Survival and Recovery Following Wildfire in the Southern Range of Survival and Recovery Following Wildfire in the Southern Range of the Coast Redwood Forest the Coast Redwood Forest

Fire plays a central role in determining structure, composition, and recruitment in many forest types. In coast redwood forests, the role of fire is not well under - stood and scant literature exists on post-fire response, particularly in the south - ern part of the range. In order to better understand patterns of survival and recruitment following fire for coast red - wood ( Sequoia sempervirens [lamb. ex D. Don] Endl.) and associated tree species, three sites in the Santa Cruz Moun-tains, California, USA, were sampled following wildfire. Randomly selected 10 m diameter plots were used to collect data on survivorship and post fire regeneration in order to analyze short-term responses including mortality, crown retention, basal sprouting, canopy regeneration, and seedling production. Results indicated that coast red - wood had the lowest percent mortality (11.98 %) and highest mean canopy retention (43.10 %) of all species sampled, followed by Douglas-fir ( Pseudotsuga menziesii [Mirb.] Franco) (25.54 %), tanoak ( Notholithocarpus densiflorus [Hook. & Arn.] Manos) (23.27 %), combined oak species ( Quercus sp.) (6.67 %), and Pacific madrone ( Arbutus


INTRODUCTION
Disturbance events are important in determining the composition and structure of biotic communities.Coast redwood (coast redwood [Lamb.ex D. Don] Endl.) forests are generally considered not to be disturbance dependent (Olson Jr. et al. 1990, Arno andAllison-Bunnell 2002).Gap phase succession, as a result of individual tree mortality, is considered the primary successional driver, rather than largescale disturbance caused by fire or flood (Busing and Fujimori 2002).Large stand-replacing events are exceedingly rare in this forest type due to a suite of adaptive features that allow coast redwood to survive and regenerate prolifically following fire events.Fire does, however, play an important role in shaping the coast redwood community by increasing the relative dominance of coast redwood over other less fire-adapted species.
A variety of species interface with coast redwood within the coastal mixed evergreen forest mosaic in the Santa Cruz Mountains, located in coastal central California, USA.Douglas-fir (Pseudotsuga menziesii [Mirb.]Franco) is a common co-dominant and exhib-its similar fire resistant properties as coast redwood, including thick insulating bark and high crown (Hermann and Lavender 1990).However, Douglas-fir has no ability for basal or epicormic sprouting, so that complete loss of foliage for an individual results in mortality without exception.The same is true for ponderosa pine (Pinus ponderosa Lawson & C. Lawson) and knobcone pine (Pinus attenuata Lemmon).However, both pine species seed prolifically following fire, particularly knobcone pine-a true fire-dependent species with serotinous cones (Lanner 1999, Barbour 2007).Several hardwood species are associated with coast redwood forests in the southern range, including Pacific madrone (Arbutus menziesii Pursh); a variety of oak species, including California live oak (Quercus agrifolia Née), canyon live oak (Q.chrysolepis Liebm.), and interior live oak (Q.wislizeni A. DC.); tanoak (Notholithocarpus densiflorus Hook.); and California bay (Umbellularia californica [Hook & Arn.] Nutt.).These species present little above-ground fire resistance with relatively thin bark, but exhibit substantial postfire sprouting ability.Other than coast redwood, California nutmeg (Torreya californica Torr.) is the only other sprouting conifer associated with the southern range of coast redwood forests.California nutmeg is a minor component of the forest mix, however, and has only modest sprouting ability.
Naturally ignited fires are rare in the coastal forests where coast redwood thrives due to high atmospheric moisture content along the coast (summer fog) and the rarity of lightning strikes in the dry season (Brown et al. 1999, Brown andBaxter 2003).When fire does occur, intensity tends to be relatively low, resulting in low mortality, although flame lengths can vary significantly.Bark thickness and height to live crown both increase with age and maturity, increasing the ability of older coast redwood trees to survive fires (Olson Jr. et al. 1990).Agee (1993) and Scanlon (2007) observed that size and age of trees, as well as fire intensity, play a role in coast redwood's ability to survive fire.Their observations, in addition to quantitative data from Finney and Martin (1993) and Ramage et al. (2010), suggest that larger, mature coast redwoods are able to survive moderate to high intensity fire events, while smaller immature trees can be killed by relatively low intensity fire events.
Following fire, coast redwood trees regenerate vigorously from both basal sprouts and seed (Agee 1993, Scanlon 2007), although seeds often have a short viability period and are typically less numerous than sprouts (Olson Jr. et al. 1990).As a result, 40 % to 81 % of coast redwood trees originate as sprouts (Douhovnikoff et al. 2004).Additionally, epicormic sprouts can regenerate quickly from stem and branch tissue to replace the crowns of damaged trees (Olson Jr. et al. 1990).Finney and Martin (1993) found that only a few trees smaller than 15 cm diameter at breast height (dbh) did not regenerate either through epicormic or basal sprouts following prescribed fire, suggesting that larger mature trees also resprout better than smaller immature trees.
Much of the previous research in the coast redwood region has focused on fire history and fire frequency, and a majority of this research has focused on the periods of Native American and Spanish occupation.During these time periods, fire return intervals were estimated to be between 6 years and 135 years (Finney and Martin 1989, Greenlee and Langenheim 1990, Finney and Martin 1992, Brown and Swetnam 1994, Brown et al. 1999, Brown and Baxter 2003, Stephens and Fry 2005).Due to the loss of Native American ignitions and the beginning of active fire suppression in the early twentieth century, fire frequency was markedly reduced (Brown andBaxter 2003, Stephens andFry 2005).Lorimer et al. (2009) state that there is a lack of research on the response of coast redwoods and associated species to fire, with specific regard to sprouting and seedling establishment.Additionally, the bulk of previous research has been conducted in the northern and central regions, which differ from the southern coast redwood region in terms of climate, growing conditions, and forest composition.Compared to coast redwood forests in the northern part of the range, stands in the southern extent are smaller and less continuous, receive less annual precipitation, and vary from northern coast redwoods in genetics, forest composition, and forest ecology (Noss 2000).Limited understanding of how coast redwoods and their forest associates respond to fire can limit fire management, conservation, and restoration efforts.The objectives of this research were to: 1) quantify post-fire mortality and crown retention for coast redwoods and associated species following fire; and 2) quantify post-fire regeneration and recruitment of coast redwoods and associated species.Based on previous research, we predicted that coast redwood would exhibit the lowest levels of mortality, and the highest levels of canopy retention, regeneration, and recruitment.

Study Sites
We collected data at three sites that experienced wildfire within two years prior to sampling.Fire severity varied within and between sites, with pockets of crown fire present on all sites.Study sites were located in Santa Clara and Santa Cruz counties of California, USA, in the southern range of the coast redwood (Figure 1, Table 1).Elevation within the Santa Cruz Mountains ranges from sea level to about 1000 m, and climate is considered temperate to hot, with a heavy marine influence in the form of summer fog.Winter precipitation ranges from 40 cm to 152 cm, and falls predominately as rain.Past land management in-  Table 1.Site descriptions and locations of three fires in the Santa Cruz Mountains, California, USA.
Prior to the institution of fire suppression in the early part of the twentieth century, there were frequent fires from mills, railroads, and other human activity (Stephens andFry 2005, Stephens et al. 2007).Today, large human populations create a growing wildland-urban interface necessitating greater fire control (Alexander and Sawyer 1997), and while there have been several large wildfires in the two counties over the past century, none of the study sites had burned within the past 50 years.
The Martin Fire Site The Martin Fire occurred 11 to 15 June 2008 in the Bonny Doon Ecological Reserve (Figure 1).It burned 210 ha, 17.4 ha of which consisted of coast redwood forest, ranging in elevation from 365 m to 518 m.Sample plots ranged from 418 m to 514 m.National Oceanic and Atmospheric Administration (NOAA) stations in Felton and Santa Cruz, California, provided temperature and precipitation data.Temperatures on the day the fire started were 33.9 °C in Felton and 32.8 °C in Santa Cruz.Over the next four days, daytime temperatures ranged from 25 °C to 33.3 °C in Felton and 18.3 °C to 20.6 °C in Santa Cruz.There was no recorded precipitation at either weather station for the month of June, and the area had experienced drought conditions for the previous two years.While unconfirmed, the fire was suspected to be human caused.Once started, a northerly wind quickly pushed the fire south through much of the reserve (CAL FIRE 2008).The reserve, which lies 5.75 km from the ocean, is the largest unfragmented area of Santa Cruz sandhills (McGraw and Levin 1998).Due to this geological diversity, the Martin Fire site was the most diverse in terms of vegetation types and included coast redwood, closed cone pine, ponderosa pine, oak, and Santa Cruz cypress communities.The impact of the fire varied dramatically across the site, with the highest severity standreplacing fire occurring in the closed cone pine community, and the lowest severity fire occurring in the coast redwood-dominated stands.

The Whitehurst Fire Site
The Whitehurst Fire burned in Mt.Madonna County Park (Figure 1) from 21 to 26 June 2008.The park is 14 km from the ocean, and burned within an elevation range from 195 m to about 490 m above sea level.Sample plots ranged from 424 m to 487 m.The two closest NOAA weather stations are in Gilroy and Watsonville, California, which recorded temperatures of 41.7 °C and 40 °C, respectively, on the first day of the fire.Daytime temperatures ranged from 22.2 °C to 39.4 °C in Gilroy and 16.7 °C to 35.6 °C in Watsonville for the duration of the fire.Gilroy received trace amounts of precipitation on 22 June and Watsonville received none for the month of June.This area had also been experiencing drought conditions.The lightning-ignited fire burned 103 ha, nearly all of which were coast redwood or mixed evergreen forests.While large-scale stand replacement did not occur on the Whithurst Fire site, occasional crowning activity did occur in areas along the western flank where CAL FIRE personnel performed a back burn.There was little wind, so the heat from the back burn went up into the canopy (personal communication, Battalion Chief D. McLean, CAL FIRE, Santa Clara, California, USA, 2010).

The Lockheed Fire Site
From 12 to 23 August 2009, the Lockheed Fire burned 3163 ha in the Swanton area north of Santa Cruz, California.An unattended campfire was determined to be the source of the fire.NOAA recorded a temperature of 22.2 °C in Santa Cruz the day the fire started and ranged from 20 °C to 27.2 °C for the dura-tion of the fire.Drought conditions continued from 2007 and 2008, and the only precipitation recorded was 0.051 cm on 2 August 2009.Additionally, the marine fog layer was absent during the first two days of the fire and there were northerly winds present, which allowed the fire to spread quickly (CAL FIRE 2009a).Elevations ranged from 61 m to 335 m above sea level, and sampling occurred between the elevations of 115 m and 276 m .Data collection took place in the Little Creek Management Unit of Swanton Pacific Ranch (Figure 1), which is actively managed for timber and is dominated by young second growth forests, including some recently harvested stands (Swanton Pacific Ranch Nonindustrial Timber Management Plan, http://www.spranch.org/forest_management.ldml).
Coast redwood dominated much of the fire area, with associated Douglas-fir, tanoak, and Pacific madrone common on ridge tops.Fire behavior was highly variable across the site, ranging from small areas of low severity surface fire to large areas of moderate intensity fire, and pockets of higher intensity crown fire.

Plot Design and Measurements
We collected data between January 2009 and September 2010.At each site, we used 30 randomly selected circular plots with a 10 m diameter for a total of 90 plots.At the center of each plot, we measured aspect and slope with a handheld clinometer to determine if post-fire conditions were correlated to physical parameters.Canopy retention (foliage retained through fire) and regenerated canopy (new foliage produced after fire from epicormic sprouts) for each species were also measured at plot center using a convex spherical densiometer and ocular assessment based on foliage color.In most cases, the distinction between retained and regenerated canopy was quite apparent, as new foliage was noticeably lighter in hue.
The following assessments were made for each individual tree: tree species (Quercus species were combined as those without foliage were difficult to distinguish); dbh (diameter at breast height); tree height, measured with a hand held clinometer; mortality (defined as the lack of residual or sprouted green tissue on base, bole, or branch); bole char height, measured with a clinometer; presence or absence of retained canopy (defined as foliage retained through fire); and presence or absence of regenerated canopy (defined as epicormic sprouting that occurred following the fire event).
Post-fire regeneration was assessed on each 10 m diameter plot by counting the number of seedlings of each species and the number of basal sprouts for all sprouting trees (coast redwood, tanoak, oak, and Pacific madrone) in three height classes (small: 0.0 m to 0.5 m; medium: 0.5 m to 1.0 m; large: >1.0 m).

Analysis
One-way ANOVAs (analysis of variance), two sample t-tests, and linear regression analyses were conducted on a Systat 13 platform (Systat Software Inc., Chicago, Illinois, USA) with an alpha level of 0.05.The ANOVA was used to conduct comparisons between species for several independent variables, including mortality, canopy retention, and the number of basal sprouts.Data were tested for homogeneity using the Bartlett's chi-square statistic, and post-hoc analyses were conducted using the Bonferroni test for pair-wise differences be- tween groups.Two sample t-tests were used to compare differences in tree size (dbh and height) between trees that exhibited canopy retention versus those that did not, and for trees that exhibited canopy regeneration versus those that did not.Linear regression analysis was used to analyze correlations between bole char height versus canopy retention, and overall mortality versus the density of coast redwood stems.

RESULTS
Fire severity varied from plot to plot, with bole char heights ranging from 0 m to >50 m.The highest average bole char heights occurred on Pacific madrone, Douglas-fir, and combined oak species (10.92 m ± 1.53 m SE; 10.32 ± 1.72 m SE; 9.24 m ± 1.32 m SE), followed by coast redwood and tanoak (7.95 m ± 0.49 m SE and 2.00 m ± 0.25 m SE, respectively).However, seven of the eight highest bole char heights occurred on coast redwood, with the maximum height at 52.78 m.The next highest maximum bole char heights were found for Douglas-fir (38.58 m) and ponderosa pine (32.41 m).Bole char height varied substantially from site to site, with the highest levels found on the Lockheed Fire site and the lowest on the Whithurst Fire site (Figure 2).

Post-Fire Survival
Coast redwood trees exhibited the lowest mortality across all sites (F 4 = 13.889,P < 0.001), followed by Pacific madrone, oak, tanoak, and Douglas-fir (Table 2).Variation occurred between sites, with the highest mortality for all species found on the Lockheed Fire site and the lowest found on the Whitehurst Fire site (Figure 3).Coast redwood trees also had the highest mean canopy retention across all three sites (43.10 %) (F 4 = 7.234, P < 0.001), followed by Douglas-fir (25.54 %), tanoak (23.27 %), combined oak species (6.67 %), and Pacific madrone (4.13 %).
The size of individual trees (height and diameter) was associated with post-fire tree survival, with coast redwood exhibiting 100 %   survival for all individual trees greater than 7.5 cm dbh.Coast redwood, tanoak, and combined oak species each exhibited significantly larger mean dbh for trees that retained residual canopy than for trees that did not (Table 3).
Trees that retained canopy following fire were significantly taller than those that did not.In addition, regression analysis indicated a significant negative correlation between residual canopy retention and bole char height for all species combined (P < 0.001, R 2 = 0.466).This relationship was not as clear for coast redwood, which experienced the highest maxi-mum bole char heights, as well as the greatest survivorship.The bark of coast redwood is less compact than associated species such as Douglas-fir, however, and may facilitate greater flame lengths up the bole of the tree.
In an unexpected result, the presence of coast redwood appeared to afford associated trees a better chance of survival.Linear regression indicated a significant negative correlation between mortality across all species and the density (stems ha -1 ) of coast redwood trees on individual plots (P = 0.036, R 2 = 0.206).
Overall, coast redwood trees had higher (F 3 = 5.774, P = 0.001) basal sprout density than any other species, with the exception of Pacific madrone, which was statistically equivalent (P = 0.145).In the smallest height class (0.0 m to 0.5 m), there was no difference in sprout production among the four species capable of basal sprouting.In both the medium (0.5 m to 1.0 m) and large (>1.0 m) size classes, coast redwood had higher mean numbers of basal sprouts per tree than other species (Figure 4).
Coast redwood also exhibited the highest overall regeneration through seed (Table 4), although seedling density was highly variable and site specific.At the Lockheed Fire site, a mean of 450 ± 175 SE coast redwood seedlings ha -1 were found, with a mean of 4.18 × 10 4 ± 748.7 SE coast redwood seedlings ha -1 at the Whitehurst Fire site.In contrast, no coast redwood seedlings were found at the Martin Fire site.Seedling density was variable between plots as well, with a maximum of 2.09 × 10 5 seedlings ha -1 found on a single plot at the Whitehurst Fire site, while many plots contained no coast redwood seedlings at all.There was no correlation found between seedling numbers and slope, aspect, or canopy cover, and the seasonality of the fires was nearly equivalent for the site with the greatest number of seedlings and the site with the least number of seedlings.

DISCUSSION
As predicted, the results of this study indicated that coast redwood in the southern range was unusually well adapted to fire both in terms of survival and recovery.Coast redwood exhibited the lowest levels of post-fire mortality and canopy loss, in addition to the highest levels of post-fire regeneration and recruitment of all tree species sampled.In addition, larger trees appeared to be better able to withstand fire than smaller trees.These results are consistent with the findings of research in the  Coast redwood, as well as all associated species, experienced the highest average mortality on the Lockheed Fire site, as compared to the other two fire sites.The Lockheed Fire site also exhibited the highest average bole char height, suggesting that fire severity may have been higher on that site compared to the other two.However, the Martin Fire site had very similar bole char heights with much lower mortality, suggesting that bole char height was not a very good predictor of mortality.These results may have been confounded to some degree by differences in management history between sites.While all sites had been logged in the late 1800s and early 1900s, the Lockheed Fire site had experienced more recent logging activity, with the evidence of selective harvest apparent on some of the plots.
Interestingly, Pacific madrone had the second lowest average level of mortality after coast redwood, even though the majority (86 %) of plots containing madrone experienced no crown retention.Thin bark and relatively low crown heights make madrone susceptible to above-ground damage (McDonald and Tappeiner 1990); however, its ability to regenerate vigorously through basal sprouting lowered the rate of mortality even for trees with complete canopy loss.
Coast redwood also had an apparent mitigating effect on the mortality of nearby associated tree species.This result was somewhat counterintuitive as fire severity (estimated using bole char height) reached maximum heights on coast redwood trees, and although coast redwoods are adapted to withstand extreme flame lengths, many associated species are not.It may be that the elevated flame lengths associated with coast redwood were the result of highly concentrated duff (Nives 1989), or dense sprout thickets at the base of coast redwoods, and did not spread to other nearby trees.It should also be noted that, while coast redwood exhibited the highest maximum bole char height, the average bole char height for the species was modest.
The prodigious germination of coast redwood seedlings on two of the fire sites (Whitehurst and Lockheed) was also unexpected.While recruitment of coast redwoods is known to occur through both the production of seed and basal sprouts, basal sprouts are generally considered the dominant method (Douhovnikoff et al. 2004).And although basal sprouting was quite common, seedling populations outnumbered sprouts on many of the Whitehurst Fire and Lockheed Fire sample plots, while there was a total lack of coast redwood seedlings measured on the Martin Fire site.No correlation was found between seedling recruitment and any of the other sampled variables, leaving the question of the cause of this variation unanswered.However, at the time of sampling, many of the seedlings were lacking vigor in comparison to nearby basal sprouts, suggesting that the long-term viability of the seedlings is doubtful.

Management Implications
The ability of coast redwood to survive fire, even when all foliage has been lost, suggests that post-fire management techniques such as salvage logging are not recommended, especially if conducted to salvage fire-killed trees.While crown fires in most coniferous forests result in stand replacement (Agee 1993, Sugihara et al. 2006), coast redwoods rapidly produce sprouts from scorched bole and branch regenerating foliar tissue so that the original stand is retained.Additionally, residual and regenerated canopies provide interception surfaces for precipitation (Dawson 1998, Burgess andDawson 2004), which reduces soil erosion (Biswell 1989).Presence of a canopy also provides shade to the forest floor and prevents drying of the fine surface fuels.In contrast, removing these trees through salvage logging can increase the potential for soil erosion in the short term (Kattelmann 1996;Karr et al.

Figure 1 .
Figure 1.Site locations of three fires in the Santa Cruz Mountains, California, USA.

Figure 3 .
Figure 3. Post-fire mortality (mean % ± SE) of five most common tree species across three sites in the Santa Cruz Mountains, California, USA.Figure 2. Bole char height (mean ± SE) of five most common species across three wildfire sites in the Santa Cruz Mountains, California, USA.

Figure 2 .
Figure 3. Post-fire mortality (mean % ± SE) of five most common tree species across three sites in the Santa Cruz Mountains, California, USA.Figure 2. Bole char height (mean ± SE) of five most common species across three wildfire sites in the Santa Cruz Mountains, California, USA.

Figure 4 .
Figure 4. Number of basal sprouts per tree (mean ± SE) in three height classes (small: 0 to 0.5 m; medium: 0.5 m to 1.0 m; large: >1.0 m) on three wildfire sites in the Santa Cruz Mountains, California, USA.

Table 2 .
Average mortality of five most common tree species across three sites in the Santa Cruz Mountains, California, USA.

Table 3 .
Paired t-test comparisons of height and diameter of trees with and without canopy retention, and height and diameter of trees with and without regenerated canopy, following three wildfires in the Santa Cruz Mountains, California, USA.