The Bor Forest Island Fire Experiment
Fire Research Campaign Asia-North (FIRESCAN)
II. Fire History, Ecology, and Short-term Fire Effects
1. Fire Ecology of Pinus sylvestris Forests of the Sym Plain
The Bor Forest Island experimental site is typical of pine-lichen forests of Western Siberia that have been described in the literature (Tkachenko 1952; Korchagin 1954; Shanin 1965; Komin 1967; Popov 1982; Furyaev and Zlobina 1985, and others). Generally, in the taiga zone, dominant pine stand age is dependent on the time period since the last intense fire. Pine stands on sands are represented primarily by pine-lichen forest types. Species composition of postfire woody plant regeneration is typically similar to prefire composition. The time it takes for a young stand canopy to become closed after fire depends on burned area size, seed sources available, and seed production in the years following the fire. Because of insufficient surface fuel loads, fires typically are patchy and cover relatively small areas. As a result, seed sources are generally available nearby. Only young and pole stands, and sometimes middle-aged stands, tend to burn out completely.
Popov (1982) described the regeneration processes in pine stands of the southern taiga zone in the Angara region of Central Siberia. These stands occurred on sandy and podzolic soils on high terraces and in river valleys, and experienced periodic fires. They were the same forest type as the Bor Forest Island experimental site. Popov found that grass cover following burns in pine-lichen forests was usually extremely low and poorly developed; soil often remained bare in considerable parts of burned areas. Pine, either in pure stands or with a very small admixture of birch, typically began to regenerate in high densities (usually several hundred thousand pine seedlings per hectare) after the first year of good seed crop. Popov described the following developmental stages for these stands:
young pine stands with red whortleberry (Vaccinium vitis idaea)-lichen surface cover, and birch as a minor component
pole pine stands with litter and lichen surface cover;
middle-aged, pre-mature, mature and old pine stands with the surface cover dominated by red whortleberry; and
pine – mixed grass – red whortleberry forest.
Popov found that forest regeneration typically occurred without major changes in woody or herbaceous species composition. Some of the typical understory species for this type, along with their importance in the different regeneration stages, are listed in Table 1. A continuous cover of lichens, Pleurozium schreberi and V. vitis idaea develops under the canopy of dense young pine stands. An understory composed of alder (Alnus fruticosa) and dogrose (Rosa acicularis) is restricted to glades (clearings). Projected cover of the grass-low shrub layer does not exceed 0.4. This layer is dominated by V. vitis idaea, Majanthenum bifolium, Antennaria dioica, Festuca ovina, and Lycopodium annotinum. Lichens and Pleurozium schreberi account for some 0.5 of the area. Several species not listed in the table occur primarily in the first stage of forest development, and are relatively uncommon even then. These include: Chamaenerium angustifolium, Calamagrostis obtusata, and Polygala hybrida. Viola uniflora is found throughout forest development, but typically only as single, scattered individuals.
Tab.1. Understory vegetation typical of different forest development stages in pine stands on sandy podzolic soils
(according to Popov 1982).
un = single individual;
sol = up to 10 percent cover;
sp = few individuals, 10 to 30 percent cover;
cop = 30 to 90 percent cover.
During pole stand formation, a fragmented understory of alder develops, dogrose begins to develop in places, and Ledum clusters are observed in clearings. Only isolated clusters of lichens and V. vitis idaea occur in extremely dense pole stands. Pole stands form with surface cover dominated by litter. If a surface fire occurs, it results in the formation of a lichen layer under the canopy of thinned pole stands. Rapid drying out of lichens promotes frequent low-intensity fires. Postfire changes in living surface cover occur at a low rate. Lichens that cover fire-exposed soil are gradually replaced by P. schreberi and V. vitis idaea.
Density of pure pine stands decreases with time and by the age of 80-100 they become rather open. Single individuals of alder, dogrose and Ledum are sparsely distributed in the understory. Pine regrowth occurs in groups and is viable only in clearings. The grass-low shrub layer is composed of V. vitis idaea with Majanthenum bifolium, Antennaria dioica, Festuca ovina, and Lycopodium annotinum. Lichens and P. schreberi account for 0.6 of the area. In recently burned pine stands, regeneration is very abundant (more than 1,000,000/ha), but underdeveloped. Understory is represented by low growing dogrose and young alder. The low shrub layer is absent. The soil is partly bare and partly covered by lichens.
Thus, according to Popov (1982), regeneration patterns for pine stands on sandy and podzolic soils in the Angara region are very simple, with one forest type, which is characterized by four short-term regeneration stages.
Similar patterns have been described for post fire dynamics of pine forests of the southern taiga subzone in the West-Siberian Plain (Furyaev and Kireyev 1979; Furyaev and Zlobina 1985). They identified ecodynamic series based on descriptions of pine sites of differing ages following fire, and emphasized more the spatial dynamics and fire patterns typical of stands at different stages of development. Six regeneration (succession) stages differing in fire resistance were described for pine stands on fresh sands, which are mainly pine-lichen forest type similar to the forest on our experimental site.
Stage 1: recently burned areas with no signs of regeneration; partial or full tree mortality as a result of fire. Pine regrowth, understory and lichen layer completely removed by fire. Litter is the fuel type characteristic of burned sites in pine stands with lichen-dominated surface cover, and high fire danger is maintained due to the presence of downed wood and snags. Recurrent fires can hamper the development of living surface cover and result in destruction of all recent seedlings. Post fire grass cover (before a new young pine stand is formed) is remarkably undeveloped and represented by sparsely distributed sedges and Calamagrostis.
Stage 2: young pine stands with surface cover composed of V. vitis idaea and lichens. Pine is mixed with birch, understory is absent, and the grass-low shrub layer is poorly developed and represented by sedge and V. vitis idaea.. Sixty percent of the area is covered by a lichen layer 2 cm deep.
Stage 3: pole pine stands with litter and lichens as surface cover. Pine stands are pure or mixed with birch. The fuel type is characterized by litter or lichens. The time when a young stand canopy becomes closed after fire depends on specific site and environmental conditions. Repeated fires that promote stand thinning are common.
Stage 4: middle-aged and pre-mature (120-160-yr.old) pine stands. These are uneven-aged pure pine stands. The fuel complex includes litter or lichens. The lichen layer continues to develop and grow in depth; it covers 0.6-0.7 of the ground.
Stage 5: pine-lichen or pine-whortleberry-lichen forests (120-160-yr.old). These stands experience many repeated fires of low and moderate intensity. They are uneven-aged. Regrowth is sparse. Low shrubs account for up to 0.4 of the area. Living surface cover is dominated by lichens.
Stage 6: old pine stands with V.vitis idaea and lichens dominant in the surface cover. Age class patches are easy to identify. These are subject to recurrent surface fires. Pine regrowth is sparse. Lichens account for 0.6-0.7 of the area.
To conclude, post fire regeneration stages for pine stands growing on recent sand deposits consist largely of uneven-aged pure pine stands with poorly developed pine regrowth, sparse woody understory, and a lichen layer that develops gradually in depth and cover after fire. The process of regeneration of pine-lichen forests does not include species replacement. After a year with abundant seed yield, one can expect abundant pine seedlings to occur. However, the great annual variability in seed production in taiga pine forests leads to considerable uncertainty as to the timing of postfire regeneration. Furthermore, because fire danger resulting from large amounts of downed wood, snags, and deteriorating trees remains high for several years in areas that have experienced stand-replacement fires, forest regeneration may be interrupted by repeat fires.
Mesoclimatic and hydrologic regimes characteristic of the sites studied by both Popov (1982) and Furyaev (Furyaev and Kireyev 1979; Furyaev and Zlobina 1985) undoubtedly differ somewhat from those of our experimental stand; nonetheless, we expect the processes of regeneration to be similar, with differences primarily in the duration of regeneration stages or in exact patterns of understory development.
Postfire insect infestations: Although it is clear from casual observation that severe insect infestation often occurs following fire in P. sylvestris, there have been no studies of insect complexes on burned areas in pine forests of the middle taiga subzone of the Siberian plain. We speculate that there may be some similarities to insect populations in clearcut areas of neighbouring regions. Fifty-eight species of stemwood insects have been recorded from pines in areas near the Bor Forest Island study site. These are mainly capricorn beetles, bark beetles, Buprestidae, snout beetles, and Siricidae. Population levels of these insects may increase greatly due to stress from disturbances such as drought, changing water tables, infestations by needle-eating insects, fungi, and fire. Some of those that may be most likely to invade following fire include:
Ips sexdentatus, the stenographer beetle, which is a widespread insect in pine stands. It is especially common in cut-over areas where pines have been damaged by slow-moving surface fires.
Phaenops cyanea F., is a widely distributed Buprestid species, and is one of the first beetles to attack trees that have been weakened by fire.
Ancyclochiera novemmaculata L. is less common than P. cyanea. However, it is well-adapted to invading after fire, as it has been observed to fly great distances toward fires during the night by following smoke plumes.
Monochamus species. M. galloprovincialis pistor Germ. and M. sutor L. are both serious pests of pines. They cause large-scale drying of crowns and larval damage to the wood causes serious loss of wood quality.
While information from unburned and cutover areas can provide some insights into possible insect pests following stand-replacement fires, we have little specific information from burned areas. Clearly studies of insect populations following fires in this forest type are sorely needed.
It should be noted that the fire effects research studies referred to in this paper were conducted by comparing burned areas of different ages. Long-term observation of a permanent site burned by a high-intensity fire has never been undertaken in Siberia before.
2. Physical Characteristics of Bor Forest Island Study Site
The study site is a nearly level, slightly elevated, sandy island, about 50 ha in size, which is surrounded by bogs dominated by mixed-grass/sphagnum and tall sedge. The central portion of Bor Forest Island is about 6 m above the bog surface. Soils are homogeneous across the island. The soil is a ferric podsol, with a coarse sand texture. The A horizon, of mixed mineral and organic matter, is very thin. Occasional weakly cemented patches occur in the B horizon. Characteristics of the soil profile are described in Table 2. The humus layer contained 34.6 t ha-1 of organic matter and 19.9 kg ha-1 of carbon, with 18.1 kg ha-1 of carbon in the mineral soil (0-70 cm).
Tab.2. Characteristics of the soil profile of Bor Forest Island.
3. Fire History
Fire records for the region are incomplete and cover only the last 20 years at best; therefore we must turn to records such as those in tree rings and lake deposits to obtain long-term records of fire history (Valendik and Ivanova 1990).
(a) Long-term Pollen and Sediment Records
Sediment charcoal provides evidence of the long-term importance of fire. Unfortunately there have been no comparisons of particle accumulation rates in sediments with fluxes to the ground that occur during burns. During the Bor Forest Island Experimental Fire the spatial pattern of charcoal accumulation at ground level was determined using water traps (see Section III), and compared to a 4500 yr record of charcoal accumulation in sediments of a nearby lake. A core of lake sediment was obtained from Bor Lake, ca. 25 km east of Bor Forest Island, for purposes of reconstructing Holocene fire regimes. Soils are coarse alluvial sands of undetermined age. The lake is approximately 5 ha in area and 2 m deep. A 1-m thick fringe of Sphagnum encircles the entire lake. The lake is a closed basin, possibly an ancient oxbow of the Yenisei River. The lake catchment is dominated by P.sylvestris, but also includes scattered aspen groves. The catchment was clearcut 1 year prior to coring. A 1-m piston core was extracted and shipped to the laboratory for analysis. The core is largely organic with sand at the base. The core has been sampled for 14C dating, pollen analysis, and thermogravimetric analysis of sediment charcoal.
The pollen record shows a slight increase in Pinus relative to Betula since 4000 years ago, with a concurrent decrease in Picea pollen (Figures 3a and 3b), patterns typical of western Siberia (Peterson 1993). Although changes in the composition of vegetation may have been modest over the last 5000 years, regional fire importance appears to have changed substantially, as evidenced by a dramatic decline in small charcoal particles in lake sediments between 4500 and 2500 BP (Figure 3c). Such changes may reflect a combination of decreases in area burned or decreases in intensity of fires. In modern times, Picea -dominated forests are more likely to burn in high-intensity crown fires than are Pinus-dominated forests, where a higher percentage of area burns in low-intensity surface fires. The large particles that respond to more local fires indicate two periods, 5000 to 4200 BP and 3400 to 2800 BP, when nearby fires appear to have occurred. Particle size distributions in these sections of the core were nearly identical to those observed in particle traps at the experimental burn, but core samples contained order-of-magnitude higher values than observed during our experimental burn. The sediment distribution of sieve samples is continuous with that from the smaller particles observed on pollen slides, suggesting that distribution data are relatively unbiased by method. Sieve samples and airborne samples were analyzed by identical methods, and results were identical (Figure 3c). The decline in fire importance over the last 5000 years suggests that boreal fire regimes are sensitive to climate changes such as those that might occur with global warming. Before 4600 BP, western Siberia was about 2o C warmer than today, a relatively small increase compared to the 5o C predicted for boreal regions in coming decades by some Global Circulation Models. Our data suggest cause for concern over the impacts of such changes on fire regimes in the boreal zone.
Fig.3. Pollen counts (% of total pollen) from Bor Lake core for, a.Betula and Pinus, b.Picea and Abies.c. Accumulation rates of carbon particles in sediment core. Period covered is 5000 years before present (BP) to 600 years BP.
To reconstruct the fire history of Bor Forest island and compare it to surrounding areas, we sampled and analyzed scarred Pinus sylvestris trees. The goal was to determine frequency, seasonality, and size of past fires in the pine forests of this region. Ultimately, we plan to use the fire history of Bor Forest Island, in combination with fire histories from many other stands in west-central Siberia to investigate the interactions of climate, human land-use practices, and fire regimes.
Cross sections were obtained from eight trees on the island (trees EXB 03 to EXB 13) and from seven additional trees on a larger forested “mainland” to the northeast of the island (trees EXB 15 to EXB 101) with a chainsaw. The mainland was located a short distance across a bog, and the maximum distance between sampled trees on the island and the mainland was about 2 km. Sampled trees on the island were located primarily on the southern and western side of the island. Standard dendrochronology techniques (e.g. Graybill 1979, Swetnam and Dieterich 1983) were used to crossdate tree rings among the fire scar specimens. The calendar year dates of fires and the approximate season of occurrence of the fires were then determined by microscopically observing the position of the fire scars (lesions) within the exactly dated annual rings (Dieterich and Swetnam 1984). In addition to sampling fire-scarred trees, increment cores were taken from 20 dominant trees on Bor Forest Island to assess maximum ages of overstory trees. This was an informal sampling (not based on plots or transects), so the ages of overstory trees and cohorts reported here are preliminary and should be confirmed by further sampling.
Before the experimental burn of 1993 at least six fires burned portions of the island during the past six centuries (AD 1481, 1638, 1753, 1796, 1867, and 1956). Intervals between these fires ranged from 43 to 157 years, with a mean fire interval (MFI) of 95 years. Fire-scarred samples in pine forests adjacent to the island, but separated by wet bog, recorded nearly three times more fire dates during the same time period (Figure 4a). Only two of the six fire dates on the island coincided with fire dates on the mainland forest. As in other pine forest sites on the Dubches Plain, mainland forest MFI ranged from about 25 to 40 years. Preliminary stand age structure estimates on Bor Island, derived from increment cores taken from mature trees (Figure 4b), suggest that the overstory is composed of at least two major cohorts that established approximately 180 and 130 years ago. We speculate that these cohorts established following the fires of 1796 and 1867, respectively. An earlier cohort, about 320 to 340 years old, was also suggested by the fire-scarred samples, although the number of trees sampled was too low to assign much confidence to this estimate. In addition to dating fires to the calendar year, microscopic analysis also enabled us to estimate relative seasons of past fires. The largest percentage of fire scars from Bor Island and the mainland appeared in the latewood portion of the tree ring, while the next largest percentage was within the first one-third of earlywood. Although we lack specific knowledge of the cambial phenology of P. sylvestris from this area, it is likely that the latewood fire scars represent burns toward the end of the growing season (possibly August or September), while the earlywood scars probably represent fires that burned in June or July. Our findings suggest that relatively small stands of pine forest surrounded by bogs, such as Bor Forest Island, sustain lower fire frequencies because of they are isolated from fires spreading across the larger, more continuous fuels of surrounding forests. Differences in fire sources and frequency suggest that significant differences in forest age structure and species composition might also be expected in landscape patches of different sizes and varying degrees if isolation within the matrix of bogs and river drainages on the Sym Plain.
Fig.4. a. Fire scar chronology for Bor Forest Island and nearby areas. Site locations are described in the text. b. Dates of major regeneration episodes on the Island.