1Deserts are outstanding sources of information about past geomorphic processes and former climatic fluctuations. Much of the evidence for past climatic changes is preserved in sediments inherited from both wetter and drier episodes, although interpreting this evidence is not always easy. More subtle and much harder to interpret is the legacy of former phases of erosion preserved in rock-cut terraces, pediments and glacis. Another striking feature of many desert landscapes is the close juxtaposition of very ancient and very recent rock formations. Few deserts show this contrast more starkly than the Flinders Ranges of South Australia (fig. 1).
2The streams within the main massif are seldom active today but flow for a few hours to a few days during rare, mostly summer rainstorms, when they transport and deposit boulders, gravel and coarse sand. They are characteristic desert streams : ephemeral or highly seasonal, with coarse traction loads of sand and gravel which stand in marked contrast to the very fine-grained late Pleistocene valley-fill deposits which are widespread within the ranges. This late Pleistocene alluvium consists of clay, silt and very fine sand with occasional lenses of coarse alluvial gravel and now forms terraces and terrace remnants. The fine-grained valley-fill overlies several metres of coarse gravel and is capped locally by up to two metres of gravel. The ephemeral traction-load streams are entrenched up to 18 m into the late Pleistocene valley-fill clays. Such clay-rich deposits are not accumulating today and were laid down under a climatic regime very different to that prevailing at the present time.
3This paper discusses the age, origin and significance of these late Pleistocene valley-fill deposits in relation to wider issues of landform inheritance and complex hydrologic and geomorphic responses to climatic change in desert environments. The geographical focus is the central part of the ranges for which we have the most detailed chronology. New geochemical and isotopic data are used to show that aeolian dust is recycled between mountain and playa during wet and dry climatic intervals. A simple water balance model indicates that reduced evaporation during the LGM would enable a permanent wetland to persist in the central Flinders Ranges during a time of peak regional aridity.
4The Flinders Ranges consist of uplifted, folded and weakly metamorphosed Proterozoic and Cambrian sedimentary rocks that form a series of north-south trending strike-ridges extending for about 400 km. The Ranges are located east of a major tectonic lineament, the Torrens Hinge Zone (Preiss, 1999), and extend from 30ºS to 33ºS and from 138ºE to 140ºE (fig. 1). The Ranges rise abruptly from the plains to over 600 m in elevation. The present-day climate is semi-arid to arid with steep gradients of temperature and precipitation. Daily temperatures vary between 18 ºC and 35 ºC in summer (December, January, February) and between 6 ºC and 17 ºC in winter (June, July, August). Annual precipitation ranges from less than 200 mm on the lower ridges to over 400 mm on the high ridges. There is a significant orographic enhancement of precipitation, especially during the winter months when atmospheric condensation levels and the cloud base coincide in elevation with the higher peaks. Much of the heavy summer rainfall arises from occasional incursions of tropical summer monsoon air masses, whereas winter rains are generally of relatively low intensity (Schwerdtfeger and Curran, 1996). The annual evaporation exceeds 2000 mm with summer monthly maxima up to 300-400 mm.
Fig.1 – Location map.
Fig. 1 – Carte de localisation.
1 : bedrock upland (> 400 m a.s.l.) ; 2 : wetland sediment ; 3 : alluvial sediment ; 4 : tufa bench ; 5 : rock-cut bench or gently sloping bedrock ; 6 : major drainage ; 7 : land above 300 m ; 8 : isohet (mm).
1 : hautes terres rocheuses (> 400 m a.s.l.) ; 2 : sédiments palustres ; 3 : sédiments alluviaux ; 4 : banquette de tufs ; 5 : replat rocheux sur versant en roche-mère à pente faible ; 6 : réseau hydrographique principal ; 7 : relief au-dessus de 300 m ; 8 : isohyète (mm).
5The basement geology comprises a variety of steeply dipping Neoproterozoic and Cambrian units that include resistant quartzite and mature sandstone, more easily eroded fine grained shale and siltstone, and dolomitic and limestone units of intermediate resistance to weathering and erosion (Drexel and Preiss, 1995 ; Lemon, 1996 ; Preiss, 1999). The Ranges are structurally controlled ; the high quartzite and sandstone ridges follow large-scale anticlines and synclines. Smaller rolling hills and valleys, which consist of the more readily weathered shale, siltstone and limestone units, generally separate the high ridges. Alluvial fans flank the western edge of the Ranges with a number of ephemeral stream channels dissecting them on their way towards Lake Torrens to the west (fig. 1).
6Lake Torrens is a structurally controlled playa located to the west of the Flinders Ranges and has been the site of accumulation for over 300 metres of sediments since the Eocene. The Flinders Ranges to the east are the major source of sediments for Lake Torrens. The upper 75 metres of sediments are of Quaternary age and consist mainly of brown, red and grey clays interbedded with fine sand layers and white powdery gypsum. The playa remained a dry to ephemeral depositional environment throughout the Quaternary.
7The streams draining the Flinders Ranges are ephemeral or highly seasonal coarse bed-load desert streams. During sporadic but intense summer storms, these streams may flow for several hours to depths of several metres, and are competent to move rocks up to a metre or more in diameter. Peak flows occur during heavy summer monsoonal downpours. Transport of large cobbles, gravels and coarse sands is characteristic of the contemporary flow regime. The sediments carried by the present-day ephemeral streams are in marked contrast to the late Pleistocene valley-fills within the ranges, which consist of very fine sandy clays up to 18 m thick, with intercalated minor cut-and-fill structures of fine to medium gravel close to the main ridges. Almost all of these valley-fill clays are now deeply dissected by the modern drainage system.
8Extensive fine-grained valley fills are now exposed in stream-bank sections within the ranges as well as along the piedmont margins. Field textures range from fine sandy clays to sandy clay loams across the full vertical heights of the profiles, with lenses of coarse gravel becoming increasingly abundant with proximity to the quartzite and sandstone ridges. Laminated clays and very fine sands of late Pleistocene age are exposed near the confluence of Brachina and Etina Creeks in bank sections up to 14 m in height (fig. 2). Individual laminations can be traced laterally for several hundred metres in the vicinity of section 1 (figs. 1 and 2). The laminae are of variable thickness, but usually range from a minimum of 1-2 cm to a maximum of 10-15 cm. The laminae range from nearly horizontal to gently sloping to slightly undulating with a wave height up to 10-15 cm. The associated ostracod and shell fauna are consistent with sluggish, shallow water flow under freshwater to saline conditions, as is the diatom assemblage (Williams et al., 2001).
9Initially, the laminated sediments were considered to be lake sediments (Cock et al., 1999). Williams et al. (2001) subsequently completed a detailed theodolite survey of the exposed sediments along their entire length and demonstrated that the sediments accumulated in a fluvial wetland that drained towards the west with a mean gradient of 1 in 80 over a distance of 14 km. The surface slope of the wetland was paralleled to a height of 70 m above the bedrock floor of the modern stream channel by much older rock-cut strath-terraces of similar gradient, thereby ruling out the possibility of any recent tectonic tilting of the wetland sediments. Figure 1 shows the approximate minimum area occupied by the wetland. Grey clays underlie the alluvial plain upstream of section 1 and merge progressively eastwards into red-brown clays of similar age. The wetland clays reach an elevation of nearly 340 m a.s.l., at section 1 and cover an area of 3.4 km2 to a maximum observed depth of ~14 m at this site and ~18 m further downstream. The catchment area of the wetland is roughly 222 km2 so that the system would have been highly sensitive to even slight changes in runoff.
10The wetland clays comprise a massive lower facies containing at least seven species of ostracods and at least two species of freshwater gastropod shells (Austropyrgus, Glyptophysa) ; and a laminated upper facies consisting of alternating dark and light bands. The dark laminae contain more organic carbon and less calcium carbonate than the lighter bands. A distinct palaeosol separates the two facies at section 1 (fig. 2, section 1). This palaeosol is roughly horizontal and is laterally continuous over at least 200 m. It has a strong coarse pedal structure with abundant root channels and/or insect burrows, and contains more organic carbon than the unmodified wetland sediments. In some sections, the clays contain bands of calcareous tufa up to 15 cm thick and several metres wide, containing herbaceous plant macrofossils and gastropod shells (fig. 2, section 2).
11Fourteen Accelerator Mass Spectrometer radiocarbon dates (AMS 14C) obtained from charcoal and shell fragments and six Optically Stimulated Luminescence (OSL) dates indicate that the deposits are of late Pleistocene age and were laid down between ~33 ka and ~17 ka (fig. 2 and tab. 1). The ages obtained for the wetland span the Last Glacial Maximum or LGM, which was a time of glacial aridity when many lakes and wetlands in southern and central Australia were becoming dry. The Flinders Ranges are presently arid to semi-arid, with very high rates of evaporation in summer, and there are no lakes or wetlands within the ranges today. Three questions therefore need to be answered. First, under what conditions did the fine-grained valley-fills accumulate? Second, how did the late Pleistocene wetland survive during this time in spite of a substantial reduction in precipitation in southern Australia? Third, why are the late Pleistocene valley-fill sediments so fine-grained when the present-day ephemeral streams throughout the region carry and deposit mostly coarse sand and gravel? The first aim of this paper is to offer a working hypothesis for the origin of the fine-grained valley-fills. The second aim is to consider how a late Pleistocene wetland persisted in the arid Flinders Ranges of South Australia during a time of widespread regional aridity. We turn now to the origin of the fine-grained sediments that make up the bulk of the late Pleistocene valley-fills.
Fig. 2 – Stratigraphic sections.
Fig. 2 – Coupes stratigraphiques.
1 : reddish-brown sandy clay ; 2 : gray silty clay ; 3 : brown silty clay ; 4 : pale yellow silty clay ; 5 : reddish-brown silty clay ; 6 : calcareous bed ; 7 : thick horizontal/sub-horizontal bedding ; 8 : thin horizontal/sub-horizontal bedding ; 9 : thin/medium horizontal laminations ; 10 : massive bedding with sub-vertical structure ; 11 : poorly developed horizontal/sub-horizontal bedding ; 12 : horizontally-bedded tufa with reed or root casts ; 13 : shell or shell fragments ; 14 : bioturbation ; 15 : gypsum crystals ; 16 : calcium carbonate concretion ; 17 : erosional break ; 18 : pebble lense.
1 : argile sableuse brun-rougeâtre ; 2 : argile limoneuse grise ; 3 : argile limoneuse brune ; 4 : argile limoneuse jaune pâle ; 5 : argile limoneuse brun-rougeâtre ; 6 : lit calcaire ; 7 : épais litage horizontal à sub-horizontal ; 8 : litage horizontal à sub-horizontal fin ; 9 : laminations horizontales fines à moyennes ; 10 : litage massif à structure subverticale ; 11 : litage horizontal à sub-horizontal mal défini ; 12 : tufs à litage horizontal contenant des figures de manchon racinaire ou de roseau ; 13 : coquillage ou fragments de coquille ; 14 : bioturbation ; 15 : cristaux de gypse ; 16 : concrétion de carbonate de calcium ; 17 : rupture de pente d’origine érosive ; 18 : lentille de galets.
Table 1 – List of OSL and AMS 14C ages.
Tableau 1 – Liste des datations OSL et AMS 14C.
12Discontinuous layers of red-brown very fine sandy or silty clay up to 30 cm thick crop out on the summits of almost every quartzite and argillite ridge examined in the central Flinders Ranges. Irrespective of underlying lithology, these red-brown clays form shallow mantles across the slopes. The uniform colour, lithology, field texture and widespread distribution provided strong circumstantial evidence that the clays are the remnants of once continuous wind-blown dust mantles blown in from the west during times of greater dust flux in the late Quaternary. Aeolian dust or parna mantles are widespread in southeastern Australia but are seldom very thick (Butler and Hutton, 1956 ; Butler, 1982). The aeolian dust particles often become incorporated into the underlying soil or weathering profile and so are not always obvious (Chartres et al. 1988 ; Greene et al., 2001 ; Gatehouse et al., 2001 ; Mee et al., 2003). We therefore used a variety of analytical methods to test two hypotheses : (i) The discontinuous patches of red-brown clay on the ridge tops and along the slopes are aeolian dust ; (ii) The fine-grained valley-fill deposits consist at least in part of reworked aeolian dust.
13In order to test these two hypotheses, we sampled the argillite bedrock, the valley-fill sediments, the red-brown clays on the summits of quartzite, limestone and argillite hills and the uppermost beds in the Lake Torrens playa west of the Ranges (fig. 1). Particle size analysis confirmed our field observations and showed that the ridge-top samples are well-sorted silty to fine sandy clays, with most particles in the 2-63 µm size range, regardless of the underlying bedrock lithology (Williams and Nitschke, 2005). X-ray diffraction (XRD) analysis showed that the dominant clay mineral suite in the ridge-top samples consists of kaolinite and illite. The trace elements Zr, Ti, Th, La, Ce, Y, Cr and Nd are enriched in the ridge-top samples compared to the underlying bedrock. 143Nd/144Nd ratios of the ridge-top, valley-fill and Lake Torrens samples show clear correlation with the argillite Bunyeroo and Brachina Formations and 87Sr/86Sr residue ratios show a similar pattern, with ridge-top, valley-fill and Lake Torrens samples similar to the Bunyeroo and Brachina Formations. 87Sr/86Sr leachate ratios are notably lower, ranging from 0.710 to 0.715, reflecting the influence of the limestone rock units in the Flinders Ranges as well as calcareous aeolian dust of marine origin. Except for one sample, the Lake Torrens samples have the same clay mineral suite, similar major and trace element geochemistry patterns and similar 87Sr/86Sr and 143Nd/144Nd ratios as the ridge-top and valley-fill samples along with the Bunyeroo and Brachina Formations (Williams and Nitschke, 2005).
14The particle size, geochemical and isotopic data all imply a significant degree of sediment recycling between the Flinders Ranges and Lake Torrens (fig. 1). We suggest that during regionally drier intervals during the late Pleistocene (and no doubt before then) fine sediment was blown eastwards from the Lake Torrens playa to form the ridge-top and hillslope aeolian dust mantles, later reworked by slope wash and incorporated into the fine-grained valley-fill deposits. Further support for this inference comes from the rapid rates of sedimentation estimated as ~0.7 m ka-1 revealed by the AMS 14C dates obtained for section 2 (fig. 2). This rate is much higher than the 6 to 14 mm ka-1 rates of long-term erosion in the catchment obtained by measuring in situ cosmogenic 10Be (Williams et al., 2001). During more humid phases fine sediment was ferried westwards in streams flowing out from the western ranges and laid down in the Lake Torrens playa. We conclude that the fine-grained late Pleistocene valley-fills in the central Ranges contain some material derived from the weathered argillites in the wetland drainage basin and some reworked material originally blown in from the west as aeolian dust.
15The experimental observations of Yaïr (1994) on loess-mantles in the northern Negev Desert of Israel prompted us to ask what impact thin aeolian dust mantles might have had upon local runoff in the Flinders Ranges. For example, did the presence of more or less continuous aeolian dust mantles alter the hydrologic response of the previously bare slopes to precipitation events during the late Quaternary? During the late Quaternary (and earlier) extensive wind-blown dust mantles accumulated on desert margin slopes in every continent, especially Africa, Asia and Australia (Williams et al., 1998). The most recent phase of aeolian dust accumulation spans the LGM between 24 ka and 18 ka and may have lasted from ~35 ka to ~15 ka. Pockets of this aeolian dust are common within cracks in the bedrock along the ridge crests and valley sides. Along the lower slopes remnants of these reworked dust mantles are seen to absorb gentle winter rains today. Based on these observations, we hypothesise, but cannot prove, that these aeolian mantles may have altered hillslope runoff sufficiently to mask the direct influence of late Quaternary climatic fluctuations upon the landscape. The effect would only be noticeable during times of low intensity winter precipitation, prior to ~15 ka, since the clay mantles would soon become eroded by the intense downpours that heralded the return of the summer monsoon rains at ~15 ka. The renewal of rapid erosion at this time is evident in the sudden change from accumulation of clay to that of gravel and sand at the top of sections 1 and 2 (fig. 2), followed by incision to bedrock. The hypothetical sequence of events is illustrated in fig. 3 and is as follows : (1) Coarse sands and gravels are transported along the valley floor by ephemeral stream channels during intense summer rainstorms and alluvial fans are active along the valley margins. (2) The summer monsoon rains weaken and gentle winter rains become dominant as the regional climate becomes colder, drier and windier. (3) Between ~33 ka and ~17 ka a thin permeable mantle of aeolian silts and sandy clays blown in from the west covers the previously bare impermeable rock surfaces in the Flinders Ranges. (4) Hillslope infiltration is increased and runoff is reduced during a long interval of low intensity mainly winter rains. (5) Base flow increases and flash floods become increasingly rare. (6) Previously ephemeral traction load streams become perennial suspension load streams transporting fine-grained sediment. (7) As the progressive influx of reworked wind-blown dust leads to an increase in the ratio of load to discharge, there is widespread alluviation in the valley bottoms and progressive accumulation of fine-grained valley-fills. (8) With the return of the high intensity summer monsoon rains at ~15 ka, alluvial fans become active once more along the valley sides, depositing a thin mantle of coarse sand and gravel across the margins of the former wetland. Fluvial erosion during short intervals of torrential flow in the now ephemeral stream channels leads to entrenchment of the late Pleistocene valley-fill down to bedrock.
Fig. 3 – Late Quaternary geomorphic evolution.
Fig. 3 – Évolution géomorphologique au Quaternaire récent.
A : > 33 ka, alluvial fans active along valley margins ; ephemeral streams deposit sand and gravel bars during rare torrential flows ; B : 33-17 ka, gentle winter rains become dominant as the summer monsoon weakens. Fine sediments accumulate along the valley bottoms as aeolian dust mantles are washed off the slopes. A permanent wetland develops under a colder climate with reduced evaporation ; C : < 17 ka, with the return of high intensity summer rains, alluvial fans become active once more along the valley sides and there is an initial phase of gravel deposition along the surface of the former wetland adjacent to the upland slopes ; D : subsequently, as the summer rainfall regime becomes established, ephemeral streams incise the late Pleistocene valley-fill down to bedrock during brief intervals of torrential flow.
A : avant 33 ka, des cônes alluviaux sont actifs le long des versants, et des rivières à régime torrentiel et éphémère déposent des bancs de sable et de graviers ; B : entre 33 et 17 ka, les pluies fines hivernales dominent et les pluies estivales s’affaiblissent. Des sédiments fins s’accumulent le long des vallées à mesure que les poussières éoliennes déposées sur les versants sont érodées. Un marécage permanent se forme sous un climat plus froid et moins évaporant ; C : avec le retour de la mousson d’été après 17 ka, les cônes alluviaux redeviennent actifs et un épandage de galets s’accumule sur la surface de l’ancien marécage à proximité des pieds de versant ; D : une fois que le régime des pluies estivales est bien établi, le remblaiement pléistocène récent est entaillé jusqu’au socle et un régime fluviatile torrentiel mais éphémère reprend.
16The end result of the processes outlined above was a change from deposition of coarse sand and gravel bed-loads by ephemeral or highly seasonal late Quaternary desert rivers prior to ~33 ka [stage 1] to a depositional regime characterised by widespread accumulation of sediments in the valley bottoms between ~33 ka and ~17 ka [stage 7]. If this hypothesis is correct, we should expect to find similar fine-grained late Pleistocene valley-fills well beyond the confines of the central Flinders Ranges. A growing body of evidence from now semi-arid localities fine-grained up to 500 km east of the Flinders Ranges suggests that this may indeed be so (Williams et al., 1991 ; Fanning, 2002 ; Keating, 2002). We now use a simple water balance model to show how it was possible for a wetland to persist during a time of peak regional aridity.
17We noted above that the clay-rich late Pleistocene valley-fills accumulated more or less continuously during an interval lasting about 15 000 years between ~33 ka and ~17 ka. There was widespread aridity and regional desiccation in Australia at this time, with desert dunes active and large lakes drying out throughout Australia, including Lake Eyre in central Australia and the Willandra Lakes in semi-arid western New South Wales (Magee, 1997 ; Bowler, 1998). This time interval spans the Last Glacial Maximum (LGM), here taken to be the 6000-year interval between 18,000 and 24,000 calendar years, equivalent to 15,000-21,000 radiocarbon years BP (Bard, 1999). The LGM was a time of lower temperatures and enhanced aridity within the sub-tropics in both hemispheres (Williams et al., 1998). How did a wetland persist under such conditions? We tackled this question in several stages. We started our analysis using a simple water balance model in which we began by assuming no losses to or inputs from groundwater (Chor et al., 2003). For the wetland to remain saturated, water losses from evaporation must balance water inputs from runoff and direct precipitation onto the wetland surface. Thus :
18Ac Pc k + Aw Pw = Aw E (1)
19where Ac is the catchment area, Pc is the mean annual precipitation over the catchment, k is the runoff coefficient, Aw is the surface area of the wetland, Pw is mean annual precipitation over the wetland and E is the mean annual surface evaporation from the wetland.
20We first examined the relationship between present-day mean monthly temperatures and mean monthly evaporation and found a statistically significant relationship between these two parameters. We went on to produce a set of regression equations relating present-day mean monthly evaporation to mean monthly temperature for a number of stations in and around the Flinders Ranges, two of which are given here for the town of Hawker (fig. 1) :
21E = 55.906e0.1123 Tmin (2)
22E = 17.717e0.0931 Tmax (3)
23E is mean monthly evaporation in mm, Tmin is mean minimum monthly temperature and Tmax is mean maximum monthly temperature. The R2 value for regression (2) was 0.868 and for (3) was 0.9375. In every case we found a closer correlation between maximum temperatures and evaporation than minimum temperatures. The next step was to evaluate the possible values of late Pleistocene evaporation in relation to best estimates of glacial age temperatures inferred from proxy data (Galloway, 1965) including lower limits of periglacial solifluction mantles dated using cosmogenic nuclides (Barrows et al., 2001) and amino acid racemization of emu eggshells dated by AMS 14C (Miller et al., 1997). Regional temperatures during the LGM were at least 6 °C lower than at present (Galloway, 1965) and may have been as much as 8 °C to 10 °C colder (Miller et al., 1997 ; Barrows et al., 2001). In a topographic setting relatively sheltered from wind, such as the central Flinders Ranges, our model showed that the lower temperatures would result in greatly reduced evaporation.
24We then introduced further modification to our water balance model by first removing the more extreme summer precipitation events and then removing the summer precipitation altogether from mean annual precipitation. The rationale behind this came from several studies that showed a substantial weakening of the summer monsoon rains in Australia during the LGM. For example, the carbon isotopic studies of Johnson et al. (1999) on sub-fossil emu eggshells in central Australia revealed an absence C4 grasses from the emu diet at this time. These grasses are dependent on summer rainfall and their absence would imply a strong reduction in summer rainfall over central Australia at that time. Wyrwoll and Miller (2001) analysed biogenic materials in flood deposits from drainage basins in the Kimberley region of northwestern Australia and concluded that the onset of an active summer monsoon began about 14,000 years ago following a weak to absent summer monsoon before then.
25We tested our provisional conclusions against the water balance equation used by Bowler (1981, 1986) in his analysis of Australian lakes, substituting wetland area for lake area. The runoff coefficients adopted were based on Australian basin data provided by Bowler (1986) and runoff studies in arid western New South Wales where precipitation events in excess of 20 mm produce surface runoff but runoff values seldom exceed 10% (Pilgrim et al., 1979). For a runoff coefficient of 0.05 (5%), a temperature 10 ºC cooler than present is required to enable a perennial wetland to exist, but temperatures would only need to fall by 6 ºC if the runoff value was 10%. The outcome of this analysis was to show that a perennial late Pleistocene wetland could be maintained provided evaporation was reduced to 25% of present, which a drop in mean maximum temperature of only 6 ºC would allow.
26Our simple hydrological model showed that substantially lowered evaporation values could have been sufficient to enable a wetland to persist in the central Flinders Ranges even during a time of regional aridity and much reduced summer rainfall. We will need to refine our model by taking into account possible groundwater inputs and outputs, a wider range of runoff values, and the potential impact of evapotranspiration from the different plant communities in the past.
27Geochemical and isotopic analyses of samples collected from Lake Torrens and from within the ranges indicate that during wetter phases fine sediment was transported westwards from the mountains to accumulate in the Lake Torrens playa and during drier phases aeolian dust was blown from the playa eastwards into the massif, where it was eventually incorporated into the late Pleistocene fine-grained valley-fill deposits.
28The fine silt and clay-rich valley fills accumulated for a period of 15,000 years between ~33 ka and ~17 ka, including the very arid Last Glacial Maximum. For a permanent wetland to exist during peak regional aridity when the majority of lakes were drying throughout Australia required a substantial reduction in high intensity summer rainfall events as well as greatly reduced evaporation. A late Pleistocene wetland could be maintained even with lower precipitation provided evaporation is reduced to 25% of present. A simple water balance model showed that this would be possible if mean maximum monthly temperatures were only 6 ºC cooler.