Author for correspondence: Charles Musil Tel: +27 021799 8710 Fax: +27 021797 6903 Email: firstname.lastname@example.org
• Here we examine the response of succulents in a global biodiversity hot spot to experimental warming consistent with a future African climate scenario.
• Passive daytime warming (averaging 5.5°C above ambient) of the natural vegetation was achieved with 18 transparent hexagonal open-top chamber arrays randomized in three different quartz-field communities.
• After 4-months summer treatment, the specialized-dwarf and shrubby succulents displayed between 2.1 and 4.9 times greater plant and canopy mortalities in the open-top chambers than in the control plots. Those surviving in cooler ventilated areas and shaded refuges in the chambers had lower starch concentrations and water contents; the shrubby succulents also exhibited diminished chlorophyll concentrations.
• It is concluded that current thermal regimes are likely to be closely proximate to tolerable extremes for many endemic succulents in the region, and that anthropogenic warming could significantly exceed their thermal thresholds. Further investigation is required to elucidate the importance of associated moisture deficits in these warming experiments, a potential consequence of supplementary (fog and dew) precipitation interception by open-top chambers and higher evaporation therein, on plant mortalities.
Changes to the global climate system are anticipated from the accumulation of CO2 and other greenhouse gases in the atmosphere since preindustrial times. This accumulation has had a discernible influence on global temperature and is predicted to cause further warming in this century (IPCC, 2001). Direct and dramatic ecological responses to global warming are expected (Peters & Lovejoy, 1992; Thomas et al., 2004), as are feedback effects whereby ecological responses generate additional climatic impacts by modifying transfer rates of energy, water, and trace greenhouse gases at the planetary surface (Rosenberg et al., 1983; OIES, 1992). These prospects are supported by long-term monitoring studies, which indicate that recent climatic and atmospheric trends are inconsistent with past climatic variation and are already affecting the phenology, physiology and distribution of plant species (Hughes, 2000; Parmesan & Yohe, 2003). Concerns have been amplified because rates of vegetation change are expected to occur much faster than past successional processes and species dispersal rates (Pastor & Post, 1988; Overpeck et al., 1991).
Various climate futures focusing on regional mean temperature and rainfall changes in different seasons have been predicted for the African continent (Hulme et al., 2001). They draw upon different draft emission scenarios prepared for the Intergovernmental Panel on Climate Change (IPCC, 2001) and have been incorporated into bioclimatic models to predict ecosystem responses to climate change (Rutherford et al., 1999). The application of these models in vulnerability and adaptation assessments in a South African Country Study on Climate Change predicted future warming and aridity trends sufficient to cause large reductions in species richness in Mediterranean-climate Fynbos and Succulent Karoo biomes (Midgley et al., 2002, 2003). These biodiversity changes accompanied by the displacement of the Succulent Karoo biome southward along the west-coast and interior coastal plain (Hannah et al., 2002), a feature of glacial-interglacial climatic oscillations of the Pleistocene (Midgley et al., 2001). The predictions concur with the purported particular sensitivity of Mediterranean-climate ecosystems globally to changes in biodiversity induced by the five major drivers of biodiversity at the global scale, with climate change rated only second to land use as the driver with the largest effect on biodiversity when all ecosystems are averaged (Sala et al., 2000). They are also supported by local observations, which indicate that the population demographies of some large succulents, such as Aloe dichotoma, have already begun responding to anthropogenic induced climate change in a manner projected by bioclimatic models (Foden, 2002). Such changes are of concern for biodiversity conservation in the region, since both the Fynbos and Succulent Karoo biomes are characterized by exceptionally high species richness and endemism (Cowling et al., 1989, 1998; Hilton-Taylor, 1996), the Succulent Karoo biome especially possessing the highest species diversity recorded for an arid vegetation type worldwide (Hilton-Taylor, 1996) and listed among 25 global biodiversity hot spots (Myers et al., 2000). Despite these features, these biomes are not yet represented in a network of 32 ecosystem warming research sites currently representing Forest, Grassland, high and low latitude/altitude Tundra biomes (Rustad et al., 2001).
The ability of bioclimatic models to elucidate biodiversity responses to climate change has been questioned (Woodward & Beerling, 1997), and is certainly limited by a paucity of empirical information from field and laboratory trials. Experimental tests are urgently required to validate and refine bioclimatic model extrapolations, since this remains one of the few methods able to generate predictions of climate impacts on large numbers of individual species (Thomas et al., 2004). In this paper, we report on initial responses of some endemic succulent species in a southern African biodiversity hotspot to experimental warming approximating a future climate scenario (Hulme et al., 2001). For this and logistic and engineering reasons, we focused on specialized dwarf and shrubby leaf succulents occurring on sandy-loam substrates covered by quartz gravel in the Knersvlakte centre of endemism in the Succulent Karoo biome, an area of great ecological significance and conservation value (Schmiedel & Jürgens, 1999; Schmiedel, 2001, 2002).
Materials and Methods
Study sites, treatments and species
Two study sites, roughly 10 km apart, were located on the farms Quagga Kop and Luiper Kop, situated in the Knersvlakte (elevation ± 160 m) north of the town of Vanrhynsdorp (31°36′ S, 18°44′ E), South Africa. Daytime passive heating of the natural quartz field vegetation at each site was achieved with hexagonal open-top chambers constructed of clear (transmission down to 280 nm) acrylic. The open-top chambers were 46 cm high with distances between parallel sides of 120 cm at their bases and 72 cm at their apices. The efficacy of such open-top chambers as analogues of regional climatic warming has received biotic validation (Hollister & Weber, 2000). Eighteen open-top chamber arrays including controls (demarcated plots equivalent to open-top chamber basal dimensions) were randomized during late spring (November 2002) in three communities (six open-top chamber arrays and six control plots per community). The first community, incorporated in a Conophytum calculus (A. Berger) N.E.Br. ssp. calculus Community Group (Schmiedel, 2002), was present at the Luiper Kop site and contained Ruschia burtoniae L. Bolus as the dominant, in which Cephalophyllum spissum H.E.K. Hartmann occurred with a high constancy. The second community, incorporated in a Salsola (Chenopodiaceae) spp. –Argyroderma pearsonii (N.E.Br.) Schwantes Alliance (Schmiedel, 2002), was also present at the Luiper Kop site and contained C. spissum as the dominant in which Drosanthemum diversifolium L. Bolus occurred with a high constancy. The third community, also incorporated in the above Alliance, occurred at the Quagga Kop site and contained A. pearsonii as the dominant in which Dactylopsis digitata (Aiton) N.E. Brown occurred with a high constancy.
Ambient and open-top chamber microenvironments
Air temperatures at the two study sites were recorded at 30-min intervals by synchronized miniature thermocouple data loggers (Spectrum Technologies Inc., Plainfield, Illinois, USA) installed in ventilated radiation shields positioned 12 cm above the ground at northern, central and southern points within the open-top chambers (Fig. 2a) and control plots (ambient air). These were compared with historical records of ambient air temperatures obtained from two meteorological stations located at similar elevations and degree of longitude proximate to the experimental site (Climate of South Africa, 1986). These included Clanwilliam (32°11′ S, 18°54′ E) at an elevation of 75 m located c. 56 km south of the experimental site and Garies (30°34′ S, 18°00′ E) at an elevation of 227 m located c. 99 km north of the experimental site.
Plant and canopy fatalities
In mid-autumn (April) 4 months after placement of open top chambers, the numbers of deceased and live plants of each species, i.e. those with completely and partially lifeless canopies, respectively, present in the open-top chambers and control plots were counted and the fractions of deceased plants expressed as percentages. Also, the numbers of dry lifeless and fresh leaves present in the canopies of live plants in the open-top chambers and control plots were counted and the fractions of lifeless leaves expressed as percentages. Following onset of winter precipitation, deceased plants and canopies were re-examined to confirm their moribund states were irreversible, and plants classified as live were substantiated by their production of new leaves. In A. pearsonii, the ages of the deceased and live plants present in the open-top chambers and control plots were also determined from counts of the numbers of residual leaf pairs present on each plant. This species produces a new set of leaves annually, the deceased leaf pairs from previous years remaining permanently attached to the plant.
Leaf physical properties
Fresh intact leaves were randomly excised at midday 4-months after commencement of the study from each surviving succulent species present in the open-top chambers and control plots. In A. pearsonii, the entire plant comprising a single pair of leaves was removed for analysis.
Fresh leaf samples were weighed, dried in a forced draft oven at 60°C to a constant mass and reweighed. Their water contents were calculated from the differences between their fresh and dry masses and expressed as percentages of their fresh masses.
Leaf chemical properties
Photosynthetic pigments were extracted from fresh leaf samples ground at a low light intensity in 10 ml of 100% methanol at 2°C. Absorbances of centrifuged extracts were measured with a spectrophotometer (Beckman DU 640, Beckman Instruments Inc., Fullerton, USA) at specified wavelengths required for computation of chlorophyll a, chlorophyll b, and total carotenoid (xanthophylls +β-carotene) concentrations from published formulae (Lichtenthaler, 1987). Leaf pellets remaining after centrifuging were dried at 60°C in a forced draft oven and weighed. Photosynthetic pigment concentrations were expressed as µg mg−1 leaf dry mass.
Total soluble sugars (sucrose, glucose and fructose) were extracted from fresh leaf samples ground in two, 10-ml volumes of 80% ethanol (80 : 20, v : v, ethanol : water). The extracts were centrifuged, and supernatants adjusted to 25 ml in volumetric flasks for spectrophotometric determination of total soluble sugars (Buysse & Merckx, 1993). Residues were dried at 60°C and weighed, before analysis of starch concentrations by hydrolysing for 3 h in 5 ml of 3.6% HCl at 100°C, centrifuging and analysing the resultant sugars in the extracts (Buysse & Merckx, 1993). Soluble sugar and starch (non-structural carbohydrate) concentrations were expressed as µg mg−1 leaf dry mass.
Phenylpropanoids (flavonoids and anthocyanins) were extracted from fresh leaf samples ground in 10 ml of acidified methanol (79 : 20 : 1, v : v, methanol : water : HCl). Absorbances (Ab) of appropriately diluted centrifuged extracts were measured with the spectrophotometer at 300 nm, 530 nm and 657 nm. Flavonoid concentrations were computed as Ab at 300 nm g−1 dry mass (Mirecki & Teramura, 1984), and anthocyanins as Ab 530 nm −1/3 Ab 657 nm g−1 dry mass (Lindoo & Caldwell, 1978).
A single-factor nested analysis of variance tested differences in population and canopy mortalities, leaf physical and chemical properties between open-top chambers and control plots. Percentage values were arc-sine transformed to correct non-normality in proportions prior to statistical analysis. A two-factor nested analysis of variance tested differences in ages of deceased and live A. pearsonii between open-top chambers and control plot and their interaction.
Ambient and open-top chamber microenvironments
Measured average daily maximum air temperatures at 1400 SAST in the control plots (summertime monthly range: 33.2–35.6°C) were virtually identical (> 1.3°C difference) to those spanning a 28-yr recording period (summertime monthly range: 33.3–35.4°C) at the closest meteorological station (Clanwilliam) to our experimental sites (Fig. 1a). Noteworthy, was that the highest monthly air temperature of 46.1°C recorded at Clanwilliam was exceeded in our control plots on 3 d over the 4-month summer treatment period (Fig. 1d), 1 d of which comprised temperatures in excess of 2°C and 2 d in excess of 1°C of the highest recorded over a 28-yr span at Clanwilliam (Fig. 1a).
Inside the open-top chambers, there was strong diurnal air temperature fluctuation with most of the heating above ambient occurring during the daytime period (Fig. 1b). However, there was considerable day-to-day variation with temperature increases of the largest magnitude evident on clear-sky days with high solar radiation, and those of the smallest magnitude evident on overcast days. Notable, was that the scale of temperature increases and the timing of diurnal temperature maxima displayed gradients along north-to-south diagonals inside the open-top chambers (Fig. 1c). Temperature increases of smallest magnitude were measured at northern points inside the open-top chambers, with diurnal temperature maxima also occurring earlier in the day than those at southern points inside the open-top chambers (Fig. 1c). These temperature gradients inside the open-top chambers concurred with similar trends, but of opposing direction, reported for open-top chambers sited in the northern hemisphere where multiple regression analyses revealed that the scale of temperature increases were significantly linked to both wind speed and global radiation intensity (Marion et al., 1997). In the centres of the open-top chambers daily maximum air temperatures over the 4-month treatment period (summertime monthly range: 38.3–41.6°C) averaged 5.5°C above those in the control plots (Fig. 1c). These were still 1.0–4.0°C below corresponding mean monthly maxima recorded over a 28- yr span at the Clanwilliam metereological station (Fig. 1a), though the highest daily temperature of 53°C measured in the centres of the open-top chambers (Fig. 1d) did exceed that recorded at Clanwilliam by 7°C. Noteworthy, was a close correspondence between ratios of highest recorded monthly temperature against corresponding average daily temperature maxima in our open-top chambers (summertime monthly range: 1.279–1.335) control plots (summertime monthly range: 1.303–1.327) and at the Clanwilliam meteorological station (summertime monthly range: 1.291–1.333).
Plant and canopy mortalities
The fractions of A. pearsonii and C. spissum populations that deceased were significantly (P ≤ 0.05) greater (3.5–4.9 times) in the open-top chambers than in the control plots (Table 1). Also, the fractions of deceased leaves in the canopies of surviving live C. spissum and D. diversifolium were significantly (P ≤ 0.05) greater (2.1–4.8 times) in the open-top chambers than in the control plots.
Table 1. Impact of diurnal temperature maxima averaging 5.5°C above ambient (Control) in open-top chambers (OTC) over a 4-month summer treatment period on population and canopy mortalities, leaf physical and chemical properties of species of dwarf and shrubby leaf succulents on quartz fields
Leaf fresh and dry masses of live A. pearsonii were significantly (P ≤ 0.05) larger in the open-top chambers than control plots (Table 1). These concurred with the significantly (F1,87 = 10.8, P≤ 0.01) greater ages of live than deceased A. pearsonii in both the open-top chambers and control plots (Fig. 2d). There was no significant interaction (F1,87 = 0.8, P = 0.05) between plant age and warming treatment. Significantly (P ≤ 0.001) smaller leaf fresh and dry masses were measured in the canopies of live D. diversifolium in the open-top chambers than in the control plots (Table 1). D. diversifolium and C. spissum also displayed slightly, yet significantly (P ≤ 0.05) lower water contents in the open-top chambers than control plots.
Leaf chemical properties
D. diversifolium displayed significantly (P ≤ 0.01) higher foliar concentrations of chlorophyll b and R. burtoniae significantly (P ≤ 0.05) lower foliar concentrations of both chlorophyll a and b in the open-top chambers than control plots, but foliar carotenoid concentrations in all study species were unaffected by the elevated temperatures in the open-top chambers (Table 1). Noteworthy, were the low carotenoid concentrations, particularly in C. spissum and R. burtoniae. Slight instrumental deviation from the tabulated wavelengths λmax for chlorophyll b (Lichtenthaler, 1987), and/or carotenoid decomposition during pigment extraction under the influence of high endogenous organic acids in the CAM (crassulacean acid metabolism) succulents might provide an explanation.
Only A. pearsonii and D. diversifolium exhibited significantly (P ≤ 0.01) lower starch concentrations in the open-top chambers than control plots, though C. spissum and R. burtoniae displayed similar but statistically insignificant (P = 0.05) tendencies. Also, a slightly, yet significantly (P ≤ 0.01) higher soluble sugar concentration was measured in R. burtoniae leaves in the open-top chambers than control plots, this tendency apparent in A. pearsonii and C. spissum leaves as well but statistically insignificant (P = 0.05).
Foliar flavonoid concentrations in all study species were unaffected by the elevated temperatures in the open-top chambers. However, foliar anthocyanin concentrations were significantly (P ≤ 0.05) lower in R. burtoniae but significantly (P ≤ 0.05) higher in D. diversifolium in the open-top chambers than control plots.
Ambient air temperatures at our experimental sites during the 4-month treatment period did essentially represent average conditions for study area. This supported by the similar average daily maximum air temperatures recorded in our control plots and at the closest meteorological station at Clanwilliam. However, there were 3 d with abnormally high ambient temperatures at our experimental sites in which monthly diurnal air temperature maxima in the control plots did slightly exceed those recorded over a 28-yr span at the Clanwilliam meteorological station.
Daily maximum air temperatures in the centres of our open-top chambers, which averaged 5.5°C above ambient over the 4-month treatment period (Fig. 1c), approximated temperature increases of between 4.5 and 5.0°C (means of 7-GCM experiments) predicted (SRES A2-high climate sensitivity scenario) for the geographic coordinates 30° S to 31° S, 19° E to 20° E in the year 2080 (Hulme et al., 2001). The observed close correspondence between ratios of highest recorded monthly temperature against corresponding average daily temperature maxima in our open-top chambers, control plots and at the Clanwilliam meteorological station indicated that the temperature extremes accompanying the average daily maxima in our open-top chambers also provided a rational analogue of the predicted climate change.
The elevated temperatures in our open top chambers were associated with a 3.5- to 4.9-fold increase in plant and canopy mortality among the specialized dwarf succulents A. pearsonii and C. spissum, and a 2.1-fold increase in canopy mortality in the shrubby succulent D. diversifolium (Table 1). These substantial reductions in live standing succulent biomass seemingly contrasted with results of a meta-analysis of plant productivity responses to experimental warming in 20 of 32 global sites representing Forest, Grassland, high and low latitude/altitude Tundra biomes (Rustad et al., 2001). The meta-analysis found a 19% productivity increase in response to an average 2.4°C of experimental warming among the 20 sites distributed in cooler regions at latitudes above 35° N (Rustad et al., 2001). However, relative productivity responses to experimental warming declined with increasing site mean annual temperature (Rustad et al., 2001) implying that productivity could be expected to decrease with experimental warming at lower latitude subtropical and tropical sites not included in the meta-analysis. Noteworthy, was the abrupt rather than gradual temperature increase in our open top chambers on commencement of the study, which may have precluded natural plant acclimation to the increased heat stress, thereby contributing to the extraordinarily high mortalities observed. Indeed, an analysis of the age and size distribution of surviving and deceased A. pearsonii in the control plots and open-top chambers revealed that the live plants were significantly older (Fig. 2d) and larger than the deceased plants (Table 1); possibly indicative of past selection for heat resistant ecotypes.
An examination of the distribution patterns of surviving dwarf succulents in our open-top chambers revealed that live C. spissum and A. pearsonii individuals particularly were restricted to cooler microhabitats in more ventilated areas at the edges of the chambers (Fig. 2b) and shaded refuges beneath the skeletons of larger succulents such as Salsola spp. (Fig. 2c). These distribution patterns supported our perception that lethal temperature thresholds were exceeded in the open-top chambers, though ostensibly less severely in the cooler microhabitats. Indeed, diurnal temperature extremes in our open-top chambers were closely proximate to the upper temperature limit of 55°C considered tolerable by most vascular plants (Larcher, 1980; Kappen, 1981). However, there are reports of much higher lethal temperature thresholds in a diverse array of dwarf succulents. These, based on the ability of chlorenchyma cells to uptake a vital stain (Onwueme, 1979), ranging from 66.4°C to 66.9°C in rosette leafed Haworthia species (H. retusa and H. turgida), 68.3°C to 68.7°C in spherical leafed Lithops species (L. leslie and L. turbiniformis), and 69°C to 70°C in seedlings of Ferocactus (F. covillei and F. wislizenii) and in detached stem segments of Opuntia (O. ficus-indica) species (Smith et al., 1984; Nobel et al., 1986; Nobel, 1989). Nevertheless, lethal temperature thresholds in the majority of southern African dwarf succulent taxa may be lower, since they occur at high densities virtually exclusively on quartz-fields with milder thermal regimes (Schmiedel & Jürgens, 2002, 2004). Indeed, maximum daily soil temperatures during summer have been measured up to 10°C lower on highly reflective quartz substrates than on adjacent brown shales, and up to 3°C lower leaf temperatures have been measured in A. pearsonii growing inside quartz fields compared with the same species growing on neighbouring soils without quartz cover (Schmiedel & Jürgens, 2004). Similarly, another earlier study reported that leaf surface temperatures in a dwarf Argyroderma species on a quartz-covered substrate remained close to an ambient air temperature of 35°C whereas those of a shrubby Ruschia species on adjacent brown shale exceeded 45°C (von Willert et al., 1992).
Common physiological responses to experimental warming observed among the surviving dwarf and shrubby succulents in the open-top chambers included diminished foliar water contents and starch concentrations, the shrubby succulents also exhibiting altered foliar chlorophyll levels. Relocation of carbon reserves to secondary phenylpropanoid compounds was apparent in D. diversifolium where reduced foliar starch concentrations corresponded with increased anthocayanin levels. Nevertheless, the physiological changes observed in the surviving succulents did imply moisture limitations in the open-top chambers. Moisture deficits modify leaf conductance, transpiration and carbon assimilation rates (Ni & Pallardy, 1992) resulting in a degradation of stored non-structural carbohydrate reserves (Dunn et al., 1987), with small stature species with limited root extension and carbon reserves particularly sensitive to low soil water potentials (Donovan & Ehleringer, 1991; Flanagan et al., 1992). Indeed, a higher fraction of small leaves with a lower mass leaves were observed in the canopy of live D. diversifolium (Table 1), suggesting an acclimation to reduce transpiration loss under the warmer conditions in the open-top chambers. In this regard, it is known that shallow rooted dwarf succulents of the Mesembryanthemaceae require frequent though small amounts of water for survival, and the role of supplementary precipitation by fog and dew, estimated at as much as 38% of annual hydrological input (Dawson, 1998), is an important factor in ameliorating summer water deficits in semiarid and arid Mediterranean-climate ecosystems and preventing thermoregulation problems due to reduced transpiration (von Willert et al., 1992; Turner & Picker, 1993). This supplementary precipitation accrued on leaf surfaces drips, or is funnelled via stem-flow, onto the soil (Hutley et al., 1997) where it can be absorbed by plant root systems or directly by leaves from their wetted surfaces (Yates & Hutley, 1995; Martin & von Willert, 2000) or from the vapour-saturated atmosphere (Breazeale et al., 1950). Indeed, the interception of such supplementary precipitation by the open-top chambers and its channelling to the chamber edges could also partly explain the observed prevalence of surviving dwarf succulents in these locale.
In conclusion, it does seem likely that current thermal regimes are closely proximate to tolerable extremes for many of the 1563 almost all endemic succulent species included in the subfamily Ruschioideae which diversified rapidly in the region during the cool Pleistocene (Klak et al., 2004). Anthropogenic warming could therefore significantly exceed their thermal thresholds resulting in localized extinctions of particularly those specialized species range-restricted to specific habitats. However, further investigation is required to elucidate the importance of associated moisture deficits in these passive warming experiments, a potential consequence of supplementary precipitation interception by open-top chambers and higher evaporation therein, to allow reliable assessment of potential extinction losses under a range of climate change scenarios, and to match these with the predictions on the increasingly useful bioclimatic modelling approach.
We thank Mr S. Snyders and Mrs J. Arnolds for technical support. The study was partly funded by the German Federal Ministry of Education and Research (Promotion Number: 01LC0024).