Spatial and temporal patterns of vegetation recovery following sequences of forest fires in a Mediterranean landscape, Mt. Carmel Israel

Catena 71 (2007) 76 – 83 www.elsevier.com/locate/catena

Spatial and temporal patterns of vegetation recovery following sequences of forest fires in a Mediterranean landscape, Mt. Carmel Israel Lea Wittenberg a,⁎, Dan Malkinson a,b , Ofer Beeri c , Alon Halutzy a , Naama Tesler a a

c

Department of Geography and Environmental Studies, University of Haifa, Haifa 31905, Israel b Golan Research Institute, University of Haifa, Haifa 31905, Israel Earth Science Systems and Policy, University of North Dakota, Box 9011 Grand Forks, ND 58202 USA Received 1 January 2006; accepted 1 June 2006

Abstract The Mediterranean ecosystem of Mt. Carmel is subjected to increasing number of forest fires at various extents and severities due to increasing human activities. Accordingly, we tested whether in areas exposed to different fire histories vegetation regeneration is different in north versus south facing slopes, and the potential impact on erosion processes. Using remote sensing techniques we evaluated the Enhanced Vegetation Index (EVI) to monitor vegetation recovery following a single fire and three successive fires, using a series of Landsat images taken between 1985–2002. Following a single fire, vegetation cover reached pre-disturbance values within less than 5 years. Repeated fires caused further reduction of EVI values, especially at south facing slopes (SFS). The effects of three successive fires within 10 years, followed by a three year recovery period, however, are negligible when considering vegetation cover values. This was deduced as north facing slope EVI values returned to pre-disturbance conditions at the end of the 3 years and SFS EVI values to 80% of the pre-disturbance conditions. Our results indicate that Mediterranean eco-geomorphic systems are quite resilient, showing quick response, at least in terms of return to predisturbance states of vegetation cover, and hence of soil erosion rates. This is true not only in response to a disturbance caused by a single fire, but also for repetitive fire incidents. © 2006 Elsevier B.V. All rights reserved. Keywords: Mediterranean ecosystem; Forest fire; Remote sensing; Vegetation regeneration; Soil erosion

1. Introduction Fires in Mediterranean ecosystems have a complex effect on geomorphological processes and vegetation regeneration due to the complexity of landscape structures as well as differential responses of such systems to various types of fire regimes. Specifically, different fire regimes are manifested by different fire intensities, seasonalities, recurrence probabilities and the extents of these events (Naveh, 1973). At the landscape level, post-fire regeneration would depend mainly on the initial vegetation and onsite environmental factors — climatic and terrain parameters (Pausas and Vallejo, 1999).

⁎ Corresponding author. E-mail address: [email protected] (L. Wittenberg). 0341-8162/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.catena.2006.10.007

Vegetation cover plays one of the key factors affecting soil erosion and land degradation processes (Thornes, 1990; Shakesby et al., 1993). Assuming a rapid and considerable intensification of the erosive processes following fires (Inbar et al., 1997, 1998) vegetation recovery normally leads to a decline in post-fire runoff and soil erosion rates. Thornes (1990) suggested that a minimum value of 30% projective plant cover is sufficient for protecting the soil against water erosion. Risks of post-fire soil erosion are higher when the time required for the vegetation to reach this minimal vegetation threshold cover is longer. This has been widely demonstrated at various spatial scales and under different ecological conditions (Shakesby et al., 1993; Inbar et al., 1998; Cerda, 1998a,b). Apparently, much of the sediment loss occurs during the first year following the fire occurrence (DeBano et al., 1998; Inbar et al., 1998; Cerda and Doerr,

L. Wittenberg et al. / Catena 71 (2007) 76–83

2005). Therefore, it is essential to evaluate vegetation recovery rates immediately after fires. Most studies addressing post-fire geomorphological processes and vegetation dynamics, at various temporal and spatial scales, were conducted following a single fire event. At the landscape level, where burning occurs and regeneration takes place, both short and long-term effects of a single event are inevitably related to the long-term fire regime which creates a landscape mosaic consisting of patches with different burning history. The interrelations between these patch types largely determine the recovery patterns of frequently burnt areas, which are common in the Mediterranean landscape (Mouillot et al., 2003). In combination with ancillary spatial data such as soil properties and terrain characteristics, accurate estimates of vegetation cover and regeneration can be valuable for identifying areas of elevated erosion risks following fires. Two principal methods are commonly used to estimate post-fire vegetation dynamics — experimental vegetation/ runoff plots and remote-sensed image analysis. The utilization of plots enables relatively accurate and detailed measurements of vegetation cover and plant community properties, as well as runoff and erosion rates, yet within restricted areas and limited time scales. Application of remote sensing indices is more appropriate at the landscape scale, yet endures many limitations. A number of vegetation indices have been developed and used for monitoring vegetation structure and function. Among these, is the Normalized Difference Vegetation Index (NDVI), which uses spectral information in red and near infrared bands, and is commonly used to estimate net primary production and fire effects on vegetation (Paltridge and Barber, 1988; Illera et al., 1996). Nevertheless, NDVI has several limitations, including sensitivity to atmospheric conditions (Holben, 1986) and sensitivity to soil background (Huete, 1987). To account for residual atmospheric contamination (e.g., aerosols) and variable soil background reflectance, the Enhanced Vegetation Index (EVI) was proposed, which directly adjusts the reflectance in the red spectral band as a function of the reflectance in the blue band (Liu and Huete, 1995; Huete et al., 1997). Many remote sensing fire recovery studies have been conducted in environments with Mediterranean climates (Jakubauskas et al., 1990; Marchetti et al., 1995; Viedma et al., 1997; Díaz-Delgado et al., 1998; Henry and Hope, 1998; Ricotta et al., 1998). Validation methods to correct for external factors affecting image quality include comparison with non-burnt sites characterized by similar environmental conditions located within the extent of the same image (Díaz-Delgado et al., 1998), validation with aerial photography and field studies (Kushla and Ripple, 1998). While using remote sensing techniques enables the investigation of such effects at large spatial scales, such studies are limited by the ability to accurately detect high resolution changes in the structure of vegetation communities. Studies, based on both approaches have shown that postfire recovery is quick for most species, due to their resprouting abilities, or the persistence of their seed bank. Rapid

77

regeneration occurs within the first 2 years following fires (Trabaud, 1981; Kutiel, 1994; Inbar et al., 1998), with differential recovery rates at the north and the south facing slopes (Cerda and Doerr, 2005). Pausas and Vallejo (1999) noted that in the Iberian Peninsula, within a year after single fire event vegetation cover reached 52.4% on the north facing slope and 32% on the south facing slope. Similar trends were also found at the Mt. Carmel region, Israel, following the 1988 fire (Kutiel, 1994; Inbar et al., 1998). Notwithstanding, Díaz-Delgado et al. (2002) observed lower NDVI values after the second of two successive fires occurring within an 11 year interval, i.e. the green biomass diminishes significantly when disturbances occur within short time intervals. They concluded that increased fire frequency may reduce ecosystem resilience — the ability of the system to recover to a pre-disturbance state. In spite of the considerable efforts invested in fire research, the ability to predict the impact of fires on the landscape is still limited (Moreno et al., 1998; De Luis et al., 2004). In light of the evident increase in the number of fires and burnt areas in the Mediterranean basin (Pausas and Vallejo, 1999) the need to evaluate fire effects on vegetation and geomorphological patterns and processes is essential. The objective of the study presented herein was to assess vegetation recovery rates under different fire occurrences and in different slope aspects. Specifically, we compared recovery rates in areas repeatedly burnt — three fires within a ten year period — to an area burnt only once during this period. Further, within those regions, we investigated whether differential vegetation recovery patterns are associated with slope aspects. The conceptual model of soil loss recovery (Inbar et al., 1998) was reexamined in light of the long-term vegetation cover monitoring, to examine the interrelations between vegetation and land degradation processes in a fire-prone environment. 2. Methods 2.1. Study site and data acquisition The area studied is located at the north-western part of the Carmel Mountain ridge (35°W, 32°N). Mt. Carmel is an isolated mountain ridge, rising from the northern Mediterranean Sea shore of Israel to a height of 500 m above MSL (Fig. 1). The Mediterranean type climate at Mt. Carmel is characterized by dry and hot summers and rainy winters. In Northern Israel precipitation commonly commences during October and ends in May; where most rainstorms usually occur between November and March. Autumn precipitation is often convective in nature with relatively high rainfall intensities while winter rainfalls (December–February) are mainly a result of frontal activities related to wide synoptic systems. The average annual rainfall in Mt. Carmel ranges from 550 mm near the coastal plane to 750 mm at the highest elevations. Owing to long period of dry spells and the increased climatic uncertainty (Paz and Kutiel, 2003), Naveh (1973) characterized the region as a Mediterranean

78

L. Wittenberg et al. / Catena 71 (2007) 76–83

Fig. 1. Location map, fires sites and the study plots.

fire bio-climate. Vegetation on the terrarosa soils is characterized by a complex of Pinus halepensis–Pistacia palestina–Cistus sp. associations on south facing slopes and Quarcus calliprinos–Pistacia palastina association on north facing slopes, forming a Mediterranean evergreen sclerophyll forest which is highly flammable during summer time (Inbar et al., 1998). Due to intense human activities in the region, wildfires commonly occur on the Carmel Mountain ridge. We have recorded eight large wildfires that occurred in the study region during the past 27 years, which consumed areas of 80–530 ha each, and dozens of other smaller ones. According to the fire department reports during the dry months there are 11 fire events on average in the Carmel region. However, only under certain climatic and vegetation conditions combustion might spread and consume large areas. Although all fires in the region are of anthropogenic sources, weather conditions play an important role in determining fire intensity and its spread rates. Long dry spells of 5–6 months (May–October) coupled with easterly hot winds lead to rapid drying of the vegetation. Thus, during May–October there are abundant combustible matters. Most large fires in the Mt. Carmel region occurred in this period following long dry spells coupled with easterly hot and dry winds (Kutiel, 1992). In addition, annual rainfall and maximum temperature were found to be positively correlated

with both the size of the burnt areas and fires frequencies (Levin and Saaroni, 1999). 2.2. Data processing and analysis To assess the regeneration rates of vegetation we delineated areas which were burnt various numbers of times, and using satellite images we analyzed vegetation cover within them. Specifically, we located sites which were either burnt once, three times, or did not burn at all during the period of 1985–2002, hence creating “treatments” of zero, one or three fires; see Table 1 for details. During this period the three fires partially burnt the area, in September 1989, October 1998 and December 1999. Further, within each of the burnt sites EVI values were compared between north and south facing slopes. Aspect analysis was conducted using a Digital Elevation Model of the region with ARC/GIS Version 9.1. Cells with values ranging from 135°–225° Table 1 Properties of the fires studied Year

Date

Area burnt (hectares)

1989 1998 1999

19 September 11 October 4 December

530 161 158

L. Wittenberg et al. / Catena 71 (2007) 76–83

79

Table 2 Mean EVI and SD values at the control, 1 fire and 3 fires sites within the north and the south facing slopes of Mt. Carmel, 1985–2002 South facing slope Control

1985 1990 1995 2000 2002

North facing slope 1 fire

3 fires

Control

1 fire

3 fires

Avg.

SD

Avg.

SD

Avg.

SD

Avg.

SD

Avg.

SD

Avg.

SD

0.79 0.78 0.71 0.76 0.89

0.05 0.06 0.05 0.07 0.07

0.82 0.43 0.67 0.65 0.75

0.07 0.07 0.08 0.07 0.07

0.75 0.51 0.69 0.24 0.69

0.07 0.09 0.05 0.06 0.12

0.76 0.95 0.66 0.81 0.94

0.05 0.06 0.04 0.05 0.06

0.76 0.62 0.71 0.69 0.77

0.10 0.14 0.16 0.13 0.08

0.65 0.58 0.60 0.44 0.68

0.07 0.06 0.11 0.08 0.09

were classified as south facing slopes, and cells with values ranging from 315°–45° were classified as north facing slopes. Cells not falling within these value ranges were excluded from the analysis. Two additional control sites were chosen for calibration purposes: an evergreen vegetation site located at a stream bottom used for uppermost values of productivity, and a ground data site — a gravel mining pit for the minimal values. We visually inspected each area at each image, and selected cells that were at least 60–90 m inside the burnt area. The Landsat images acquired for analyses were from 8 August 1985, 26 May 1990, 21 September 1995, 22 June 2000 and 18 October 2002. The 1985 and 1995 images are from Landsat 5, 1990 image is from Landsat 4, and the 2000 and 2002 images are from Landsat 7. We chose these as they were the available images representing 5 year time step intervals and they were cloud free (except the 2002 image). Satellite image preprocessing entailed three steps: georeferencing, calculation of ground reflectance and masking clouds and clouds shadows. The Landsat images were subset to cover a 22 * 24 km region, and then locally georeferenced using a local Digital Elevation Model (DEM). The images were geo-referenced using 15 ground control points from a 25 m resolution DEM, which accounts for the topographic variability of the mountainous terrain. The root-mean-square error was smaller then half a pixel, 15 m, and we used the nearest neighborhood algorithm to resample each image. In the second stage, we applied the ERDAS imagine ATCOR 2.1 package to correct for atmospheric interferences. Since this did not produce sufficient results, as only a subset of the image was used for the analysis, further correction was applied. An empirical line technique was applied using deep sea and the gravel control site's spectra. By applying this technique, we used the darkest and brightest targets in each image, the sea and the gravel respectively, as the lowermost and uppermost spectra to further enhance the atmospheric corrections. As 2002 image included small clouds and shadows two methods were used to mask them: (1) To identify cloud coverage, a duplicated criteria that used the blue and the thermal Landsat spectra, with less then 20 °C and band 1 ground reflectance greater then 0.19 was used. (2) To depict the cloud shadows, the infrared bands, bands 4, 5 and 7 were summed, and all pixels with values less than 0.24 were masked. For each site, and for each image, the EVI values were calculated as it provides an established proxy for vegetation

productivity and it is less sensitive to atmospheric interferences (Huete et al., 2002). The vegetation cover at the stream bottom control plot, which is dominated by evergreen trees, was assumed to be constant throughout the study period. Therefore, within each image the study plots' EVI indices were standardized against the stream bottom mean EVI value. Following the standardization procedures a MANOVA was performed to evaluate significant differences among the different plots using the Bonferroni correction. Depending on the year, a 1 × 2, 2 × 2 or a 2 × 3 MANOVA was performed, where two levels of aspect were considered (north, south). The number of times an area was burnt was the second factor analyzed. In the 1985 image, which served as control, only one level was considered. In the 1990 and 1995 images, two levels were considered, non-burnt areas, and areas burnt once. In the last two images three levels of fire were considered, non-burnt areas, and areas burnt once or three times. Finally, the specific hypotheses described above were tested. 3. Results 3.1. EVI — single fire Due to the limitations of comparing images among years, the analyses we conducted compared EVI values only within years. Vegetation cover and regeneration as manifested by the EVI values (Table 2, Fig. 2) indicate that prior to the fire (1985: Table 2, Fig. 2) EVI's for all sites ranged between 0.65 in the NFS (north facing slopes) to 0.82 in the SFS (south facing slopes). The MANOVA revealed that mean

Fig. 2. 1985–2002 annual EVI values at the different sites.

80

L. Wittenberg et al. / Catena 71 (2007) 76–83

EVI values were significantly different overall (p b 0.001), but no apparent trends existed, as some of the south facing slopes yielded higher EVI values compared to north facing slopes, and vice versa. Eight months following the 1989 fire, EVI values of north facing slopes (NFS) (0.58–0.62) were significantly higher compared to the south facing (SFS) burnt sites (0.43–0.51) (p b 0.001, Table 2); the burnt sites of both aspects had lower values compared to non-burnt sites (p b 0.001). There was no significant interaction, however, between aspect and burnt condition (p = 0.818). Five years later, and 6 years after the first fire (1995, Fig. 2) EVI values ranged 0.64–0.71, and no significant differences among the sites were found. Specifically, no differences between north and south facing slopes existed (p = 0.26), nor between burnt and non-burnt areas (p = 0.816). 3.2. EVI — repeated fires The 2000 image includes in addition to the 1989 fire sites areas that were burnt in 1998 and 1999, hence three times within a period of 10 years. In June 2000 EVI values in the 3-times burnt sites were significantly lower compared to non-burnt sites (p b 0.0001) — the burnt SFS had lower EVI values (0.24) compared to the 3-times burnt NFS (0.44) (p b 0.001). EVI values for the 3-times burnt sites were significantly lower within aspects and between aspects (p b 0.01). South facing slopes that were burnt three times yielded the lowest EVI values — 0.27, whilst the once-burnt site in the SFS was 0.65. Within the NFS, EVI for the 1 and 3-times burnt sites were 0.70 and 0.41 respectively. This indicates that recovery of the vegetation on south facing slopes in areas frequently burnt was much slower. In 2002, following two winters, no significant differences were apparent among the aspects (p = 0.343). The differences in EVI values were not apparent any more, suggesting that at least overall vegetation cover returned to pre-fire values. 3.3. Aspects Generally, in both aspects, repeated fires within 10 years decrease post-fire vegetation regeneration rates (Fig. 2). EVIsð1990Þ ¼ 0:55 EVIcð1990Þ

EVIsð2000Þ ¼ 0:32 EVIcð2000Þ

EVInð1990Þ ¼ 0:65 EVIcð1990Þ

EVInð2000Þ ¼ 0:54 EVIcð2000Þ

where EVIs represent SFS vegetation values, and EVIn — NFS values and EVIc — non-burnt sites vegetation values. These values do not provide for statistical comparisons between the different years. They do provide, however, relative values within years, which enables for comparisons of relative EVI values in sites burnt once or three times.

However, this trend is more profound at the SFS, when comparing between south north facing aspects EVI values. Using the South–North Ratio (SNR), we compared recovery rates in response to different number of fires: EVIsð2000Þ ¼ 0:54 EVInð2000Þ

EVIsð1990Þ ¼ 0:69 EVInð1990Þ

A repeatedly burnt SFS — in relation to NSF had a lower ratio (0.54), compared to the sites which burnt only once (0.69). 4. Discussion The use of time series satellite images, albeit their well known limitations, may provide further insights to post-fire vegetation dynamics over large regions and long time periods. The EVI, which is also used to assess vegetation cover, is responsive to canopy structural variations, including leaf area index, canopy type, plant physiognomy, and canopy architecture (Gao et al., 2000). The index was found to be sensitive to seasonal vegetation variations, land cover variations, and biophysical parameter variations. Therefore it provides a good dynamic measure for monitoring and assessing spatial and temporal variations of vegetation cover and condition. Moreover, the EVI minimizes much of the contamination problems present in the NDVI (Huete et al., 2002). Given the limitations of this index, studies using EVI values provide knowledge merely on vegetation cover at large scales with no detailed information about the vegetation, such as species composition. Therefore, high resolution data (e.g. based on small-scale plots), is essential for validating and complementing information obtained from remote sensing sources. Addressing vegetation recovery patterns under multiple fire regimes the EVI provides an essential tool to assess the long-term–large scale temporal perspective. Notwithstanding, this tool has not yet been extensively applied in the study of Mediterranean ecosystems relative to numerous field-based-studies. Moreover, the limited literature concerning vegetation recovery patterns following single and multiple forest fires in this region is largely based on the NDVI index (e.g. Ricotta et al., 1998; DíazDelgado and Pons, 2001; Díaz-Delgado et al., 2002). We used the EVI to detect both temporal and spatial changes in vegetation cover following a single and multiple fires. This approach comes to complement the long-term field-based study conducted in the restricted burnt area following the 1989 fire (Inbar et al., 1998). 4.1. Spatial patterns In Mediterranean systems of the northern hemisphere, south facing slopes are commonly characterized by more arid conditions due to exposure to higher annual radiance. Under these conditions vegetation recovers more slowly following disturbance compared to north facing slopes, and hence regeneration is slower (Mouillot et al., 2003). Owing

L. Wittenberg et al. / Catena 71 (2007) 76–83

to environmental differences between aspects post-fire regeneration rates are assumed to vary between SFS and NFS, resulting in slower recovery on SFS. The South/North ratio was assessed to evaluate revegetation rates between aspects, and test this assumption. The results — 0.69 in 1990 and 0.54 in 2000 indicate that not only the vegetation cover is lower, a year after the fire on the SFS, but that repeated burning further reduce EVI values on southern slopes compared to the NFS. However, in 2002 (13 years following the first fire and 3 years following the last one) vegetation cover on both NFS were similar to that of the control EVI and higher compared to SFS. 4.2. Temporal patterns In the first year following fires, at both aspects, EVIs were significantly lower compared to the control plots. Moreover, repeated fires further reduced recovery rates. Relative burnt/ un-burnt ratio indicates that the lowest values were obtained at the three times burnt SFS (0.3). Also, in the NFS the ratio decreased with the increased number of fires, but to a lesser extent — 0.77 following a single fire and 0.61 following three events. Further information collected following the 1989 fire confirms our results. Ne'eman et al. (1995) indicated a significant increase in species richness during the first 3 years after the fire. Temporal patterns of revegetation are strongly related to mean annual rainfall amounts. In the thermomediterranean shrubland area in south-eastern Spain, 3 years after a fire event, vegetation cover was significantly lower on the burnt site than in the non-burnt site (Gotzenberger et al., 2003). In Catalonia, Díaz-Delgado et al. (2002) demonstrated, using NDVI measurements, that a positive correlation between NDVI and mean annual rainfall, exists and a negative correlation between post-fire NDVI and solar radiation also exists. They argued, however, that direct rainfall during the first year after the fire was not significantly correlated with resilience. Repeated fires decrease ecosystem resilience. Vegetation re-growth after 70 months was significantly lower after the second fire than after the first. This difference, however, was not significant at earlier stages of recovery (Díaz-Delgado et al., 2002). Our results conform to the assumption that south facing slopes are less resilient in comparison to north facing slopes. Similarly to Díaz-Delgado et al. (2002) we were able to detect a rapid recovery of vegetation cover, within a time period of 2–3 years. Rapid regeneration of vegetation cover is a key factor in the rehabilitation of the landscape and an indicator for the resilience of the eco-geomorphic system. The coupled effect of vegetation destruction and changes in soil properties (Cerda and Doerr, 2005) inevitably leads to a rapid increase in soil loss and runoff processes. This increase, however, peaks during the first year following a fire. Decrease in soil erosion rates within 3–4 years is largely attributed to the regeneration of vegetation as has been demonstrated in another study conducted following the 1989 fire. Within the burnt areas studied in this research, a hillslope plot study was conducted to monitor erosion rates and vegetation recovery

81

Table 3 Mt. Carmel experiment plots following 1989 fire, annual sediment yield (g/m2/yr). After Inbar, 2002 Plot/year

1 (north) 2 (north) 3 (north) 4 (control-north) 11 (south) 12 (south) 13 (south) 14 (south) 16 (control-south)

1989–1990

1990–1991

1991–1992

1992–1993

Sediment g/m2/yr

Sediment g/m2/yr

Sediment g/m2/yr

Sediment g/m2/yr

202.5 18.5 50.0 0.001 420.0 875.0 440.0 515.0 0.02

23.3 2.0 4.5 0.01 0.25 12.0 7.0 100.0 0.005

0.07 0.11 0.23 0.012 0.03 0.04 0.20 0.07 0.055

0.03 0.16 0.10 0.035 0.02 0.02 0.11

(Inbar et al., 1997; Inbar et al., 1998). Their results (Table 3) complement the results presented in the current study, as the authors noted that the fire increased sediment yield rates at the hillslope plots by four to five orders of magnitude as compared to control plots (Inbar, 2002). A sharp decrease of one to three orders of magnitude occurred during the second year, in spite of exceptionally high rainfall values. It was suggested that rapid vegetation recovery played a key role in the overall rehabilitation of the burnt site, in terms of erosion rates. Post fire sediment yield is commonly estimated using hillslope plots. Therefore, local micro-conditions such as soil and bedrock type, roughness, stoniness, gradients and direct rainfall largely control erosion rates (Kutiel et al., 1995; Pausas et al., 1999) and thus, the variation between adjacent plots is inevitably great (see Inbar et al., 1998). Moreover, upscaling micro-plot results for estimating soil loss at larger scales such as slopes or basins are considerably complicated. Similarly, vegetation regeneration estimated using smallscale methods might be influenced by local site conditions. Applying remotely sensed indices over large spatio-temporal scales overrides, and averages over the local micro-conditions, and provides additional information. Coupling our results based on EVI values with Inbar et al.'s (1998) findings provide a stronger inference about the dynamics of Mediterranean eco-geomorphic systems. Both methods signified a rapid regeneration of vegetation cover following the fire and also similarly identified variations in regeneration rates between different slope aspects. This emphasizes that the results characterize the entire burnt area and not only the randomly chosen small plots. Using EVI analyses enable to extend not only the spatial scale but also the time span of the investigation and the inclusion of the effects of recurring fires. 5. Conclusions The Mediterranean ecosystem of Mt. Carmel is subjected to an increasing number of forest fires at various extents and severities. The effects of three successive fires within a 10 years period, followed by a recovery period of 3 years are negligible. With respect vegetation cover NFS EVI values

82

L. Wittenberg et al. / Catena 71 (2007) 76–83

returned to pre-disturbance condition and SFS EVI to 80% of the pre-disturbance condition. EVI is not applicable for identifying species composition and hence the index is not appropriate to assess the ecological states of the system following the disturbance. It is, however, of great importance to assess post-fire geomorphological processes associated with vegetation recovery. The implications of these results should therefore be viewed in context of two different frameworks, geomorphological and ecological processes. In terms of runoff and erosion processes our results, coupled with Inbar et al.'s (1997, 1998) results, indicate enhanced post-fire soil erosion, especially in the SFS, up to four orders of magnitudes higher than the non-burnt adjacent sites due to the elimination of vegetation cover. This phase is followed by a rapid recovery of the system namely, returning to predisturbance erosion rates within 3–5 years, due to the regeneration of vegetation. Considerable differences in sediment yield between neighboring plots are associated with the effects of the local micro-conditions, which are more profound during the first year following the fire when vegetation is absent or minimal. Using remote sensing techniques we were able to detect that vegetation cover achieved pre-fire values within 3 years following a fire event. Using these techniques within an ecological framework poses some caveats however. EVI values provide an indication for vegetation cover. It is impossible to use this index to identify vegetation communities, nor specific species. Hence, it is impossible to detect whether the vegetation community is returning to pre-fire states, and what are the long-term implications of changing disturbance regimes to the structure of vegetation communities. Therefore, we propose that the methodology used in this study be coupled with field studies to assess the recovery of ecological systems in Mediterranean climates. Our results indicate that Mediterranean systems are quite resilient, showing quick response, at least in terms of return to previous states of soil erosion rates and vegetation cover not only in a response to disturbance caused by a single fire effect, but also to repetitive fire incidents. Acknowledgements We would like to thank two anonymous reviewers for providing critical comments, which greatly improved the manuscript. References DeBano, L.F., Neary, D.G., Ffolliott, P.F., 1998. Fire's Effects on Ecosystems. John Wiley and Sons, New York, p. 333. Cerda, A., 1998a. Changes in overland flow and infiltration after a rangeland fire in a Mediterranean scrubland. Hydrological Processes 12, 1031–1042. Cerda, A., 1998b. Postfire dynamics of erosional processes under Mediterranean climatic conditions. Zeitschrift für Geomorphologie 42, 373–398. Cerda, A., Doerr, S.H., 2005. The influence of vegetation recovery on soil hydrology and erodibility following fire: an eleven year investigation. International Journal of Wildland Fire 14 (4), 423–437.

De Luis, M., Baeza, M.J., Raventos, J., Gonzales-Hidalgo, J.C., 2004. Fuel characteristics and fire behavior in mature Mediterranean gorse shrub land. International Journal of Wildland Fire 13, 79–87. Díaz-Delgado, R., Pons, X., 2001. Spatial pattern of forest fires in Catelonia (NE of Spain) along the period 1975–1995. Analysis of vegetation recovery after fire. Forest Ecology and Management 147, 67–74. Díaz-Delgado, R., Salvador, R., Pons, X., 1998. Monitoring of plant community regeneration after fire by remote sensing. In: Traboud, L. (Ed.), Fire management and landscape ecology. International Association of Wildland Fire, Fairfield, WA, pp. 315–324. Díaz-Delgado, R., Lloret, F., Pons, X., Terradas, J., 2002. Satellite evidence of decreasing resilience in Mediterranean plant communities after recurrent wildfires. Ecology 83 (8), 2293–2303. Gao, X., Huete, A.R., Ni, W., Miura, T., 2000. Optical–biophysical relationships of vegetation spectra without background contamination. Remote Sensing of Environment 74, 609–620. Gotzenberger, L., Constanze, O., Hensen, I., Gomez, P.S., Wesche, K., 2003. Postfire regeneration of a thermomediterranean shrubland area in southeastern Spain. Anales de Biologia 25, 21–28. Henry, M.C., Hope, A.S., 1998. Monitoring post-burn recovery of chaparral vegetation in southern California using multitemporal satellite data. International Journal of Remote Sensing 19 (16), 3097–3107. Holben, B., 1986. Characteristics of maximum-value composite images from temporal AVHRR data. International Journal of Remote Sensing 7, 1417–1434. Huete, A.R., 1987. Soil influence in remote sensed vegetation–canopy spectra. In: Elachi, C. (Ed.), Introduction to the physics and techniques of remote sensing. Wiley–Interscience, New York, pp. 107–141. Huete, A.R., Liu, H.Q., Batchily, K., van Leeuwen, W., 1997. A comparison of vegetation indices over a global set of TM images for EOS-MODIS. Remote Sensing of Environment 59, 440–451. Huete, A.R., Didan, K., Miura, T., Rodriguez, E.P., Gao, X., Ferreira, L.G., 2002. Overview of the radiometric and biophysical performance of the MODIS vegetation indices. Remote Sensing of Environment 83, 195–213. Illera, P., Fernandez, A., Delgado, J.A., 1996. Temporal evolution of the NDVI as an indicator of forest fire danger. International Journal of Remote Sensing 17 (6), 1093–1105. Inbar, M., 2002. Land management planning and the recovery of landscape after forest fire in Mediterranean land-forms. In: Pardini, G., Pinto, J. (Eds.), Fire, Landscape and Biodiversity: An Appraisal of the Effects and Effectiveness. Universitat de Girona, pp. 153–166. Inbar, M., Wittenberg, L., Tamir, M., 1997. Soil erosion and forestry management after wildfire in a Mediterranean woodland, Mt. Carmel, Israel. International Journal of Wildland Fire 7, 285–294. Inbar, M., Tamir, M., Wittenberg, L., 1998. Runoff and erosion processes after a forest fire in Mount Carmel, a Mediterranean area. Geomorphology 24 (1), 17–33. Jakubauskas, M.E., Lulla, K.P., Mausel, P.W., 1990. Assessment of vegetation change in a fire-altered forest landscape. Photogrammetric Engineering and Remote Sensing 56 (3), 371–377. Kushla, J.D., Ripple, W.J., 1998. Assessing wildfire effects with Landsat Thematic Mapper data. International Journal of Remote Sensing 19 (13), 2493–2507. Kutiel, H., 1992. Weather conditions and suppression of forest fires in Israel. Horizons in Geography 35–36, 35–42 (In Hebrew with an English abstract). Kutiel, P., 1994. Fire and ecosystem heterogeneity: a Mediterranean case study. Earth Surface Processes and Landforms 19, 187–194. Kutiel, P., Lavee, H., Segev, M., Benyamini, Y., 1995. The effect of fireinduced surface heterogeneity on rainfall–runoff–erosion relationships in an eastern Mediterranean ecosystem, Israel. Catena 25, 1–4. Levin, N., Saaroni, H., 1999. Fire weather in Israel — synoptic climatological analysis. GeoJournal 47 (4), 523–538. Liu, H.Q., Huete, A.R., 1995. A feedback based modification of the NDVI to minimize canopy background and atmospheric noise. IEEE Transactions on Geoscience and Remote Sensing 33, 457–465. Marchetti, M., Ricotta, C., Volpe, F., 1995. A qualitative approach to the mapping of post-fire regrowth in Mediterranean vegetation with

L. Wittenberg et al. / Catena 71 (2007) 76–83 Landsat-TM data. International Journal of Remote Sensing 16 (13), 2487–2494. Moreno, J.M., Vazquez, A., Velez, R., 1998. Recent history of forest fires in Spain. In: Moreno, J.M. (Ed.), Large Forest Fires. Backhuys Publishers, Leiden, The Netherlands, pp. 159–186. Mouillot, F., Ratte, J.P., Joffre, R., Moreno, J.M., Rambal, S., 2003. Some determinants of the spatio-temporal fire cycle in a Mediterranean landscape (Corsica, France). Landscape Ecology 18, 665–674. Naveh, Z., 1973. The ecology of fire in Israel. Proc. 13th Annual Tall Timbers Fire Ecol. Conf., Tallahassee, Florida, pp. 133–170. Ne'eman, G., Lahav, H., Izhaki, I., 1995. Recovery of vegetation in a natural east Mediterranean pine forest on Mount Carmel, Israel as affected by management strategies. Forest Ecology and Management 75, 17–26. Paltridge, G.W., Barber, J., 1988. Monitoring grassland dryness and fire potential in Australia with NOAA–AVHRR data. Remote Sensing of Environment 25, 381–394. Pausas, G.J., Vallejo, V.R., 1999. The role of fire in European Mediterranean Ecosystems. In: Chuvieco, E. (Ed.), Remote sensing of large wildfires in the European Mediterranean basin. Springer–Verlag, pp. 3–16. Pausas, G.J., Carbo, E., Caturla, R.N., Gil, J.M., Vallejo, R., 1999. Post fire revegetation patterns in the Eastern Iberian Peninsula. Acta Oecologica 20 (5), 499–508.

83

Paz, S., Kutiel, H., 2003. Rainfall Regime Uncertainty (RRU) in a Mediterranean Region. The Israel Journal of Earth Sciences 52 (2), 47–63. Ricotta, C., Avena, G.C., Olsen, E.R., Ramsey, R.D., Winn, D.S., 1998. Monitoring the landscape stability of the Mediterranean vegetation in relation to fire with a fractal algorithm. International Journal of Remote Sensing 19 (5), 871–881. Shakesby, R.A, Coelho, C.O.A., Ferreira, A.D., Terry, J.P., Walsh, R.P.D., 1993. Wildfire impacts on soil erosion and hydrology in wet Mediterranean forest, Portugal. International Journal of Wildland Fire 3, 95–110. Thornes, J., 1990. The interaction of erosion and vegetation dynamics in land degradation: spatial outcomes. In: Thornes, J. (Ed.), Vegetation and Erosion. Wiley, New York, pp. 41–54. Trabaud, L., 1981. Man and fire: impacts on Mediterranean vegetation. In: di Castri, F., Goodall, D.W., Specht, R.L. (Eds.), Mediterranean-Type Shrublands. Elsevier, Amsterdam, pp. 523–537. Viedma, O., Meliá, J., Segarra, D., Garcıa-Haro, J., 1997. Modeling rates of ecosystem recovery after fires using Landsat TM data. Remote Sensing of Environment 61, 383–398.