Seasonal growth of the overstorey and understorey in mediterranean-type shrublands and heathlands in the south eastern Cape, South Africa

The phenological model of Specht et a/. (1983) was tested by determining the seasonality of shoot growth and leaf loss of the overstorey and understorey components of vegetation on nutrientpoor, base-rich and calcium-rich soil suites in the mediterranean climate region of the south eastern Cape. Results were largely consistent with the predictions of the model although certain inconsistencies were noted. The model is discussed critically in terms of its assumptions and predictions.


Introduction
The current interest in the structure and functioning of mediterranean-type vegetation has led to the assertion that the global physiognomic similarity of these formations makes it probable that they have similar trends of phenophase sequences (Kummerow 1983). There is some evidence from northern hemisphere mediterranean vegetation to support this assertion (Kummerow 1983).
In mediterranean-type ecosystems a distinction is made between shrubtands on relatively fertile soils (Specht 1979;Di Castri 1980;) and heathtands on infertile soils (Specht 1979). In the latter it is postulated that low levels of soil nutrients are of overriding importance in determining plant structure (Specht 1979;Cowling & Campbell 1980), leading to a global heathland concept associated with a particular soil suite rather than any single climatic regime. Whereas most species of mediterranean shrub lands show growth flushes in spring (Kummerow 1983), dominant heathland shrubs in mediterranean climate regions of Australia and South Africa grow mainly in summer (Kruger 1981;Specht et at. 1981).
Recently Specht (Specht et at. 1981;) developed a model to explain and predict the growth of overstorey and understorey species in mediterranean-type vegetation on base-rich, nutrient-poor and calcium-rich soil suites in southern Australia and South Africa. This model predicts that the overstorey and understorey species of shrublands on base-rich soils would show spring (Sep.-Nov.) and/or autumn (Mar. -May) growth while leaf fall would be delayed until the dry season of the year (in late spring or summer, Dec. -Feb.). Similar phenophases are predicted for understorey heathland species on nutrient-poor soils; overstorey species would grow in late spring or summer with leaf fall occurring at the same time.  further predict that because of nutrient stresses induced by the high pH of calcium-rich soils, communities on this soil suite would show a similar seasonality to heathlands.  argue that the timing of shoot growth and leaf fall on the different soil suites confers ecophysiological advantages. On base-rich soils, where nutrients are assumed to be non-limiting, shoot growth precedes leaf fall and the additional new foliage would increase photosynthetic production during the favourable spring and autumn periods. In heathlands on nutrient-poor soils 'shoot growth in late spring and summer follows the release of nutrients from decomposing litter during spring' . In this way nutrient cycling is maximized both externally via decomposing leaf litter and internally from abscising leaves. Furthermore the absence of winter growth in both heathlands and shrublands means that soil moisture is conserved during the wet period to be used later in the 'more favourable warmer months' .
In this paper we test the phenological predictions of Specht's model ) by measuring and observing shoot growth and leaf loss periodicity in overstorey and understorey species over a period of 14-16 months in communities on nutrient-poor, base-rich and calcium-rich soil suites in the mediterranean-type climate region of the south eastern Cape. We did not measure changes in soil moisture and aspects of nutrient cycling which would provide a true test of Specht's model and explain the 'ecophysiological advantages' of the shoot growth and leaf loss periodicities predicted by the model. A detailed account of the phenology of the communities together with information on the study sites and methods is presented in Pierce & Cowling (1984).

Study area
The study area is located in the Humansdorp region of the south eastern Cape on a level coastal plain which cuts across two geological formations: sandstone of the Table Mountain Group (TMG) and shales of the Bokkeveld Group. Along the coast there are deposits of recent calcareous sands. It was therefore possible to study eight communities on different substrates or soil suites all of which experienced similar mesoclimatic conditions. Details on the floristics, structure, soils and land-use of the communities and an account of the climate of the region are given in Pierce & Cowling (1984). Table 1 shows the correspondence of the communities we studied with  soil suites.  use the terms highly leached and moderately leached instead of nutrient-poor and base-rich, respectively. We prefer Specht's (1979) earlier terms (latter above, Table 1) since in South Africa, many heathland (fynbos) soils which are infertile are not leached (Bond 1981;Carnpbell1983;Cowling 1983) and do not have an eluvial horizon which characterizes the zone of maximum leaching (Brady 1974). In a recent symposium (Day 1983), nutrient-poor soils were defined as having a pH of less than 6,0, a total nitrogen of less than 0,120Jo and a sum of exchangeable cations (S-value) of less than 7 me/100 g soil (see also . S.-Afr. Tydskr. Plantk., 1984, 3(1) Calcium-rich soils have a high pH ( > 7) and large amounts of exchangeable calcium. According to , in the South African mediterranean-type region nutrient-poor soils support fynbos (a heathland) while base-rich soils support shrublands, including '0/ea-Sideroxy/on open scrub' and disclimax 'renosterveld'. An important feature of the shrublands is the presence of a 'savanna understorey' of seasonal grasses and forbs, in contrast to the evergreen sclerophyllous understory of heathlands. Calcium-rich soils support 'a stunted overstorey of evergreen sclerophyllous trees and shrubs over a ground stratum showing a gradation from seasonal grasses and herbs to evergreen hemicryptophytes' .

Soil suites and shrubland types
In our study area heathlands on nutrient-poor soils included communities 1 and 2 in Table 1. They differ from Specht's (1979) heathland concept in that the understorey includes many seasonal grasses. In the mediterranean-type region of South Africa the most widespread shrubland on base-rich soils is renosterveld (Boucher & Moll1980) which has an understorey of seasonal grasses and forbs and an overstorey of small leaved sclerophyllous and semi-succulent shrubs.  regard the '0/ea-Sideroxylon' open scrub (Kaffrarian thicket in Table 1) as the true South African mediterranean-type shrubland analogous to the maquis, mattoral, chaparral and mallee of other mediterranean-type regions (Di Castri 1980). Cowling (1983) has argued that in the Cape region thicket is restricted to special edaphic sites and was certainly never widespread in the past. We sampled renosterveld (communities 3 and 4) and thicket (community 5) on base-rich soils. On calcium-rich soils we sampled a heathland (dune fynbos), a grassy heathland (dune grassland) and a shrubland (dune thicket) (communities 6-8 in Table 1 ).
Both Cowling (1983) and Campbell (1983) have criticized Specht's shrubland concepts as applied to the South African mediterranean-type region. Many non-heathland types (e.g. renosterveld) occur on soils which are nutrient-poor. Moreover shrublands on true base-rich soils are not 'open scrubs with a savanna understorey' (Specht & Molll983) but rather closed shrublands with a sparse understorey of shade tolerant herbs.

Methods
Details on the sampling methods and the list of species are given in Pierce & Cowling (1984) where 173 species were studied. However, the occurrence of the same in more than one community meant that 185 species in all were observed. We monitored active growth of species in the eight communities (Table 1) and expressed the results in bar graphs depicting the percentage of total species of each stratum actively growing in each month for 12, and for some communities 16 months from January 1981 to March 1982. Leaf fall was measured for selected species (Pierce & Cowling 1984) and qualitative observations were made for the rest.

Results
The results showing percentage species in the overstorey and in the understorey actively growing each month for nutrientpoor, base-rich and calcium-rich soil suites, are shown in Figures 1, 2 & 3 respectively.
On nutrient-poor soils, most species of the restioid grassland (understorey only) grew in early spring with a lesser peak in autumn (Figure 1a). Leaf senescence of geophytes occurred in late spring after or immediately prior to flowering. Leaves of grasses senesced in late spring (Oct/Nov) but remain attached to the parent shoot. Growth periodicity was consistent with Specht's predictions but patterns of leaf loss appeared to be anomalous. Grassy fynbos had a distinctive summer growth peak (Figure 1 b & c). In general litter fall was not entirely synchronous with shoot growth. Leaf loss in Leucadendron salignum was variable although there was evidence of a January peak, while Leucospermum cuneiforme shed most leaves in autumn (Pierce & Cowling 1984). This latter pattern could be consistent with Specht's wide prediction of 'late spring (or) . .. increase markedly towards the end of the dry period' for leaf loss of the overstorey on nutrient-poor soils (Specht eta/. 1983).
Both ericoid overstorey species showed non-seasonal leaf loss (Pierce & Cowling 1984) although Erica pectinifolia did show a slight mid-to late summer as well as autumn peak. The overstorey shoot growth and, to a lesser extent, leaf loss conformed with Specht's predictions. However, the grassy fynbos understorey showed maximum growth in summer which is at variance with the model. Two understorey hemicryptophytes (Restio triticeus, Tetraria involucrata) showed spring and/ or autumn peaks but the C 4 grass, Diheteropogon filifolius, grew in summer (Pierce & Cowling 1984). All chaemaephytes (e.g. He/ichrysum spp., Selago g/omerata) grew in summer. There were no geophytes present in the grassy fynbos site. Specht's model predicts spring or autumn growth and postgrowth leaf loss for overstorey species on base-rich soils. The Kromme River thicket did show maximum shoot growth in spring ( Figure 2a) and maximum leaf loss in midsummer (Pierce & Cowling 1984). Most shale grassland species (understorey only) grew in late winter/early spring as well as late summer/autumn (Figure 2b). Specht's predictions for growth periodicity of this stratum on base-rich soils are vague and therefore difficult to test. Specht et a/. (I 983) do suggest that on base-rich soils seasonal grasses and herbs complete their annual cycle by the end of spring; our data for the shale Figure 2 Shoot growth of understorey and overstorey species in base-rich soils. n = number of species observed; .6 indicates start of study of one community. 20 grassland shows a marked drop in growth in late spring/ early summer (Figure 2b). Late summer growth could be the result of the anomalously high January and February rainfall during the sampling period (Pierce & Cowling 1984). The renosterveld overstorey showed clear spring and late summer/ early autumn growth peaks (Figure 2c) . Again, heavy late summer rainfall may explain the early onset of autumn growth. The renosterveld understorey showed late winter I early spring growth (Figure 2c). Overall shoot growth and leaf loss periodicity on base-rich soils was consistent with Specht's predictions.
On the calcium-rich soils, dune grassland (understorey only) (Figure 3a) and the dune fynbos understorey (Figure 3d) showed spring growth peaks. Specht's predictions for this stratum on calcium-rich soils are vague although it is suggested that phenophases should be similar to those on nutrient-poor soils (spring and autumn peaks) (Specht eta/. 1983). Our data are consistent with this prediction. Overstorey growth and leaf loss periodicity on calcium-rich soils should be similar to the phenophases on nutrient-poor soils (Specht eta/. 1983). Most dune fynbos overstorey species, however, grow in late spring and late summer/autumn (Figure 3c). Moreover maximum leaf loss in the two dominant overstorey species, Passerina vulgaris and Agathosma apicu/ata, does not coincide with their growth peaks (Pierce & Cowling 1984). Most dune thicket overstorey species grew in spring and late summer/ autumn ( Figure 3b) and showed maximum leaf fall in midsummer (Pierce & Cowling 1984). The results are not consistent with Specht's predictions. Both thicket communities showed similar phenophases and had similar compositions of subtropical trees and shrubs.

Discussion
We found evidence which both confutes and supports Specht's predictions for shoot growth and leaf loss phenophases in vegetation on different soil suites in a South African mediterranean- S. -Afr. Tydskr. Plantk. , 1984, 3(1) type region. Much of the confusion in testing Specht's model results from inexplicit assumptions and vague predictions. There are problems with the overstorey/understorey characterization. No explicit definition of the strata in terms of readily observable growth form or life form attributes is given. In heathlands, nanophanerophytes could either be overstorey or understorey elements depending on the stratification of the community. The assertion that understorey species are shortlived and overstorey species long-lived  certainly does not apply to fynbos where understorey Restionaceae may persist long after seed regenerating overstorey Proteaceae have senesced and died . (cf van Wilgen 1982).  give Euclea, Rhus, Olea, Pterocelastrus and Sideroxylon , all subtropical genera with few or no species endemic to the South African mediterranean-type region, as typical of the overstorey component on base-rich soils. No mention is made of the renosterveld overstorey comprising small leaved endemic shrub genera (Elytropappus, Relhania, Aspalathus, Euryops). Certainly renosterveld is the most widespread shrubland on base-rich soils and although some renosterveld has been derived in historical times from a grassland (Cowling 1983) it cannot be ignored in the guise of a 'disclimax' community cf. Cowling 1983).
Overstorey shoot growth on base-rich soils 'normally occurs in spring, sometimes in autumn, while leaf fall follows shoot growth as soil drought increases in late spring and summer' (Specht eta/. 1983). If shoot growth occurs in autumn, when will leaf fall occur? The base-rich understorey is predicted to grow any time between autumn and spring. This component is assumed to be entirely 'seasonal' and to die back during the unfavourable midsummer season. Not all of our understorey species showed this seasonality. Predictions for vegetation on calcium-rich soils are especially vague and we doubt whether high pH limits nutrient uptake in organically enriched dune thicket soils.
Our findings may be explained by temperature and soil moisture, the factors suggested by Kummerow (1983) as determinants of phenophases. Shoot growth and flowering are temperature dependent while root growth and litter fall are soil moisture dependent (Kummerow 1983). He attributes the time difference between spring growers and late growing species to the availability of soil moisture to the former shallow rooted species, while the latter are deep rooted and able to tap deeper underground water sources. However, Kurnmerow (1983) considers soil moisture thresholds and the accumulation of warming hours of prime importance but feels his own model is not yet satisfactory.
The model of  assumes that ecophysiological factors (e.g. water and nutrient availability) are of overriding importance in determining phenophases. Although the seasonal regime of the habitat will impose major constraints on plant growth, biological constraints are also important. For example, in many species of Restionaceae shoot growth must precede flowering since inflorescences are terminal. Flowering time, in turn may be affected by interrelations with disperser seasonality (cf. Bond & Slingsby 1983;Thompson 1981).
Biogeographical and historical factors can impose further constraints on the model. The South African mediterraneantype region is not isolated from the subtropical summer rainfall region but grades into it along the south and south east Cape coastal forelands. The area is thus exposed to migration of a subtropical flora. Many of the subtropical species are C 4 grasses which have high temperature optima for growth (e.g. S. Afr. J . Bot., 1984, 3(1) Diheteropogon filifolius in the grassy fynbos understorey).
Referring to our data, it is not always possible to give ad hoc explanations for the deviations from Specht's predictions. Specht eta/. (1983) explain their phenophase predictions in terms of seasonality in the availability to plants of water and nutrients. Implicit in the model is the assumption that nutrient uptake occurs concurrently with shoot growth. However, shoot and leaf growth may be the result of relocation of stored metabolites (Groves 1965). We lack these data and therefore could not provide a real test of the model. We question whether the implied seasonality of nutrient availability can be tested, given the difficulties of measuring plant-available soil nutrients (W. Stock, pers. comm.).
Specht is to be commended for developing so bold a phenological model. Our data were consistent with many of his predictions. However, on the basis of the inconsistencies, the model should be re-evaluated. Perhaps the notion of polarizing mediterranean-type shrublands and heathlands into overstorey and understorey components should be questioned. Is it not more realistic to expect species to show periodic behaviour as a continuum between the rhythms predicted for overstorey and understorey strata?