The Dry Valleys of Barbados: a Critical Review of Their Pattern and Origin Author(S): J

The Dry Valleys of Barbados: A Critical Review of Their Pattern and Origin Author(s): J. Fermor Source: Transactions of the Institute of British Geographers, No. 57 (Nov., 1972), pp. 153-165 Published by: Blackwell Publishing on behalf of The Royal Geographical Society (with the Institute of British Geographers) Stable URL: http://www.jstor.org/stable/621559 Accessed: 07/10/2009 17:21

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/action/showPublisher?publisherCode=black.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission.

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Blackwell Publishing and The Royal Geographical Society (with the Institute of British Geographers) are collaborating with JSTOR to digitize, preserve and extend access to Transactions of the Institute of British Geographers.

http://www.jstor.org The dry valleys of Barbados

A criticalreview of their patternand origin J. FERMOR FormerlyLecturer in Geography,University of the WestIndies

Revised MS received i March i972

ABSTRACT.The major erosional feature on the emerged coral reefs of Barbadosconsists of a network of valleys. These valleys are generallydry, and the main problem discussed is that of their origin and subsequent desiccation. A secon- dary problem is provided by the courses of many valleys which have sections at right-angles to past and present shorelines. Of the various hypotheses reviewed, that of surface run-off after heavy storms receives the most support. The possibility of climatic change is discussed and the contributory role of karstificationconsidered. The valley pattern is judged to be a consequent one, its peculiarities resulting from primary depressions on the reefs.

THErock outcropping over six-seventhsof Barbadosis a corallinelimestone often referred to as the CoralCap. In the north-eastof the island this has been breachedto revealthe underlyingsandstones and clays of the Scotlanddistrict (Fig. i.) The CoralCap has never been overlainby other strataand its surfacedescends west and south from elevationsof over 350 m nearthe escarpmentin a seriesof terraces.The originof these terraceshas been explained by K. J. Mesolella (i967), whose stratigraphicalstudies reveal an ordered sequenceof fore reef, reef crest and back-reefdeposits across each terrace.This indicates that the terracesare separateraised reefs, ratherthan wave-cutplatforms on a single tilted reef or the productof a series of step faults. E. T. Price (1958) concludedthat the decliningsilica/aluminium ratios with heighton the limestonesoils indicatea uniformrate of uplift. Mesolellaet al. (i969) confirmthis for the lowerterraces by Thorium-Uraniumdating, and they find an overallassociation of the height spacingof reefs with the time spacingof radiationmaxima predicted by considering the changingtilt of the earthand the precessionof the equinoxes.An almostuniform rate of uplift of 0-3 m per thousandyears is indicated.This steadyuplift has separatedthe reefs formedby the successionof high warminterglacial sea-levels in the Pleistocene. There has been some foldingduring the uplift (Fig. i) and the large-scalerelief of the CoralCap, with anticlinaldomes and synclinalvales, faithfullyreflects this. Two types of erosionalfeature are commonon the limestone.The first consistsof small dolines,usually discrete, predominantlycircular in form and seldom greaterthan I5 m in depth, many containingsinkholes and manyothers ponds. The secondand morestriking features are the valleys which radiatefrom the high ground aroundthe Scotland district and from the Golden Ridge dome. They are often incised to depths of 30 m and their sides slope commonly at more than 30° and rarelyless than I5°. The valley patternis pinnatewith some trellising. When analysedwith respect to past and present shorelines,the trend of the formershown by the raised fringingreefs, 79 per cent of the total length makes angles greaterthan 6o0°to the shore,7-3 per cent is between30° and 6o0°and 13'7 per cent is less I53 I54 J. FERMOR than 30°; in the lattercategory, many segments are parallelto the old shorelines.Many of the valleys have been beheadedby the retreatof the escarpmentaround the Scotland district.Most valleysreach the sea but some end in the Sweetvaleand St. George'svale synclines.These may terminatein definitesinks or they becomeshallower and fade away

1. Crab Hill Anticline 2. Mt. Poyer Syncline 3. Clairmont Anticline 4. Sweet Vale Syncline 5- Golden Ridge Anticline 6. St. Georges Vale Syncline 7. Christ Church Anticline

.4/ Escarpment of the Coral Cap \ Dry Valleys (from 1: 50000 map)

FIGURE i. Structure and valley pattern on the Coral Cap

(Fig. i). This imposingvalley system is dry savefor shortstretches near the mouthsof some valleys,and infrequentephemeral flows after heavy storms. The hydrologyof the CoralCap has been thoroughlyinvestigated by A. Senn (1946). He divides the cap into two areaswith respect to phreaticconditions, naming these the stream-waterand sheet-waterzones (Fig. 2). In the firstcase, waterpercolates through the coraluntil arrestedby the Oceanicsbeneath. These Oceanicsare deep-waterlimestones of low permeability,and water flows seaward above them in the basalAmphisteginal horizon of Dry valleysof Barbados I55 the CoralCap, either diffuselyor in cave systems. The form of the watertable is closely tied to the top of the Oceanicsand, where this is irregular,perched water tables may be impounded.In a few areasthe Oceanicsare brought near to the floorsof surfacevalleys and intermittentsprings may rise and flow for short distances.However, in generalthe water table in this zone is over 30 m below the valley floors. Near the coast the interfacebetween the coralrock and the Oceanicspasses below sea level, and the usual relationshipbetween fresh and seawatercomes into play. The slope of the watertable depends on the permeabilityof the coralrock and the amountof percolation. This slope is extremelylow; indeed, Senn found it lowerthan the i m closingerror of his surveyingof well tops, so that he considersthe sheet-waterzone to be virtuallyhorizontal at sea level. This low gradientis confirmedby the low seasonalrange which can be measured with greateraccuracy and which does not exceed I.5 m within this zone despite the markedseasonal regime of precipitation.Pumping tests show only fractionaldrawdowns with high yields as a furtherindication of high permeability.

W E 8km Sheet r- Stream Water Zone hreatic FWterI Zone I Divide

Sea Lvel . S

I- Sait water/Fresh water L- Fresh water lens - Water table interface

FIGURE2. Diagrammaticsection to show the hydrology of the Coral Cap (after A. Senn and P. Martin-Kaye)

Some groundwateris routedinto the Scotlanddistrict. Though no springsissue at the junction of the Coral Cap and the Oceanics,it appearsthat the latter are locally more permeableat the outcropthan when buriedbeneath the CoralCap. Waterpasses through both beds to outlets at the foot of the Oceanics.The phreaticdivide occurs at a variable distance not exceeding i km behind the scarp (P. Martin-Kayeand J. Badcock, i962; E. R. Cumberbatch, I967).

THE PROBLEM OF THE VALLEY PATTERN The pinnate form of the valley patternis to be expected of youthful drainageon high overallslopes-commonly 300 m in 8 km. The right-angledbends and the stretchesparallel to the presentor past shorelinesare less readilyexplained. The drainagehas presumably extendedacross each new reef as it emerged.Live fringingreefs are often distinctlylower at the backthan at the crest (T. F. Goreaux,1959), so that on emergencethey may present a barrierto streamextension which diverts the drainageparallel to the shoreuntil some low point or formersubmarine channel through the reef crest allowspassage to the sea. I. G. Macintyre(i967) recordsa living and a drownedbarrier reef off the west coastof Barbados today, these being respectively20 and io m high. If similarreefs were raisedin the past, as the combinedtectonic and eustatic history suggests, then trellisingof the drainagepattern is made yet more understandable. I56 J. FERMOR The ChristChurch dome is withoutmarked valleys, and in this areait is possibleto find troughsbehind raised reefs priorto valleycutting. The best exampleis foundbetween Bridgetownand GraemeHall behindthe First High Cliff,where a stringof largeelongate depressionsruns parallelto the south coast.

THE PROBLEM OF THE ORIGIN AND DESICCATION OF THE VALLEYS R. H. Schombergk(I848) concludedthat the catchmentsare too smallto provideenough run-offto cut the valleys. He consideredthem crackscaused by an earthquakeand cites their radialpattern in evidenceof this. Subsequentopinion is unanimouslyin favourof streamcutting as the principalagent, but some authorsintroduce additional factors. C. T. Trechmann(1955) suggeststhat the streamsinherited emerged submarine features, while J. Tricart(i968) believesthat the deepestsections of valleysthrough the fronts of raised reefs were cut as creeksby coastalerosion. The explanationof the right-angledbends in some valleysgiven aboveis in accordwith Trechmann'sgeneral statement. A few valleys, notablythe valleypast Moncrieff,have one side higherthan the other,suggesting that they followrecent fault lines, so providingsome support for Schombergk'sideas. Tricart's views are madeless credibleby two facts.First, deep inlets are absentfrom cliffed sections of the presentcoastline and secondly,many of the inlandcliffs, cut throughso deeplyby valleys, are practicallyunaltered reef frontsnever subjectto prolongedmarine erosion. Once streamcutting is accepted,the problemis shifted to the cause of the present desiccation.Hypotheses divide into those involvinga loweringof the watertable and those accountingfor decreasedrun-off. The implicationis that valleycutting was achievedeither by springsapping and base flow or by immediaterun-off. In normalstreams these agents are complementary.In anothersense they are competitive,since they share a common sourceof supply,and any increasein one is a denialto the other.Since in this case present conditionsreduce both to near-extinctionit is easierto imagineone of them increasedto a significantlevel than both together. Hypotheses emphasizing base flow In theory the water table within the Coral Cap could be raised to the valley floors by loweredpermeability or by increasedpercolation. In practicethe magnitudeof the change requiredmakes either explanation unlikely. At the backof the sheet-waterzone the water tableis typically50 m belowthe valleyfloors, and about i m abovesea level. In wet seasons it may risea furthermetre or two. FromDarcy's law we see that a conservativeestimate of the increasein percolationneeded to makethe valleyslines of influentseepage would be twentyto thirtytimes the presentwet seasonsupply. In the stream-waterzone the hydraulic gradientsare no guide to permeability,but the cave systemssuggest that this permeability is still very high while the streamgradients to be attainedare much greater.With the local exceptionsalready mentioned, it is inconceivablethat higherrainfall alone could ever have caused base flow throughoutthe valleys, let alone reach the originalsurface to initiate valleycutting. An increasein permeabilitymight seem a moreplausible proposition if only becauseof the developmentof cave systems.It must be noted, however,that the low hydraulic gradientsof the sheet-waterzone occur in reefs dated as young as 82ooo BP. Consequentlygreatly reduced levels of permeabilitycannot have existed in any considerable part of the CoralCap at the same time and the watertable cannot have been raised simultaneouslyto long sectionsof the valley floorsby such means. Dry valleysof Barbados I57 The uplift of Barbados provides a much more convincing reason for lower water tables now than in the past. Such uplift leads directly and indirectly to the creation of springs at lower levels which may compete with higher springs and which will lower the water table generally. An example of the indirect effect of uplift is provided by the inversion of relief in the Scotland district. Formerly this area must have constituted the highest land in Barbados with groundwater moving outward from it. Now it has a centripetal drainage with springs underdraining the infacing coral scarp. However, as by far the greater part of the percolation entering the Coral Cap still flows down the dip of the Oceanics, attention has focused on the more direct lowering of springs fed by this flow to the west and south. J. B. Harrison and A. J. Jukes-Browne (I890) point to the short perennial streams occupying the lowest sections of some valleys at the present time. They assume that such streams have always existed near the coast through the island's history. Each period of emergence saw an extension of such streams to the new coastline, followed first by the deepening and then by the desiccation of the previously occupied sector of the valley as both the stream and the water table feeding it became adjusted to the change in base level. In this way a lengthening dry valley would be left behind, 'each part of which has been at sometime occupied by a surface stream, though probably the whole length has never been so occupied at one and the same time' (p. 40). There is every reason to accept this view in so far as it relates to the water table. At each fall in sea level the fresh-water lens will slide down the dip-slope of the Oceanics together with the sea water on which it floats. The fall in the water table will be confined to the sheet-water zone and to the latest addition to the stream-water zone, as elsewhere base level is provided by the top of the Oceanics both before and after. If the mechanism is considered in reverse, it is clearly capable of raising spring levels to any point on the valley floors or to the surface existing before valley cutting took place. To suppose that such springs and their outflows cut the valleys is much more conjec- tural. Harrison and Jukes-Browne's account contains a paradox. Granted that stream incision and falling water tables are both expected reactions to uplift, why should first one and then the other set the pace ? Today many coastal wells in Barbados show tidal fluctuations, a semi-diurnal demonstration that the height of the sheet-water table not only depends on sea level, but that any change in that level is transmitted inland far more quickly than a knick point could be. Yet at each extension of the streams, considerable incision has taken place for the valleys cut deeply through the emerged reef crests. In the case of the First High Cliff (the face of an emerged reef) the larger valleys have cut down to levels below the base of the cliff for as much as 0o5 km behind the cliff face, creating valleys 30 m deep. Present sea level is only 5-10 m below these valley floors as they emerge from the cliff and the sea is about i km distant, but perennial streams occupy only a fraction of the sections between the First High Cliff and the sea. If streams behind the present uncliffed coastline are so short, a fall in sea level which exposed the face of the present fringing reef might be expected to lead to new springs or seepage zones from this face and to the swift desiccation of the previous stream sections, xwithouttime for the period of incision needed to provide integrated extensions of the valleys. Valley networks originating as Harrison and Jukes-Browne propose should have some distinctive properties. The uniform rate of tectonic uplift suggests that every altitude above the present stand of the sea has had the same time within the migratory zone of valley cutting. In the course of time, Barbados has increased in height and size, changes likely to I58 J. FERMOR increase the discharge of marginal springs and hence their power to erode. As the streams extended to keep pace with falling sea levels, many confluences have occurred especially in connection with the trellised section previously described. New streams then arose in the widened interfluves. This pattern repeats itself at all elevations. Considering only the uniform duration of valley cutting at each height, the density of the valley pattern should not change. Considering also the likely increase of discharge with time, the valley pattern might well increase in density at lower elevations. This would certainly be a distinctive feature in a monolithological catchment with more rainfall on high ground. To test this prediction, an area of western Barbados was chosen between northings 456 and 468. No valleys enter or leave this area, nor is there any great loss or gain of groundwater, from either north or south. The kilometre grid squares of the i :io ooo map series were used to divide this region. Fifty-eight squares were compared for height and valley density, the latter from blue line data (dry valleys as well as streams are shown thus on these maps). A positive correlation of 0-49 resulted, this being significant at the i per cent level. This disagrees sharply with deductions made from Harrison and Jukes-Browne's hypothesis.

Hypotheses emphasizing surface run-off Overland flow occurs when rainfall intensities exceed infiltration capacities. Such flow may concentrate and cut valleys or it may collect in depressions. A. Coleman and W. G. V. Balchin (1959) propose a general model of karstificationwhich includes a youthful stage of valley cutting followed by desiccation linked with increasing sink-hole drainage and development of depressions. Barbados supplies an intriguing test case for this model since the age of any surface can be estimated roughly by equating each metre in elevation with approximately 3000 years of emergence. The first possibility is that depressions enlarge with age. The size of each depression was estimated from the i :io ooo map series with its contours at 6 m interval, using the highest closed contour. The results were grouped into classes by elevation. Several difficulties arose. The smaller depressions were difficult to measure accurately with a planimeter, and many sinkholes are marked without an enclosing contour. A further complication is that the size range is not normally distributed but includes a few larger depressions which, moreover, are unlikely to be products of karstificationbut seem rather to be relics of former submarine topographies. These survive only at low elevations and run counter to the general size trend. To overcome these difficulties, all the depressions were compared with an arbitrary standard size and the proportion larger than this size was noted for each height range of 30 m, sinkholes without enclosing contours being counted as small depressions. The results are summarized in Table I and suggest that depressions tend to increase in size as surfaces become older up to about 400 ooo years, thereafter remaining stable. A further check was made to ascertain if the frequency of depressions increases with the age of the surface. Table I shows this to be so at first but that later a decline sets in. These two lines of evidence support a primafacie case for supposing that the infiltration capacity of the Coral Cap has increased during the first 300 ooo years of emergence. R. K. Matthews (I968) and N. E. Pingitore (I969) reveal that the issue is more com- plex. These authors demonstrate that a lithological change is occurring in the raised reefs of Barbados which runs its course in the first 200 000 to 500 ooo years of emergence. Under the conditions of deposition, aragonite is the stable form of calcium carbonate. Subaerial Dry valleysof Barbados I59

TABLE I Density and size of depressionsat differentheights

Height (m): 0-30 30-6o 6o-90 90-120 I20-I50 I50 -I8 i8 -210 210-240 240 -270

Number per km2 i-6 4-4 6'8 9'4 7'3 8'7 6-8 4'3 4'8 Per cent o-i8 ha or more I8 22 29 36 40 42 39 42 39 Valley density 1I2 1-5 2-2 2-4 3-2 3-I 3'3 2'7 4'5 conditions cause a gradual conversion to calcite, and this conversion involves an 8 per cent increase in volume. The hydrological implications are uncertain. A loss of primary porosity owing to void filling has been demonstrated, but since the conversion requires solution and reprecipitation the outcome depends on the rapidity of groundwater movement which might flush out the solution before reprecipitation takes place. The rate of such movement must depend on the size and degree of connection of the cavities. One can see in this the mechanism by which water transport in the vadose zone is concentrated into sink-hole drainage, as pathways below a critical size are further reduced and those above it enlarged. In view of the loss in general porosity, however, it is less certain that the infiltration capacity of large areas is increased. Tricart (i968) remarks on the ability of low-level surfaces to absorb high rainfalls without inundation in the absence of valleys or sinks, and we have remarked on the high permeability of coastal areas which may also result from factors giving high infiltration capacities. So far developments which might reduce overland flow into the valleys have been considered. There is also the possibility of sinks dismembering the valleys themselves. A great many valleys do terminate in sinks in synclinal areas, but this is no explanation of valley desiccation since the streams never did continue farther. Open sinks within main valleys are rare. Small depressions and filled solution pipes occur more frequently, but would not interrupt even small flows. There do appear to be cases of sinks capturing small tributaries, but it should be emphasized that in most cases these are the merest nibblings at the valley network. In some instances, indeed, it is hard to tell if a sink has diverted a small tributary just before its junction with the main valley, or if widening of this main valley has wasted one wall of an elongate depression prior to the complete capture of the depression by the valley. G. Lasserre (i96I) describes so-called dry valleys on the limestone of Guadeloupe and maintains that these are intermittently active and in the process of consuming sink-hole plains. In Barbados this view is supported by the decline in the frequency of depressions above elevations of I20 m, which is coupled with increasing valley densities (Table I). Nevertheless, it appears that karstification is making some small inroads on the valley network especially in those drier areas where valleys exist, but the competiton between sink-hole and valley drainage seems tipped only slightly and recently in favour of the former and seems to be the result of valley desiccation rather than its cause. Tricart (I968) considers that the valleys were cut by surface run-off during periods of more frequent intense rains, probably associated with a higher incidence of tropical cyclones. Figure i shows that valley density is not everywhere the same. If Tricart's views are correct this variability should be related to a variable distribution of the frequency and I6o J. FERMOR

TABLE II Rainfall intensity, slope and valley density

Station Frequencyof annual extreme24 hr Numberof 6 m Valley length (km) rainfall, I953-70 contourscrossed within 0.5 km > Ioo mm > I50 mm > 200 mm of station

Lears 7 2 2 5 2-8 Balls 4 o o0 5 o-9 Searles 4 3 0 2 o1I Seawell 8 2 I 3 0'2 Groves 6 3 2 7 2-4 District B 4 I I 8 2.0 (Police Station) Jordans 5 2 I 8 3-0 Congo Road 8 2 o 3 0-5 Ruby 6 2 I 2 o-7 Edgecumbe 5 3 2 4 o'9 Home Agriculture 7 2 0 2 0'o Claybury 9 4 4 8 4'o Kendal 5 3 2 4 2'0 Lion Castle 7 4 3 8 3'9 Vaucluse 7 4 2 5 2'3 District D 5 3 2 I2 3'2 Apes Hill Io 4 3 8 4'6 Holetown 5 3 3 8 2-9 Lancaster 7 3 2 7 3.I Alleynedale 9 3 2 3 2'5 Warleigh i 5 2 8 3-8 District E 7 4 2 7 I15 Pickerings I 0 4 I II I I Castle Grant Io 4 4 7 3'3 Easy Hall 7 3 3 9 3'7 Colleton 6 3 I 4 0'7

Rank correlationof intensity and valley length = + o-8i Rank correlationof slope and residual valley length = + 0-45 duration of intense rains, and perhaps also to slope variations. The expected correlation might still hold for present weather conditions, even if these bear only a muted resemblance to those of the valley cutting era. Somewhat fragmentary records of the highest daily rainfalls experienced each year exist for numerous stations in Barbados. Only nine stations have complete records from 1953, but further records can be completed in an approximate manner by comparison since certain years were without intense rains. This restriction on the precision of the data enforces a rather crude method of analysis. Some twenty-five stations were ranked for rainfall intensity in the following manner. For the period 1953-70, each station was checked for the number of years in which its daily extreme exceeded 200, 150 and o00 mm. Ranking was effected first by reference to the 200 mm rains and ties were resolved by referring successively to the I50 and o00 mm rains. This rank order was compared with the valley density within a radius of 0-5 km of the station. At Holetown and Groves the circles were shifted upslope and downslope respectively, as the first station is coastal and the second is so close to the crest of the Golden Ridge dome that overland flow would not be expected to have become organized into valleys. A rough measure of slope was also obtained for each area by noting the height range exclusive of the valleys, since Dry valleysof Barbados I6I this will have influenced the balance of run-off and percolation. The data are presented in Table II and a positive correlation appears between valley density and rainfall intensity. If the residual variation in density is ranked and compared with rank slope, a further positive correlation is observed. These results support the view that the valleys were cut following intense rains, with high slopes as an added influence in favour of run-off.

CLIMATIC VARIABILITY AND CLIMATIC CHANGE The present climatic regime of Barbados can be viewed in terms of probability. Any site has storms of increasing intensity with lengthening recurrence intervals. Any valley section has some critical intensity/duration of rain beyond which flows occur. These rela- tionships should be investigated before climatic change is invoked. On 2 October 1970, Barbados suffered rains totalling 500 mm in 24 hours at some stations. Considerable flows occurred in the vast majority of the valleys at this time, the only exception found by the author being the lower section of the Botanical Gardens in Welchman Hall Gully. This valley has been beheaded by scarp retreat and is partly blocked by falls from its near-vertical side walls. The extreme restriction of catchment seems to explain the exception. The records from 1953-70 show nothing comparable to this storm; the nearest approach gives totals from one-half to two-thirds the daily rain at most stations. E. Stoute (1970) compares the 1970 storm to storms in 1949 and I898. Interviews at two sites, each where a road embankment blocked a valley, established that the last flow prior to 1970 was in 1949. As any flows would lead to large impounded lakes this information should be reliable. An attempt was made to use road crossings as direct evidence, on the assumption that if embankments were constructed without culverts this reflects the rarity of flows. Figure 3 illustrates this interplay of physical and human factors. The physical factors influencing the size of flood at any recurrence interval are shown by S. T. Wong (I971) to be the catchment area and average land slope above the gauging site. However, the decision whether or not to construct a culvert involves a perception of risk, and perceptions may differ where- as risks do not. Until recently roads were a parochial responsibility, and the change along highway 4B from St George to St Phillip should be noticed. Furthermore, acceptable risks on country roads may be intolerable in built-up areas, so that the culverts at Black Rock and Wildey may in part reflect land use. Some unculverted embankments showed signs of deposition behind them, while others have drainage wells sunk just above the crossing, perhaps to remedy an earlier miscalculation of flood risks. Despite these complications, two generalizations are ventured. The first is that the frequency of flows in some valleys, and perhaps most valleys in their upper reaches, is very low, probably of the order of once in 20 years. Secondly, many valley sections have far more frequent flows, and a broad division can be made between valleys east and west of the Clairmont anticline. Valleys to the west generally have steeper courses and intense rains are more frequent over a greater proportion of their catchments. The testimony of local observers and the frequency of culverts show that these are ephemeral stream courses. Can Tricart's hypothesis of climatic change be discarded and valley cutting be referred to present conditions ? The evidence from the eastern region discounts such a claim. The absence of culverts across valleys far from their heads, the paucity of flows observed in some valleys, and the lack of marked signs of erosion along some valley floors inspected only three months after I62 J. FERMOR the last of these flows, argues the need for increased storm frequencies and intensities if valley cutting on this scale is to result. But Barbados is too small for a radical climatic change to affect the east and not the west. Hence a consideration of climatic change with respect to the whole valley network is required. Before this is attempted, evidence for the duration of the period of valley cutting is considered. Almost all the valleys that reach the sea do so along the west coast. Here most

FIGURE 3. Perception of flood risks by road builders are usually blocked by small sand bars, but these were breached after the storm of October 1970. I. G. Macintyre (I967) presents evidence suggesting that such sand bars have long characterized this coast. He describes sediments off the west coast, shoreward from the first barrier reef, consisting in part of well rounded and polished quartz grains. He sup- poses that these are derived from a belt of sandy soils mapped by K. C. Vernon and D. M. Dry valleysof Barbados I63 Carroll(I965) I-2 km inland at heights between40 and 75 m. The offshoredeposits lie on a slopingshelf probablyproduced by wave actionduring the slow rise in sea level in the last 8000 years. Streamtransport of sedimentthroughout this periodis indicated. At the otherend of the system,many valleys both east and west of the Clairmontanti- clineare beheadedwith wind gaps over I 5 m in depth.Valleys on the southernflank of the unbreachedGolden Ridge anticlinedo not reachsuch a depth of incisionuntil over i km from the crest. This comparisonignores any differencesin slope, stormfrequencies or the erodibilityof the valley floorsbetween the two areas.Nevertheless, it arguesfor considerable escarpmentrecession following the cutting of the valley headswhich are now decapi- tated. H. Blume (I968) claims greaterrecession during glacial periods when enhanced percolation(increased rain, decreasedevaporation) would accelerateslumping and spring sappingat the Scotlandescarpment. This seemsto placethe cuttingof valleyheads prior to the last glacial. The reconstructionof the climatic regime over this period is possible only on a regionalbasis, and in a sketchyand tentativemanner. Nevertheless some of the evidenceis suggestive,and allowsTricart's brief referenceto be elaborated.Barbados emerged during the Pleistocene,a time of both climaticand eustaticoscillations. Oxygen isotope studies of sedimentarypelagic foraminiferafrom the sea floor of the Caribbeanand its surrounds allow surfacewater temperaturesto be estimated.There is disagreementon the results, however; C. Emiliani (I964) estimatesa swing of 6°C between glacial and interglacial times, while N. Shackleton(I967) limits the swing to less than 2°C, and G. W. Lynts and J. B. Judd (I971) favour an intermediateestimate of 3.60 C. Emiliani'sproposal is for temperaturesof 23°C in the glacialmaxima, and 29°C duringinterglacial and early post- glacialtimes comparedwith 27-28°C today.A fall from29°C to 23°C involvesa 30 per cent declinein saturationmixing ratios and hence loweredlevels of rainfallintensity. Moreover, the formationof tropicalcyclones is inhibitedby surfacewater temperatures below 260C (E. Palmen, 1948). Twenty-four hour rainfall intensities generally increase with height on Barbados today. Loweredsea levels duringglacial advances would probablyincrease rainfall totals but, for the reasonsgiven above,storm intensities may nonetheless have been reduced.The long-termtectonic uplift would not be subjectto such synchronizedconflicting influences. Sea-level temperaturesbeing equal, each successive interglacialwould have an added orographicinfluence in favourof intenserains. It is in these interglacialsand in earlypost- glacialtimes that conditionsappear most favourablefor valley cutting, and evidence has been given above for river erosionduring these periods.

CONCLUSION This paperprovides an explanationof the valley patternon Barbadosby referenceto the initialsurface conditions on the emergingreefs, and firmlysupports surface run-off rather thanbase flow as the agentof erosion.Present run-off appears insufficient to accountfor the valleys, this being most obvious east of the Clairmontanticline. In this easternregion, sink-hole developmentmay well be outpacingvalley downcuttingtoday. Karstificationof the valleys,however, is nowherefar advanced and can be only a partialand secondarycause of their desiccation.Valley density varies widely over the CoralCap. This can be relatedto differencesin the frequencyof intense rainstoday, but it may equallywell reflectparallel spatial variationin past climates brought about by persistent differencesin relief and I64 J. FERMOR aspect. Some areas,notably the ChristChurch dome, are virtuallywithout valleys, so that much of Barbadosmust at the best of times have presentedmarginal conditions for valley incision.In such a settingclimatic change offers a plausibleexplanation for the desiccation of some valleys and the reduction of flows in others. Independentevidence indicates stormierpalaeoclimates but this needs to be more firmlyestablished. The severalhypotheses reviewed have their advocatesin other areas. Harrisonand Jukes-Browneanticipated a whole family of hypothesesrelating dry valleys to lowered water tables followinglowered sea levels. Karstificationof a valley system or desiccation followingclimatic change are likewiseapplicable elsewhere. The causes of valley desiccation may well differin other areas,but the modes of testing commonhypotheses may be transferable.

ACKNOWLEDGEMENTS The writer is indebted to Dr R. J. Small for valuable comments on the paper. He thanks Mr J. Lirious of the World MeteorologicalOffice and Mrs F. Jordon of Queen's College, Barbados,for the informationthey supplied on rainfall intensity and road embankmentsrespectively.

REFERENCES BLUME,H. (I968) 'Zur Problematikdes Schichtstufenreliefsaud den Antillen', Geol. Rundsch.58, 82-97 COLEMAN,A. M. and W. G. V. BALCHIN(I959) 'The origin and development of surface depressions in the Mendip Hills', Proc. Geol.Ass. 70, 29I-309 CUMBERBATCH,E. R. (I967) 'Conservationand reclamationin the Scotland district of Barbados', Ministry of Agri- culturebulletin, No. 49 EMILIANI,C. (I964) 'Paleotemperatureanalysis of the Caribbeancores A254-BR-C and CP-28', Bull. geol. Soc. Am. 75, I29-44 GOREAUX,T. F. (I959) 'The ecology of Jamaican coral; species composition and zonation', Ecology40, 67-90 HARRISON,J. B. and A. J. JUKES-BROWNE(I890) The geologyof Barbados(Bridgetown) LASSERRE,G. (I961) La Guadeloupe:la natureet les hommes(Bordeaux), vol. I. LYNTS, G. W. and J. B. JUDD(i97i)'Late Pleistocene paleotemperaturesat Tongue of the Ocean, Bahamas',Science, N.Y. I7I, I 43-4 MACINTYRE,I. G. (I967) 'Recent sediments off the west coast of Barbados, W.I.', Unpubl. Ph.D. thesis, McGill University MARTIN-KAYE,P. and J. BADCOCK(I962) 'Geological backgroundto soil conservationand land rehabilitationmeasures in Barbados, W.I.', Rep. 3rd Caribbeangeol. Conf. MATTHEWS,R. K. (I968) 'Carbonate diagenesis: equilibration of sedimentary mineralogy of the subaerial environ- ment; coral cap of Barbados, W.I.', . sedim. Petrol. 38, 1110 -19 MESOLELLA,K. J. (i967) 'Zonation of uplifted Pleistocene coral reefs in Barbados, W.I.', Science, N.Y. 156, 638-40 MESOLELLA,K. J., R. K. MATTHEWS,W. S. BROECKERand D. L. THURBER(1968) 'The astronomical theory of climatic change: Barbados data', J. Geol. 77, 250-74 PALMEN,E. (I948) 'On the formation and structure of tropical hurricanes',Geophysica 3, 26-38 PINGITORE,N. E. (I970) 'Diagenesis and porosity in Acropora Palmata, Pleistocene of Barbados, W.I.', J. sedim. Petrol.40, 712-2I PRICE,E. T. (I958) 'Notes on the geography of Barbados', Dept. of Geography, Univ. of California, Berkeley, mimeo SENN,A. (I946) Report of the British Union Oil Co. Ltd., on the geologicalinvestigation of groundwaterresources of Barbados,B.W.I. SCHOMBERGK,R. H. (1848) The history of Barbados(London) SHACKLETON,N. (I967) 'Oxygen isotope analyses and Pleistocene temperaturesreassessed', Nature, Lond. 2I5, 15-I7 STOUTE,E. (I970) 'Rain, rain and more rain-a flashbackto the flood of I949 and the hurricane of I898', Barbados Advocate, i October I970 TRECHMANN,C. T. (1955) 'A new explanation of mountain uplift', British Associationfor the Suppressionof Science (West Hartlepool) TRICART,J. (I968) 'Notes geomorphologiques sur la karstificationen Barbade (Antilles)', Centre de Rechercheset DocumentationCartographique et Geographique,Memoirs et Documents,4, 329-34 VERNON,K. C. and D. M. CARROLL(I965) 'Soil and land use surveys, No. i8: Barbados',Imperial College of Tropi- cal Agriculture, U.W.I., Trinidad Dry valleysof Barbados I65 WONG,S. T. (1971) 'Effect of stream size and land slope on the recurrence of floods in New England', Geogr. Anal. 3, 77-82

RESUME.Les vallees sechesdes Barbades: un examencritique de leur dessinet origine.Le trait caracteristique,en ce qui concerne l'erosion des recifs coralliens des Barbades qui emergent de l'eau, nous est donn6 par un reseau de vallees. Ces vallees sont generalement seches et les problemes majeurs que sont leur origine et leur d6ssechement ulterieur sont d6battus. Un probleme de seconde importance est celui pose par un ensemble de nombreuses vallees qui sont a angle droit aux rives de nos jours et a ceux d'autrefois. Apres avoir envisage de nombreuses hypotheses, nous avons retenu celle de l'elimination de la surface apres de nombreux orages. La possibilite d'une alteration graduelle est discutee et on considere aussi le r6oleque le phenomene karstique a pu jouer. Le dessin des vallees est considere comme etant une consequence et ayant comme particularitesles resultats de depressions primaires sur les recifs.

FIG. i. Structure et dessin de la vallee sur le sommet de corail FIG. 2. Coupe schematique pour mettre en evidence l'hydrologie du sommet de corail (selon A. Senn et P. Martin- Kaye) FIG. 3. Risques d'inondation envisages par les constructeurs de routes

ZUSAMMENFASSUNG.Die trockeneTdler von Barbados: eine kritischeRundschau ihre Gestaltungund ihres Ursprungs. Die grosste Verwitterung an den hervorragendenKorallenriffen in Barbadoszeigt sich an einem Netzwerk von Talern. Diese Taler sind im allgemeinen trocken und das Hauptproblem bei der Debatte war ihr Ursprung und die nachfolgende Austrocknung. Ein zweites Problem waren der Verlauf vieler der Taler, die rechtwinkelige Sektionen zu vergangenen und gegenwartigen Kiistenlinien hatten. Von den verschiedenen das das Ansichten, die besprochen wurden, zieht man jetzt vor, die von dem herunterlaufendenWasser von den Oberflachen bei schweren Sturmen. Die Moglichkeit klimatischerWechsel wurde diskutiertund eine beitragendeRolle fur karstartigesGebiet in Betracht gezogen. Die Talgestaltung mit seinen Eigenheiten beurteilte man als logisches Ergebnis der ersten Senkungen der Riffe.

ABB. i. Struktur und Talgestaltung der Korallenspitze ABB. 2. Diagrammatische Durchschnitte, um die Hydrologie der Korallspitzen zu zeigen (nach A. Senn und P. Martin-Kaye) ABB. 3. Wahrnehmungder Uberschwemmungsgefahrdurch Strassenbauer