Un aquifère karstique s'est développé dans des carbonates triasiques (Fig. 1). La grotte est située à une profondeur de 4 à 90 m [2]. Au toit des galeries, les points de suintement sont rares et à faible débit.

Soixante-dix-neuf fractions de pluie ont été collectées (Fig. 1) entre 1995 et 1999. Le débit des eaux de suintement (stalactite située sous 30 m de marbres) est mesuré par un pluviographe ; 43 échantillons mensuels ont été pris.

Pluviométrie et débit de suintement

Les 490 mm de précipitations annuelles sont concentrés entre octobre et février (Fig. 2). Deux épisodes majeurs ont été enregistrés (Fig. 3) : 428 mm (12/95–01/96) et 481 mm (12/96–01/97). Au point de suintement du Rincón del Órgano, considéré comme représentatif de la grotte, le débit moyen est de 89 cm³ j⁻¹ [28]. La relation chronique de pluie–débit du suintement montre l'existence de deux modalités d'infiltration : l'une lente, au long de l'année, et l'autre rapide, qui ne se produit que de manière discontinue. L'évolution du débit de suintement (Fig. 3), avec des maxima d'été–automne et des minima d'hiver–printemps, reflète l'existence d'une circulation lente. L'utilisation des teneurs en K⁺ naturel comme traceur de l'infiltration a permis d'évaluer ce temps de transit [5,14] à 5–8 mois, (selon la hauteur et l'intensité de la pluie et selon le transit dans les fissures). Ce déphasage corrobore les résultats d'une analyse corrélatoire et spectrale effectuée sur les chroniques journalières précipitations/débit [4] : la zone non saturée a un temps de régulation de 85 jours, le déphasage entrée/sortie est de 6 mois pour des séquences à périodicité annuelle (Fig. 4). Pendant les périodes de recharge importante (début 1997), l'infiltration rapide dépend du montant de la recharge et de l'eau stockée dans l'épikarst avant l'épisode. Le fonctionnement est moins inertiel en année humide qu'en année sèche.

Contenu isotopique de l'eau de pluie et de suintement

Eau de pluie

Le diagramme δH2–δO18 (Fig. 5) montre que la pluie est alignée au voisinage de la droite météorique mondiale (LMM) [18], sauf certains échantillons évaporés (pente 5,5, A sur la Fig. 5), qui n'engendrent pas d'infiltration (selon calcul par bilan de Thornthwaite avec RFU = 25 mm) [28]. L' « excès » en deutérium (−4,8‰ à 20,8‰) démontre la prédominance atlantique des précipitations [3,19]. Les pluies des mois chauds (Fig. 6) ont un δO18⩾−3‰ (faible hauteur et température élevée) [20,32,34]. En mai (grande lame d'eau, Fig. 6), les valeurs de δO18 de la pluie sont ⩽−4‰. Il en est de même pour les précipitations des mois froids, appauvries (δO18⩽−4‰).

Eau de suintement de la grotte

Les échantillons s'alignent sur la LMM (origine météorique) [13,14] ; il apparaît des échantillons évaporés (débits <19cm3j−1), qui s'ajustent sur une pente 5,5 (B sur la Fig. 5). La composition du suintement est plus homogène que celle de la pluie. Les eaux les moins négatives sont observées pendant l'hiver et le printemps, les plus négatives pendant l'été et l'automne (Fig. 7). La teneur moyenne du suintement pondéré par son débit (−4,4‰) est moins appauvrie que celle de la pluie efficace pondérée par sa hauteur (−5,7‰, Tableau 1). Cet enrichissement est attribuable à l'évaporation (sol, épikarst et intérieur de la grotte), surtout pour les débits très faibles (temps de contact prolongé avec l'atmosphère de la cavité).

Flux d'eau à travers la zone non saturée

Les pluies d'octobre à février engendrent l'essentiel de la recharge. Sur la Fig. 8, le ¹⁸O révèle un flux d'entrée en novembre–décembre. Les maxima de flux de sortie dans la cavité s'observent en août–septembre, et occasionnellement à des périodes (décembre–mars) associées à des épisodes pluvieux exceptionnels (fin 1996–début 1997). La relation entrée/sortie du δO18 montre un déphasage de 8 mois (1 sur la Fig. 8). Ce temps nécessaire à traverser 30 m de zone non saturée par un écoulement de type piston suppose une matrice à faible teneur en eau, résultat cohérent avec la teneur en eau mesurée de 0,37%, moyenne de 27 mesures sur échantillons de marbre prélevés dans une carrière située au voisinage. Cependant, lorsqu'une recharge significative a lieu, le flux de sortie répond avec un déphasage n'excédant pas 1 à 3 mois (2 sur la Fig. 8).

La grotte de Nerja est située dans la zone non saturée d'un karst [2], épaisse de 4 à 90 m. L'étude des isotopes stables a démontré que les pluies hivernales, qui engendrent la recharge, sont plus pauvres en isotopes lourds, alors que les pluies estivales, moins abondantes, sont plus enrichies. À l'intérieur de la grotte, le débit des points de suintement est plus fort pendant les mois d'été–automne que pendant ceux d'hiver–printemps, à l'exception des pluies très intenses. Les teneurs de l'eau de suintement sont appauvries pendant les mois d'été–automne, et sont plus riches en hiver–printemps (déphasage entre l'entrée et les suintements). Il existe une différence significative entre la teneur isotopique moyenne de la pluie et celle des eaux de suintement, due à l'existence de processus évaporatoires (épikarst, zone non saturée et intérieur de la cavité). Il existe deux modalités d'infiltration, l'une lente et l'autre rapide. L'infiltration lente, qui prédomine toute l'année, se produit dans les petites fissures de l'épikarst et de la zone non saturée, ce qui permet accumulation de l'eau, existence de processus d'évaporation, et homogénéisation des flux d'infiltration. L'infiltration rapide se produit de manière localisée dans le temps, en réponse à des précipitations importantes. La comparaison des maxima de flux d'entrée et de sortie en δO18 démontre que la circulation lente a un temps de transit de 8 mois, la circulation rapide de 1 à 3 mois. L'approche isotopique valide les résultats de l'hydrodynamique et l'hydrogéochimie. Ces résultats rappellent l'existence d'un réservoir permanent dans la zone dite « non saturée » d'un système karstique, avec un temps de séjour moyen plurimensuel, qui contribue au débit des sources karstiques.

The mean precipitation in the area is 490 mm yr⁻¹, with a distinct wet season between October and February and a dry season during the summer months (Fig. 2).

Two periods of significant rainfall occurred during the study period (Fig. 3); the first of these was during December 1995 and January 1996, with 428 mm of precipitation, thus ending a long period of drought in the region. The second episode of considerable rainfall took place during December 1996 and January 1997, with a total precipitation of 481 mm. From this date until the end of the study period, there has been a pattern of declining precipitation.

There are few drip points in the Nerja cave and the drip water flow is, in general, very low. At a point considered representative of the cave (Rincón del Órgano), the mean drip water flow has been calculated as amounting to 89 cm³ d⁻¹ [28]. The thickness of the unsaturated zone situated above this dripping point is about 30 m. The relation between evolution of rainfall and drip discharge versus time displays two infiltration modalities through the epikarst and the unsaturated zone: a slow infiltration, which predominates along the year, and a rapid infiltration, which occurs during short time periods.

Taking into account the yearly distribution of rainfall, the general evolution of the drip water rate (Fig. 3), with maximum values during summer and autumn, and minimum ones during winter and spring, reveals the existence of a slow rainwater flow from its input within the aquifer until its output at the dripping points of the cave. Using natural K⁺ contents as a tracer of seepage water enabled to compute approximately this transit time [5,14]: 5 to 8 months, depending on the value and intensity of rain, which fell before, during and after the K-rich rainfall, and from the pathway that is followed through the fissures of the marble. This seasonal outphasing is corroborated by results of a Correlatory and Spectral Analysis performed with daily precipitation and discharge data, corresponding to five hydrologic years [4]. These demonstrate that the unsaturated zone of the Nerja cave displays a very inertial behaviour; it takes much time to respond to a rain input, and has a high modulating power of the input. The memory effect is very high (about half a year), the regulation time is long (85 days), the spectral band is narrow, with a cut-off frequency of 0.10 (10 days) and the input/output outphasing is about 6 months for sequences of yearly periodicity (Fig. 4). The quantity of water that is stored within the epikarst before precipitations plays a prominent role in the hydrodynamic functioning of the system, which displays a less inertial behaviour during humid years than during dry ones.

During high-recharge periods, both in terms of quantity and of intensity (e.g., at the beginning of 1997), we can observe the existence of a rapid infiltration, which is demonstrated by a clear and rapid increase of the drip flow rate within the cave, whose magnitude depends on the amount of the recharge, but also on the water volume that is stored within the cave epikarst, before the rainfall.

The maximum values for drip water flow are recorded during summer and autumn, and minimum values correspond to winter and spring (Fig. 3). This observation confirms the slow circulation of rainwater through the unsaturated zone, thus producing a seasonal lag from when the water enters the aquifer until it appears at the drip points. During the significant recharge periods (e.g., early 1997), the increase in drip water flow occurs sooner; the magnitude of this increase depends on the quantity of the recharge and on the quantity of water stored in the epikarst of the cave before the precipitation occurs.

Table 1 shows the results of the isotopic contents obtained from the analysed samples. Results are expressed as parts per thousand of the internationally accepted standard deviation [18]. The analytical error was ±0.1‰ for the δO18 and ±1‰ for the δH2.

The δO18 of the rainwater ranged from −10.4 to +1.2‰, with a mean value of −4.9‰, while the δH2 ranged from −69.4 to +13.7‰, with a mean value of −29.6‰. In the drip water, the measured δO18 was between −5.1 and −0.3‰, with a mean value of −3.8‰, while the δH2 ranged from −32.9 to +3.0‰, with a mean value of −22.1‰. The variation coefficient of the isotopic composition of drip water was lower than that observed in the rainwater samples.

By plotting the results obtained on a δO18–δH2 graph, we see that the rainwater samples analysed are aligned according to the world meteoric water line (LMM) established by Craig [18], except several clearly evaporated samples, which fit a straight line with a slope of 5.5 (A in Fig. 5), and that they do not produce any effective infiltration in the aquifer, as it has been estimated from computation of effective rainfall by the Thornthwaite method with a daily time-step, with a 25-mm soil reserve [29]. The deuterium excess (d=δH2−8×δO18, [20]) is a characteristic parameter of the source area of the water vapour from which rainfall originates. In the study area, this parameter varied from −4.8 to +20.8‰, with a mean value of +9.8‰. These data reflect the fact that precipitation of both Atlantic and Mediterranean origin falls on the region, although most rainfall is of Atlantic origin, as has been shown by various authors [3,19].

In general, the rainwater samples corresponding to the warmest months (Fig. 6) present less negative values of δO18 (with δO18 figures higher than −3‰) and δH2, due to the low quantity of water collected and to the higher air temperature. Thus, the rainwater drops tend to evaporate during their descent through a very dry atmosphere; there is also an enrichment of the isotopic content, depending on the quantity effect and on the seasonal effect [20,32,34]. On the contrary, during the winter months (and also in May), rainfall presents more negative values. A similar distribution was obtained by Celle et al. [15,16] for precipitation in the Avignon region (southern France). In May, due to the higher rain value that has been recorded (Fig. 6), the δO18 figures of rainwater are more depleted, lower than −4‰. In addition, precipitations of colder months are more depleted in heavy isotopes, with figures lower than −4‰.

The drip water samples fit well to the world meteoric line (LMM), which reveals their meteoric origin, as reported by Caballero et al. [13] and Carrasco et al. [14].

As in the case of rainwater, evaporated samples appear clearly; they fit on an evaporation line with a slope of 5.5 (B in Fig. 5). In every case, these are samples with a very low drip flow rate, lower than 19 cm³ d⁻¹. The isotopic composition of the drip water in the cave is more homogeneous than that of rainwater. Moreover, the highest isotopic contents recorded for the drip water correspond to the samples obtained during winter and spring, while the most negative values for the isotopes correspond to the samples obtained during summer and autumn (Fig. 7).

At other sites, the mean value for the δO18 of the drip water in a karstic cave has been utilised as the mean value of the rainfall in the area in which the cave is located [37]. In the present study, however, the mean value obtained for δO18 within the cave is less negative (−3.8‰) than that corresponding to rainwater (−4.9‰). A similar result is obtained when comparing the average isotopic composition of drip water with the drip water flow (−4.4‰) and the weighted rainfall composition, weighted with the quantity of precipitation (−5.1‰), although in the latter case the difference is smaller. This isotopic enrichment, from when the rain falls onto the cave exterior until its appearance as seepage inside the cave, is related to evaporation processes within the soil and in the epikarst.

On the contrary, the differences are more marked if we compare the average δO18 of the effective rainfall (−5.7‰) with that of dripping water (Table 1). This isotopic enrichment can be attributed to water enrichment in the soil, in the epikarst and inside the cave (the clearly evaporated samples correspond to low drip flow rates, in order that the drop of water remains longer in contact with the cave atmosphere). Nevertheless, evaporation measured at the drip point is lower in summer than in winter, and the differences are tiny, about 0.2 mm d⁻¹. A larger difference exists in evaporation at the ground surface (1–6 mm d⁻¹), demonstrating that the isotopic enrichment should mainly occur within the epikarst. Moreover, a slight modification of the isotopic composition can occur inside the cave; it is accentuated in some samples taken during low-discharge periods.

The flow of ¹⁸O through the unsaturated zone of the aquifer was determined (Fig. 8). The input flow was calculated from the effective rainfall (Thornthwaite's method, daily time step, with a 25-mm soil reserve), multiplied by δO18. In general, the winter rainfall episodes are the ones that produce recharge, and so the isotopic inputs to the unsaturated zone are between October and February, together with those that occasionally occur in the spring (April and May). In Fig. 8, we have represented the ¹⁸O fluxes of the input, gathered by rain periods (assigning the obtained sum to the centre of gravity of the period) in order to obtain the time distribution of the major recharge episodes of the aquifer. As a general case, we observe the existence of an input flux to the aquifer, during November and December.

The output flow of ¹⁸O into the cavity was determined by multiplying the drip water flow by its δO18 value. Maximum values are obtained in summer and autumn (August and September), and occasionally in other periods (December–March) associated with significant rainfall episodes (e.g., end of 1996 and early 1997).

The ¹⁸O input/output ratio (Fig. 8) for rainwater and drip water reveals the existence of a seasonal lag of about 8 months (1 in Fig. 8) between the input of rainfall and the output of seepage within the cave. This lag, observed for a 30 m flow through the unsaturated zone, implies a low moisture content in the limestone matrix, according to the piston flow model applicable to this medium. This result is consistent with the average of moisture content of 27 marble samples taken in a quarry situated in the vicinity of the Nerja Cave: 0.38%, measured by Prof. J. Motyka at the University of Krakow (Poland). Nevertheless, when an important recharge episode occurs (such as in early 1997), the drip water response is faster, the lag never exceeding 1 to 3 months (2 in Fig. 8).

According to the high Total Organic Carbon (TOC) contents measured in the soil water situated above the cave, the TOC range observed in the dripwater (0.95–5 mg l⁻¹, average: 2.23 mg l⁻¹) demonstrates that the drip cannot be several years old, at most several months [1,9,10,21].