Juvenile periods of proteoids ranged from 4 to 9 yr in our study, which are comparable to those at Outeniqua Nature Reserve, immediately west of the study area, those reported for the western part of the CFK (Table 1; CapeNature unpublished data; Kruger & Bigalke 1984; Le Maitre 1992), and those of proteoids in SE and SW Australia (Cowling et al. 1987; Richardson et al. 1990; Lamont et al. 1991b; Enright et al. 1996, 1998; Gill & McCarthy 1998; Bell 2001; Bradstock & Kenny 2003). However, there is usually a lag between the age at which the first plants flower and that at which the majority flower (Kruger & Bigalke 1984). Early inflorescences furthermore do not necessarily produce fertile cones (Enright et al. 1996). In western Australian scrub heath, a detailed study of the canopy seed bank dynamics of Banksia hookeriana showed that even though the species started flowering at 3–4 yr of age (similar to our findings), FRIs optimizing the likelihood of successful recruitment were estimated at 15–18 yr (Enright et al. 1996).
In North American temperate serotinous pine forests, under regimes of frequent disturbance, maturation ages optimizing plant fitness were estimated to be 0.4 times the average disturbance interval (Clarke 1991). Median FRIs recorded in the study area since 1980 varied considerably (8.3–26.2 yr) depending on the method of estimation (T. Kraaij, unpublished data), and predict optimal juvenile periods of 3.3–10.5 yr, which are in line with those we observed.
There was large variability in the degree of flowering of populations at given plant ages among sites and within species. In B. hookeriana (in western Australia), within and between year variability in cone production increased with plant age (Enright et al. 1996). In our study, the number of flowerheads produced per flowering plant did not vary widely within or among years in Protea species, although it was more variable in the single year that L. eucalyptifolium flowering was recorded. Results from the two sites where we undertook recurrent surveys suggest that flowering status is more closely related to plant height than to plant age (cf. Le Maitre & Midgley 1991). The substantial difference between the two sites in the plant age to maturation is unlikely to be due to species differences, given the lack of a species effect observed at the moist site as well as in our one-off surveys of flowering status. Site differences thus appear to play a key role in plant growth and maturation rates. The most noticeable abiotic difference between the two sites was aspect, with plants on the relatively dry, western slope being slower to grow and mature than those on the relatively moist, southern slope. Evidence for the effects of aspect (dry, northern or western slopes vs moist, southern or eastern slopes) on flowering status as measured in one-off surveys was, however, ambiguous.
Recruitment was always above replacement levels following fires that burned in vegetation of ≥7 yr post-fire age. The lack of a relationship between recruitment success and pre-fire vegetation age suggests that once a critical post-fire age (and by implication, seed bank size) is attained, factors other than seed abundance affect recruitment success. These factors may include season of fire, slope and aspect in relation to moisture regimes, pre-fire parent density and interspecific competition (Bond et al. 1984, 1995; Le Maitre 1988a,b; Mustart & Cowling 1993; Laurie & Cowling 1994; Heelemann et al. 2008, 2010).
There was considerable variation in recruitment success for any given FRI, species or site (cf. Bond et al. 1984, 1995; Midgley 1989; Laurie & Cowling 1994). Average to very good (including the highest of all records) recruitment occurred at the site with the longest FRI (38 yr). This finding is at variance with that of Bond (1980) in the Swartberg Mountains (ca. 100 km inland from our study area), where senescence (40–45 yr of age) negatively affected post-fire recruitment of fynbos. We concur with van Wilgen et al. (2011) that the occurrence of very old vegetation is not a key concern in the ecological management of fire in montane proteoid fynbos, both because it is very limited in extent (Bond 1980; T. Kraaij, unpublished data), and because recruitment does not appear to be negatively affected by relatively long inter-fire periods.
Minimum fire return interval
Our results on post-fire recruitment success of proteoids suggest that FRIs of <6 yr would result in the local extirpation of this guild from eastern coastal fynbos, whereas recruitment above replacement levels consistently should occur after fires at ≥7-yr intervals. Application of Kruger & Lamb's (1978) rule of thumb (that 50% of the individuals in a population should have flowered for at least three seasons) to observed proteoid flowering status (50% of plants flowered once by 5–6 yr of age; Fig. 2a), and assuming that plants flower every year after first flowering and that seeds take 7 mo to ripen (van Staden 1978), implies a minimum FRI of 9 yr for Tsitsikamma proteoid fynbos. We do not have data on the flowering status of populations at a post-fire age of 9 yr in order to support this estimate, but most, although not all, species surveyed had reached the required level of flowering at 11 yr post-fire (Fig. 2c; Table 1). Using twice the primary juvenile period (in our case, twice >4 yr; Figs 2a and 3) as a guide for minimum FRIs (Bell 2001), suggests a lower threshold in excess of 8 yr. Substantial variation, both in flowering status and post-fire recruitment, as well as disparity among estimates of minimum FRI based on these measures, emphasize the need to empirically verify rules of thumb that are currently used as fire management guidelines. Verification may be done by relating pre-fire flowering status of proteoid populations to their post-fire recruitment response at corresponding sites (as done for the site that burned at 5 yr of age).
All sites from which we collected data to estimate juvenile periods were located in the Tsitsikamma region. Juvenile periods (and by implication, minimum FRIs) are likely to be considerably longer in dry habitats where plant growth rates are lower (Le Maitre & Midgley 1992). This is substantiated by the difference observed in growth rates and flowering status between the sites on a relatively dry western and the moist southern slope, respectively. At a regional scale, rainfall is lower in the Outeniqua than in the Tsitsikamma region, and at Outeniqua Nature Reserve (immediately west of the study area) the time needed for 50% of individuals to flower once and three times was longer than that in the Tsitsikamma region (Table 1). Median FRIs, recorded since 1980, have accordingly been shorter in the Tsitsikamma than in the Outeniqua region (T. Kraaij, unpublished data). Overall, maturation rates of proteoids are comparable between the eastern and western CFK (Table 1), suggesting that variation is related to local moisture regimes (Kruger & Bigalke 1984) rather than a general east–west gradient within the CFK. Seydack et al. (2007) accordingly found an inverse relationship between FRI in fynbos shrublands and rainfall (related to plant productivity) along an altitudinal gradient in the Swartberg Mountains.
Range-restricted species or habitat specialists may have very specific FRI requirements (Kruger & Bigalke 1984). For instance, P. grandiceps (Near Threatened; Raimondo et al. 2009), a slow-growing, high-altitude species from the study area and elsewhere in the CFK, is slow to mature (>10 yr; J.H.J. Vlok, pers. comm., 2012, local botanist) and favours rocky outcrops and steep slopes where it is relatively safe from frequent fires (Rebelo 2001). Some subpopulations of this species have been exterminated as a result of too frequent fires (J.H.J. Vlok, pers. comm., 2012, local botanist). A survey done on the Kammanassie Mountains (30-km inland of our Outeniqua region) showed that only ten out of 200 plants flowered for the first time at 9 yr post-fire (CapeNature unpublished data).
Bradstock & Kenny (2003) discuss the limitations of using juvenile periods and life spans of the most sensitive plant guilds to inform boundaries for FRIs, and inter alia state that consistent favouring of one guild may in the long term lead to a loss of biodiversity. However, retention (and even dominance) of proteoids was shown to be key to the conservation of diversity in fynbos overall (Vlok & Yeaton 1999). Experience furthermore suggests that rigid control over fire regimes is largely unattainable (Keeley et al. 1999; Moritz 2003; van Wilgen et al. 2010), with wildfires providing sufficient variation to preclude consistent favouring of a particular plant guild.
In the same way that plant characteristics can be used to establish thresholds for FRIs, the post-fire development of habitat and fuel attributes (Haslem et al. 2011), and the life cycles and behaviour of selected animal species, may be used (Gill & McCarthy 1998). In the mediterranean shrublands of SW Australia, population modelling based on the demography and behaviour of rare, poorly-dispersing, ground-dwelling bird species suggested optimum FRIs in excess of plant maturation rates (Brooker & Brooker 1994; Gill & McCarthy 1998). South African fynbos does not have many endemic bird species, but some may be adversely affected by short FRIs. Both the Cape Sugarbird (Promerops cafer) and Protea Seedeater (Crithagra leucopterus) require mature proteoid fynbos for feeding and breeding, and it may take ≥10 yr post-fire for fynbos to attract breeding birds (Milewski 1978; Martin & Mortimer 1991). The existence of adequate areas of mature fynbos in the landscape is thus a requisite for these birds' persistence. While the focus of this paper has been on establishing minimum FRIs based on plant attributes, it would also be important to ensure that at least a proportion of the vegetation is in the age classes of 10–20 yr to conserve these faunal elements. This implies that the minimum age for burning would have to be >10 yr for a proportion of the area.
The suggested minimum FRI of 9 yr in relatively moist and productive Tsitsikamma fynbos is not intended to prescribe rigid management of fire according to a fixed burning rotation. Neither does it negate the need to consider site- or species-specific requirements. Instead, it provides a lower threshold for a range of acceptable FRIs beyond which a significant decline of species populations is predicted (Bradstock & Kenny 2003; van Wilgen et al. 2011). Where two or more species of slow-maturing reseeders co-exist, the FRI for maximizing survival or abundance may be different (Gill & McCarthy 1998). Variation in fire regimes is therefore necessary to maintain plant diversity in the landscape (Cowling 1987; Gill & McCarthy 1998; Thuiller et al. 2007). In light of climate change and the associated increases in fire frequency that have been recorded locally (Forsyth & Wilgen 2008; Wilson et al. 2010; T. Kraaij, unpublished data) and in temperate shrublands globally (Piñol et al. 1998; Williams et al. 2001; Keeley & Zedler 2009), managers attempting to maintain fire regimes within ecological thresholds should follow a precautionary approach, particularly at the lower end of the FRI range. In addition to allowing for variation in the fire regime, they should aim for mean FRIs towards the middle of the ecologically acceptable range, rather than at the lower end, as the increasing occurrence of unplanned fires is likely to reduce mean FRIs overall.
Both frequent and low-intensity fires in fynbos and other temperate shrublands favour resprouters over slow-maturing, serotinous or myrmechocorous reseeders, leading to a loss in diversity overall (Haidinger & Keeley 1993; Vlok & Yeaton 1999, 2000). In large parts of the GRCMs, graminoid sprouters dominate, while proteoids are sparse or absent (Heelemann et al. 2010), notably in areas near plantations of alien pine trees (T. Kraaij, pers. obs.). This likely resulted from frequent and low-intensity burning in the past, aimed at protecting timber plantations from fire (Kraaij et al. 2011). Facilitation of FRIs that would ensure the long-term persistence of proteoids in the GRNP is therefore a priority for fynbos conservation management, particularly where this guild has persisted in the landscape. Short interval fires, which alter vegetation structure (from woody to herbaceous; Kruger 1984; Lloret et al. 2003) and thus fuel dynamics, may set up negative feedback loops whereby short FRIs persist in the landscape (Haidinger & Keeley 1993; Milton 2004).
Fynbos vegetation in the GRCMs is currently severely threatened by widespread invasion by pine trees (Pinus pinaster and P. radiata) grown in plantations that are scattered throughout the landscape (Cowling et al. 2009; Kraaij et al. 2011). Like the proteoids, pine trees are serotinous and fire-adapted, and repeated fires drive their rapid spread and proliferation (Richardson 1998). Prolonging the FRI (i.e. reducing the fire frequency) would thus generally be in the interest of invasive plant control by curbing the rate of spread of pines. On the other hand, in areas where the proteoids have already been lost due to past management practices, application of a single FRI that is shorter than the juvenile period of these pines (ca. 5–6 yr; Richardson et al. 1990) may in some instances provide an inexpensive means of substantially reducing dense infestations of young pine recruits.