Forest fires directly modify hydrologic and geomorphic dynamics in drainage basins, particularly water and sediment production and yield. Literature on the effects of fire on hydrology is abundant (see reviews on Anderson et al., 1976; Tiedemann et al., 1979), although it still lacks a comprehensive knowledge of the processes and response of the environment even in long-term monitored areas (Inbar et al., 1998). Some of the effects of fire on hydrology that have been reported include the reduction of infiltration rates and the increment of overland flow (e.g. DeBano et al., 1967), the increment of total discharge and peak flows (Brown, 1972; Lavabre et al., 1993) and the acceleration of soil erosion and, consequently, the increment of sediment river load (Brown, 1972; Batalla & Sala, 1998).

Hydrological effects of forest fires can be studied at a different spatial scales (slope or catchment) and temporal scales (annual, seasonal, monthly, daily or individual floods). Several factors control the effects of fire: a) fire characteristics (temperature and duration - intensity, wildfire or prescribed burning), b) type of soil, c) rainfall regime, d) topography, e) percentage of affected basin area, f) vegetation characteristics and burnt strata, and g) time after fire (López et al., 1999).

Disappearance of vegetal cover modifies the processes of interception and evapotranspiration, seriously affecting the hydrological cycle. Forest fires can also affect hydrological processes indirectly, altering the physical chemical properties of the soil, converting organic ground cover to soluble ash and giving rise to phenomena such as water repellency (e.g. Chandler et al., 1983). Water repellency is an abnormality in soils which results from the coating of soil particles with organic substances which reduce the affinity shown by the soil for water (DeBano et al., 1967). Water repellency is an important factor increasing surface runoff, especially in soils affected by high-intensity fires; those soils present a barrier to water infiltration so that overland flow may result. In this context, Campbell et al., (1977) described a 66% reduction of the infiltration rate in the soil of a pine forest area in Arizona, resulting in a 800% increase in stream flow from the burnt catchment in the first wet season following the fire. Other studies indicated that physical properties of the soil are scarcely affected by fires of light to moderate intensity (Hudson et al., 1983); in such cases, time taken for surface runoff to return to pre-fire levels will be largely governed by the rate of recovery of vegetation cover and by the interception characteristics of vegetation.

Runoff increments have been reported mainly from Mediterranean areas (California, Israel and the Iberian Peninsula), varying from 11% (Anderson et al., 1976) up to 300% (Nasseri, 1988) in large catchments, to 800% in small basins (Campbell et al., 1977), and up to 50000% at plot scale (Inbar et al., 1998). Flood peak increments range from 45% (Anderson et al., 1976) up to 600% (Nasseri, 1988) in large catchments, to 5700% in small basins (Campbell et al., 1977). The shape of flood hydrographs are also profoundly altered, with secondary sharp fast peaks replacing smooth recession limbs prior the fire (Brown, 1972).

Scott & Van Wyk (1990) showed a 200% increase in the annual runoff and up to 300% for the peak discharges during the year following the forest fires in a Southafrican basin. López & Batalla (2001) described an increment of 30% in flood runoff and 120% of peak discharges for the same rainfall input during events six month after the 1994 wildfire in the Arbúcies basin. Lavabre et al. (1993) studied the hydrological effects of a forest fire that burnt 85% of an experimental evergreen-oak Mediterranean drainage basin at a different temporal scale. Annual runoff increased 25% for the same annual precipitation, related to the reduction in evapotranspiration due to destruction of vegetation cover, while increase on monthly and daily runoff reached 30%. Flood frequency was greatly increased: the ten-year flood estimated before the fire occurred three times in the year after the fire, although rainfall events did not exceed the 1-year return period.

The increment of river sediment load is one of the most dramatic responses associated with fire. Sediment result from overland flow and soil erosion, channel scouring because of increased discharge and greater sediment delivery area, and bank and mass erosion. There are not many studies reporting downstream effects on sediment and solute load at catchment scale. Batalla & Sala (1998) analysed the effects of the 1994 wildfire on sediment load in the granitic Mediterranean Arbúcies basin, in which approximately 1000 ha (10% of the catchment area) were completely burnt, mainly evergreen-oak woodland and pines.

Analysis of suspended sediment results indicated that, a) concentrations during the flood immediately after the fire rose in an average of two orders of magnitude compared with pre-fire mean values, increasing from 22 mg l^-1 to more than 2000 mg l^-1 for a mean discharge of 0.8 m³ s^-1, b) mean increment of suspended load ranged from 600% to 2300% during consecutive floods in relation with pre-fire concentrations, c) increments of suspended load at peak flows were especially acute, rising mean concentrations up to 1000% in relation to those obtained for similar discharges prior to fire (figure 2) and, d) maximum measured suspended sediment concentration during the study period was 27 gr l^-1 (for a discharge of 11.5 m³ s^-1 during the recession of October 12^th flood event), a value that represents more than fifteen times the pre-fire concentration for the same discharge.

Solute load was also affected, although showing moderate increments compared to suspended sediment load. Total dissolved load rose in 60% during the flood event immediate after the fire in comparison with pre-fire values. Increments in Potassium and Sulphate concentrations were particularly remarkable (700% and 250%, respectively). After five consecutive events, mean solute concentrations were found to reach similar levels than those obtained before the fire.

Fire produced, thus, acute effects on sediment load of the Arbúcies river, taking into account that 'only' 10% of the area was seriously affected by fire. Two facts appeared to be responsible for the rapid and strong impact of fire on river load: a) On the one side, the proximity of burnt areas to the sampling point at the basin outlet and, b) On the other, the concatenation of six flood events, the largest with a recurrence period of 8 years, during immediate weeks after the fire.