special paths made by the forest workers.
2.3. Soil Sampling
Four soil samples were collected and pooled together from each of the four plots assigned to each post-fire practice: salvage logging (B-SL); unlogged (B-UL) and from four unburned control plots (UB). Sampling began in autumn (October) 2007, after logging ended (t0), and continued seasonally (autumn, winter, spring, summer) until summer 2010 (about 3 years after logging practice began). Each soil sample was immediately placed in an insulated container and taken to the laboratory. The soil samples were kept in cold storage at 4˚C until processing. Before biological and chemical analyses, they were sieved (2-mm mesh size) in order to remove root particles and other organic debris.
2.4. Laboratory Analysis
The following analyses were performed on each soil sample:
2.4.1. Soil Moisture
Soil moisture (SM) was determined gravimetrically as percentage of dry mass by drying the samples to a constant weight at 105˚C .
2.4.2. Soil Organic Carbon
Soil organic carbon (OC) content was determined by oxidation with 1 N potassium dichromate in acidic medium, according to Rowell .
2.4.3. Total Soluble Nitrogen (TSN) and Dissolved Organic Carbon (DOC)
Total Soluble Nitrogen (TSN) and Dissolved Organic Carbon (DOC) were determined by chemical extraction and color reactions using a Skalar autoanalyzer. Dried soil samples were extracted in a 1:10 (w/v) ratio with 0.01 M calcium chloride solution. Total soluble N: The determination of total N content was based on the following reaction: a sample was mixed with a borax buffer, excess potassium persulfate solution was added, and the mixture was placed in an ultraviolet digester. Nitrate was determined by the Griess reaction after oxidation of ammonium and reduction of nitrate to nitrite by cadmium copper redactor. The color was measured at 540 nm . Dissolved organic C:A sample was acidified and the inorganic carbon removed with nitrogen. The sample was mixed with persulfate buffer and irradiated in a UV destructor. Hydroxylamine was added and the sample entered into a dialyzer. The generated CO2 diffused through a gas-permeable silicone membrane. A weakly buffered phenolphthalein indicator solution was used as the recipient stream, and the color intensity of this solution decreased in proportion to the change in pH caused by the absorbed CO2 gas. The color was measured at 550 nm .
2.4.4. Microbial Nitrogen (MBN)
Microbial nitrogen (MBN) was determined using the chloroform fumigation-extraction (CFE) technique . Soil samples with the addition of moisture (40% WHC) were fumigated and extracted at a 1:10 (w/v) ratio with 0.01 M calcium chloride solution. The same procedure was repeated with nonfumigated soil samples. Measurements of TSN were determined using a Skalar autoanalyzer by the same procedure as for TSN. Microbial N was calculated as the difference between fumigated and nonfumigated samples divided by the KEN value (extractable fraction of microbial biomass N), 0.54 .
2.4.5. CO2 Evolution (SR) and Microbial Biomass (MB)
CO2 evolution (SR) and microbial biomass (MB) were determined using the MicroRespTM system , by measuring carbon dioxide evolution detected by dye plates—a colorimetric reaction using absorbent alkali with the ability to measure carbon dioxide evolution. Water was added to whole-soil samples in deep-well plates covered by dye plates in order to measure respiretion. Glucose was added to determine microbial biomass according to the SIR method .
Both CO2 evolution and microbial biomass were used for calculating the metabolic efficiency of the soil microbial population under environmental conditions. The coefficient (qCO2) was calculated according to the following equation :
2.5. Statistical Analysis
All data obtained in the study were subjected to statistical analysis of variance combined with post hoc Duncan’s multiple range test (SPSS for Windows, version 10, SPSS Inc., Chicago IL) to evaluate differences between means. Treatment effects were tested by computing redundancy analysis (RDA) (Canoco version 4.5 for Windows, PRI Wageningen, the Netherlands) in order to evaluate its contribution to data variance. Differences at P < 0.05 were considered statistically significant.
2.6. Relative Percentage Calculation
The data obtained in the current study reflected two main effects: the wildfire effect and the post-fire practice effect. The wildfire effect is much stronger than the logging effect  thus, in order to avoid the wildfire effect, the data were presented as a relative percentage calculated by the following equation:
Statistical analysis was performed on the raw data before calculating the relative percentage, as specified above.
Post-Fire Forest Management
Soil moisture (SM) (Table 1) was higher in the unburned control area (UB: 12.9%) than in the unlogged area (B-UL: 9.7%), and lowest in the logged area (B-SL: 8.7%) during the entire study period. This was also indicated by RDA analysis, where B-UL practice accounted for 33% of the SM variance (Table 2), as elucidated by the length and direction of the SM arrow in Figure 1. The differences in SM values between sampled areas were found mostly during winter and summer (Figure 2A), the two extreme moisture-related seasons. One and a half years after the initiation of logging, the differences between the two post-fire practices diminished and became non-significant (Figure 2A).
Organic carbon (OC) was higher in the UB and B-UL areas than in the B-SL areas throughout the study period (Table 1). However, RDA analysis showed a non-significant contribution of both practices (B-SL and B-UL) to the variance of organic C values (Table 2). It is also important to add that overall, OC exhibited minimal variation throughout the research period and was higher in the B-UL than in the B-SL areas only in autumn 2008 (Figure 2B), one year after salvage logging took place.
Figure 1. Redundancy discriminate analysis (RDA) of soil parameters between two post-fire forest practices, using 499 premutations. Soil parameters: SM = soil moisture (%); OC = organic carbon (%); TSN = total soluble nitrogen (mg/kg); DOC = dissolved organic carbon (mg/kg); SR = soil respiration (μgCO2/g soil/h); MB = microbial biomass (μgC/g soil); qCO2 = metabolic qoutient (μgCO2-C/(gCmic× h) × 103); MBN = microbial nitrogen (mg/kg). Post-fire practice: B-UL = unlogged-natural regeneration; B-SL = salvage logged.
When dissolved organic carbon (DOC) variations were examined throughout the study period, higher values were found in the B-SL and B-UL (60 and 60.1 mg/kg, respectively) than in the UB areas (45.2 mg/kg) (Table 1). This pattern was also recorded by RDA analysis, which showed non-significant contributions of both practices to the total explained variance of DOC (Table 2). In addition, the DOC arrow in Figure 1 is positioned in the middle between the B-SL and B-UL arrows, which indicates their equal influence. This equal influence could be the result of a turnover in the DOC patterning that occurred a year and a half after logging was initiated. This turnover was found when results were examined on a seasonal time scale (Figure 2(c)). From the time that logging was initiated until 1.5 years after initiation, higher DOC values (exhibited as relative percentage) were found in the B-SL area compared to the B-UL area during the summer and winter seasons. After this period, the pattern was reversed, and higher DOC values were found in the B-UL compared to the B-SL areas (Figure 2(c)) during spring and summer 2010.
Total soluble nitrogen (TSN) was lower in the UB plots than in the B-UL and B-SL plots, which had similar values, throughout the study period (Table 1). Such findings were mostly due to variation between the burned and the unburned plots that was related to wildfire effect and will not be discussed here. RDA analysis indicates a greater impact of B-UL practice on TSN values (60% of its variance) compared to B-SL practice (30%) (Table 2). However, the total variance explained by both practices was low, as indicated by the short length of the TSN arrow in Figure 1. When examined on a seasonal time scale (Figure 2(d)), differences in patterns between the two post-fire practices resembled those found in DOC but were not as strong. During winter 2008 (Figure 2(d)), higher TSN values were found in the B-SL (16.2 mg/kg) compared to the B-UL (9.9 mg/kg) areas. In spring and summer 2010, almost 3 years after logging was initiated, higher TSN values were found in the B-UL (4.8 mg/kg) compared to the B-SL (2.9 mg/kg) areas.
Table 1. Statistical analysis of differences between means of soil parameters at each of the sampling locations throughout the research period (autumn 2007-summer 2010).
Table 2. Summary of redundancy discriminante analysis (RDA) estimating the effect of two post-fire forest practices on soil parameters using 499 premutations.
Figure 2. Changes in seasonal relative percentage of soil abiotic parameters: a) Soil moisture (SM) %; b) Organic carbon (OC) %; c) Dissolved organic carbon (DOC) mg/kg soil; d) Total soluble nitrogen (TSN) mg/kg soil; between the two post-fire forest practices: unlogged-natural regeneration (B-UL); and salvage logged (B-SL) in relation to the control unburned state represented by black line. Asterisks represent significant differences between the two practices at the p < 0.05 level.
Soil respiration (SR) was higher in the B-UL and B-SL areas compared to the UB area, with no apparent effect of post-fire practices throughout the study period (Table 1). This pattern is related to wildfire effects on soil and will not be discussed this study. Although no apparent effect of post-fire practice was found, multivariable RDA analysis revealed that B-SL practice accounted for 63% of SR variance compared to 32% attributed to B-UL practice (Table 2). When examined on a time scale (Figure 3(a)), the differences between the two post-fire practices were detected even 3 years after logging. Higher SR values found in the B-UL practice compared to the B-SL practice were found almost consistently during all summer and autumn seasons (except summer 2009), whereas an opposite pattern was found during spring 2008 and 2009, and winter 2010 (Figure 3(a)).
Similarly, microbial biomass (MB) was higher in the burned areas (B-UL and B-SL) than in the unburned areas, with a significant effect of post-fire practice found as lower biomass in the B-SL (96 μgC/g soil) compared to
Figure 3. Changes in seasonal relative percentage of soil biotic parameters: a) Soil respiration (SR) mgCO2/g soil/h; b) Microbial biomass (MB) mgC/g soil; c) Metabolic quotient (qCO2) mgCO2-C/(gCmic×h) × 103; d) Microbial nitrogen (MBN) mg/Kg soil, between the two post-fire forest practices: unlogged-natural regeneration (B-UL); and salvage logged (B-SL) in relation to the control unburned state represented by black line. Asterisks represent significant differences between the two practices at the p < 0.05 level. During summer 2009 values were 5 times higher for both soil respiration and microbial biomass and were left out so that the differences could be more obvious. No significant differences were found between the two post-fire forest practices at this time point.
the B-UL (124.3 μgC/g soil) throughout the study period (Table 1). Accordingly, RDA analysis (Table 2) showed that 56% of MB variance was attributed to B-UL practice compared to 26% attributed to B-SL practice. Time scale analysis (Figure 3(b)) showed similar patterns to SR changes, where MB values were found to be higher in the B-UL areas compared to B-SL areas during summer 2008, autumn 2009, and spring 2010. An opposite pattern (B-SL > B-UL) was again found during spring 2009 and winter 2010 (Figure 3(b)).
The metabolic quotient index (qCO2) is based on the ratio between SR and MB, and indicates microbial metabolic efficiency . High ratios indicate that the microbial community is under stress conditions. During the study period, a general pattern was found in which the highest qCO2 values were found in the B-SL > B-UL > UB areas (Table 1). In addition, multivariable RDA analysis showed a stronger connection of qCO2 values to B-SL practice (Figure 1), which accounted for 90% of its variance (Table 2). When examining the seasonaltemporal variation of qCO2 (Figure 3(c)), significantly higher values were found under salvage logging (B-SL) compared to unlogged practice (B-UL) during the first 9 months after logging (Figure 3(c)). This pattern was reversed during autumn 2008 and winter 2009, when qCO2 was higher in the B-UL than in the B-SL areas (Figure 3(c)).
Microbial nitrogen (MBN) was higher in the UB compared to both the B-UL and B-SL areas throughout the study period, with no apparent effect of post-fire practices (Table 1). These differences are the result of the wildfire effect and will not be discussed here. Similarly, RDA analysis showed that B-UL practice accounted for 95% of MBN variance (Table 2), which is also exhibited by the direction of the MBN arrow (Figure 1), however, the length of the arrow is small, indicating a minor influence. This was also the case when seasonal variation in MBN values were examined (Figure 3(d)), with no significant differences found at all sampling times between the two post-fire practices.
Post-fire forest management has received much attention recently [ 4-6,20] and as the frequency of forest wildfires increases, debate on the proper manner of forest rehabilitation continues. In this research, an attempt was made to examine the consequences of two different post-fire forest practices—human intervention by salvage logging (B-SL) and natural regeneration without logging (B-UL)—on the chemical and biological aspects of soil. Previous studies indicated the detrimental effect of logging on the soil system [ 10,17,18] was usually related to soil erosion and compaction. However, there is a lack of data on the effect of salvage logging on a soil microbial community . In the current study, some undesirable effects on soil chemical properties and microbial communities were also recorded, but in limited extent and time.
Soil moisture was one of the soil parameters strongly affected by the different post-fire forest practices in this study. It decreased significantly under salvage logging compared to the natural regeneration practice. These results are in agreement with the results of Greene et al. , who investigated the survival of Sphagnum in logged sites and found lower survival, indicative of drier conditions at these sites. Martínez-Sánchez et al.  also found a reduction in seedling height in logged plots and related it to water stress due to increased sunlight exposure. According to our results, seasonal variations and time influence the differences between the two practices. Most of the differences between logged and unlogged areas were observed during the two water-extreme seasons: summer (dry) and winter (moist). Furthermore, in the second year after logging, the differences between the two practices diminished and became non-significant. By that time, the logged areas were covered with dense vegetation of annuals and shrubs (personal observation of authors), which may increase soil water retention and reduce its loss. Increase in grass cover in clear-cut, harvested plots was also reported by Hannam et al. [ 33] .
A positive impact of salvage-logging practice on soil organic carbon (OC) was observed in this study, with values remaining higher in logged areas compared to unlogged areas almost a year after logging. Fernández et al. [ 12] also found limited perturbation of logging operations on the soil organic layer and the upper centimeters of mineral soil. Also, according to Beghin et al. , salvage-logged, non-planted areas exhibit a dense herb layer that, in time, might compensate for organic matter losses. Parallel to the increase in organic carbon, an increase in dissolved organic-carbon (DOC) levels in logged areas was also recorded in this study. This increase was time-limited (lasted only until 1.5 years after logging), and then decreased to lower values than those found in the unlogged, naturally regenerated areas. Johnson and Curtis  explained the higher soil C levels in logged plots by the incorporation of slash into the mineral soil, especially in coniferous forests, which, when harvested, produce more soil C and N than hardwood. However, in this study, there was less impact of logging on total soluble nitrogen (TSN) levels, which increased in logged compared to unlogged areas only 3 months after logging, and then decreased to non-signifycant levels. Decreases in DOC and TSN levels in the salvaged-logged areas in this study could be a result of plant establishment in the exposed, unshaded areas of logged plots, which quickly utilize available N and C, on the one hand, and increase organic matter on the other.
Measurements of soil microbial activity such as soil respiration and microbial biomass are indicative of proper soil function . Since there are less data available on the salvage-logging effect on the soil microbial community , information is drawn from traditional forest-harvesting investigations [ 21,36,37] . Most of the harvesting operations cause major changes related to soil compaction [ 36] . Schnurr-Pütz et al.  indicated that soil compaction alters the structure and function of the microbial community and favors anaerobic prokaryotes. Siira-Pietikäinen et al.  reported that the decomposer community of coniferous-forest soils is resistant to initial environmental changes induced by forest harvesting. However, in the current study, a clear influence of salvage logging on soil microbial parameters was observed. Consequently, microbial activity—soil respiration and microbial biomass—was lower in the logged compared to unlogged, naturally regenerated areas.
These results correspond with the 18% decrease in microbial biomass carbon observed by Smith et al.  under harvest treatment. Moreover, the differences in microbial-activity measurements in this study were timelimited and extended only until a year after logging began. Higher microbial activity in the unlogged, naturally regenerated areas could be a result of the creation of new microsites by fallen trees and uprooted stumps that are more shaded and moist, with stabilizing microclimatic conditions . The observed lower microbial activity in the logged areas found in this study was followed by an increase in metabolic quotient values (qCO2) that lasted up to 9 months after logging and then decreased. Such higher qCO2 values are indicative of stressed environmental conditions that impact microbial activity . Additionally, Smith et al.  reported a 25% decrease of microbial N under harvest treatment, which could also be indicative of changes in microbial composition and function. However, in this study, no differences between microbial nitrogen in the two practices were found. The reason for this could be the rapid decrease in TSN levels in both the logged and unlogged, naturally regenerated areas found in this study.
Post-Fire Practices: Recommendations
Post-fire management activities such as salvage logging sometimes result in significant mortality of the managed areas, despite their restorative purposes, with timing of logging playing a critical role . Fernández et al. [ 12] stated that postponement of harvesting by one year may lessen the negative impact of this practice on seedling recruitment. According to Izhaki and Ne’eman , this corresponds well with the time of pine-seed germination, which usually takes place in the post-fire wet season. By examining the response of the soil biotic and abiotic parameters to salvage-logging and naturalregeneration (no logging) post-fire forest practices, the results of the current study support the postponement of salvage-logging operation. As observed, most of the unfavorable changes in the logged areas were short-lived and diminished a year after logging operations. Taking into consideration that in this study logging operations were already postponed for a year after the wildfire, a longer time period of logging practice postponement should be considered. Moreover, postponement of logging operations will not completely eliminate negative consequences such as soil compaction, therefore, the means to diminish these consequences must be examined. For example, Greene et al.  suggested leaving a mosaic of logged and unlogged areas or belts, i.e., naturally regenerated sites around logged areas. In this way, naturally regenerated and logged sites will compensate each other and might ameliorate the damage caused by logging operation. With respect to soil function, we believe that such post-fire forest-management actions will be the most effective, however, this needs to be further investigated.
We wish to thank Mrs. Gineta Barness for technical assistance and Ms. Sharon Victor for useful comments. Special thanks to Dr. Marcelo Sternberg for constructive advice. This research is part of the Ph.D. thesis of Orit Ginzburg and was funded by the KKL organization. The funding source had no involvement in the study design, collection, analysis, and interpretation of data, in the writing of the paper, or in the decision to submit the paper for publication.