This work explains the most appropriate techniques and methods for conducting a fire history study from beginning site selection to final analysis of fire-climate relationship.
Annual tree-ring patterns are a source of ecological and environmental information including the history of fires in forested areas. Tree-ring based fire histories include three fundamental phases: field collection, laboratory methods (preparation and dating), and data analysis. Here we provide step-by-step instructions and issues to consider, including the process for selecting the study area, sampling sites, plus how and which fire-scarred trees to sample. In addition, we describe fire-scar sample preparation and dating which are done in the laboratory. Finally, we describe basic analysis and relevant results, including examples from studies that have reconstructed fire history patterns. These studies allow us to understand the historical fire frequency, changes in those frequencies related to anthropogenic factors, and analyzes of how climate influences fire occurrence over time. The description of these methods and techniques should provide a greater understanding of fire history studies that will benefit researchers, educators, technicians, and students interested in this field. These detailed methods will allow new researchers to this field, a resource to start their own work and achieve greater success. This resource will provide a greater integration of tree-ring aspects within other studies and lead to a better understanding of natural processes with forested ecosystems.
Forest fires, ignited by natural or anthropogenic causes, are considered one of the most common ecological disturbance factors that influence terrestrial ecosystems1. For example, fire and more specifically fire regimes, influence plant species composition and structure2. Fire is also a fundamental process linking biogeochemical cycles and climatic variability3,4. In some areas, fire contributes to degradation and deforestation, while in other areas, fire is fundamental for regeneration and sustaining open forest structures5,6. As a result, understanding the ecological role of forest fires is essential to management and conservation programs.
Fire regimes are defined as the pattern of fire events over time characterized by the frequency and its variability in type, extent, intensity, seasonality and severity7,8. Forest fire regimes can be studied through direct observation, reports, satellite images, oral history, age structure and species composition, and through the use of dendrochronological methods9. Dendrochronology uses tree-rings, dated with annual precision, to study climatic and ecological events10. One of the branches of Dendrochronology is fire history reconstruction or Dendropyrochronology which uses tree-rings to determine the spatial and temporal patterns of past and contemporary fires thereby reconstructing the fire regime within a study area11,12. Dendrochronological methods, provide precision and resolution advantages compared to other dating methods, because they allow dating of ecological events, with annual to intra-annual (i.e., seasonal) precision, at long temporal scales, sometimes up to several thousand years13,14.
Fire history reconstructions are also critical in understanding how general climate circulation patterns at regional scales have influenced fire spread. These analyses of the climate-fire relationship are novel because they provide insight into how climate influences fire frequencies over long periods of time, which is not possible with the modern instrumental climate records4. In order to facilitate reconstructing fire histories, we provide a field and laboratory protocol that describes dendrochronological methods and techniques that will allow researchers, teachers, technicians, and students interested in this field of study to initiate their own projects and studies.
In this protocol, we provide the tools to develop a greater understanding and answers to different ecological questions in the field of forest ecology such as: 1) What is the fire regime? 2) Have fire regimes changed in recent decades or have fire frequencies continued without significant change? or 3) Have there been changes attributed to anthropogenic influence? 4) How are fire frequency patterns related to climate variability?
1. Sampling strategy
Figure 1: Pinus hartwegii forests. (A) Topographic variability of the site in terms of slope, forest cover, orographic barriers, fuel, among others. (B) Broader landscape perspective on the terrain and forest conditions, variables that influence fire behavior, and the selection of study sites. Please click here to view a larger version of this figure.
Figure 2: Study sites with and without potential for fire history reconstruction. (A) Pine forest that has been affected (scorched) by a recent fire, but trees show no evidence of scarring; such sites are not useful for this type of study because they lack fire-scarred trees. (B) Pine forest with evidence of past fires, the trees have visible charred section at the base of the trunk in the shape of a triangle, known as “cat face”, formed as the tree heals after repeated fire events. Such sites are considered to have potential for fire history reconstruction. (C) Close-up view of the base of a fire-scarred tree that appears to have recorded numerous fires. Each of the different layers represent a fire scar. In this case, 11 fire scars are visible. Please click here to view a larger version of this figure.
Figure 3: Fire scar sampling process. (A) A tree with a fire scars is selected and (B) a close-up view of the cat face (areas with exposed fire scars at the base of the tree) shows numerous fire-scars and would be an example of a tree that could be selected for sampling. (C) Extraction of a fire-scarred sample from a log. In the case of logs, extraction of partial or complete section is easier because cutting can be done vertically. In the case of live trees and snag, the process is more difficult and includes the following steps: (D) to extract fire scars from live trees, the first step is to select the face with the clearest records, and make two horizontal cuts at the base of the tree trunk. (E,F) To extract the sample, perform an plunge cut, where the tip of the chainsaw is pushed vertically along the back end of the two horizontal cuts, from the bark toward the center of the tree to break off the sample, (G) the sample is then extracted and (H) labeled (study area, site and tree number, sample number, coordinates), and finally (I) the sample is wrapped in plastic to avoid damage while it is transported to the laboratory. Please click here to view a larger version of this figure.
Figure 4: Sampling fire-scarred trees by extracting growth cores (increment cores) with a Pressler drill. To successfully execute this sampling technique, it is important to consider the angle of the extraction in relation to the scar. 1) The sample core that crosses the fire scar will be incomplete because all the rings after the scar will be missing, 2) in the second core the first rings after the scar will may also be missing, but 3) ideally a third core will have all the growth rings and will allow the identification and dating of the fire scar to the exact year and 4) a fourth core far from the fire scar, therefore, with all the growth rings will be obtained, but it will not serve to identify and date of the fire. However, the latter can serve as a reference chronology for the tree. Please click here to view a larger version of this figure.
2. Sample preparation in the laboratory
Figure 5: Fire-scarred Pinus hartwegii sample after preparation or sanding. The initial tree-ring count marked in by blue dots indicates the age of the sample (121 years). The dated annual rings are shown in black (1891–2011). Direct dating is possible in samples collected from live trees where the year of the outermost ring is known (2011 in this case), the rings are clear, and there are no growth problems (missing and false rings) or such problems can be easily distinguished. Please click here to view a larger version of this figure.
3. Tree-ring dating
4. Fire scar dating
Figure 6: Fire scar position and seasonality within the tree-ring and corresponding calendar year. Panel A is an example of a fire-scarred cross-section with individual fire scars indicated by the red arrow and preceded by the year in which each fire occurred between 1902 and 2003. Panels B, C and D show magnified examples of fire scars in the dormant (D), early-earlywood (EE) and middle-earlywood (ME) within the annual tree-ring, respectively. Please click here to view a larger version of this figure.
5. Determining fire scar seasonality
6. Data analysis
7. Climate-fire analysis
When a surface fire burns in a forest, the tree trunks of some trees are often damaged, causing injury that subsequently heals (Figure 7A). These scars form when the fire is intense enough or has a long enough residence time to penetrate the bark and kill part of the cambium. Historically, such fires occurred often enough to prevent the accumulation of fuels; therefore, most of these fires would not be able to reach the tree canopies. As a result, most mature trees survived and continued growing, allowing the damaged portion to partially heal before the next fire (Figure 7B). This recurring process resulted in the recording of a fire-scar within the tree-rings (Figure 7C). The open wound facilitates scarring by future fires and thus the history of past fire events can be reconstructed by selecting the best individuals and making an appropriate collection of the samples, as suggested in section 1.
Figure 7: Fire scar formation within a tree. (A) As a fire burns at the base of a tree, it damages the bark and part of the cambium on the upslope of the tree, where there is greater fuel accumulation and the fire is protected from the wind. The longer residence time allows the heat to penetrate the bark and damage the cambium (Photo taken by Dante A. Rodríguez-Trejo), (B) As a result of the heat, that portion of the tree is no longer functional, creating a scar, (C) In time the scar is incrementally covered by growth for areas adjacent to the scar. However, recurring fires create new scars at the base of the tree stem. The correct extraction of the sample, the dating of the annual growth tree-rings and fire scars (indicated by the arrows in red), allow the reconstruction of the historical fire frequency in forested areas. Please click here to view a larger version of this figure.
Using these same methods here, we provide an example of a fire history study conducted within a watershed. The forests in the upper part of the watershed were divided into lower part (LP) and upper part (UP). A total of 68 fire-scar samples were collected from the following species: Pinus arizonica Engel., Pinus strobiformis Engelm., Pinus theocote Schlecht. & Cham., and Pseudotsuga menziesii (Mirb.) Franco. Of the 68 fire scar trees, 46 were collected in LP and 22 in UP, using section 1 (steps 1.4.6.1–1.4.6.7). Most samples (74%) were taken from dead trees (snags or logs) and the rest (26%) from live trees (Table 1). Following sections 2 and 3, it was possible to date 50 samples (74%) and using section 4, it was possible to identify 596 scars. It was not possible to date 18 samples (26%) due to deterioration or insufficient number of rings to allow reliable dating.
Site | Samples collected | Used in the study | Living | Snag or log | Cut stump | Species | Average diameter |
Lower | 46 | 33 | 10 | 16 | 7 | Par, Pst, Pte, Psm | 45.9 |
Upper | 22 | 17 | 0 | 7 | 10 | Par, Pst | 46.4 |
Note: Tree species are Pinus arizonica (Par), Pinus strobiformis (Pst), Pinus teocote (Pte) and Psedotsuga menziesii (Psm). |
Table 1: Characteristics of sampled trees. This table has been modified from Cerano-Paredes et al., 201930.
Of the 596 scars dated, it was possible to determine the fire-scar position (seasonality) within the tree-ring on 560 scars (94%), based on sections 5 and 6 (steps 6.4 and 6.5). The most common intra-ring position was EE (91.0% and 97.8%), followed by ME (8.7% and 1.8%) and less than 1% (0.3% and 0.4%) in LE for the LP site (Figure 8B) and UP (Figure 8A), respectively. No scars were found in D and L portions of the tree-rings. Of all the fire scars, 91% and 97.8% were determined to have occurred in the spring, 9% and 2.2% in summer, for LP and UP, respectively.
Figure 8: Fire seasonality (number and percent) based on the position of the fire scar within the tree-ring between 1575 and 2008. (A) Seasonality of fire occurence for the UP and (B) LP sites. Most fire-scars were identified early within the growing season. More than 90% of the scars occurred during the spring season. Please click here to view a larger version of this figure.
A fire history record was reconstructed following section 6 (steps 6.1 to 6.3), from 1700s to the early 1950s, when fires occurred frequently at both sites (Figure 9). The pattern of frequent fires was interruption in the mid-20th century. The UP site shows a change in fire frequency starting in the early 20th century. In general, fire frequencies have been altered at both sites in recent decades.
Figure 9: Fire history chart for low and high elevation sites (LP and UP) along the elevation gradient within the watershed for the period 1575–2008. Each horizontal line represents the lifespan of a sample, vertical black lines represent fire-scars, and the gray shaded lines highlight widespread fires affecting both sites (years when fires were recorded at both sites within the same year). The pink shaded area indicates a long period (50 years) with an absence of large fires (lack of synchrony among fire scars between trees), and the blue shaded area is a period when fire frequencies began to be altered, one hundred years ago at the higher elevation site. This figure has been modified from Cerano-Paredes et al., 201932. Please click here to view a larger version of this figure.
The mean frequency intervals (MFI) were generated following section 6 (steps 6.1, 6.2, and 6.4). The results show that, during the last centuries, fires occurred at intervals of every 3-years for both sites (LP and UP) considering all scar filters and at intervals of 9 and 6 years for the most extensive fires (10% filter) in LP and UP, respectively. However, this frequency changed dramatically after 1951, with current extensive fire-free intervals for LP and UP, of 27 and 48 years, respectively (Table 2). Fire return intervals were described using three filters: 1) all scars, which included every fire year that was recorded in at least one sample, 2) ≥10% scars, which included only fire years recorded by at least 10% of the recording samples, and 3) ≥ 25% scars, which included only fire years recorded by ≥25% of the recording samples. The ≥25% filter is widely used in the literature as an estimate of fire frequency for large fires.
Site/analysis period | Categor of analysis | No. intervals | MFI | Min | Max | WMPI |
Lower site 1739-1982 | All scars | 77 | 3.16 | 1 | 16 | 2.69 |
10% scarred | 56 | 4.34 | 1 | 20 | 3.73 | |
25% scarred | 28 | 8.68 | 1 | 31 | 7.22 | |
Upper site 1739-1954 | All scars | 76 | 2.63 | 1 | 7 | 2.45 |
10% scarred | 60 | 3.33 | 1 | 9 | 3.14 | |
25% acarred | 32 | 6.25 | 1 | 19 | 5.44 |
Table 2: Fire interval statistics. This table has been modified from Cerano-Paredes et al., 201930.
The influence of climate on fire occurrence was obtained following section 7. The Superposed Epoch Analysis (SEA) shows that years in which fires occurred were dry and preceded by wet years (Figure 10). In the last 300 years, there has been a significant relationship (P < 0.01) between the fire occurrences and lower than normal rainfall (Figure 10A). The SEA also showed that fire years occurred when El Niño Southern Oscillation (ENSO) NIÑO 3 indices were negative. This suggests tropical climate patterns indicated by the NIÑO 3 indices has had a significant effect (P < 0.05) on the fire occurrences within this study area (Figure 10B). In addition, both indices (precipitation and NIÑO 3) were significantly (P < 0.01) greater than normal 1 year prior to the fire year, suggesting wetter than normal conditions on years prior to the fire events.
Figure 10: Superposed Epoch Analysis (SEA) showing the relationship between climatic variability [reconstructed precipitation33, ENSO indices (NIÑO 3)]34 and reconstructed fire frequency, for both the LP and UP sites. The year when the fire occurred is indicated as year 0 (gray bar), while years prior to the fire year are indicated as negative and years following the fire as positive numbers along the X-axis. In this example, average weather conditions 5 years prior and 2 years after the fire are shown. Climate conditions are indicated along the Y-axis, where values below zero are below average and values above zero represent conditions above average. The upper and lower horizontal lines on each graph indicate the confidence intervals (dotted, P < 0.05; dashed, P < 0.01; and solid, P < 0.001). This figure has been modified from Cerano-Paredes et al., 201932. Please click here to view a larger version of this figure.
The relationship between climate and fire frequency over time can be analyzed graphically by comparing the climate variability of the study region (employing a tree-rings chronology, reconstructed precipitation, ENSO index, PDSI index, among others) and the fire reconstruction (Figure 11). However, it will always be very important to know the statistical relationship between both variables.
Figure 11: Relationship between the climate variability and the fire history. (A) Represents winter-spring precipitation reconstructed, the bottom blue line represents the annual variability; the flexible blue curve is a smoothing spline at 10-year intervals (spline) to detect dry and wet events; and the dotted horizontal line indicates average precipitation and (B) represents the fire history reconstruction. The yellow vertical line allows analyzing the relationship between fire frequencies and decreasing precipitation below average (droughts). This figure has been modified from Cerano-Paredes et al., 201539. Please click here to view a larger version of this figure.
In forested ecosystems, fire is a key ecological process; therefore, reconstructing historical fire regimes is important toward understanding the frequency, seasonality, and variability of fires overtime. Changes to the historical fire regime can potentially lead to unintended consequences in regards to forest structure and health; therefore, such information is critical in forest management. This methodological approach focuses on the importance of selecting the study area and sites, collecting the best fire-scarred trees, as well as the laboratory sample preparation and dating. Likewise, we describe step-by-step analytical procedures to successfully reconstruct the fire history in a forested study area. Such detail procedures are generally summarized and not as well-described in typical fire history study publications. This protocol can be implemented in different ecosystems where trees form annual rings and fires play an important role in forest dynamics.
Forest fire regimes, specifically fire return intervals, frequency, extent, and seasonality, vary over space and time; therefore, it is important to understand these patterns in regions and forests of interest. In some mixed-conifer forests, fire frequencies have been altered by fire suppression efforts since the beginning of the 19th century25,35. While in other regions, fire regime changes occurred later in the mid-20th century36,37,24,38,32, whereas in some sites fire frequencies have remained unchanged39,40,41,42. Conversely, anthropogenic factors have increased fire frequency at other areas43. In most instances, changes to the natural fire regimes have brought about major alteration to the forest and fuel structure, culminating its un-natural fire behavior and stand replacing fires in forests that are not adapted to such events.
In the case study presented here, fires were very frequent prior to 1951 (Figure 9). Moreover, the fact that these fires scarred trees but did not kill them suggests that these were low severity surface fires. That is, the high fire frequencies maintained low fuel loads and tree densities, preventing high-severity fires. However, the process of fuel reduction by frequent fires ceased with fire suppression after 1951. As a result, fuel loads have increased and become more homogeneous within the study area. In the future, this could potentially lead to stand-replacing fires, particularly during extreme climatic conditions (drought), increasing the risk of deforestation, loss of wildlife habitat and affecting the services these forest provide44,45. Fire suppression in forests with a frequent surface fire regime is not a recommended management strategy, given that it can lead to changes in forest stand density, fuel accumulation, forest health issues, and an increased risk of high severity stand-replacing fires5,46,47,48. Whenever possible, fire should be used to restore the regime of frequent surface fires and reduce the risk of severe stand-replacing fires49,38.
Dendro-based fire history reconstructions do have a number of limitations that are worth mentioning. First, of course is that such studies can only be applied in ecosystems with annual tree-rings. Moreover, tree-rings also need to be cross-datable. In dry forests, for example, trees can often have annual tree-rings but may not be cross-datable due to missing or double tree-rings as mentioned previously. To ensure tree-rings within a site cross-date, we suggest collecting and cross-dating core samples prior to sampling fire-scarred trees in a study area. Another potential limitation could be the lack of fire scars within a study area. Although this can suggest that fire is not common in such systems, fire-scars can also be healed over or “buried” within a tree, particularly in fast growing trees and/or when fire intervals are long, thereby allowing the trees to heal or cover the wounded area. In such cases, trees with buried fire-scars may have non-uniform or depression along the trunk. Using these abnormalities as cues, it may be worth cutting into such trees in search of buried fire-scars.
Another limitation of dendro-based fire history studies is that they only provide a limited record of the fire histories because most trees live, die, and decompose within a few centuries, at best. Therefore, the fire history records are short compared to charcoal-based fire histories, for example. However, the main advantage of tree-ring based studies is the annual to sub-annual temporal resolution. One of the advantages of the annual resolution is that forest fire dynamics can then be related to annual climate variability50,51,24,38,50. In general, large fires occur during dry years caused by atmospheric circulation climatic phenomena such as El Niño Southern Oscillation (ENSO)24,38,50,39,47,32. Understanding historical climate-fire relationships allows us to use contemporary weather information from buoys and satellites in the tropical Pacific to monitor and predict the evolution of the ENSO and other climate patterns. These forecasts, paired with region-specific information on historical fire regimes, could allow us to improve management strategies in order to mitigate the impact of shifting trends on fire behavior at multiple scales32.
The results generated by this protocol and associated fire history reports and studies offer the forest managers´ greater understanding of the role of fire within a specific study area and/or region. This information can then be used to design fire management and prevention plans that allow for maintaining or restoring historical fire regimes into the future with the goal of forest sustainability and increasing the quality of ecosystem services.
The authors have nothing to disclose.
The research project was carried out thanks to the financing through the project: Study of the climate-fires relation in north-central Mexico, financed by the SEP-CONACYT fund.
Belt Sander | Dewalt Dwp352vs-b3 3×21 PuLG | For sanding samples | |
Chain Saw Boots | Forestry Suppliers | There is no any specific characteristic | https://www.forestry-suppliers.com/Search.php?stext=Chain%20Saw%20Boots |
Chain Saw Chaps | Forestry Suppliers | PGI 5-Ply Para-Aramid | https://www.forestry-suppliers.com/Search.php?stext=Chain%20Saw%20Chaps |
Chainsaw | Stihl or Husqvarna for example | MS 660 | Essential equipment for taking samples (Example: 18-24 inch bar) |
Clinometer | Forestry Suppliers | Suunto PM5/360PC with Percent and Degree Scales | https://www.forestry-suppliers.com/Search.php?stext=Clinometer |
COFECHA Software | https://www.ldeo.columbia.edu/tree-ring-laboratory/resources/software | ||
Compass | Forestry Suppliers | Suunto MC2 Navigator Mirror Sighting | https://www.forestry-suppliers.com/Search.php?stext=compass |
Dendroecological fieldwork program | Program where dating skills can be acquired or honed | http://dendrolab.indstate.edu/NADEF.htm | |
Diameter tape | Forestry Suppliers | Model 283D/10M Fabric or Steel. | https://www.forestry-suppliers.com/Search.php?stext=Diameter%20tape |
Digital camera | CANON | EOS 90D DSLR | To take pictures of the site and the samples collected (https://www.canon.com.mx/productos/fotografia/camaras-eos-reflex) |
Digital camera for microscope | OLYMPUS | DP27 | https://www.olympus-ims.com/es/microscope/dp27/ |
Electrical tape or Plastic wrap to protect samples | uline.com | https://www.uline.com/Product/Detail/S-6140/Mini-Stretch-Wrap-Rolls/ | |
FHAES Software | https://www.frames.gov/partner-sites/fhaes/fhaes-home/ | ||
Field format | There is no any specific characteristic | To collect information from each of the samples | |
Field notebook | To take notes on study site information | ||
Gloves | For field protection | ||
Hearing protection | Forestry Suppliers | There is no any specific characteristic | https://www.forestry-suppliers.com/Search.php?stext=Hearing%20protection |
Large backpacks | There is no any specific characteristic | Strong backpack for transporting samples in the field | |
Safety Glasses | Forestry Suppliers | There is no any specific characteristic | https://www.forestry-suppliers.com/Search.php?stext=Safety%20Glasses |
Sandpaper | From 40 to 1200 grit | ||
Software CDendro/ CooRecorder | Tree-ring-measurements and dating can also be done using scanned images of the cross-sections | https://www.cybis.se/forfun/dendro/ | |
Software Measure J2X | Version 4.2 | ttp://www.voortech.dreamhosters.com/projectj2x/tringSubscribeV2.html | |
Stereomicroscope | OLYMPUS | SZX10 | https://www.olympus-ims.com/en/microscope/szx10/ |
Topographic map, land cover map | Obtained from a public institution or generated in a first phase of research | ||
Velmex equipment | Velmex, Inc. | 0.001 mm precision | www.velmex.com |
Wildland Fire Helmet | Forestry Suppliers | There is no any specific characteristic | https://www.forestry-suppliers.com/Search.php?stext=Wildland%20Fire%20Helmet |
.