Oxygen isotope compositions of phosphate from vertebrate tooth enamel were measured to determine the evolution of tropical sea surface (<∼200 m depth) temperatures in the western Tethys during the Middle-Late Jurassic. On the basis of a high-resolution stratigraphic framework with a 1 Myr time resolution, vertebrate teeth were sampled on Aalenian to Portlandian isochrons over the Anglo-Paris Basin. Asteracanthus sharks and Pycnodontidae teleosteans, identified as sea surface dwellers, have enamel with δ18O values that range from 18.5 to 22.3‰. Thermal variations of tropical surface waters, with amplitudes of a few degrees per few million years, suggest that Middle to Late Jurassic climates were quite variable. Assuming a seawater δ18O value of 0‰ for surface tropical waters in the absence of polar ice caps, temperatures increased from 25 to 29°C from the mid-Bajocian to mid-Bathonian. During the middle to late Bathonian, a strong geographic zonation in isotopic compositions is observed between the eastern and western parts of the basin. High δ18O values of fish tooth enamel (up to 22.3‰) could reflect the arrival of a cold water current from the Arctic during the opening of the North Sea rift. An apparent large drop of temperatures from 28 to 21°C is identified at the Callovian-Oxfordian boundary over no more than ≈2–3 Myr. This cooling is compatible with previous paleobotanical and geochemical studies and can be precisely correlated with the migration of boreal ammonites into the Tethyan domain. Because isotopic sea surface temperatures are probably too low to be compatible with tropical climatic conditions, the δ18O value of seawater could have been >0‰ owing to limited growth of continental ice during the early middle Oxfordian. The resulting sea level fall is estimated to be at least 50 m and is compatible with a global regression stage. The middle Oxfordian thermal minimum is followed by a new warming stage of 3–4°C from the middle to the late Oxfordian.
 Before the Latest Jurassic-Early Cretaceous radiation of carbonate plankton [Roth, 1989], ectotherm vertebrates living in surface waters, defined here as about the top 100 m of the water column, can be considered as potential substitutes of paleorecorders of sea surface temperatures [Picard et al., 1998]. To characterize the living depth of Jurassic fish, their δ18O values were compared with a previously obtained data set of coexisting brachiopods [Picard et al., 1998]. The oxygen isotope compositions of brachiopods record bottom-water temperatures from the shallowest to the deepest depositional environments of the Anglo-Paris Basin. The oxygen isotope compositions of brachiopods appeared to be in good agreement with independent hydrodynamic and sedimentary criteria [Guillocheau, 1991].
 Well documented sedimentological and paleogeographic studies of the 600 km × 800 km Anglo-Paris Basin [Ziegler, 1988; Guillocheau, 1991; Robin et al., 2000] and the high-resolution stratigraphic framework, set up by Garcia et al. , Gaumet et al.  and Garcia and Dromart , may help us to understand changes in the hydrologic budget and oceanic circulation patterns that affected its water masses. This Basin was located at tropical latitudes (20°–30°N) during the Middle and Late Jurassic [Dercourt et al., 1985; Ziegler, 1988] with water depths down to about 200 m [Guillocheau, 1991; Picard et al., 1998]. This epicontinental sea, representing the occidental part of the equatorial Tethyan ocean, was connected to the opening central Atlantic ocean. Since the middle Bathonian, the development of seaways to the north probably enabled efficient communication of surface waters between the Tethyan Basin and boreal realms.
 The Anglo-Paris Basin, with its high-resolution stratigraphic framework, thus constitutes an especially interesting case study to apply oxygen isotope compositions of Middle to Late Jurassic fish tooth enamel to (1) reconstruct tropical surface seawater temperature variations with time, (2) document the existence of paleocurrents and their influence on faunal migration at the few million year timescale, and (3) explore the consequences of tropical temperature variations in terms of sea level variations and the possible presence of a polar ice cap.
2. Depositional Environments
 Depositional environments were dominated by three main carbonate platform domains (Burgundy, Ardennes and Normandy), Normandy being separated from the two others by a marly furrow and bordered by land masses during the Bathonian and the middle and late Oxfordian (Figure 1). The Callovian is characterized by the drowning of relatively high areas by clayey and sandy wedges with a maximum flooding surface in the middle Callovian [Gaumet et al., 1996]. A regressive-transgressive second-order sequence was recorded in the late Bajocian to middle Callovian carbonate series [Guillocheau, 1991; Garcia et al., 1996]. This sequence includes (1) a regressive hemicycle from the upper Bajocian to the lower-middle Bathonian boundary corresponding to an overall infill of the basin by progradation of carbonate platforms until subaerial exposure, and (2) a transgressive hemicycle from the middle Bathonian until the transgressive peak of middle Callovian age (Jason zone). Siliciclastic sands and silts were deposited at the same time in southern England and over the Channel (between England and France). The next regressive-transgressive cycle corresponds to the progradation of siliciclastic (middle Callovian to middle Oxfordian) then carbonate wedges from northwest to southeast.
 The water depth of the fair weather and storm wave limits may fluctuate depending on both coastal morphology and occurrence of submarine topographic breaks. Nonetheless, these hydrodynamic parameters can be used to define four main depositional environments in the Anglo-Paris basin. (1) The protected environment where wave action is limited by topographic barriers. These environments can sometimes correspond to relatively deep locations compared to emersive lagoonal environments. Shallow environments show evidence of limited subaerial exposure such as mud cracks or birds' eyes, but without evaporites (excluding dolomite) or significant siliciclastics. (2) The shoreface above the fair weather wave base forms barriers that were oolitic and skeletal in the Bathonian and Callovian, and were often made up of reefal mounds in the Oxfordian. (3) The upper offshore located between the fair weather and the storm wave bases, and (4) the lower offshore occurs below the storm wave base.
3. Sample Collection
 Eighty nine teeth (including seventy eight enamels and eleven whole teeth) and two bones of Chondrichtyan and Osteichtyan Jurassic fish were selected from shallow to deep marine environments of the Anglo-Paris Basin (Table 1; Figure 1). In addition, eleven bulk teeth and one tooth enamel of reptiles, mainly marine crocodilians, were selected when found associated with fish teeth. Vertebrate remains were sampled along sedimentary isochron profiles [Garcia et al., 1996; Gaumet et al., 1996] with a resolution of about 1 Myr from the Bajocian to the Tithonian, with one Aalenian tooth. Isochrons consist of sedimentary beds defined by monospecific or plurispecific brachiopod associations that are laterally continuous throughout the basin and coincide with maximum platform floodings [Garcia and Dromart, 1997]. These brachiopod shell beds were accurately identified on the basis of one hundred and forty exploratory oil wells that were used to construct high frequency cycle correlations [Garcia and Dromart, 1997]. They were also calibrated with ammonite standard zonation and Jurassic timescale [Gradstein et al., 1994].
Table 1. Oxygen Isotope Compositions of Aalenian to Portlandian Vertebrate Phosphates Along With Taxon Determination, Sample Locations, Ages, and Environmentsa
Note that Microdon, Mesodon, Pycnodus genera are all Pycnodontidae. Stratigraphic ages for isochrons were reported on the ammonite biostratigraphic scale and on the absolute scale proposed by Gradstein et al. . P2O5 wt % of biogenic apatites were calculated considering a mean value of 80 ± 3% for the silver phosphate chemical yield obtained with NIST NBS120c. Depositional environments: P, protected; S, = shoreface; O, offshore; U, upper; L, lower; PZ, photic zone; BPZ, below photic zone; p, proximal; d, distal. Samples: B, bone; T, tooth; b, bulk; d, dentine; e, enamelloïd.
 Chondrichtyan fish are sharks represented by four main genera that are Asteracanthus (crushing fish), Notidanus, Sphenodus and Odontaspis (tearing fish). Asteracanthus feeds on shells and is characterized by platy and massive teeth with an enamel structure [Cuny, 1998]. Jurassic Osteichtyan fish are represented by durophageous fish, Pycnodontidae with tabular teeth, except one genus, Sphaerodus, having hemispherical teeth, and Gyrodontidae (Gyrodus). Ecology and, especially, living depths of Jurassic fish, cannot simply be determined from paleontological or paleoenvironmental criteria. Depositional environments, constituting a condensed image of fauna and flora activity occurring from surface to bottom waters, also cannot provide information on living depths in the water column. Previous studies, however, indicate that pycnodont fish have been mostly marine, and only in some cases estuarine to freshwater dwellers [Martill, 1990; Nursall, 1996; Poyato-Ariza et al., 1998]. Pycnodonts must have been an important component of tropical and subtropical coral reef communities, analogous to balistid fish in modern oceans [Nursall, 1996]. Asteracanthus sharks are considered as being essentially marine, but their ecology is poorly known [Cappetta, 1987]. Other sharks, with lamniform teeth, were probably piscivorous [Cappetta, 1987; Martill, 1990].
 Enamel is more resistant than dentine or bone to post-depositional oxygen isotope exchange [e.g., Sharp et al., 2000] because of its high degree of mineralization, therefore this phosphatic tissue was preferentially selected and separated mechanically with a microdrill under a binocular (Table 1). In the case of the smallest teeth, a few bulk isotopic analyses were performed because it was not possible to separate enough enamel.
 Similarly, carbonate ions that are present in the apatite structure are supposed to be less resistant than phosphate ions to oxygen isotope exchange during diagenetic alteration [Iacumin et al., 1996; McArthur and Herczeg, 1990]. To test the state of preservation of Jurassic teeth, we have measured the oxygen isotope compositions of both phosphate and carbonate in fifteen fish and six reptile teeth.
4. Analytical Methods
 Carbonate apatite samples were first treated with a 3% sodium hypochlorite solution for 4 hours to remove organic matter and rinsed three times with distilled water. Then they were treated with a 1M acetic acid-acetate buffer for two days to remove exogenous carbonates according to the method presented by Bocherens et al. . Oxygen and carbon isotope ratios from the carbonate in substitution in the apatite were determined using the phosphoric acid method [McCrea, 1950] with a reaction time of 48 hours at 30°C. Reproducibility for carbon and oxygen isotope measurements in carbonates were ±0.05‰ and ±0.1‰, respectively.
 Phosphate radicals were isolated from biogenic apatites as Ag3PO4 crystals using the method of Crowson et al.  modified by Lécuyer et al. . Silver phosphate was reacted with graphite at 1100°C to produce pure CO2 [Lécuyer et al., 1998] (modified after O'Neil et al. ). The quality of the reaction, marked by the absence of CO production, was monitored by the carbon isotope composition of CO2 derived from the graphite (Figure 2a). Carbon dioxide was analyzed with a VG Prism™ mass spectrometer. All data are quoted in the δ notation relatively to SMOW, via calibration on NBS18 and NBS19 international standards. Several tooth enamels, including all those with extreme δ values were duplicated. Repeated analyses of the phosphorite NBS120c gave a mean δ18O value of 21.68 ± 0.19‰ (Figure 2b). The CO2 gave a mean δ13C value of −23.51 ± 0.11‰ (Figure 2c), similar to that of graphite analyzed by the CuO method (δ13C = −23.5 ± 0.05‰). The P2O5 contents of samples were calculated from their chemical phosphate yields using the average yield of 80 ± 3% obtained for the NBS120c phosphorite that contains 33.33% of P2O5 (Table 1).
 Diagenetic alteration of fossil apatite was detected in eleven samples during the chemical procedures using the following criteria: (1) recovery of silver phosphate down to 60% (Table 1), because of the replacement of phosphates by carbonates, silica or metal oxides (samples A1-L, 93355, L5b, 93471), (2) solutions that turned from red to yellow during neutralization with KOH (O1, N3, D22, 93355, D28), revealing the presence of metal oxides (Fe, Mn, Al), and (3) both low δ18O values of PO43− and δ13C values of graphite (Figure 2a) for samples L11, 92154, D8 and D28, indicating the presence of exotic organic molecules that were not totally eliminated during the wet chemical procedures [Lécuyer, 2003].
 Oxygen isotope compositions of phosphate from Jurassic fish and reptiles are reported in Table 1. Analyses of associated dentine and enamel from two teeth of Chondrichtyan lamnidae (samples L8 and L1) gave chemical phosphate yields that are about 10% lower for the dentine than for the enamel. The dentine has δ18O values from 1.7 to 2.8‰ lower than the enamel (Table 1), which indicates strong isotopic disequilibria between the two coexisting phosphatic tissues. Similar observations were made by Sharp et al.  and Puceat et al. . Massive enamel of Asteracanthus and pycnodonts have high mean P2O5 percentages with relatively low variance (36.7 ± 3.6% and 39.3 ± 3.4%, respectively), bracketing those obtained on enamels from tearing teeth (38.2 ± 6.8%; n = 13) that have a microstructure close to that of modern lamnidae. By contrast, bulk fish teeth and bones have variable and low P2O5 contents (31.6 ± 6.9%) and constitute a fossil material that commonly suffered chemical and isotopic alterations.
 The range and mean δ18O values of all well-preserved vertebrate taxa (where no indicators of diagenesis were observed) are reported in Table 2. Similar ranges and mean δ18O values were obtained with Asteracanthus and Pycnodontidae enamels from the Bathonian-Callovian (19.5 ± 0.5 and 19.6 ± 0.7‰) and Oxfordian-Tithonian (20.1 ± 0.6 and 20.7 ± 0.5‰) periods. Mean δ18O values for other vertebrates like Notidanus (20 ± 0.7‰) and Odontaspis (20 ± 0.7‰) sharks, though based on a very limited number of samples, have similar values to those documented for Asteracanthus and Pycnodontidae. Mean δ18O values for Chondrichtyan Sphenodus (20.8 ± 0.2‰) and Osteichtyan Sphaerodus (21.2 ± 0.6‰) are higher than those for Asteracanthus and Pycnodontidae sensu stricto, but they are mainly restricted to the Late Jurassic, a period characterized by higher δ18O values than those documented throughout the Middle Jurassic. The more positive δ18O values for Sphenodus and Sphaerodus may indicate that they lived in deeper waters and do not record surface temperatures unlike Asteracanthus seems to do (Figure 3). However, we note that Pycnodontidae have δ18O values indistinguishable from those of Asteracanthus during the Bathonian-Callovian and they constitute the most abundant fossil fauna analyzed during the Oxfordian-Tithonian period (Table 2). Therefore the high δ18O values of Oxfordian fish teeth can be considered as also recording the temperature of the upper part of the water column.
Table 2. Statistics of Mean δ18O Values Calculated for Dental Apatite From Asteracanthus, Pycnodont, and Sphenodus for the Bathonian-Callovian and Oxfordian-Tithonian Ontervals
Asteracanthus, n = 19
Pycnodont, n = 7
Sphenodus, n = 1
Asteracanthus, n = 7
Pycnodont, n = 11
Sphenodus, n = 4
 Mean δ18O values of middle Bathonian to late Callovian (168–160 Myr) Asteracanthus have been calculated for the various depositional paleoenvironments documented in the eastern part of the Basin (Figure 3). Note the uniformity of the δ18O values of Asteracanthus teeth (from 19.2 ± 0.2 to 19.7 ± 0.6‰) that are preserved in inner to outer depositional environments. Middle to late Bathonian samples from the western and northwestern part of the basin (Boulonnais, Normandy, and Poitou regions) have δ18O values that are on average 1.5‰ higher than their southeastern counterparts (Table 1). Samples from the central region (Ja1, Bi, SERA, Table 1) have intermediate oxygen isotope ratios, confirming the existence of an isotopic gradient across the whole basin.
 The Bajocian to the Tithonian fish tooth samples from the eastern part of the basin show δ18O variations over 20 Myr (Figure 4). These isotopic variations define an envelope from 0.5 to 1.5‰ wide for the various isochrons. This isotopic range is comparable to the one measured by Vennemann et al.  on recent fish teeth, which is typically between 0.6 and 1.1‰ for different teeth from a single shark. This naturally occurring variation in oxygen isotope compositions recorded in fish teeth may reflect the diversity of the living environments which includes: (1) thermal gradients within the water column; (2) seasonal thermal variations since the growth of a fish tooth represents less than a year (several weeks to several months depending on species), but they should remain limited under tropical latitudes; (3) coastline proximity and the influence of freshwater inputs; and (4) the possible occurrence of surficial oceanic currents.
 Three major stages may be distinguished: (1) a period of decreasing δ18O values from 20 to 19‰ for the Bajocian to Bathonian interval, (2) a rapid increase of δ18O values up to 21‰ initiated during the late Callovian and reaching a maximum during the early-middle Oxfordian, and (3) another stage of decreasing δ18O values, similar in amplitude to stage 1, starting in the late Oxfordian to reach values of 20‰ around the Oxfordian-Kimmeridgian boundary.
 Oxygen isotope compositions of carbonates from Jurassic fish teeth range from 24.5 to 30.0‰ whilst those of reptiles range from 23.6 to 30.4‰ (Table 3). For fish teeth, differences between the δ18O values measured in apatite carbonate and its associated phosphate oxygen (Δc-p) range from 7.0 to 9.5‰, except for two samples; a whole tooth (L12; Δc-p = 5.0‰) and one altered sample (93355; Δc-p = 5.7‰). These values are within or close to the known fractionations (Figure 5) between apatite carbonate and phosphate oxygen [Longinelli and Nuti, 1973; O'Neil et al., 1969; Iacumin et al., 1996].
Table 3. Oxygen Isotope Compositions and Mass Fractions of Structural Carbonate From a Selection of Fish and Reptile Apatites
 The amount of carbonate substituted in the apatite structure of Jurassic fish tooth enamel ranges from 3.9 to 5%, except for samples D21P and 93357 (Table 3). These values match those of 4–5% that were measured in present-day vertebrate teeth [LeGeros and LeGeros, 1984; Michel et al., 1995]. The bulk tooth L12 has a low CO32− content of 2.9‰ and a low Δc-p of +5 both diagnostics of a diagenetic alteration of the carbonate component of the apatite. The reptile bulk teeth have CO32− contents that range from 3.8 to 7%.
6.1. Diagenetic Perturbations
 We conclude from our observations during the chemical procedures that seven vertebrate whole teeth (i.e., <8% of the enamel samples) and the two fish bone fragments have been chemically and isotopically modified by diagenetic processes. Only one enamel from a Sphenodus tooth (93355) and one enamel from a massive tooth (93471) are suspect on the basis of the selected criteria. Some of these phosphate samples are depleted in 18O relative to coexisting or associated enamel (D28; D22; A1-L; Table 1). Moreover, only two fish teeth (enamel 93355 and whole tooth L12) plot outside the domain of apparent oxygen isotopic equilibrium between phosphate and carbonate, limited by previously proposed fractionation equations (Figure 5). This suggests that most enamels from Jurassic teeth have conserved their primary isotopic signatures.
 On the basis of oxygen isotope analyses of modern teeth, dentines are depleted in 18O by up to 1.7‰ relative to associated enamel [Vennemann et al., 2001]. For comparison, Jurassic dentines from two lamniform teeth show δ18O values up to 2.8‰ lower than coexisting enamel, suggesting that these dentines have been significantly altered. Consequently, we do not consider bulk samples for paleoenvironmental studies even if some of them have δ18O values in the range of contemporaneous fish enamel (W22-b, L126-b, L12; Table 1). The greater resistance of Jurassic enamels to diagenetic processes than either dentine or bone supports earlier studies realized on Mesozoic vertebrate teeth [Kolodny et al., 1996; Sharp et al., 2000].
6.2. Vertebrate δ18O Values and Sea Surface Temperatures
 Oxygen isotope analysis of apatite from modern selachians has shown that isotopic temperatures reflect the average temperature of the seawater layer in which the fish lives [Picard et al., 1998]. Although the living depths of Jurassic vertebrate taxa are not known, estimates can be made. Marine bottom temperatures can be derived from the δ18O values of brachiopods giving a lower limit for the water column [Picard et al., 1998]. Isotopic temperatures inferred from the δ18O values of brachiopods sampled from known different depositional environments can be compared with the temperatures derived from the associated vertebrate δ18O values. Note that if the vertebrates and brachiopods were living at the same time and place (associated fossils), they were not necessarily present in the same water layer.
 Mean δ18O values, calculated for both middle Bathonian to late Callovian Asteracanthus sharks from the eastern part of the Anglo-Paris basin, are reported in Figure 3 as a function of their depositional environment. The isotopic values, ranging from 19.2 ± 0.2 to 19.7 ± 0.6‰, remain relatively constant independent of the deepening of the marine basin from the shoreface to lower offshore environments at about 200 m depth. This pattern of isotopic distribution strongly contrasts with the δ18O values of coexisting brachiopods that increase with depth and record thermal differences of up to 12°C between surface and lower offshore bottom waters [Picard et al., 1998]. These observations suggest that the oxygen isotope compositions of Asteracanthus sharks record temperatures close to sea surface temperatures.
 Isotopic temperatures are calculated from the fractionation equation determined by Kolodny et al.  on the basis of various fresh and seawater fish species. A similar equation (Figure 6) is deduced from present-day shark and water isotopic data published in the work of Picard et al. .
 Assuming a δ18O value of 0‰ compatible with both the absence of continental ice caps and the high-salinity of low-latitude surface sea waters, isotopic temperatures of 27 to 29°C are calculated, and are comparable to surface temperatures measured in modern tropical platform waters [Adlis et al., 1988]. Similar δ 18O values, and therefore isotopic temperatures, of contemporaneous sharks (Asteracanthus, Notidanus, Odontaspis and Sphenodus) and Pycnodontidae (Tables 1 and 2) strongly suggest that these fish were principally restricted to surface waters. They are used to reconstitute sea surface temperatures.
6.3. Basin-Scale Oxygen Isotope Variations
 Three different interpretations may be formulated to explain the δ18O spread of 3‰ identified throughout the Anglo-Paris basin from the Middle to Late Jurassic. In terms of salinity changes, it would correspond to variations of up to 7.5 ppt over the basin, assuming a 2.5 ppt salinity increase for 1‰ increase of δ18O value of seawater under subtropical latitudes [Craig and Gordon, 1965]. Such salinity gradients seem unrealistically large without any known physical barrier isolating the western part of the basin from the east. Additionally, no evidence for excessive evaporation over precipitation is provided by the sedimentary record in the western domain of the Anglo-Paris basin. The highest δ18O fish values are observed in the Boulonnais (Samples D35, D71, P12; Table 1; Figures 1 and 4) where estimated water depths are 50–200 m for the upper to lower offshore deposits [Vidier et al., 1995; Garcia et al., 1996].
 An apparent variation in the isotopic temperatures of 15°C is observed over a 3° change of paleolatitudes between the Boulonnais and the Burgundy platform (Figures 4 and 7). Such a thermal gradient of 5°C/°Lat. is unrealistically too high when compared to modern thermal gradients that are generally lower than 1°C/°Lat [Savin, 1977] (see also http://ingrid.ldeo.columbia.edu)
 For the middle-late Bathonian, we observe that the δ18O fish values progressively decrease along a north-south direction (Figure 7). This distribution may be due to the influx of a cool current that progressively becomes warmer southward. This water mass of boreal origin probably had a δ18O value lower than that more typical of tropical waters, taken here to be 0‰, consequently, calculated temperatures must be considered as maximal values. During the late Bathonian, isotopic temperatures may have been close to (1) 19°C in the northern part of the basin, in the Boulonnais and Normandy, (2) 24°C in the central part and, (3) 28–31°C in the southeast. Such temperature gradients of 10°C over a few hundred kilometers have modern analogues such as the area between the east U.S. coast and Bermuda where the Gulf Stream heads northeast from the Sargasso Sea [Brown et al., 1989]. The proposed cool current may have resulted from the opening of the North Sea rift that was initiated during the middle-late Bathonian [Ziegler, 1988]. Cool waters, formed at high paleolatitudes in the Boreal Arctic domain, could have flowed southward to mix progressively with the western Tethys. The connection between the basins was probably limited during the middle Bathonian; during this period, corridors were filled with protected sandy environments and deltaic shales, both characterizing water depths less than 30 m [Ziegler, 1988]. The late-Bathonian Eudesia isochron corresponds to the most extensive flooded surface and coincides with the first brief excursion of boreal ammonites into the Anglo-Paris Basin [Marchand and Thierry, 1974, 1997; Cariou et al., 1985], confirming a real communication between the Tethyan and Boreal basins and the influx of cold waters.
6.4. Middle-Late Jurassic Evolution of Tropical SST in the Anglo-Paris Basin
 At tropical latitudes, the evaporation/precipitation ratio tends to be higher than 1 resulting in increased salinity and δ18O values of surface waters relative to mean ocean water. 18O-enrichment of about 1‰ relative to SMOW has been documented for present-day tropical surface waters [e.g., Craig and Gordon, 1965; Carpenter and Lohmann, 1995; Gonzalez and Lohmann, 1985]. Therefore isotopic temperatures of surface waters have been derived from the δ18O record of biogenic phosphates through the Middle-Late Jurassic (≈20 Myr), assuming a δ18O seawater of 0‰. This hypothesis has been retained considering both the tropical paleolatitudes of western Europe and the absence of polar ice caps as a first approximation. Sea surface temperatures (SST) inferred from surface dwellers (Asteracanthus and Pycnodontidae) increased from 25°C to 30°C on average over ≈7 Myr from late Bajocian to late Bathonian then decreased down to 20°C in the middle Oxfordian boundary in ≈8 Myr (Figure 8). This thermal minimum is followed by a warming of 5°C of surface waters during the late Oxfordian and early Kimmeridgian.
 The apparent cooling of surface waters that was initiated during the late Callovian Lamberti zone and that attained a maximum during the early-middle Oxfordian relates with the colonization of the western Tethys by the boreal ammonites of the Cardioceratidae family [Marchand and Thierry, 1974, 1997; Cariou et al., 1985]. Such a cooling episode may have corresponded to the southward propagation of a boreal oceanic current and/or to a global climate change. There is no available evidence for the existence of a cold oceanic current throughout the major part of the Oxfordian. Studies of the distribution of the wood genus Xenoxylon and palynomorphs argue for a temperate and wet climate during the early Oxfordian in western Europe [Philippe and Thévenard, 1996; Abbink et al., 2001]. All these data support the existence of a cooling episode that affected both marine and terrestrial environments of the western Tethyan domain.
 Perturbations of the carbon cycle due to volcanism or changes in the rates of burying organic matter could have induced temperature oscillations at the timescale of a few million years. According to Dromart et al. , the warm Bathonian-Callovian phase could be related to large CO2 fluxes released by contemporaneous arc-related volcanism in Patagonia at 164.1 Myr [Féraud et al., 1999] and South China at 164.6 Myr [Davis et al., 1997]. Organic rich-layers deposited during the middle Callovian could have sequestered carbon and lowered atmospheric CO2 levels, thus initiating the global cooling stage at the Callovian-Oxfordian boundary.
 With the hypothesis of a constant isotopic composition of seawater of 0‰, we calculated tropical surface temperatures ranging from 25°C to 30°C for the eastern basin during the Bajocian throughout the Callovian. This temperature range is slightly higher than modern thermal variations (19°C to 27°C) measured in shallow waters (≈50 m) under similar latitudes (20–30°) in analogous carbonate platforms [Adlis et al., 1988] (Ocean Atlas data, Scripps Institute of Oceanography). By contrast, middle Oxfordian isotopic temperatures of 20°C are rather low considering the presence of coral reefs in the Swiss Jura [e.g., Gygi and Persoz, 1986]. The apparent low sea surface temperatures during the Oxfordian could result from an erroneous estimate of the oxygen isotope composition of ambient seawater. A cooling stage could have been accompanied by the growth of a polar continental ice cap. The formation of a polar ice cap, even limited in size, could have led to a positive shift in the δ18O of seawater in the Oxfordian relative to the Callovian. This change would be recorded in the δ18O values of fish tooth enamel. The possible ice volume component of the δ18O fish record cannot be simply determined without a knowledge concerning the extent of glacial sediments, the relative sea-level change, and the δ18O value of Jurassic continental ice. However, basic calculations can provide first-order estimates. During Quaternary peak glacial times, most of the tropical surface temperatures cooled no more than 2–3°C with local exceptions around 5°C [e.g., Broecker, 1992]. Relative to the observed isotopic shift of 1.5‰ observed during early-middle Oxfordian, a δ18O variation of 0.5‰ could be attributed to the ice volume component, assuming similar thermal variations for Jurassic and modern tropical surface waters. Considering an oxygen isotope composition for continental ice in the range −40 to −30‰, the mass of continental ice would range between 2 × 1019 kg and 3 × 1019 kg, and a fall in sea-level of about 50 m. Dromart et al.  recently suggested the formation of polar ice to account for large-scale sea-level change, documented worldwide and of eustatic origin, with a maximum in the late Callovian. The geological evidence are (1) the occurrence of glendonites in the sediments of northern Asia, (2) deposition of deep-sea fans in Oman, (3) seaward migration of shorelines (North Sea), (4) valley incision in south Portugal and Saudi Arabia, and (5) subaerial exposures of marine sediments in Portugal, Israel, Alaska and Argentina. As noted by Dromart et al. , the abrupt and global marine regression of the late Callovian follows the decrease of sea surface temperatures along with a southward migration of boreal ammonites. Therefore a glacial event is presented as a plausible scenario that could conciliate isotopic, paleontological and geological data.
 Similarly to biocarbonate systems, interpretation of the bioapatite δ18O value in terms of temperature depends critically on the choice of the δ18O seawater value. Our proposed temperatures may have to be revised, by probably less than three or four degrees, if independent evidence is found for a value other than 0‰ for δ18O seawater. The presence or absence of continental polar ice caps is a crucial element in this discussion. From a purely isotopic point of view for the period from the Bajocian through to the Thithonian (Figure 8), a tentative case can only be made for the presence of a significant continental ice reservoir during the Oxfordian, the Thithonian and perhaps the Bajocian-Bathonian, the three stages with relatively high δ18O apatite values. These propositions are supported, in part, by the documented periods of major regression [Guillocheau, 1991; Garcia et al., 1996] and occurrence of possible glacial phenomena in other part of the world [Price, 2000].
 Oxygen isotope compositions of marine vertebrate tooth enamel, that were not diagenetically altered, vary from 18.5 to 22.5‰ for the Middle-Late Jurassic of the western Tethyan basin. These data are interpreted as evidence for thermal variations of sea surface (≤100 m) temperatures over a few Myr scale. A drop of isotopic temperatures (about 3°C to 5°C) occurred around the Callovian-Oxfordian boundary. This cooling of surface waters is associated with the southward migration of boreal ammonites and the development of temperate flora in the Anglo-Paris Basin. This cooler episode precedes a thermal increase that began during the late Oxfordian until the Kimmeridgian. Apparent isotopic temperatures not exceeding 20°C are low for tropical latitudes even when assuming an isotopic composition of seawater of 0‰ compatible with 18O-enriched low-latitude sea surface waters and the absence of permanent well-developed polar ice caps. Therefore we cannot exclude the existence of an Oxfordian glaciation event that could explain the contemporaneous southward migration of boreal ammonites and the global sea-level fall.
 Strong oxygen isotope gradients (3‰) over 500 km from the northwest to the southeast of the basin suggest the development of a cool oceanic current during the middle-late Bathonian that could be related to the opening of the North Sea rift. According to paleontological and paleobotanical studies, we suggest that the late Callovian-early Oxfordian thermal decrease could be the expression of a global event rather than a more local geographic effect of these cool oceanic currents. In the absence of isotopic records provided by foraminifera, those available from fish phosphatic remains reveal that Jurassic climates were not so consistently warm and equable as previously thought.
 Sampling would not have been possible without the help of A. Prieur (Université Claude Bernard de Lyon), B. Laurin, J.-H. Delance (Université de Bourgogne), M. Weidmann (Université de Lausanne), A. Boullier (Université de Besançon), A. Léna, G. Gallio (Muséum de la Citadelle, Besançon), J. P. Goujon (MNHN, Paris), Muséum de Bâle, J. M. Mazin (Université de Poitiers), and G. Breton (Muséum du Havre), who provided some of the vertebrate samples. We also thank M. Emery for her help in sample analyses. The authors thank the reviewers David Dettman and Christina Hartman, who helped us to improve the scientific content of this manuscript. This work was supported by the CNRS through the programs “Intérieur de la Terre” (Comportement dynamique des plates-formes carbonatées) and “ECLIPSE.”