et al. 1992). In contrast, little is known about the Lübbe WA (1990) Recovery of Ocotea bullata (Lauraceae) after fire impact of fire on Afromontane forests. Our study indicates damage. South African Forestry Journal 152: 1–6 that at least two fires influenced the recent history of PWB; Midgley JJ, Cowling RM, Seydack AHW, Van Wyk GF (1997) the 1869 fire and the 1996 fire. We suggest that fires similar Forest. In: Cowling RM, Richardson DM, Pierce SM (eds) to the 1869 fire are relatively common in mountain forest. In Vegetation of Southern Africa. Cambridge University Press, this type of fire; fire-sensitive species in the forest margin are Cambridge, UK killed but there is little effect on fire-resistant species, and Richardson DM, Cowling RM (1992) Why is mountain fynbos inva- the latter thus dominate the margin; fire has little influence sible and which species invade? In: Van Wilgen BW, Richardson on species in the forest core and the core thus has greater DM, Kruger FJ, Van Hensbergen HJ (eds) Fire in South African species diversity than the margin. The study of Watson and Mountain Fynbos: Ecosystem, Community and Species Response at Swartboskloof. Springer-Verlag, Berlin, Germany Cameron (2001) and inspection of mountain forest influ- Seydack AHW (1995) An unconventional approach to timber yield enced by fire (M Cameron pers. obs.) supports this argu- regulation for multi-aged, multispecies forests. I. Fundamental ment. considerations. Forest Ecology and Management 77: 139–153 In contrast to the fires described above, we consider the Scholtz A (1983) Houtskool weerspieël die geskiedenis van fire of 1996 to be unprecedented in the recent history of inheemse woude. Bosbounuus 3: 18–19 PWB. This fire changed the structure of O. bullata in the mar- Toerien DK (1979) The Geology of the Oudtshoorn Area. gin and reduced the number of C. capensis in the margin by Department of Mines: Geological Survey. Government Printer, 68%. Using these species as indicators, we suggest that Pretoria, South Africa under similar fire conditions described here, fire will not only Van Wilgen BW, Bond WJ, Richardson DM (1992) Ecosystem man- reduce the structure and diversity of the forest margin, but agement. In: Cowling RM (ed) The Ecology of Fynbos: Nutrients, Fire and Diversity. Oxford University Press, Cape Town, South will erode the forest core and ultimately reduce the structure Africa and diversity of mountain forests as a whole. Given that large Van Wilgen BW, Higgins KB, Bellstedt DU (1990) The role of vege- areas of mountain fynbos are invaded by woody aliens tation structure and fuel chemistry in excluding fire from forest (Richardson and Cowling 1992), that large areas of southern patches in the fire-prone fynbos shrublands of South Africa. and western Cape have been afforested and that global cli- Journal of Ecology 78: 210–222 mate change is likely to lead to an increase in the conditions Van Wilgen BW, Richardson DC (1985) The effects of alien shrub favouring intense fires, we consider effective management of invasions on vegetation structure and fire behaviour in South mountain forests essential. We consider the following priori- African fynbos shrublands: a simulation study. Journal of ties: control of woody aliens in mountain fynbos; burning of Applied Ecology 22: 955–966 mountain fynbos at appropriate frequencies; and the mainte- Van Wilgen BW, Van Hensbergen HJ (1992) Fuel properties of vegetation in Swartboskloof. In: Van Wilgen BW, Richardson DM, nance of adequate ecotones around mountain forests. South African Journal of Botany 2002, 68: 357–361 361

Kruger FJ, Van Hensbergen HJ (eds) Fire in South African Walker BH (1976) An approach to the monitoring of changes in the Mountain Fynbos: Ecosystem, Community and Species composition and utilization of woodland and savanna vegetation. Response at Swartboskloof. Springer-Verlag, Berlin, Germany South African Journal of Wildlife Research 6: 1–32 Versfeld DB, Van Wilgen BW (1986) Impacts of woody aliens on Watson LH, Cameron MJ (2001) The influence of fire on a southern ecosystem properties. In: MacDonald IAW, Kruger FJ, Ferrar AA Cape mountain forest. South African Forestry Journal 191: (eds) The Ecology and Management of Biological Invasions in 39–42 Southern Africa. Oxford University Press, Cape Town, South Africa Zar JH (1984) Biostatistical Analysis. Prentice-Hall, New Jersey, Von Breitenbach F (1968) Groenkop Indigenous Forest USA Management Plan (1967/68–76/77). Department of Forestry, Pretoria, South Africa

Edited by RM Cowling